How to evaluate vulnerability in changing environmental conditions

How to Evaluate Vulnerability in
Changing Environmental
Conditions?
Edited by Roger A. Pielke, Sr. and Lelys Bravo de Guenni

Introduction
Roger A. Pielke, Sr.
The prevailing paradigm has been that climate varia- existing paradigm for predicting the future starts from
bility and change can be projected decades or even a the global scale and then attempts to downscale to the
century or more into the future (e.g. IPCC 1996, 2001). regional and local scales; it focuses on averages, is long-
However, for several reasons -e.g. imperfect representerm, and depends on skilful prediction of the future. It
tation of the full complexity of the Earth system, non- is therefore much less useful to policy-makers (Sarewitz
linear spatial and temporal feedbacks, and imperfect et al. 2000). The in-depth discussion of why we need a
foresight of human behaviour, it may not be possible to vulnerability approach that then follows will be illus-
assess the range of potential future climate change actrated using water as an example.
curately.
Water is an essential component of life both in terms
Therefore, a newer paradigm requires that our vul- of its quality and quantity; however, this resource is
nerability to the entire spectrum of environmental threats threatened. For example, as reported in the Economist
be assessed, including the ranking of their seriousness. (May 29, 1999, p. 102), while 90% of the world's popula-
Once vulnerabilities have been so evaluated, whether or tion have enough water at present, it is estimated that
not they can be quantitatively predicted for the future by 2050 more than 40% of the population will face some
can then be tested. This part of the BAHC synthesis dis- water shortage (see also Sect. E.6.1). The lack of access
cusses this perspective to assess risk associated with en- to safe water is even more serious. According to the same
vironmental variability and change. In particular, we article in the Economist, developing countries often have
should start the analysis by first assessing vulnerability, very limited access to safe water supplies. Only about
instead of projecting possible climate change scenarios, 30% of the residents in rural Brazil, for example, cur-
and then assessing vulnerabilities based on that subset rently have access to safe water.
of possible future climates.
There is increasing recognition that threats to future
This is a broader definition of vulnerability than that water quality and quantity are influenced by a wide va-
used by the impacts community. As discussed later, this riety of environmental concerns. These concerns involve
approach involves first assessing all of the vulnerabilities effects both from natural and human origin. Stahle et al.
of an environmental (or other) resource to environmen- (2000), for example, document a 16th century megatal
variability and change. Once these vulnerabilities are drought in western North America that dwarfs any
determined, estimates (with probabilities, if they can be drought since then. Wilhite (2000) presents a series of
quantified) of what scenarios would cause a vulnerabil- articles that demonstrates the extensive effect of drought
ity threshold to be exceeded are specified. Also, unlike on human society. Kunkel et al. (1999) describe the im-
the more narrowly defined concept of vUlnerability, nonportance of vulnerability to weather and climate exlinear
feedbacks between the resource being affected and tremes, and how this vulnerability changes in response
the forcings are included.
to increases in human exposure, even when extreme
We will discuss first the relationship between pre- weather statistics remain unchanged. Non-weather rediction
and vulnerability in Chapt. E.2 and E.3. The scelated influences include land-use change effects on runnario
approach will be reviewed later in Chapt. E.4 and off and stream flow, and deliberate engineering of wa-
its shortcomings illustrated. The reasons why a vulnerter flow whereas human activities change the distribuability
approach as presented in Chapt. E.5 is more aption and availability of water through engineering works.
propriate for environmental assessments is because it is Examples of this kind of human activity include rerout-
regional and local in scale; it involves the evaluation of ing of rivers, creating artificial surfaces (reservoirs),
thresholds and the threat associated with extreme con- changing surface water content through irrigation of
ditions; it can include a spectrum of threats, and can drylands or drainage of wetlands, and through lifting
consider the effect of abrupt changes. In contrast, the groundwater to the surface.

Predictability and Uncertainty
Roger A. Pielke, Sr. .Gerhard Petschel-Held .Pavel Kabat. Brad Bass. Michael F. Hutchinson
Vijay Gupta. Roger A. Pielke, Jr. .Martin Claussen. Dennis Shoji Ojima
It is appropriate to consider water quantity and water
quality as two facets of water resources. Both facets are
intimately connected to the hydrological cycle, which itself
is a component of the Earth's climate system. Since
human activities and health are so connected to water
resources, it is essential to determine how far into the
future we can predict the condition of the water resources
of a region with a sufficient level of confidence. When
predictions are not possible, resilience must be built into
a water system so that human (and natural) needs are
not negatively affected. Even when skillful predictions
are possible, they are seldom completely accurate. Thus
uncertainty needs to be included when scientific analyses
and predictions are used for water resource planning
and management, particularly for issues such as adaptation
or risk management. The prediction of weather and
climate are essential aspects of planning for water resources
in changing environmental conditions.
Lorenz (1979) proposed the concept of forced and free
variations of weather and climate. He refers to forced
variations as those caused by external conditions, such
as changes in solar irradiance. Volcanic aerosols also
cause forced variations. He refers to free variations as
those which "are generally assumed to take place independently
of any changes in external conditions". Dayto-day
weather variation is presented as an example of
free variations. He also suggests that "free climatic variations
in which the underlying surface plays an essential
role may therefore be physically possible".
However, if the ocean surface and/or land-surface
changes over the same time period as the atmosphere
changes, then the non-linear feedbacks (i.e. two-way
fluxes) between the air, land and water, eliminate an interpretation
of the ocean-atmosphere and land-atmosphere
interfaces as boundaries. Rather than "boundaries",
these interfaces become interactive media (Pielke 1998a,
2001). The two-way fluxes that occur between the atmosphere
and ocean, and the atmosphere and the land- surface
(as detailed in Part A of this book), must therefore
necessarily be considered as part of the predictive system.
On the time scale of what we typically call shortterm
weather prediction (days), important feedbacks
include biophysical (e.g. vegetation controls on the Bowen~
,
ratio), snow cover, clouds (e.g. in their effect on the surface
energy budget), and precipitation (e.g. which changes
the soil moisture) processes. This time scale is already
considered to be an initial value problem (Sivillo et al.
1997> since operational numerical weather prediction
models are routinely reinitialised twice daily. Seasonal
and interannual weather prediction include the following
feedbacks: biogeochemical (e.g. vegetation growth and
senescence); anthropogenic and natural aerosols (e.g.
through their effect on the long- and short -wave radiative
fluxes and their effects on cloud microphysics and hence
the hydrological cycle); sea ice; and ocean sea surface temperature
(e.g. changes in upwelling such as those associated
with an El Nino) effects. For even longer time periods
(of years to decades and longer), the additional feedbacks
include biogeographical processes (e.g. changes in
vegetation species composition and distribution), anthropogenic-caused
land-use changes, and deep ocean circulation
effects on the ocean surface temperature and salinity.In
the context of Lorenz's (1979) terminology, each
of these feedbacks are free variations.
We begin to tackle this problem by using a hierarchy
of models. We will consider two examples to illustrate
this important point. First example is the O-dimensional
dynamical model (having no spatial dimensions, i.e.
o-th order in space) which fully and non-linearly couples
radiation, biota, and the hydrological cycle, with
other components of the Earth climate system. This step
is necessary to obtain a fundamental theoretical understanding
of the first-order effects on planetary climate.
These first-order effects tend to be associated with both
positive and negative feedbacks. It seems particularly
important to include negative feedbacks, or the "homeostatic"
mechanisms in the language of Watson and Lovelock
(1983), in low-dimensional dynamical models.
Negative feedbacks tend to be underrepresented in more
complex infinite dimensional dynamical models, involving
one or more spatial dimensions, and therefore are
less well understood.
Insight from simple non-linear dynamical models
should serve as foundations for the development of more
complex models (Shackelyet al. 1998; Ghil and Childress
1987>. Negative feedbacks coming from coupling with~

486
CHAPTER E.2 .Predictability and Uncertainty
biota have been explored in models such as Daisyworld tionary (ergodic) or can change with time, respectively.
(Meszaros and Palvolgyi 1990; Von Bloh et al. 1997, 1999; So far, all model simulaltions (e.g. Cubasch et al. 1994)
Nevison et al. 1999; Weber 2001), though full coupling have shown that the global climate system seems to re-
with other components of an Earth-like climate system spond almost linearly to greenhouse-gas forcing, if the
has yet to be explored. Still, even in simple models, new next several decades, perhaps up to a century, are con-
non-linear effects are only now being discovered (Nordsidered. However, since these are sensitivity model restrom
et al. 2004), a point which serves to emphasize the sults, we do not know if this linearity will remain when
importance of understanding the role of positive and the entire spectrum of natural- and human-forcings on
negative feedbacks on the climate system.
these time scales are included. At the regional scale, the
The second example is the so-called EMICs (Earth Sys- climate clearly exhibits intransitive behaviour as shown
tem Models of Intermediate Complexity; Claussen 2001). for the thermohaline circulation in the North Atlantic
These models explicitly simulate the interactions among (e.g. Ganopolski and RaJl1mstorf 2001), Sahelian rainfall
as many components of the natural Earth system as pos- (Wang and Eltahir 2000), and Northern African deserts
sible. They include most of the processes described in (e.g. Claussen et al. 199fl). Thus at the global scale, in-
comprehensive models of atmospheric and oceanic cirtransitive behaviour Ca11tnot be excluded, because most
culation -usually referred to as "climate models" -al- models have not yet inc:orporated all feedbacks of the
beit in a more reduced, i.e. a more parameterised form. climate system. This argument further supports the use
Therefore, EMICs are considered to test scientific ideas of the term "sensitivity I~xperiment" instead of"projec-
in a geographically explicit model environment, not to tion", when referring to the IPCC results.
make the most detailed and realistic prediction.
There are actually tv.ro types of prediction with re-
Regarding predictability we can distinguish between spect to water resources. The first type of prediction in-
prediction, or forecast of the first and the second kind volves an equilibrium impact of environmental change,
(Lorenz 1975). An example of a commonly known forecast
of the first kind is short-term weather forecasts, i.e.
01. We can write this ma1:hematically as
the weather forecast for several days into the future is
predicted given accurately monitored initial and bound-
01= fJ(OA,OB, OC,. (E.!)
ary conditions. A prediction of the second kind occurs
when boundary conditions determine the state of the
or in some cases as an implicit function
system, and initial conditions are no longer important.
Currently, longer term weather predictions of the first
SI= h(SA, SB, SC, ..., S1) (E.2)
kind have been successful only in the case of forecast- where OA, on, etc., repl:esent a set of environmental
ing seasonal weather such as a six month forecast of EI perturbations from a ref~~rence state of basic variables A,
Nino (Landsea and Knaff 2000) when the oceanographic B, etc. For example, 01 colwd be the effect on river flow at
monitoring system had already indicated an eastward a stream gauge (i.e. 1 is the river flow itself). oA could
moving Kelvin wave of tropical warm water.
then be the radiative effi~ct of increased CO2 and other
The Intergovernmental Panel on Climate Change anthropogenically-emit1:ed greenhouse gases with re-
(IPCC) uses the term "projection" to indicate that a clispect to the pre-industri~l1level. oB could be the biologimate
forecast, of the second kind is meant. However, the cal effect of increased CIJ2; oC could be human-caused
climate system of the future has not been shown to be land-use change; oD the direct radiative effect of anthro-
independent, for example, of the initial (current) Earth's pogenic aerosols; oE the indirect effect of these aerosols
land cover. Moreover, we conclude that the term "pro- on cloud microphysics (cloud condensation nuclei, ice
jection" is misleading, because it suggests some more or nuclei), etc. The choice of the perturbation depends on
less complete prediction of the future. However, most what is the specific impal~t of concern. 01 can either rep-
climate projections of the IPCC include only changes in resent a state variable or the statistics of a state variable
the composition of the atmosphere, whereas other natu- such as a probability density function. 01 can be time
ral and anthropogenic forcings, such as solar variabil- dependent (i.e. in noneqlillibrium as a result of a change
ity, vegetation dynamics and land use (see Part A), are in.!;). When just one perturbation or a small number of
likely to affect future climate. We therefore propose to perturbations on a subs~~t of perturbations is imposed,
use the term "sensitivity experiments", when referring the effect on 01 is said to be a sensitivity experiment.
to the IPCC model results.
When all significant (on 01) perturbations are included,
Predictability of climate can be limited owing to the the effect on 01 is said to be a realisation. If a spectrum of
non-linearity of the Earth system. Following Lorenz's perturbations are perforlmed, which represent all possi-
(1968) terminology, non-linear systems, even without ble situations, the experiment is referred to as an ensem-
any external unsteady forcing, can be "transitive" or "inble. Anyone realisation selected from this ensemble is
transitive", i.e. the statistics of the system can be sta-
called a scenario. The uncertainty in terms of 01 will be

"L"3"6!;j
-eueldx3 lO}) UO!JJ!P3ld}o
s3ssep JU3l3JJ!P }O J!Jew3qJS
(i:OOi: 3J[l3!d
wall p3Jdope ~JX3J 33S SUO!J
[~
xog ~AOqv JO ~ZJS=V
JO J..~!{!qllqold pUll) UO!~nq!l~S!P 3q~ wall p3U!Wl3~3p
wall p3~dllpll 'rH 3m8!d "SO!l1IU3:>S 3q1 JO (3:>U3lln:>:>0
"UO!~:>!P3ld Jo 3A!~:>3dsl3d S!q1 S3~lIl~Snm '('lOO'l) 3}{{3!d
-8y 3q1 U! p3~lIl~Snm 3lll UO!~:>!P3ld Jo S3SSlI{:> lnod
3ln~nJ 3{q!ssod {Jt! Jo 3Sl3A!Un 3q1 ':>!~lIW3q:>S S!q1 uI "3ln
-{!q1l3q13{!qM 'n J..q U3A!8 3lll SUO!~!puo:> Jt!~U3WUOl!AU3
-nJ 3q~ ~SlI:>3l0J o~ UO!~:>!p3ld Jo SSlI{:> lllIll:>!~llld 11 Jo l..),!
mM 3m~nJ 3q1 ~lIqM ~:>!p3ld J..{Jt!n~:>lI o~ J..{3}{!{ SS3{ '3smo:>
Jo 'S! ~! n Jo 1I3lll {Jt!ws 11 Sl3AO:> V JI "V J..q P3UY3P S! 3m~
3p!S~nO 3!{ PIlla:> V J..pn~s ~!A!~!SU3S 11 q~!M 'P33PUI "3q
-qns 11 J..JUo S3!lllA J..pn~s l..),!A!~!SU3S 11 '3m8y 3q1 uI "n Jo
-U! ~ou S30p lO/PUlI SUO!~lIqln~l3d Jt!~U3WUOl!AU3 Jo ~3S
0!lllU3:>S V "S}{:>lIqp33J W3~SJ..S q1l11H ~Ut!~lodUl! {Jt! 3pn{:>
pUt! SuO!~lIqm~l3d Jt!~U3WUOl!AU3 ~Ut!~lodw! II1I S3pn{:>U!
UO!~lIS!Jt!3l 3{8U!s 11 J..{UO S! ~nq 'S}{:>lIqp33J W3~SJ..S ~l1IH
-:>!P3ld Jo wnl~:>3ds 3q~ wall (SUO!~lIS!Jt!3l Jo ~3sqns lO)
-!Jt!3l {Jt! Jo 3{qW3su3 3q1) 3do{3AU3 3q1 U3qM J..{UO "W3~
-sJ..s q1lllH :>!~ollq:> 'l1l3U!{-UOU 3q~ wall 3{q!ssod SUO!~
S! 'P3U!1I~qO S! SUO!~!puo:> 3ln~nJ 3{q!ssod 3q~ Jo (SUO!~lIS
-sn:>s!p Jt!UO!~!Ppll) UO!~:>3fold Jt!n~:>l1 Ut! UO!~:>!p3ld 3q1
\'l00'l 3}{J3!d pUt! "lOO'l U3}{:>lIl::>
-:>lIW U! p3~U3S3ld 3lll UO!~:>!P3ld Jo S3SSlI{:> 3S3q1 Jo UO!S
-A!~!SU3S 11 J..II1In~:>lI S! (9661) ::>::>dI 3q1 '3{dWtlX3 Ut! sy
3q~ UO S~:>3JJ3 :>!u380dolq~UlI 1111 ~OU 3:>U!S J..pn~s J..~!
-ulIwnq P3S1I3l:>U! JO ~:>3JJ3 3A!~1I!plll 3q~ p3pn{:>U! ~lod
-3l ::>ddI 3qJ. "P3l3P!SUO:> 3l3M W3~SJ..S 3~1IUl!{:> s,q1l11H
3pn{:>U! ~OU P!P ~nq 'S{OSOl311 pUt! s3s1I8 3snoqu33l8 ~ndU!
::>::>dI 3q~ 3:>U!S 'l3A03l0W "}{Ooq S!q~ JO V ~llld U! p3SSn:>
-S!P SlI '38ulIq:> 3Sn-pUlI{ SlI q:>ns 'S~:>3JJ3 ~ulI~lodw! l3q10
JO S3~lIW!~S3 Jt!UO!83l U!lI~qo o~ 8U!Jt!:>SUMOP p3Sn U3~
SlI) JP/IP 'IP S3~1IUl!~S3 ~:>lIdUl! 3:>npOld S{3POW W::>D 3q1
8uowlI s~Ills3l Jt!UO!83l Jo J..~!Sl3A!P 3~ '38ulIq:> 3~lIW!{:>
V JO 3Z!S 3q1 S3)[eW q:>!qM (tX3~ 8U!MOnOJ 3q1 u! p3SSn:>S!p
Jt!qo{8 3q~ lOJ Ut!q~ '3Jt!:>S Jt!UO!83l 3q1l0J l3{Jt!WS U3A3
S1~urn°J!AU3 :!Jmnd :![q!$Od IrY=!l
q:J!M S~J~~S '~S~J~UO:> U! 'q:>~OJdd~ J..~!I!q~J3uynA 3q.L '3~:>S
'u3q~ J..IUO 'pUt! '(0 Jo 3Z!S :>!~S!~3J wnw!X1Jw 3q:J 3~~W!~S3
U~:> 3M S~ ~S3q S~) n U!q~!M S3n~A I~ JO ~U3WSS3SS~ 3q~
'SJ3q:JO
u~q:J J..I3~I 3JOW 3J~ s~:>~dw! q:>!qM 3U!WJ3~3p o~ S)f33S
-O!q '3J3qdsow~~) S3J3qds ~Jn~~u 3q~ s3pnI:>u! q:>!qM
W3~SJ..S :>!Wt!uJ..p ~ S~ p3J3P!SUO:> S! W3~SJ..S q:J~a 3q.L
-UO:>3) 3J3qdsodOJ~Ut! 3q:J pUt! (":>~3 '3J3qdsolpJ..q '3J3qds
'i..wo
-Y3P 3q.L "(8661 J3qnquII3q:>S) W3q:J U33M:j3q SUO!~:>~J3~U!
O!IBU3:>S pUt! '3Iqw3su3 'UO!~~S!~3J ~!A!~!SU3S ~ JO SUO!~!U
J..q p3SS3SS~ 3q ~snw ~:>~dw! 3q~ U3qM 3W~S 3q~ U!l!W3J
'(J)/ JO ~U3WSS3SS~ 3q:J ~3J.. W3~SJ..S :>!Wt!uJ..p ~ JO S!sJ..~Ut!
"3S~:> S!q:J U! ~yn:>YJ!P
3JOW q:>nw S3WO:>3q '3W!~ JO Uo!~:>unJ ~ S~ ~:>~dW! 3q~ "3"!
Jp
IP
'~:>3ds~ ~!:>os-oq:>J..sd '3m~yn:> 'i..~3!:>OS
«("a) «(J}/'"' " '(J);:> '(J)fl '(J)V)S =-
-U3J3JJ!P ~ S~U3S3Jd3J ("a "ba J..q U3A!~ UO!~:>unJ 3q.L
"ra "ba Ut!q~ 3AloS o~ ~yn:>YJ!P 3JOW S! q:>!qM uO!~~nb3 ~!~
("a "ba U! UO!~d!J:>S3p :>!Wt!uJ..p 3q:J snq.L "1 JoJ ~U!AIOS pUt!
0 = JP/IP ~U!~~3S J..q p3U!WJ3~3p S! wn!Jq!nnb3 ~~q~ 3~ON
JO ra "ba U! wn!Jq!I!nb3 :>!~~~S 3q~ U~q:J ~J3U3~ 3JOW S!
-SS3SS~ U~ SM°Ire UO!~d!J:>S3p 3q~ 'J~yn:>!~IBd uI "'t"a "ba
S3!~J3dOJd 3:>u3!I!S3J JO J..~!nq~~s 3q~ JO 3p~W 3q O~ ~U3W
3JOW S3WO:>3q ~U3WSS3SS~ 3q.L "s~u!od wn!Jq!I~b3 Jo
'J~3U!I -uou S! S UO!~:>unJ 3q:J U3qM ~yn:>YJ!p
~UW°.l!JUg 3.11T1fIJ [ImIJ8
3q) }U:lS3Jd;J O} V Jo A}!I!QV
3q~ pUt! (":>~3
s3m~~3J 3:>U!S
SO!J~U3:>S UO!~~:>JnJ!q x3Idwo:> JO UO!~OW :>!~o~q:> s~ q:>ns
3W!~ J3AO In:>:>o U~:> "S3s!JdJns" S3S~:> 3S3q:J uI "3S!J~ Ut!:>
3Iq~~s ~ '3Idw~x3 JOd "~:>!p3Jd o~ 3Iq!ssodw! 3J~ q:>!qM
'3Iq~~sun 3WO:>3q U~:> W3~SJ..SO:>3 Ut! U!q:J!M wn!Jq!I!nb3
"W3~SJ..S 3q:J U! S3!~
-J3doJd ~Jn~:>nJ~s M3U J..I3~3Idwo:> o~ P~31 ~q~!W q:>!qM
~

488
.
CHAPTER E.2 .Predictability and Uncertainty
The accurate prediction of oJ or dr/de requires that ft,h
and g be accurate representations of reality. If, however,
there are large uncertainties in the specification of the perturbations
and/or in the form offt,h andg, the range of oJ
and dr/de that results could be quite large. The choice of just
one value of oJ or dr/de (or a limited subset of each) from a
limited set of perturbations using It, h and g will be incomplete,
even if it is assumed thatft'h andg are accurate.
As an alternative, in the case of equilibrium considerations,
the vulnerability of the water resource (or other
environmental resource) can be determined by estimating
what the maximum risks are. Equations E.l and E.2
can be rewritten as
10Ilmax= liI(OA, OR, OC, ...)Imax (E.4)
10Ilmax= Ih(OA, OR, OC, ..., oJ)lmax
where 10Ilmax is the largest effect that results from the
perturbations of the environmental conditions. In order
to determine the maxima, however, it is necessary to know
the ranges of possible values of the independent variables
OA, OR, etc. Yet one might also determine the maximum
possible values of OJ when we assume only small
changes in these variables. Mathematically this can be
achieved by calculating the gradient of OJ with respect to
the input variables, i.e.
VA,B, ...,OJ= VA,B, ...,ft(OA,OB, ...) (E.S)
and analogously for /2 (Ludeke et al. 1999).
Fig.E.2.
Ecological vulnerability!
susceptibility links in environmental
assessment as
related to water resources
(from Pielke and Guenni 1999)
~.. ~
\
\
\
\
\
I \ '
I '
,.. \
Animal and
Insect
[)ynamlcs I \
\ \
\ \
When considering a dynamic system, an analogous
analysis is possible by considering the interval G(A, B, C,
...,I) of possible values on the right hand side ofEq. E.3.
We then obtain a so-called differential inclusion (Aubin
and Cellina 1984), i.e.
dI
-E G(A,B,C,...,I) (E.6)
dt
which then allows a computation of the set of admissible
"futures" of the possible trajectories 1( t) which are realisable
by the assumed set of independent variables A, B,
C, ...This allows an evaluation of the maximum impact,
Imax' at any arbitrary point in time.
Using these approaches, as long as ii, h and g are realistic
representations of the climate system, policy-makers
can concentrate their efforts at reducing the contributions
of the perturbations that most contribute to 01
or dlldt. If these perturbations cannot be manipulated,
this information also needs to be communicated to
policy-makers.
The accurate determination of ii, hand g may not
be possible. With respect to the future climate, general
circulation models (GCMs) have been applied (e.g. IPCC
1996), however, they have been used as sensitivity experiments
since, in general, only one or two perturbations
have been initiated i.e. the radiative effect of
an anthropogenic increase in CO2 and other anthropogenic
greenhouse gases or the radiative effect of an
anthropogenic increase in aerosols. GCM and EMIC
land-cover change simulations have also been per-
'..
, ..
.. ..,
-I
I
I
1
I
I
, I
"',
I
~...
I
,
Climate
Change
& Var1ablll1y
Predictability requires:
-the adequate quantttatlve understanding of these InterocfIons
-fhat the feedbacks are not substant1ally nonlinear.
I

formed as sensitivity experiments (e.g. Brovkin et al.
1999; Claussen et al. 2001; Chase et al. 1996; Pitman et al.
1999; Bounoua et al. 2000).
An alternative approach is to determine what values
of ~I or dI/dt result in undesirable impacts. What
are the thresholds beyond which we should be concerned?
This approach involves starting with the impacts
model (what is sometimes termed an "endpoint analysis"
or "tolerable window approach") and, without using
II, h or g, determine the magnitude of ~I or j that
must occur before an undesirable effect occurs. The
impact functions lI,h and g are then used to estimate
whether such thresholds could be reached under any
possible environmental variability or change. The estimates
for what is realistic, with respect to climate, would
include the GCM results but would also utilise palaeorecords,
historical data, worst case combinations from
the historical data, and "expert" estimates. Such an
approach would provide a risk assessment for policymakers
that is not constrained by uncertain predictions.
c
§ V)
~""'
Q.
~ ~u
as
:=J
u
~
~
CHAI'TER E.2 .Predictability ilnd Uncertainty 489
Figure E.2, from Pielke and Guernu (1999), illustrates a
generic schematic as to how to assess ~I and dI/dt for
water resources.
Pielke and Uliasz (1998) and F'ielke (1998a) discuss
this type of an approach to estimate uncertainty in air
quality assessments. Lynch et al. (:~OOI) apply this technique
to assess the sensitivity of ;a land-surface model
to selected changes (plus and minus) of atmospheric variables
such as air temperature and precipitation. Hubbard
and Flores-Mendoza (1995) assess the effect on corn,
soybean, wheat and sorghum production of positive and
negative changes of precipitation alild temperature in the
United States. 16th et al. (1997> and Petschel-Held et al.
(1999a) have used this "inverse concept" in the form of
the dynamically tolerable windows approach within an
integrated assessment of climatE: change. Similarily,
Alcamo and Kreilemans (1996) apptly the general idea of
the end-point analysis within theil' "safe landing analysis"
of near term climate protection strategies. Figure E.3
illustrates an example where warmler and cooler condi-
CLIMATE
MODELS
Fig. E.3. Simulated water budget responses to uniform change scenarios in the North Fork American river basin of California. a, c, and d
illustrate mean changes under scenarios in which mean temperatures are changed, and b provides the percentage changes as function of
changes in both mean temperature and mean precipitation (contours for uniform change scenarios -dashed whl~re negative; symbols for
GCM model sensitivity experiments; from Teton et al.1999)

490
CHAPTER E.2 .Predictability and Uncertainty
tions are assessed in an impacts model to estimate the theory: individualistic, egalitarian and hierarchical.
sensitivity of water resources in this region to this as- Different parameterisations of the model may then
pect of climate variability and change.
be determined. There is also a strong water compo-
In recent years there has been a growing number of nent within this modelling framework from which
studies and modelling attempts on global environmen- to compute the basic supply and demand issues of
tal change which take uncertainty into account. Within water resources (Hoekstra 1996).
these studies one might want to distinguish between 3. There is a recent attempt to apply qualitative model-
(1) classical probabilistic approaches, (2) new, quantitative ling techniques to the analysis of environmental
approaches based on theoretical frameworks such as change. Originally :,uggested by the German Advisory
cultural theory and u) qualitative, yet formal approaches. Council on Global Change (WBGU 1994) and in cooperation
with the Council, and further developed by
1. More traditional approaches within integrated assess- Schellnhuber et al. (1997) and Petschel-Held et al.
ments of climate change are taken, for example, in the (1999b), the syndrome approach tries to identify ma-
PAGE 95 model (Plambeck and Hope 1996) or in the jor patterns of civilisation-nature interactions, which
ICAM 2.0 and 2.5 models (Dowlatabadi and Morgan govern the dynamics of environmental change. Most
1995) which use probability distribution functions to interesting in the present context is the 1997 Annual
represent uncertainties in parameters and functional Report of the Council (WBGU 1999) which focuses
relationships. The implication of learning as the major
process in reducing uncertainties is central to the
on the sustainable use of freshwater resources.
studies by Kolstad and Kelly (1999).A recent overview As discussed here, a vulnerability assessment provides
on the issue of uncertainty in climate change assess- a comprehensive framework within which to estimate
ments is given in Schellnhuber and Yohe (1998) or environmental risk. This is in contrast to starting with a
Dowlatabadi (1999).
scenario approach which limits the spectrum of estimates
2. An innovative approach is taken within the TARGETS to what can actually occur in the future. With the sce-
modelling framework (Rotmans and de Vries 1997; nario approach, for example, environmental "surprises"
Hilderink et al.1999) where uncertainties are related (Canadell 2000) will b,e missed. These two divergent ap-
to three different world views, based on cultural
proaches are discussed further in the next chapter.

Contrast between Predictive and Vulnerability Approaches
Roger A. Pielke, Jr. Thomas J. Stohlgren
While efforts to predict natural phenomena have be- These considerations suggest that the usefulness of
come an important aspect of the Earth sciences, the value scientific prediction for policy making and the reso-
of such efforts, as judged especially by their capacity to lution of societal problems dept~nds on relationships
improve decision making and achieve policy goals, has among several variables, such as the timescales under
been questioned by a number of constructive critics. The consideration, the scientific complexity of the phenom-
relationship between prediction and policy making is ena being predicted, the political and economic context
not straightforward for many reasons. In practice, con- of the problem, and the availability of alternative sciensolidative
and exploratory models are often confused tific and political approaches to the problem.
by both scientists and policy-makers alike, thus a cen- In light of the likelihood of complex interplay among
tral challenge facing the community is the appropriate these variables, decision makers and scientists would
use of such models and resulting predictions (Sarewitz benefit from criteria that would allow them to judge the
and Pielke 1999).
potential value of scientific prediction and predictive
Among the reasons for this criticism is that accurate modelling for different types of political and social prob-
prediction of phenomena may not be necessary to relems related to Earth processes arid the environment.
spond effectively to political or socio-economic problems Pielkeet al.(1999) provide the fcMowing six guidelines
created by such phenomena (for example, see Pielke et al.
1999). Indeed, phenomena or process~s of direct concern
for the effective use of prediction :in decision making.
to policy-makers may not be easily predictable or useful. 1. Predictions must be generated primarily with the
Likewise, predictive research may reflect discipllne-spe"",- needs of the user in mind. For stakeholders to parcific
scientific perspectives that do not provide "answers" ticipate usefully in this process, they must work closely
to policy problems since these may be complex mixtures and persistently with the scientists to communicate
of facts and values, and which are perceived differently their needs and problems.
by different policy-makers (for example, see Jamieson 2. Uncertainties must be clearly articulated (and under-
and Herrick 1995).
In addition, necessary political action may be deferred
stood) by the scientists, so that users understand their
implications. Failure to understand uncertainties has
in anticipation of predictive information that is not forth- contributed to poor decisions that then undermine
coming in a time frame compatible with such action. relations among scientists and decision makers. But
Similarly, policy action may be delayed when scientific merely understanding the uncertainties does not
uncertainties associated with predictions become politi- mean that the predictions will be useful. If policycally
charged as in the issue of global climate change, for makers truly understood the uncertainties associated
example (Rayner and Malone 1997>.
with predictions of, for example, global climate
Predictive information may also be subject to manip- change, they might decide that strategies for action
ulation and misuse, either because the limitations and should not depend on predictions (Rayner and Ma-
uncertainties associated with predictive models are not lone 1997>.
readily apparent or because the models are applied in a 3. Experience is an important fa(:tor in how decision
climate of political controversy and high economic makers understand and use prt~dictions.
stakes. In addition, emphasis on predictive sciences 4. Although experience is important and cannot be re-
moves both financial and intellectual resources away placed, the prediction process can be facilitated in
from other types of research that might better help to other ways, for example by fully considering alterna-
guide decision making as, for example, incremental or tive approaches to prediction, such as "no-regrets"
adaptive approaches to environmental management that
require monitoring and assessment instead of predic-
public policies, adaptation, and better planning and
engineering. Indeed, alternatives to prediction must
tion (see Lee and Black 1993).
be evaluated as a part of the prE~diction process.

492
CHAPTER E.3 .Contrast between Predictive and Vulnerability Approaches
5. To ensure an open prediction process, stakeholders
must question predictions. For this questioning to be
effective, predictions should be as transparent as
possible to the user. In particular, assumptions, model
limitations, and weaknesses in input data should be
forthrightly discussed. Even so, lack of experience
means that many types of predictions will never be
well understood by decision makers.
6. Last, predictions themselves are events that cause impacts
on society. The prediction process must include
mechanisms for the various stakeholders to fully consider
and plan what to do after a prediction is made.
Scenarios of the future such as projections of climate
change, forest production, animal migration, or disease
spread are -by definition -subsets of all possible outcomes
(Cohen 1996). Most projections have much in
common with the typical five-day weather forecast on
the nightly news:
.they are based on data from the recent past;
.they incorporate relatively few response variables (e.g.
temperature and precipitation);
.they are more accurate for some response variables
than others (i.e. minimum temperature versus preci-
pitation amounts);
.no level of certainty or probability is provided or im-
plied;
.short-term projections are typically more accurate
than long-term projections; and
.their use for planning is limited to a narrowly defined
set of questions (e.g. should snow ploughs be
positioned near highways in anticipation of a snowstorm;
Sarewitz et al. 2000).
Seasonal forecasts, such as those predicting greater
or less precipitation over the next few months, are based
on larger-scale phenomena such as ocean temperatures
and jet stream patterns, but have additional limitations
and uncertainties about the spatial and temporal accuracy
of the projections (Glantz 2001).
There is more uncertainty behind long-term climate
change scenarios with 30- to 100-year timeframes compared
to seasonal forecasts. Long-term climate scenarios
have considerably more limitations and higher levels of
uncertainty due to coarse-scale spatial resolution, natural
spatial and temporal variation, insufficient understanding
of multiple climate forcings, longer-term biological
physical feedbacks in the models, and lack of stochastic
extreme events. Globally averaged results would
be expected to be more reliable than local and regional
results. Reliance on scenarios is reliance on predictions
with all their limitations and uncertainties (Sarewitz
et al. 2000).
The value of predictions for environmental decisionmaking
therefore emerges from the complex dynamics
of the prediction process, and not simply from the technical
efforts that generate the prediction product (which
are themselves an integral part of the prediction process).
An alternative to the scenario-prediction approach
is to evaluate vulnerability based on a more comprehensive
assessment of multiple stresses, multiple response
variables, and their interactions. For example, natural
ecosystems are often affected by multiple stresses including
land-use change, loss of biodiversity, altered
disturbance regimes, invasive non-indigenous species,
pollution, and rapid climate change (Stohlgren 1999b).
Land managers are concerned about many natural and
cultural resources, local economies, and stakeholder
concerns for specific resources. The interactions of the
multiple stresses and response variables are important
and urgent considerations. A warming climate, for example,
might alter the competitive advantage of exotic
annual grasses only if nitrogen from air pollution were
available, grazing were limited, and forest canopies remained
fairly open. The effects of decreased precipitation
on aquatic biota might be more of a problem where
water was over-subscribed, low flows were adversely affected
by water diversion, and exotic fish inhabited the
streams. Human-dominated and natural ecosystems,
and their interacting components and processes, can
rarely be assessed by evaluating stresses independently.
It is often possible, however, to identify vulnerabilities
or sensitivities in cases where prediction is impractical
or impossible.
Vulnerability assessments, of course, also have limitations.
Detailed information on stresses, resources, and
interactions is often scant, resulting in increased uncertainty.
In general, the level of uncertainty is constrained
by focusing on short-term planning horizons and high
priority issues at local and regional scales. However, a
vulnerability assessment is easily made consistent with
the needs of decision makers. It is not a new concept,
and is variously referred to in the literature as "adaptive
management", "vulnerability assessment", "boundedly
rational decision making", and "successive incremental
approximations" .
An important first step in considering the implications
for research and policy is to recognise the role of
models and predictions for purposes of both science and
action. Bankes (1993) defines two types of models, consolidative
and exploratory, differentiated by their uses.
.A consolidative model seeks to include all relevant
facts into a single package and use the resulting system
as a surrogate for the actual system.
The canonical example is that of the controlled laboratory
experiment (Sarewitz and Pielke 1999). Other examples,
include weather forecast and engineering design
models. Such models are particularly relevant to deci-

E.3.1 .Societal Needs 493
sion making because the system being modelled can be emphasis on near-future responses associated with ex-
treated as being closed, i.e. one "in which all the compotreme events, information on a filII range of climate charnents
of the system are established independently and acteristics, and interactions frorn multiple environmen-
are known to be correct" (Oreskes et al.1994). The creatal stresses that they deal with on a daily basis.
tion of such a model generally follows two phases: first, This suggests a rebalancing of priorities towards ad-
model construction and evaluation and second, operaaptation and vulnerability/sensitivity assessments; away
tional usage of a final product. Such models can be used from consolidative modelling aIJd toward more explora-
to investigate diagnostics (i.e. "what happened?"), proctory efforts. Such research has the potential to contribess
("why did it happen?"), or prediction ("what will hapute to a range of important societal needs.
pen?").
When the prediction process is fostered by effective,
participatory institutions, and v~hen a healthy decision
.An exploratory model- or what Bankes (1993) calls a environment emerges from these institutions, the prod-
"prosthesis for the intellect" -is one in which all comucts of predictive science may e'ren become less imporponents
of the system being modelled are not estabtant. Earthquake prediction was once a policy priority;
lished independently or are not known to be correct. now it is considered technically infeasible, at least in the
near future. However, in California the close, institution-
In such a case, the model allows for computational alised communication among sciientists, engineers, state
experiments to investigate the consequences for mod- and local officials, and the private sector has led to conelled
outcomes of various assumptions, hypotheses, and siderable advances in earthquake preparedness and a
uncertainties associated with the creation of and inputs much decreased dependence on prediction. On the other
to the model. These experiments can contribute to at least hand, in the absence of an integrated and open decision
three important functions (Bankes 1993). First, they can environment, the scientific merit of predictions can be
shed light on the existence of unexpected properties as- rendered politically irrelevant, as has been seen with nusociated
with the interaction of basic assumptions and clear waste disposal and acid r.lin. In short, if no ad-
processes (e.g. complexity or surprises). Second, in cases equate decision environment exists for dealing with an
where explanatory knowledge is lacking, exploratory event or situation, a scientificall~'{ successful prediction
models can facilitate hypothesis generation. Third, the
model can be used to identify limiting, worst-case, or
may be no more useful than an unsuccessful one.
special scenarios under various assumptions of and un- .These recommendations fly ill the face of much curcertainty
associated with the model experiment. The limrent practice where, typically, policy-makers recogiting
"worst cases" generated in such a model are explonise a problem, scientists then go away and do rerations
of the boundaries of the universe of model outsearch to predict natural beh.lviour associated with
puts, and mayor may not have significance for under- the problem, and predictions are finally delivered to
standings of the real world. Frequently consolidative and decision makers with the expectation that they will
exploratory models are confused in this regard. Such ex- be both useful and well used. This sequence, which
periments can be motivated by observational data ~. isolates prediction research but makes policy depend-
econometric and hydrological models), scientific hypotheses
(e.g. general circulation models), or by a desire to
understand the properties of the model or class of modent
on it, rarely functions well in practice.
els independent of real-world data or hypotheses. E.3.1 Societal Needs
.A shift in the culture of research is also needed. Research
must be truly interdisciplinary (rather than
multidisciplinary).
Assessing multiple stresses requires a broad spectrum
of expertise without losing sight of the complex interactions
among the sciences. Scientists must also learn
to work interactively with stakeholders so that the highest
priority needs are met first. To date, climate change
research has been heavily focused at the global-scale
and longer timeframes, as evidenced by the reliance on
GCMs, projected changes in mean temperature and precipitation,
and too-year or double-CO2 model timeframes.
There is often a huge "disconnect" with local and
regional stakeholders who have demanded a greater
Recently, Lins and Slack (1999) published a paper showing
that in the United States in the 20th century, there
have been no significant trends U]p or down in the highest
levels of streamflow. This follows a series of papers
showing that over the same period "extreme" precipitation
in the United States has in(:reased (e.g. Karl and
Knight 1998a; Karl et al. 1995).
The differences in the two sets of findings have led
some to suggest the existence of an apparent paradox:
How can it be that on a national s(:ale extreme rainfall is
increasing while peak streamflow' is not? Resolving the
paradox is important for policy debate because the impacts
of an enhanced hydrologicill cycle are an area of
speculation under the Intergovernmental Panel on Cli-
mate Change (IPCC 1996).

494
CHAPTER E.3 .Contrast between Predictive and Vulnerability Approaches
There does exist some question as to whether comparing
the two sets of findings is appropriate. Karl and
Knight (1998b) note that "As yet, there does not appear
to be a good phys-ical explanation as to how peak flows
could show no change (other than a sampling bias), given
that there has been an across-the-board increase in extreme
precipitation for 1- to 7-day extreme and heavy
precipitation events, mean streamflows, and total and
annual precipitation. "
Karl et al. (1995) document that the increase in precipitation
occurs mostly in spring, summer, and autumn,
but not in winter. H. Lins (1999, pers. comn.) notes that
peak streamflow is closely connected to winter precipitation
and that "precipitation increases in summer and
autumn provide runoff to rivers and streams at the very
time of year when they are most able to carry the water
within their banks. Thus, we see increases in the lower
half of the streamflow distribution. "
Karl's reference to a sampling bias arises because of Furthermore,McCabe and Wolock (1997) suggest that
the differences in the areal coverage of the Lins and Slack detection of trends in runoff, a determining factor in
study and those led by Karl. Lins and Slack (1999) focus streamflow, are more difficult to observe than trends in
on streamflow in basins that are "climate sensitive" (Slack precipitation: "the probability of detecting trends in
and Landwehr 1992). Karl suggests that these basins are measured runoff (i.e. streamflow) may be very low, even
not uniformly distributed over the United States, leading
to questions of the validity of the Lins and Slack find-
if there are real underlying trends in the data such as
trends caused by climate change. " McCabe and Wolock
ings on a national scale (T. Karl 1999, pers. comn.). While focus on detection of trends in mean runoff! streamflow,
further research is clearly needed to understand the con- so there is some question as to its applicability to peak
nections between precipitation and streamflow, a study flows. If the fmdings do hold at the higher levels of run-
by Pielke and Downton (2000) on the relationship of preoff-streamflow, then this would provide another reason
cipitation and flood damages suggests that the relation- why the work of Lins and Slack is not inconsistent with
ship between precipitation and flood damages provides that of Karl et al., as it would be physically possible that
information that is useful in developing relevant hypoth- the two sets of analyses are complementary.
eses and placing the precipitation-streamflow debate into In any case, an analysis of the damage record shows
a broader policy context (cf. Changnon 1998).
that at a national level any trends in extreme hydrologi-
Pielke and Downton (2000) offer an analysis that cal floods are not large in comparison to the growth in
helps to address the apparent paradox. They relate trends societal vulnerability. Even so, there is a documented
in various measures of precipitation with trends in flood relationship between precipitation and flood damages,
damage in the United States. The study finds that the independent of growth in national population: as pre-
increase in precipitation (however measured) is insufcipitation increases, so does flood damage.
ficient to explain increasing flood damages or variabil- From these results it is possible to argue that interity
in flood damages. The study strongly suggests that pretations in the policy debate of the various recent stud-
societal factors -growth in population and wealth -are ies of precipitation and streamflow have been mislead-
partly responsible for the observed trend in flood daming. On the one hand, increasing "extreme" precipitaages.
The analysis shows that a relatively small fraction tion has not been the most important factor in docu-
of the increase in damages can be associated with the mented increases in flood damage. On the other hand,
small increasing trends in precipitation. Indeed, after evidence of a lack of trends in peak flows does not mean
adjusting damages for the change in national wealth, that policy-makers need not worry about increasing pre-
there is no significant trend in damages. This would tend cipitation or future floods. Advocates pushing either line
to support the assertion by Lins and Slack (1999) that of argument in the policy arena risk misusing what the
increasing precipitation is not inconsistent with an ab- scientific record actually shows. What has thus far been
sence of upward trends in extreme streamflow. In other largely missed in the debate is that the solutions to the
words, there is no paradox. As they write, "We suspect flood problems in the USA lie not only in a better un-
that our streamflow findings are consistent with the prederstanding of the hydrological and atmospheric ascipitation
findings of Karl and his collaborators (1995, pects of flooding, but also in a better understanding of
1998). The reported increases in precipitation are mod- the societal aspects of flood damage (see Pielke and
est, although concentrated in the higher quantiles.
Moreover, the trends described for the extreme precipitation
category (> 50.4 mm d-l) are not necessarily suf-
Downton 2000, for further discussion).
ficient to generate an increase in flooding. It would be E.3.2 Quantifying Uncertainty Using a
useful to know if there are trends in 24-hour precipitation
in the> 100 mm and larger categories. The term
Bayesian Approach
"extreme", in the context of these thresholds, may have As new information on vulnerability and changes in
more meaning with respect to changes in flood hydrol- the probability distribution of extreme values become
ogy."
better known, risk estimates should be updated. We

discuss below a procedure for handling this problem
by using a Bayesian approach to incorporate uncertainty.
Going beyond the consideration of a damaging event
in the definition of hazard. let us consider. for example.
the quantity of interest (from Chapt. E.2) to be 51. which
is defined as the amount of environmental change in
the context of water resources (e.g. amount of water recharge
to an aquifer; toxic metal content or silting degree
of a reservoir). We are interested in the thresholds
of 51 beyond which there are undesirable impacts. usually
understood as undesirable effects on the human
well-being. In probabilistic terms the values of interest
are in the tail of the corresponding probability distribution.
Since 51 is the result of different perturbations
5A. 5B. 5i and these perturbations have an inherent uncertainty
associated with each. final uncertainty of 51
will be the result of a combination of uncertainties as
defined by functions II and h given above. as well as the
inherent uncertainty associated with the validity of these
functions. All these considerations imply that 51 is a
highly-dimensional quantity dependent on several parameters
which might also be unknown. In probabilistic
terms 51 is a random variable in a highly dimensional
space.
Despite these complexities. a vulnerability assessment
of a water resource would be incomplete if a measure of
uncertainty is not given to the event of 51 being below
or above a threshold value considered as a potentially
E.3.2 .Quantifying Uncertainty Using a Bayesian Approach 495
This last step can be quite complex since it might involve
the definition of a high dimensional joint probability
distribution. However, modern computational
techniques such as Markov chain Monte Carlo methods
(Casella and George 1992) have made possible accurate
representations of the joint posterior distributions of
the parameters of complex statistical models.
The advantages of a Bayesian approach rely on the
possibility of updating the jOirlt posterior distribution
of the model parameters when new information becomes
available (something that is needed in transient risk estimation
as proposed above). A hierarchical modelling
approach can also be naturally implemented with-in the
Bayesian framework by using the powerful tool of conditioning
on the components Ol~ parameters of a previous
step of the system. This is specially useful when uncertainties
related to the spatial scale of scenarios need
to be resolved. By adding a downs~aling layer to the analysis,
under some circumstances (e"g. numerical short-term
weather prediction), it is sometirl1es possible to deal with
the uncertainty of going from a 1~rid box to a point value
to better represent sub-grid scale variability.
In order to estimate realistic thresholds of Sf associated
with environmental perturbations, coherence of
surface variability at the relativt~ly fine space and time
scales is of particular relevanct:. Since this estimation
should be done in probabilistic l:erms, calibration of at-
mospheric variability by simply parameterised spatiotemporal
stochastic models has proven to be a very useful
tool for rainfall and tempel~ature, as discussed in
Sect. C.4.1. Extension of these mt!thods within the Bayesian
paradigm (Sans6 and Guenni 1999) provide tools
for including the uncertainties dliscussed above.
A good example of how to incorporate different
sources of uncertainty by using a Bayesian framework
hazardous situation. The reason for this is that policymakers
are always interested in expected losses and these
need to be quantified in a proper way.
Bayesian statistical methods are becoming increasingly
important in evaluating uncertainty related with
environmental change (Adams et al. 1984; Pate-Cornell
1996; Tol and de Vos 1998;Wikle et al. 1998). Given a sta- is presented by Krzysztotowicz (1999). He proposes a
tistical model for a variable or a set of variables depend- Bayesian forecasting system (BFS) by which the total uning
on a given set of parameters. the Bayesian paradigm certainty about a hydrological pr,~dictant (as river stage,
involves three main steps:
"'discharge
or runoff volume) is decomposed into input
uncertainty (e.g. time series of precipitation amounts
I. Consider a prior distribution for the model param- needed as an input to a hydrological model) and hydroeters
based on prior (as for example subjective inforlogical uncertainty, which considers all sources of unmation
by experts) knowledge and before using the certainty beyond random inputs (e.g. model, parameters,
data. When prior information is not available non- estimation and measurement erl~ors).
informative (diffuse) priors could be used.
Van Noortijk et al. (1997> also use a Bayesian approach
2. Obtain an expression of the joint probability distri- to quantify different sources of uncertainty in the procbutions
of the observations conditioned on the model ess of taking optimal decisions to reduce flood damage
parameters (this is known as the likelihood function along the Meuse river in the Netherlands (near the Dutchin
classical statistics) which implies a proposed sta- Belgian border). Their approach attempts to quantify the
tistical model for the variable of interest.
expected economic losses due to flood damage at differ-
3. Obtain the posterior probability distribution of the ent discharge thresholds. This rnethodology fits very
model parameters by combining prior information nicely with the vulnerability perspective proposed in this
with the likelihood function using the Bayes rule. part of the book.

The Scenario Approach
Lelys Bravo de Guenni
The scenario approach uses a range of plausible future
environmental conditions to assess their effects on water
quality, river flows and other hydrological components.
Examples of effects on water resource availability
include the life span of a reservoir, a water supply system,
and water management-related decisions. Most of
these effects are usually assessed by using impact models
which allow for the translation of changes in the components
of the hydrological cycle (e.g. changes in river
flow) into changes in the water resource system (e.g.
hydropower generation). Changes in the environment
(including climate and land-use changes) will affect the
hydrological cycle and the vegetation patterns, which in
turn will affect the environment including the climate
system (Kite 1993). An integrated approach to assess the
impact of environmental and land-use changes on water
resources in different biomes requires a wide range of
scenarios, hydrological models and impact models involving
socio-economic aspects such as population
growth and changed water management objectives.
Most of the scenario approaches up to the present
that are used to assess impacts on water resources are
numerical and stochastic modelling-based sensitivity
experiments with a focus on atmospheric (weather)
changes and variability. Moreover, these modelling experiments
should more accurately be described as sensitivity
studies, as discussed in Chapt. E.2, since the spectrum
of human disturbances to the climate system is
not included. Such experiments have been used for impact
studies on water resources and agriculture and have
been developed to investigate the impact of climate
change where the radiative effect of anthropogenicallyincreased
greenhouse gases is the dominant environmental
perturbation of the climate system. Existing procedures
to construct scenarios are more general and include
arbitrary scenarios, historical analogues, palaeoclimatic
records, as well as General Circulation Model
(GCM) perturbation experiments, and Regional Climate
Models (RCM) (Rosenzweig and Hillel 1998).
Arbitrary scenarios assume prescribed changes in climatic
variables as for example temperature and/or precipitation.
These changes might be applied to the mean
(e.g. Rosenzweig et al.1996), variance (e.g. Mearns et al.~
1997) or other statistical characteristics and are normally
used to identify sensitivities of the system to the prescribed
changes. However, changes in one variable may
be accompanied by changes in other variable(s) and
some care must be taken to provide meteorologically
plausible conditions.
Weather scenarios from historical records are used
to construct possible realisations considering, for example,
worst case situations (cool or warm periods, dry or
wet periods). They can provide some insight on anomalous
conditions and their usefulness is conditioned by
the length of records available. Palaeoclimatic scenarios
use records of pollen, ice cores, lake-levels and others.
These records are useful to understand past atmospheric
dynamics.
Weather change sensitivity from GCMs are produced
under different forcing mechanisms (e.g. double-CO2
scenarios). These experiments are appropriately referred
to as sensitivity experiments, since only a subset of anthropogenic
perturbations to the climate system have
been evaluated using these models. Equilibrium or transient
projections can be designed depending on whether
the forcing conditions are instantaneous or are changing
with time. In the last case GCMs are used to simulate
the response of the climate system to a continuous increase
in greenhouse gases, rather than to an instantaneous
change. Although GCMs provide valuable insight
into the dynamics of the atmosphere and the oceans, it is
well known that they are not able to represent even current
climate at local scales (Cubasch et al. 1996; Risbey
and Stone 1996) and their spatial resolution is still very
coarse for impact studies. Sub-grid scale topography, land
cover and water bodies, for example, have major impacts
on local maximum and minimum temperatures, precipitation,
incident radiation and humidity. The variability
in these factors needs to be well represented when used
as inputs to water resource impact models.
Downscaling approaches have played a role as the
main procedures to provide high spatial resolution data
from the very coarse information from the GCMs (von
Storch 1994). Different approaches are available: stochastic
(Bardossy and Plate 1992; von Storch et al. 1993;
Conway and Jones 1998; Bellone et al. 2000); dynamic

498
CHAPTER E.4 .The Scenario Approach
ditions for a regional numerical weather prediction
model. This is not true with the GCMs where observed
data do not exist to influence the predictions. A regional
model cannot reinsert model skill when it is so dependent
on lateral boundary conditions, no matter how good
the regional model is.
If this conclusion is disagreed with, the first step to
demonstrate that the regional climate model has predictive
skill is to integrate an atmospheric GCM with
observed sea-surface temperatures (SSTs) for several
seasons into the future. The GCM output would then be
downscaled using the regional climate model. There is
expected to be some regional skill and this needs to be
quantified. This level of skill, however, will necessarily
represent the maximum skill theoretically possible with
AOGCMs (Atmosphere-Ocean General Circulation
Model) as applied to forecasts years and decades into
the future since SSTs must also be predicted and not
specified for these periods. Such experiments have not
been systematically completed. Indeed, does the concept
of predictive skill even make sense when we cannot
verify the models until decades into the future?
The statistical downscaling, besides requiring that the
GCMs are accurate predictions of the future, also requires
that statistical equations used for downscaling
remain invariant under changed regional atmospheric
and land-surface conditions. There is no way to test this
hypothesis. In fact, it is unlikely to be valid since the
regional climate is not passive with rt1Pect to the largerscale
climate condition but is expect~ to change over
time and feedback to the larger scales.
A critical issue in these methodologies, therefore, is
how to handle the uncertainty associated with the climate
system due to its complex non-linearity (Shukla
1998) and lack of predictability years and decades into
the future. Because of the incomplete knowledge of the
climate system and the "unknowable" knowledge inherent
in the future of the human society, scenarios are
meant to provide a range of possible climatic conditions
(Hulme and Carter 1999). Extremes and "surprises" are
an important part of this uncertainty since they can
cause substantial damages to society. Stochastic approaches
have a lot to offer on the grounds of uncertainty
evaluation since global climate models are not
actually formulated to perform an uncertainty analysis
and in most cases this is carried out as an afterthought
(Katz 1999).
Indeed, based on the different methodologies for scenario
construction (including climate) that have been
made, a quantitative assessment of what are the temporal
and spatial limits of predictive skill is required. It is
also concluded that none of the existing procedures to
construct scenarios are able to represent the spectrum
of plausible future environmental conditions on the local,
regional (and even the global) spatial scales.

The Vulnerability Approach
Lelys Bravo de Guenni .Roland E. Schulze. Roger A. Pielke, Sr. Michael F. Hutchinson
E.5.1 Risk, Hazard and Vulnerability: Concepts
Risk, in layman's terms, is the "chance of disaster" (Fairman
et aI. 1998). More formally
Risk is a quantitative measure of a defined hazard,
which combines the probability or frequency of occurrence
of the damaging event (i.e. the hazard) and
the magnitude of the consequences (i.e. expected
losses) of the occurrence.
Embedded within this defmition is the term hazard,
and implied in the phrase "consequences of the occurrence"
is the concept of vulnerability to the hazard. These
two terms are therefore described next, before re-visiting
broader issues of risk.
A hazard is commonly described as the "potential to
do harm". Defined more rigorously (Zhou 1995; Smith
1996; Fairman et al.1998; Downing et al.1999)
A hazard is a naturally occurring, or human induced,
physical process or event or situation, that in particular
circumstances has the potential to create damage
or loss. It has a magnitude, an intensity, a duration,
has a probability of occurrence and takes place within
a specified location.
The above definition serves to highlight the concept
that a physical process only becomes a hazard when it
threatens to create some sort of loss (such as loss of life
or damage to property) within the human environment
(Smith 1996). This is therefore essentially an anthropocentric
view of the concept of hazard and does not
take into account the effect that an extreme natural event
can have op an uninhabited area (Suter 1993). The assessment
of losses and the determination of the detrimental
effects on future overall sustain ability in uninhabited
areas are extremely difficult to undertake and
they generally fall under the concept of ecological risk
assessment (Suter 1993). Here, the magnitude of a hazard
is thus determined by the extent to which the physi-
cal event can disrupt the human environment, i.e. a hazard
is the combination of both the active physical exposure
to a natural process and the passive vulnerability
of the human system with which it is interacting (Plate
1996).
The physical exposure is essentially the damagecausing
potential of the natural process and is a function
of both its intensity and duration. The natural process
becomes a hazard when it produces an event that
exceeds
.thresholds, i.e. the critical limits (bounds) that the environment
can normally tolerate before a negative
impact is produced on a system or activity (Downing
et al. 1999). In the case of rainfall, too much produces
a flood hazard and too little a drought hazard. In
Fig. E.4 the shaded area represents the tolerance limits
of the variation about the average, within which a
resource such as water can be used beneficially for
social and economic activities within the human environment
(Plate 1996). The magnitude by which an
event exceeds a given threshold determines the damage-causing
potential of such an event.
.Intensity refers to the severity, or damage-causing potential,
of a natural process, e.g. rainfall at 20 mm h-l
is generally less damaging than 100 mm h-l over the
same time period. The hazard intensity is determined
by the peak deviation beyond the threshold (vertical
scale in Fig. E.4).
.Duration, the other variable determining the damagecausing
potential of an event, implies exposure to an
event, and the longer the exposure the greater the damage-causing
potential (Zhou 1995; Plate 1996; Smith
1996). Hazard duration is determined by the length of
time the threshold is exceeded (horizontal scale is
Fig. E.4).
Response to a hazard, as discussed in more detail later,
is either by
.adaptation, i.e. the long-term arrangement of human
activity to take account of natural events (e.g. becom-

500
CHAPTER E.5 .The Vulnerability Approach
Fig. E.4.
The magnitude of environmental
hazard expressed as a
function of the variability of
a physical element within
the limits of tolerance (after
Smith 1996)
ing more dependent on groundwater than on more erratic
surface water resources in more arid zones), or
.mitigation, i.e. the intentional response to cope with
a hazard (e.g. only constructing buildings beyond a
demarcated 1: 50 year flood line).
Vulnerability implies the need for protection. From
an anthropogenic viewpoint
Vulnerability is the characteristic of a person or group
or component of a natural system in terms of its capacity
to resist and!or recover from and/or anticipate and!
or cope with the impacts of an adverse event (Blaikie
et al.1994; Downing et al.1999).
Vogel (1998) describes vulnerability in terms of the
resilience and susceptibility of a system, including its
physical, social and human dimensions, while Plate
(1996) adds the reliability of the system to its attributes.
.Resilience (Vogel 1998) is the capacity of a system (e.g.
a dam) to absorb and recover from a hazardous event
(e.g. a drought). Resilience therefore implies that there
are thresholds of vulnerability.
.Reliability, on the other hand, is the probability that
the system, or a component of a system, will perform
its intended function for a specified period of time
(e.g. what is the probability that the dam will be able
to supply water to a city over the next 50 years?).
In terms of vulnerability, systems may be subjected to
.assault events (e.g. heavy rainfall; flood peak; pollution
levels above a certain concentration), in which
case the vulnerability threshold is determined by the
system absorption and redirection capacities (e.g. a
heavy rainfall saturating a soil and the soil then draining
the excess water rapidly enough, or a dam filling
to capacity and the spillway coping adequately with
Average
c~~ Band of tolerance
the flood discharge). Systems may, according to Smith
(1996), also be subjected to
.deprivation events (e.g. drought, soil erosion or leaching
of fertilisers out of the soil), in which case the
thresholds of vulnerability are determined by the retention
and replacement capacities of the system (e.g.
the buffer of deeper soil depth to storing moisture
for a plant during a drought, or the rate of weathering
to replace soil lost by erosion).
Vulnerability therefore invanably embraces an
.external dimension (Vogel 1998), i,e. the threat of an
event, that may increasingly predispose people to risk
(e.g. climate change and its impacts on water resources),
as well as an
.internal dimension, i.e. the internal capacity to withstand
or respond to an event, such as the defenselessness
to cope with a hazard (e.g. poor people living on
a floodplain) or the lack of means to cope with the
aftermath of damaging loss.
People may thus face the same potential risk, but are
not equally vulnerable because they may face different
consequences to the same hazard.
Although intuitively simple, the vulnerability concept
requires quantitative tools which are only now being developed.
The issue of vulnerability is discussed in UNEP
(2000), for example. One new aspect to the vulnerability
perspective is that the affected resource can itself
feedback to influence its environment. An attempt to
quantify vulnerability in terms of the proportion of people
affected under extreme rainfall conditions in South
America has been presented by Guenni et al. (2001).
They assumed that the proportion of people affected (or
vulnerable) to extreme rainfall conditions is a random
variable that can be calibrated by using available information
on the conseqences of extreme events on the
population in different South American countries.

Fig.E.S.
A schematic illustration in
which risk changes due to variations
in the physical system
and the socio-economic system.
In all the cases risk increases
over time (with modifications
after Smith 1996)
Risk Revisited
If risk is the probability of a specific hazard occurring
and the loss caused by that hazard in regard to the level
of vulnerability of the affected people or places, then
several possibilities exist that give rise to increased risk.
These are illustrated in Fig. E.5 (Smith 1996):
.Case A represents a scenario where the tolerance and
the variability remain constant, but there is a gradual
change over time in the mean value. In this particular
case the frequency of extreme events at one end
of the scale increases, as would be the case of a mean
decrease in rainfall, or a decrease in runoff associated
with upstream afforestation.
.Case B shows a scenario in which both the mean and
the band of tolerance remain constant, but the variability
increases. In this particular case the frequency
of potentially damage producing events increases at
both ends of the scale.
.In Case C the physical variable, e.g. runoff, does not
change, but the band of tolerance narrows, i.e. the vulnerability
of the human system increases (e.g. because
of increased water demands on a river from people
living directly in the floodplain). In this particular
situation the frequency of damage-causing events increases
at both ends of the scale (e.g. too little water
during dry periods; vulnerability to flood damage
during high flows).
.In Case D, there is an abrupt change in the mean, but
the variability remains the same. Such a rapid transition
provides little time to adopt to a change (such
as might be possible with Case A).
A
B
c
D
E.S.1 .Risk, Hazard and Vulnerability: Concepts 501
In viewing risk as having a human component, in
addition to its probabilistic one, three tiers of risk have
been identified by Zhou (1995):
.a lower band of risk, which is acceptable to the affected
people and where, for example, the benefits of
doing nothing or little outweigh the disadvantages
of carrying an unacceptable cost burden,
.a middle band of risk, where decisions have to be made
which trade off the costs of reducing the risk versus
the benefits of the risk reduction, and
.an upper band of risk, where doing nothing is completely
unacceptable, irrespective of cost.
Each of these three tiers of risk is related to the balancing
of benefits v. costs. This is usually done through
risk management, which on the one hand has to be regulated
by professional standards and legal measures while
on the other hand it contains a large element of subjec-
tivity.
The vulnerability approach described above provides
a methodology by which the public in general and the
policy-makers in particular will gain a better insight into
which thresholds and limits in environmental changes
might potentially cause stress or damage. They would
have a better estimate of the uncertainty associated with
the occurrence of these thresholds and they could better
quantify the expected losses for specific hazards.
A cost-benefit analysis must necessarily be a next step
for policy-makers to take the required actions in high
vulnerability situations. In the following sections, specific
examples of environmental variability and changes
and multiple stresses to ecosystems with their expected
impacts are presented.

502
CHAPTER E.5 .The Vulnerability Approach
Rik Leemans All these LUC-LCC aspects, interactions and feedbacks
E.5.2 Anthropogenic Land-use
and Land-cover Changes
The Importance of Land-use and Land-cover Change
Issues related to water, water quality and availability, and
water management are important on the international
political agenda. But these topics do not operate in a
vacuum. There are strong interactions between hydrological,
atmospheric and biospheric processes on the one
hand, and land use and water management on the other
hand. Land and water are thus both systemic components
of the Earth system, but simultaneously are
strongly affected by environmental change. Land use and
land-use change and its consequences for land cover
(LUC-LCC) are therefore essential components in the
development of comprehensive scenarios for the following
reasons:
.Society strongly depends on the continued availability
of renewable resources, such as food, fibre and fresh
water. Humans already dominate the use of most biological
productivity (Vitousek et al. 1997) and are a
crucial factor in defining fresh-water availability and
quality (Matthews et al. 1997; Hoekstra 1998; Arnell
1999). The continued supply of these resources can
only be guaranteed by sustainable land uses but these
are easily jeopardised by short-sighted or inappropriate
human activities (Doos 1994). An integrated approach
encompassing water, land and atmosphere
with human behaviour is thus urgently needed;
.LUC-LCC alters the Earth's surface characteristics and
thus climatic processes and patterns (e.g. Bonan 1997;
Claussen and Gayler 1997, see also Chapt. A.5 or A.8).
This also can alter water demand and supply or quality
and availability;
.LUC-LCC are important determinants of carbon, water
and energy fluxes (Braswell et al. 1997> and non-
CO2 greenhouse-gas emissions (e.g. Flint 1994; IPCC
1996). These fluxes and emissions directly influence
atmospheric composition and radiative-forcing properties
and, consequently, global and regional atmospheric
and hydrological patterns;
.LUC-LCC are an important factor determining the
response of species, ecosystems and landscapes to
environmental change and variability (Peters and
Lovejoy 1992; Leemans 1999). Land-cover modification
and conversion through, for example, nitrogen
addition by air pollution, drainage of wetlands and
deforestation change the behaviour of ecosystems;
.Finally, LUC-LCC are, in their own right, an important
component of global change (Turner et al.1995).
are important and need to be considered and addressed
in creating scenarios of Earth system dynamics.
LUC-LCC aspects, however, are difficult to define
regionally and globally. LUC-LCC is strongly dependent
on local environmental conditions and diverse societal,
economic and cultural characteristics and evolves from
diverse human activities (e.g. food and fibre production,
mining, recreation and nature conservation) that are
heterogeneous in their spatial (e.g. stand and landscape),
temporal (e.g. diurnal, seasonal, annual and interannual)
and societal dimensions (e.g. household, village, region
and country). LUC-LCC dynamics are therefore fundamentally
complex and amalgamate biogeochemical, hydrological,
atmospheric and ecological processes with
human behaviour and societal institutions. All these components
have, by definition, their specific properties and
act on a typical scale level (Fig. E.6). Regionally and globally,
only the cumulative consequences of all these diverse
local and regional LUC-LCC developments influence
the Earth system in a systemic manner. But in (global)
environmental-change or Earth-system studies it is
these coarse aggregation levels that matter.
Existing Land-use and Land-cover Datasets
and Scenarios ,
The starting point of LUC-LCC scenarios is always a description
of current or historical LUC-LCC patterns. Landcover
databases are used to initialise the models required
to develop the scenarios. However, there are large discrepancies
among different land-cover data-bases (Townshend
et al. 1991; Leemans et al. 1996). Some list only potential
natural vegetation, others include one or more
land-use classes, but none includes land use comprehen-
Fig. E.6. Interactions between different societal and ecosystems
dimensions

The
sively. Further, most databases are compiled from diverse irrigation water can easily be met when economic reand
heterogeneous sources and cannot be linked to a spe- sources are available (e.g. Rosegrant 1997). Global-change
cific time period. The determination of change is there- assessments, however, require broader LUC-LCC scefore
extremely difficult. Most of these datasets are of narios than those do. Adequate scenarios should incordubious
quality. However) more recent statistical data porate land-use activities and land-cover characteristics
sources on land use were improved and their internal in order to make comprehensive estimates of the role of
consistency is enhanced (e.g. FAO 1999). Also the high- land and land use in defining the dynamics of the Earth
resolution spatially explicit global database, DISCOVER, system. Currently) the only approaches to deliver such
has become available (Loveland and Belward 1997> and capability are Integrated Assessment Models (lAMs:
is frequently used in global-change research. This data- Weyant et al. 1996). Some of these lAMs are the Asianbase
is derived from satellite data from the early nineties Pacific Integrated Model (AIM: Matsuoka et al. 1994),
and consists of land-cover classes useful for initialising Integrated Model to Assess the Global Environment (IMland-cover
scenarios. Further, several attempts have been AGE 2: Alcamo 1994; Alcamo et al. 1998) and Integrated
made to develop historical land-use and land-cover Climate Assessment Model (ICAM: Brown and Rosenberg
databases (Klein Goldewijk and Battjes 1997; Ramankutty 1999). These models already incorporate modules to simuand
Foley 1998). These attempts use historical proxyvari- late the consequences for LUC-LCC) which generate LUCables,
such as historical maps, population-density esti- LCC scenarios on a resolution ranging from a coarse grid (IMmates)
and the location of cities and other infrastruc- AGE 2 and AIM) or for socio-economic regions (ICAM).
ture, to reconstruct likely historical land-cover patterns The starting point for most of these models are crop-profor
the last centuries. Although it is difficult to determine duction models with different degrees of precision and
the validity and reliability of these historical databases, focus. Most lAMs focus on arable agriculture and neglect
all these improved databases are of utmost importance pasturalism, forestry and other relevant land uses. None
for initialising and validating regional and global models. of them, however, satisfactorily incorporate regional wa-
Many different scenarios for LUC- LCC exist. Many ter supply and demand issues. The feedback on land-use
of them focus on local and regional issues and only a change on the overlying weather has also not been infew
are global in scope. However) most of the available cluded in these studies. Only the IMAGE-2 group is cur-
LUC-LCC scenarios are not developed to determine global
and regional environmental change, but more to
rently developing such capability (Alcamo et al. 2000).
evaluate the dynamics and environmental consequences '
of different agrosystems (e.g. Koruba et al. 1996), agricultural
policies (e.g. Moxey et al. 1995») food security
(e.g. Penning de Vries et al.1997), and projections of ag-
Construction of Land-use Change and Land-coverChange
Scenarios
ricultural production, trade and food availability (e.g. Initially the consequences of land-use change were of-
Alexandratos 1995; Rosegrant et al.1995).Moreover)these ten only depicted as causing changes in the CO2-emisstudies
do not well define the actual implications for sions from tropical deforestation. The conversion of
land-cover patterns. At best they define an aggregated these forests is one of the important human sources of
amount of arable land, pastures and other land uses. CO2. Early carbon-cycle models used simply prescribed
LUC-LCC scenario studies use different approaches. deforestation rates and emission factors to estimate fu-
Most of them are based on regression and process- based ture emissions. Land-use scenarios could only provide
simulation models. Alexandratos (1995) has combined the relevant estimates. During the last decade a more
and expanded these approaches by interactively includ- comprehensive view emerged embracing the diversity
ing expert judgement. Regional and national experts of driving forces and regional heterogeneity. The curreviewed
the model results. If such a panel determined rent driving forces of most LUC-LCC scenarios are deinconsistencies
with observed trends or likely trends) rived from population, income and productivity asthe
scenarios were changed until a satisfactory solution sumptions for agriculture and forestry. The first two are
emerged for all regions. The resulting scenario can there- commonly assumed to be exogenous variables (i.e. scefore
be interpreted as a consensus scenario. This ap- nario assumptions), while productivity is determined
proach generally leads to conservative estimates of pos- dynamically by the models used.
sible trends, because the importance of new develop- In most scenarios) population is generally split into
ments are underestimated. urban and rural; each class is characterised by its spe-
Unfortunately) most of these LUC-LCC scenario stud- cific needs and land uses. The demand for agricultural
ies do not consider changes in water supply. Some in- products (both the quantity and composition as speciclude
weather change (e.g. Alcamo et al. 1998) and oth- fied by diet) is generally assumed to be a function of iners
assume that an increased demand for, for example) come and regional preferences. With increasing incomes

504
CHAPTER E,S" The Vulnerability Approach
there is a shift from grain-based diets towards more meat- in all regions, except Africa and Asia. Pastures expandmore
based diets. Such a shift has large consequences for land rapidly than arable land. There are, however, largeregional
use (Leemans 1999). Similar functional relations are as- differences. One of the important assumptions
sumed to determine the demand for non-food products, in this scenario is that biomass will become a suitable
such as timber, fibre and biomass-based fuels. Produc- energy source. The cultivation of such biomass crops
tivity is determined by the locally or regionally pr~vail- requires additional land. Similar changes in future land
ing environmental conditions. Soil and atmospheric con- use can develop if carbon sequestration forests are plantditions
defme potential yields of each individual crop. ed. Land use will thus respond to new needs for addi-tional
In the calculation of potential yields, changes in weather
and CO2 are frequently included. The actual yields are a
fraction of the potential due to losses associated with
products and services.
improper management, limited water and nutrient availability,
pests and diseases, and pollutants (Penning de
Uncertainties in the Interpretation of Scenarios
Vries et al. 1997). During scenarios development, there- Diverse applications of LUC-LCC scenarios have been
fore, assumptions have to be made on the actual yield developed. The different approaches are all very sensi-
levels as a function of agricultural management, such as tive to the underlying assumptions of future changes in
yield increases (e.g. fertilisation and irrigation) and pro- agricultural productivity. Estimating demand is relatection
measures (e.g. weeding and pest control). Irrigatively straightforward, because it depends on the number
tion is rarely considered explicitly or dynamically evalu- of people and their diet. Productivity in most scenarios
ated from realistic water supplies, nor is the effect of ir- is calculated through potential and actual yields (i.e. the
rigation on the local weather, such as described by Stohl- interaction between environmental conditions and aggren
et al. (1998). Further, most yield calculations assume ricultural management; Rabbinge and van Oijen 1997>.
constant soil conditions. In reality, many land uses lead Large differences in productivity exist between scenarios
to land degradation, which affects yields and in turn lim- and it is not always clear if potential or actual yields are
its and changes land use (Barrow 1991). Estimating fu- used. This leads to large differences in the conclusion of
ture yield developments is thus difficult because of its different LUC-LCC scenario studies. For example, the
dependence on management. Many contrasting percep- FAO scenarios study (Alexandratos 1995) concluded that
tions therefore exist in the literature.
land availability was not a limi':ing factor in most coun-
The demand for land-use products and the calculated tries, while IMAGE-2 scenarios (Alcamo et al.1998) show
productivities are then translated into the actual land that in Asia and Africa, land rapidly becomes limited
use and corresponding land-cover patterns. In this step over the same time period. In the IMAGE-2 scenario,
large methodological difficulties emerge. Agricultural relatively fast changes towards more meat-based diets
land uses in certain regions intensify (i.e. increase lead to a rapid expansion of grazing systems on which
yields), while in others they expand in area. The causal productivity only slowly increases. This strongly influ-
factors in different regions are different and the spatial ences the simulated extent of land use. The FAO study
and temporal consequences cannot be determined un- does not consider increases in grazing systems. This exambiguously.
For example, deforestation is driven by ample shows that not clearly specifying the actual as-
timber extraction in Asia and by the conversion towards sumptions can lead to discrepancies and inconsisten-
pasture in Latin America. In Europe, increasing produccies in scenario interpretation. Unfortunately most landtivity
and a leveling demand drives the contraction of use scenario studies focus on agricultural production
agricultural land and the subsequent expansion of for- on arable land and few at the same time include grazing
ested land. Additionally, land-cover conversions are not systems. Other important land uses, such as forestry,
a continuous process. Shifting cultivation is a common infrastructure (i.e. urban land uses) and nature conser-
practice, but in many regions agricultural land has also vation, and new land uses, such as biomass and carbon
been abandoned (e.g. Foster et al.1998) or is abandoned plantations, are generally not considered, nor is the in-
regularly (Skole and Tucker 1993). Such dynamic and fluence of these changes in the climate system consid-
complex aspects of LUC-LCC make the development of ered. In most scenario studies, integration of all land
comprehensive high-resolution LUC-LCC scenarios uses is not yet accomplished.
challenging, especially when simultaneously new land Reconciling demand and supply leads to different
uses (e.g. biomass plantations) are emerging.
land-use and land-cover patterns (cf. Fig. E.7>. This is
The result of most LUC-LCC scenarios is land-cover often done by heuristic rules that link {inter)national
change. This is illustrated by an IMAGE-2 scenario socio-economic drivers with patterns of resource avail-
(Fig. E.7). This scenario illustrates some of the complexiability. Cellular automata, for example, explicitly resolve
ties in the underlying land-use dynamics. Deforestation such rules by assigning probabilities for different land
continues globally. Only in the latter half of this century uses to each cell. These probabilities can change through
is it expected that the forested area will increase again time (e.g. White et al. 1997). The amount of land used is

506
CHAPTER E.5 .The Vulnerability Approach
The purpose of the scenario development, its underlying
assumptions and approaches used should be carefully
and critically evaluated before the resulting landcover
patterns are applied in other studies. In addition
the feedbacks between LUC-LCC and other components
of the Earth system (e.g. the weather, the hydrological
cycle, etc.) needs to be investigated, as described elsewhere
in this volume. Despite the obvious limitations of
current LUC-LCC scenarios, recent scenario developments
show how useful organised thinking (i.e. detailed
consistency checks between scenario assumptions and
outcomes and explicit incorporation of interactions and
feedbacks) about the future can be. However, a better
perspective on how to use land-use and land-cover
change as a driving force to environmental change is
strongly required. This requires the further development
of high-resolution integrated assessment models and
comprehensive scenario studies, integrating other aspects
of the Earth system, such as specifically land,land
use and water issues.
Burkhard Frenzel
E.5.3 Procedures to Assess the Effect of
Environmental Conditions on Water
Resources: Natural landscape Changes
Introduction
When dealing with the long-term palaeoecological processes
that might influence present-day water budgets, it
must be noted that climatic factors may have very longlasting,
but unexpected series of consequences. An example
of this is the immigration history of forest plantspecies
into vast regions of higher latitude after the end
of the last glaciation (Bernabo and Webb 1977; Gliemeroth
1995; Huntley and Birks 1983). These migrations,
which were triggered by Late-glacial and early Holocene
climatic changes, continued until other decisive factors
such as stronger competitors or adverse geological or
climatical situations finally stopped the immigration
process.
This immigration and reorganisation process of the
various forest communities influenced the water budgets
of the regions into which they immigrated but the
decisive climatic factor was at first the transition from
the Late-glacial to the Holocene. probably nearly all the
other processes that took place during immigration were
biotic consequences, which could not be predicted exactly
if only the quality and intensity of the Late-glacial
to Holocene climatic change was measured. It is scientifically
very difficult to understand and to quantify these
long-lasting, indirect climatic influences correctly, and
demands reliable quantification of the climatic change
triggers at a regional scale and with high and reliable
temporal resolution. The feedback of the surface environmental
changes within the climate system also needs
to be assessed. However, the normal physical dating
procedures can give only approximate data. Therefore,
warve-countings, dendrochronology or age determinations
from annually-layered ice have to be used instead.
This in turn means that much more comprehensive studies
are needed. Moreover, it will often become important
to look for marker horizons that can clearly identify synchronously-formed
sediments within the ocean, in lakes
and on the continents; in general, these will be windtransported
substances like volcanic ash, etc.
Vegetation Changes: Consequences for Water Budgets
It is evident that changes in the vegetation pattern and
in the prevailing plant communities may be triggered
by atmospheric change or by biotic and pedogenetic
changes, or even by grazing and trampling animals
(Gossow 1976; Peer 2000; Samjaa et al. 2000; Schaller
1998; Turner 1975). The approach is to study the vegetation
and its ongoing or past changes on a broad regional
scale, which helps to differentiate between more regional
or local processes. This means, in general, that the various
vegetation types within a landscape, e.g. a complete
catchment, have to be studied in space and time in such
a way that it opens up starting points for ecological investigations
or calcul~ions, i.e. one has to try to map or
reconstruct the vari6us important plant communities
which existed or which had changed within the catchment
area being studied. One has to determine whether
these changes had been triggered by long-term changes
in weather or by biotic processes only, e.g. by the immigration
of stronger competitors or by pedogenesis, as
far as this is independent from vegetation (Frenzel 1994).
Plant ecologists in general determine the water budgets
of only certain plant species then, if necessary, extrapolate
to whole plant communities. Hydrologists, on
the other hand, generally investigate the water budgets
of whole catchments, which might be covered by several,
quite different, plant communities. Neither of these
approaches yields reliable data for understanding what
happens to the water budgets of a specific region, if certain
plant communities are changing. To overcome these
difficulties, at present it is only possible to rely on the
evaporation data of identical plant communities at comparable
sites, or at least to use data from related plant
communities, even though the specific soil, water and
atmospheric conditions may be different at individual
sites. These difficulties increase when reconstructing
former plant communities with no homologues (or at
least analogues) to be found in modern vegetation; or if
they do exist, do so under quite different environmental
conditions. Such an example would be the early Holo-

sa6ul!lj) adl!JspUl!lll!Jn~I!N :saJJnosaij Ja~I!M uo SUO!~!PUo) II!~UaWUOJ!AU3 :10 ~Ja:l:l3 alj~ ssassv o~ saJnpaJoJd " £'5"3
SUO!~;}l ~S1!A JO J!~S!l;}~J1!l1!qJ OS 'S~S;}lOJ qJ1!q-;}U!d ;}U;}J
AIUO ;}11! S;}n~OreU1! ;}SOqM ';}l;}qdS!W;}H U1;}q~10N ;}q~ JO
10 'SPI;}Y ;}Unp ;})(!I SUO!~!PUOJ ;}~!S re!J;}dS l;}pUn punOJ
;}Soq~ O~ SUO!~!PUOJ re~U;}WU01!AU;} ~U;}l;}JJ!P ;}~!nb l;}pun
;}~ ~1!~ ';}lOJ;}l;}q~ 'U;};}S ;}q U1!J ~I .p;}!pn~s UO!~;}l ;}q~ JO
-1!M ~U1!A;}I;}l ;}q~ ~U!~Jru~SUOJ;}l U! A~U!1!~l;}JUn JO ;};}l~;}P
;}~!nb U;}~O U1! U! ;}S1!;}JJU! A1!W S~;}~pnq-1;}~
sa:JJnosalj JaJDM U! PUD JSOJjDWJad U! Sa6UD4.J
~SOJJ~wJ~d Jo UJ~u~d uo!~nq!J~sw pu~ ~:>U~JJn:>:>O ~ql
J~qt~~M WJ~~-~UOI ~~~J~A~ ~qt uo ~u~pu~d~p J..~UOJ~S S!
JOf~w ~ s~q osre ~SOJJ~WJ~d '~W!~ J~AO ~~mq:> S~! uo pu~
~! ~J~qM SUO!~~J ~qt Jo s~~~pnq J~~~M ~qt uo ~:>u~nIJU!
-~JJ~~U! ~qt ~~J.. '(fL61 uJnqqs~M. '~'~) sd!qsUO!~~I~JJ~~U!
~s~q~ uo ~Jn~~J~~!1 ~A!su~q~Jdwo:> J..J~A ~ S! ~J~ql 'SJn:>:>o
U! SUO!~~J ~soqt Jo s~~~pnq J~~~M ~qt qt!M sd!qsUO!~~1
P~~~~!~S~AU! ~OU ~J~ ~u!J1!~dd~s!p S! ~SOJJ~wJ~d q:>!qM
-!Joq nos J~ddn ~qt ~~qt ~:>~J ~qt JO M~!A UI 'J..pU~!:>mns
%og-ov U!~~uo:> ~!J~q!S s~ q:>ns S~~J~ ~SOJJ~wJ~d U! SUOZ
-J~d JO ~U!~I~w u~ppns ~A!~~I~J ~ '~:>! s~ ~wnlOA J!~qt Jo
qs~JJ JO S~SS~W ~~nq p~~nq!J~uo:> ~A~q pynoqs ~SOJJP.W
uo '~!J~q!S UJ~~S~~ pm reJ~U~:> JO SUO!~~J J~~J~1 ~qt JOJ
p~~~W!~S~ u~~q J~A~U s~q S!ql 'SJ~A!J pu~ S~}f1!1 O~ J~~~M
J~A!J J~WJoJ U! p~WJoJ S! ~SOJJP.wJ~d J! 'pu~q J~qto ~qt
pynoqs s~~JnA!J pu~ s~}f1!1 J~WJoJ WOJJ JjounJ 'S~:>~JJ~~
-sn~:> UJn~ U! S!q~ '~wnloA U! ~S~~J:>U! ~q~ J..q p~p~dw! ~q
-JnS U! UO!~~Jn~~S-J~AO pm SUO!~~ynwn:>:>~ ~U~W!p~S ~U!
JO UO!~:>nJ~S~p q:>~3: '~Z~~JJ Osre J..~w q:>!qM J~~~M ~:>~J
~qt ~q mM ~A~~q-~SOJd 's~xnIJ ~~~q ~:>u~nIJu! J..nJ1!Ss~:>~u
n!M ~~J~ ~SOJJ~wJ~d ~ U! UO!~~~~~~A ~S~JOJ J~WJOJ ~qt
-JOJ snOJ~J!UO:> re~JoqJO J~U!1 ~q~ 'pu~q J~qto ~qt UO 'Jjo
-unJ J~dw~q U!~~~ n!M q:>!qM '(g'3: '~!d) s~:>u~nb~suo:>
SI!I{ ~S3l0d 'l!!l3q!S '1!!~no}j1!}.
J1Il~U3:> JO UO!S3l ~SOlJI!Wl3d
'8'3'6!~
3:>U3nb3SUO:>
3~1!!p3WW! 31{~ 3ll! spunow
-~SOlJ ~UI!W J.q p3l\!3p U33q
xnIJ-~113q 11 s3:JnpoJd 'q:JJ11{ PU11 3u!d J..{J11yn:J!~J11d 'S~S3
UO!Smdx3 3q~ ~ll!mOA11} snq~ 'SJ3J..11{ nos J3d33p ~ll!~11J~3
-u3d q~WJ11M J3WWnS S~U3A3Jd q:J!qM J3J..11{ ~u!p3dw!
lS0JjeWJ3d }O
sa.JJnosab' JaJDM /D!JJsaJJal
al/J uo sa.Juan/Ju/ J!al/l PUD Sa6UDl/) /aAa/-Das 'A1!M UMOU:>yUn
-03S 3~ UO PU1I '3{11:)S pUt! A~!:)°13A J!3q~ uo Su!pu3d3Q
JJounJ J3~1IMpUnOJS PU1I 3:)1IJJnS 3:)u3nyU! AI3A!~1IS3U
JO AI3A!~!sod pynOM S3SU1Iq:) 13A31-1I3S 'uo!~1In~!s {1I:)!SOI
'{11 ~3 UU1IW13~u1Ig ~'l661 '{11 ~3 nIl) SUO!S3J J3SJ1Il WOJJ
'~unO:):)11 O~U! U3}]1!~ 3q o~ 3A1Iq S3SU1Iq:) 13A31
-1I3S :)!~1I~SOS! PU1I :)!~1I~Sn3 3SJno:) JO '('lS"61 U!MpOf) ~Vg61
-Jo 3J1! SJO~:)1IJ ~U1IA313J 3~ 1{11 ~3A ~u3P!A3-JI3s S! S!q.L
Jo ~ndu! ~U3W!p3S 3~ J! 3~1I!~U3J3mp o~ ~yn:)YJ!p i..J3A U3~
13A31-1I3S 'uo!~1In~!s 3q~ SU!~1I:)!ldwo:) S! SUO!S3J 3U!J
-1IW Su!u!ofp1l 3q~ pUt! S3!J1!n~S3 J!3~ O~U! SJ3A!J J3SJ1Il
-1IJ3dw3~ 3q~ U! S3SUt!q:) Aq '31dw1Ix3 JoJ 'p3Sn1l:) S3SUt!q:)
-U! U1I:) ~11~ JO~:)1IJ J3~0Ut! 3J1I SJ3~1IM U1I3:)0 3~ Jo 3Jn~
'J3A03JOW 'SUO!S3J {1I~S1I0:) Jo s~3Spnq J3~1IM 3~ 3:)u3ny
-!AU3 {1I!J~S3JJ3~ Su!U!ofp1l Jo J..S°I°lpAq 3~ 3:)u3nyU! AI
-SUOJ~S U1I:) ~Jodsu1IJ~ ~U3W!P3S 3U!J1IW {1I~S1I0:) 3JOqS-IJo
'UO!~1Iyn:)J!:) J3~1IMpUnOJS SU!~:)3IJ1I sn~ PU1I 's3!qd1lJSOd
-o~ 3JOqS pUt! 3U!J1!wqns 3~ ~oq SU!Su1Iq:) Aq S~U3WUOJ
snOU3S0JqwO PU1I snou3Sodo~ 3q~ 3J1I SJO~:)1IJ J3~0
o~ U1I3:)0 WOJJ UO!~!sUt!J~ 3q~ ~11 WJoJ U1I:) ~11~ sSoq-~1I3d
-old AluO ~ou o~ 31q1l 3J1I A3q.L '(Vg61'{11 ~3 }]:)1ISSO)l) pu1Il
-U! OS{1I ~nq SWJO~S PU1I S3A1IM ~SU!1IS1I 3U!PS1I0:) 3~ ~:)3~
-!SOIO:)30311{11d ~S1IA 3~ ~3A 'JJOUnJ {1I!J~S3JJ3~ 3~ 3:)U3ny
-J!P MOq S3~1IJ~SUOW3p sSoq-~1I3d 3S3q~ UO 3Jn~1IJ3~!1 {11:)
-3SUO:) PU1I S3Sn1l:) U33~3q 3~1I!~U3J3JJ!P O~ S! ~! ~ln:)y
"{Vg61 '{11 ~3 }]:)1ISSO)l) UO!~1IWJOJ J!3q~ JO S3:)U3nb
Lo~

508
CHAPTER E.5 .The Vulnerability Approach
Migrating Dunes and Their Influence on Water Budgets
1\'10 types of migrating dune fields have to be taken into
consideration: those on the lowland coasts and those
within the continents. Causes and consequences of the
action of coastal dune fields are directly connected with
the processes going on along the lowland coasts. Concerning
the continental dune fields, one has to differentiate
between those of arid to semi-arid climates against
those of more temperate climates. The latter were mostly
formed during glacial times and their migrations
stopped by plant growth during the Holocene. In general,
they will become active again if the protective vegetation
cover is destroyed. This can occur due to landscape
conversion by people. In colder climates, where
plant growth is not rapid, fire and river action can cause
terrestrial sand dunes to migrate again. In arid and semiarid
regions on the other hand, continental sand dunes
will nearly always be able to migrate, provided the geological
situation is favourable. Here, migrating sand
dunes can block rivers and cause lakes to be formed.
This is not restricted to modern arid to semi-arid regions
only, but may also have happened during full-glacial
or late-glacial times of various Pleistocene glaciations
in what are now temperate climates.
During palaeoecological investigations these dunemade
lakes or swamps may easily be taken as indicators
of the weather becoming moister, though this has not
always been the case. Comparable effects can be observed
even today in temperate climates, as for example in the
lowlands of the River Leba, northern Poland (Tobolski
1982; Fig. E.9). Here, migrating dunes are invading a peatbog
or an alder carr. Again, the palaeoecological impression
would be that the weather had become moister but
Fig.E.9.
Leba-lowland, northern Poland.
A sanddune invades an
ombrogenous peatbog, causing
an inundation of the bog
in fact the drowning of the peat bog and of the alder carr
was only caused by migrating dunes impeding the runoff
and pressing groundwater out of the sediments.
Such migrating dunes, both at the coast and within
the continents, contain a lot of groundwater, thus favouring
peat-bogs to grow between the dune ridges or even
lakes to be formed (Pachur and Hoelzmann 1991). These
water masses within the dune sands can also be influenced
by permafrost. Moreover, it is inadequate to rely
on only one study site or small region. The investigation
of the history of the migrating dunes in the lowlands of
the River Leba, as already mentioned, exemplifies this
very well (Tobolski 1982). Only by a regional, comprehensive
investigation, could it be convincingly shown that
the middle to late Holocene start of the migration of these
dunes was caused by man; this in turn influenced the
morphology of the nearby sea coast, and the water budgets
and vegetation types, without any atmospheric impact!
A scientifically broad investigation of present-day
ecological and hydrological conditions in homologue
ecological situations is, therefore, of utmost importance
for understanding what has happened and for avoiding
incorrect palaeo- or ecological conclusions to be drawn.
Pedogenesis and Changing Water Resources
It is well known that pedogenesis is governed decisively
by atmospheric conditions, by the hydrological situation
and by the biosphere. With water resources, the most
important processes are the formation of finer-grained,
minerogenic particles and the narrowing of or even
blocking of pores that might contain and conduct water.
Soil formation is one long-duration process which
is often influenced by much older climatic or geological

processes. A broad literature informs about the wealth
of difficulties in dating reliably the beginning and the
various steps of pedogenesis in a given area (Morozova
1981). Where soils have been fossilised during the past
by natural or anthropogenic processes (e.g. the erection
of large burial mounds, etc.) the outlines of their formation-history
can be traced to some extent. Yet it has
always to be taken into consideration that even modern
pedogenesis can penetrate deep into the soil, thus influencing
former soil horizons. In this context it must be
stressed that the spontaneous or anthropogenic changes
to vegetation will influence soil formation intensively
thus influencing water resources indirectly.
Final Remarks
The task is to assess the consequences of natural landscape
changes on water resources. From the discussion
here, it will be seen that these natural landscape changes
are in general very complex and complicated processes,
in which the triggering mechanisms may be simple (in
general they are not) but the consequences and the interrelationships
of processes are extremely non-linear
and complicated. This being so, one should always restrain
from simple, uni-directional explanations. Instead,
the great variety of interacting processes has to be taken
into consideration. The palaeoecological and ecologicalliteratures
show that, however, a uni-directional way
of thinking is widely accepted. Very often we are a long
way from a reliable understanding of what had happened
palaeoecologic ally and thus what might have influenced
-and still influences -the environment, including
water resources. The only way to avoid these traps is
to complete a scientifically and regionally very broad,
very comprehensive, investigation of the palaeoecological
and ecological processes acting in a given landscape.
This is a very difficult task but is essential if we are to
understand how human beings alter the natural environment.
Thomas J. Stohlgren
E.5.4 An Example of the Vulnerability Approach:
Ecosystem Vulnerability
Rocky Mountain National Park and adjacent areas in the
Front Range of the Rocky Mountains in Colorado, USA,
like all bio-regions in the country, suffer from multiple
stresses simultaneously including: rapidly changing cultural
history, rapid human population growth, habitat loss
and loss of species, altered disturbance regimes, invasive
species, and air and water pollution (Stohlgren 1999a).
Rapid climate change is an additional stress that may
interact to a greater or lesser degree with these other
stresses. There is a need to assess the risks associated
E.S.4 .An Example of the Vulnerability Approach: Ecosystem Vulnerability 509
with such changes, and the resilience of the ecosystem to
resist negative impacts due to these changes.
Despite the recent attention that global climate change
and potential long-term effects of increasing greenhouse
gases in the atmosphere received, other environmental
stresses are higher priority issues and concerns for land
managers in the area (e.g. Rocky Mountain National Park
Resources Management Plan 1999). The economy of the
regions is transforming from extraction-based industries
(forestry, mining) to tourism (natural areas and wildlife).
The human population is increasing in the region at
2-3% yr-l (a doubling rate in 20 years), and annual visitation
to Rocky Mountain National Park now exceeds
3 million people (Stohlgren 1999a).Entire ponderosa pine
(Pinus ponderosa) forests, decimated byturn-of-the-century
logging and fire, have re-grown to create massive
fuel loads, that due to a half-century of fire suppression,
are ripe for catastrophic wildfire at the forest-suburb
interface (Veblen et al. 2000). There is a very strong link
of El Nino-Southern Oscillation on fire frequency and
size throughout the Front Range (Veblen et a1.2000).Also
demanding immediate management attention, are extralarge
herds of elk, without native predators (grizzly bear.
wolves). Many drainages have been altered by the nearcomplete
extirpation of beaver, altering vegetation succession,
nutrient cycling, soil erosion, and wildlife habitat.
Invasive annual grasses and noxious weeds are invading
these riparian zones, meadows, and low-elevation
forests, which as a result, urgently require control actions.
Air pollution adds more than 4 kg ha-l yr-l to terrestrial
systems in this region (Baron et al. 2000), contributes to
visibility problems, and potentially enhances the spread
of invasive species. Even without changes in climate, these
multiple stresses would remain top research and resource
management priorities.
Assessing the vulnerability of Rocky Mountain National
Park and adjacent lands to climate change requires
a full understanding of ecosystem components and processes,
existing stresses, and their interactions. Land managers
are often bombarded by so many current and nearfuture
problems and stakeholder demands, that longterm,
subtle, chronic stresses like increasing greenhouse
gases in the atmosphere are of less importance to them.
Still, the long-term conservation of natural resources and
healthy economies requires short- and long-term planning.
A common sense approach might include: (1) direct
and immediate attention to the most urgent and
costly stresses; (2) additional research on the role of ecosystem
components and processes within the climate
system, and (3) some research attention on improving
long-term scenarios. An alternative approach is to proportion
research, management, and coping efforts at
various scales (i.e. local, regional, national, global) to
assure that various stakeholder needs are met at the appropriate
scale. In any case, multiple stresses must be
assessed simultaneously.

Introduction
The central Rocky Mountains, like all bio-regions in the
USA, suffer from multiple stresses simultaneously including:
rapidly changing cultural history, rapid human
population growth, habitat loss and loss of species, altered
disturbance regimes, invasive species, and air and
water pollution (Stohlgren 1999b). Rapid large-scale atmospheric
circulation change is an additional stress that
may interact to a greater or lesser degree with these other
stresses. Assessing the immediate and long-term needs
of stakeholders, native biodiversity, the economy, and
There are strong suspicions that the episodes of
warming and cooling were abrupt rather than gradual
changes (Fall 1988; Schuster et al. 2000). Tree-ring
records nearby in the south-western US show numerous
abrupt changes in precipitation between 800 and
2000 years ago (Graumlich 1993; Hughes and Graumlich
1996), and the ice cores in Greenland (Groots et al.1993;
Alley et al. 1993) and Wyoming (Schuster et al. 2000)
have documented changes in mean annual temperature
by as much as 10 °C in a few years. Background rates of
climate change in the topographically complex Rocky
Mountains have been abrupt, bidirectional and unpredictable
(Schuster et al. 2000). The present rates of
change can be gauged against these background levels.
ecosystems requires that we understand all these stresses
and set appropriate priorities for prevention, control,
mitigation, and restoration. After a brief review of the
individual stresses, potential interactions are discussed
in this example, and strategies to assess priorities are
Rapidly Changing Cultural History
presented.
Cultural history has changed rapidly in the past 200 years
It is often helpful to evaluate climate trends in a having significant effects on the natural resources,land-
palaeoclimate context. The story that emerges in the scapes, and ecosystems in the central Rocky Mountains
central Rocky Mountains is one of tremendous tempo- (Buchholtz 1983). The discovery of gold in 1859 resulted
ral and spatial variability, and bi-directional changes in in an exponential increase in the human population,
climate and vegetation (Nichols 1982; Alexander 1985; which changed the regional economy and greatly altered
Whitlock 1993; Stohlgren 1999a,b; Schuster et al. 2000). nearly every ecosystem. Economic development began
The varying treeline locations provide evidence of to centre on mining, forestry, agriculture, recreation, and
drastic changes in climate. During the Pinedale glacia- the service industries that support these other economic
tion (22400 to 12200 years ago), the treeline in the activities (Lavender 1975). These activities required wa-
southern Rocky Mountains in Colorado was 500 m (Niter storage and re-distribution projects proportionate
chols 1982) to 1500 m (Fall 1988) below the present-day with human population growth (Stohlgren 1999a).
treeline, and annual temperatures were 7 to 13 °C cooler. The extraction-based culture in the Rocky Mountains
The lower limit of permafrost was 1 000 m lower in Colo- gained momentum by mining copper, gold, lead, morado
(Fall 1988). During the late-glacial period (15000 lybdenum, silver, tungsten and zinc, resulting in the con-
to 11000 years ago), annual temperatures were 3 to 4.5 °, tinued contamination of many lands and waterways.
cooler than today, and the treeline was about 500 m to Forestry was a major industry, though the rapid cutting
700 m below the modern treeline (Fall 1988). During the of old-growth timber, relatively slow growth rates, and
Holocene transition (12000 to 8000 years ago), subal- a shift to tourism has reduced the emphasis on forestry
pine forests probably migrated upslope, and montane in many areas. Agriculture includes dryland and irri-
forests of lodgepole pine and Douglas-fir began to exgated farming and livestock grazing. Livestock are frepand
in the lower elevations. There was a climatic optiquently moved between high-elevation summer and
mum from 9000 to 7000 years ago; mean annual tem- low-elevation winter pastures. Tourism is now the maperatures
increased 5 to 7.5 °C from the late-glacial into jor industry in the region -centered on Rocky Moun-
the Holocene. As the climate warmed between 9 000 and tain National Park, several national forests, major ski
4500 years ago, the upper treeline advanced to as high resorts, and summer vacation use in nearly all moun-
as 300 m above its present position during the warmest tain towns. It is against this backdrop of rapid cultural
period (Fall 1988). Cooling and treeline lowering oc- change that rapid human population growth has become
curred again between 4500 and 3100 years ago (Fall
1988) followed by another warming period from 3000
to 2000 years ago, and a cooling trend 2000 years ago
a major concern in many areas (Riebsame 1997).
(Elias et al. 1986). The Little Ice Age from around 1550 to
1845 A.D. ended very abruptly (Schuster et al. 2000) and
Human Population Growth
forced a descending of the upper treeline and perhaps human population grew rapidly between 1950 and
lengthened intervals between fires. Tree rings evidence 1990 with a 40-year increase of about 150% in Colorado.The
a warming trend in recent decades, but only in some population in Estes Park and visitor use of Rocky
areas (Weisberg and Baker 1995).
Mountain National Park has almost doubled since 1960

fnvasive
(Stohlgren 1999a). Current rates of human population
change are 2 to 3% for many areas in the Rocky Mountains
with urban sprawl and development in mountain
communities. Population growth increases demands for
water, power, and natural resources. Runoff and snowmelt
from the peaks supply the Arkansas, South Platte,
Colorado, and San Juan rivers with the water supply for
most western states (Riebsame 199]). Human population
growth and agriculture demands on water have
many of these rivers "over-subscribed". Evaluating the
effects of near-future weather and local land-use changes
on the quality, quantity, and timing of water are a primary
concern of stakeholders.
Range of Colorado show a tenfold increase in ponderosa
pine biomass since 1890 in many stands (M. Arbaugh,
US Forest Service, unpublished data; Veblen and Lorenz
1991). This has restored habitat for many wildlife species.
However, more than 60 years of fire suppression has created
hazardous fuels in a forest community that naturally
burned at 20-60-year intervals in many areas
(Veblen et al. 2000). Fire suppression has been effective:
perhaps too effective for some species. For example, the
number of three-toed woodpeckers increases greatly for
three to five years in burned stands, as the birds feed on
larval spruce beetles. These woodpeckers may have a profound
affect on nearby forest stands by reducing the severity
of spruce beetle epidemics. Herbaceous plant di-
Habitat Loss and Loss in Biodiversity
versity skyrockets after a burn. Seventeen years after the
390-ha Ouzel burn in Rocky Mountain National Park,
twice as many understorey species can be found in the
Another top priority environmental issue in the Rocky
Mountains, and one of the most urgent for land managers
throughout the west, is habitat loss and loss of biodiversity.
Most of the species on the endangered species
list are there because of habitat loss (mostly associated
with land exploitation) or invasive species (Flather et al.
1984). Beavers (Castor canadensis) that once played important
roles in shaping vegetation patterns in riparian
and meadow ecosystems in the Rocky Mountains are virtually
absent in many areas (Chadde and Kay 1991; Knight
burned area than can be found in adjacent unburned forest
stands (Stohlgren et al. 1997>. Species dependent on
post-disturbance stands suffer habitat loss with each suppressed
fire.
All rivers in the Rocky Mountain region have been
altered by reservoirs or other water projects (transbasin
canals, irrigation ditches, and small water impoundments).A
majortransbasin water import project in Colorado
carries about 370 Mm3 yr-l from the Colorado River
(west of the Continental Divide) through a 7.8-km tun-
1994). Top predators such as grizzly bears and grays were
hunted relentlessly by European settlers and have been
extirpated from most of the Rockies. Amphibians have
declined through habitat loss, predation by non-native
nel to the Big Thompson River (east of the Continental
Divide). Ten reservoirs were built to support the project.
In the catchments of the Arapaho-Roosevelt forests
in Colorado are approximately 40 major reservoirs
sport fishes, timber harvest, increased ultraviolet radiation,
and disease (Corn and Fogelman 1984; Bury et al.
1995). Nearly all native fisheries in the Rocky Mountains
have been compromised by introduced fishes (Trotter
1987; Behnke 1992). Populations of bighorn sheep are at
only about 2% to 8% of their populations at the time of
European settlement (Singer 1995). Long-term changes
in weather have not been implicated in the loss of any
(C. Chambers, US Forest Service, pers. comn., February
1995). Reservoir-building directly tracks human population
increases (Stohlgren 1999a). Domestic water use
accounts for less than 6% of the total water use, agriculture
accounts for about 90% of the water use (United
States Geological Survey 1990).
species in the US but abrupt changes in weather in an
already fragmented landscape, and in combination with
j Species
other stresses, may have a greater effect now than weather
changes in the past. Shifting the focus of biodiversity
] [nvasive non-indigenous species are causing severe losses
in j agriculture, forestry and fishery resources in some
preservation to the radiative effect of increased CO2 in
the atmosphere (see Sala et al. 2000) may be tangential
given that the overriding causes of species endangerment
are habitat loss (land-use change) and invasive species
(Flather et al. 1984).
regions and ecosystem processes throughout the US to
the tune of $138 billion per year (Pimentel et al. 1999;
Mack et al. 2000). Land managers, farmers, ranchers, and
1the
public consider invasive species a top priority threat
I(Mack
et al. 2000). European cheatgrass has invaded significant
portions of the western pinyon-juniper woodlands,
ponderosa pine, and Douglas-fir areas in the Rocky
Altered Disturbance Regimes
Mountains (Peters and Bunting 1994). Several rare plant
species are being displaced by introduced plants in west-
Vegetation types that are adapted to frequent fire are a
major concern in the foothills of the central Rocky Mountains.
For example, ponderosa pine forests in the Front
ern rangelands (Rosentreter 1994). A dense cover of
cheatgrass increased the fire frequency in many of these
areas. With each fire, the dominance of non-native an-

512
CHAPTER E.5 .The Vulnerability Approach
nual grasses is enhanced at the expense of native perennial
grasses. Exotic plant species such as spotted knapweed,
Kentucky bluegrass, common dandelion, and Rus-
Air and Water Pollution
sian thistle are rapidly invading many landscapes (Lang- Air and water pollution also continue to be major stresses
ner and Flather 1994; Stohlgren et al. 2000). Purple loose- throughout the region. For example, air pollution, pristrife,
another European weed, is beginning to invade marily from the combustion of fossil fuels in the Denver-
Rocky Mountain wetlands and streams ides. Purple loos- Boulder-Fort Collins metropolitan corridor, may diminestrife
spreads quickly and crowds out native plants that ish water quality, benefit exotic plant species, and affect
animals use for food and shelter. Most invaders have no nutrient cycling in Colorado (Baron et al. 1994). Chemi-
natural enemies in the United States and therefore spread cal analyses of the high-elevation Colorado snowpack are
unchecked (Thompson et al. 198]). The effects of these revealing high concentrations (about 15 microequiva-
introduced plants and links to weather are poorly unlents/litre) of sulphate and nitrate in areas north-west of
derstood. The level of soil disturbance (road building, Denver (Turk et al.1992). Remote areas of the world typi-
ploughing, small mammal burrowing, etc.) is most often cally receive less than 0.5 kg ha-l yr-l of inorganic nitro-
implicated in the spread of invasive species.
gen, whereas the high-elevation sites in the Colorado
Greenback cutthroat trout was near extinction by the Front Range now receive as many as 4.7 kg ha-l yr-l of
early 1900S because of broad-scale stocking of non-native
brown trout and rainbow trout, land and water ex-
inorganic nitrogen (Williams et al. 1996). In February
1995, the Colorado Air Quality Commission increased the
ploitation, mining, and logging (Colorado Division of Denver metropolitan area's particulate pollution limit
Wildlife 1986; Henry and Henry 1991). Three of the four from the current 41.2 t d-l to 44 t d-l in the next 20 years.
other native subspecies of cutthroat trout are extinct Terrestrial biota of high-elevation areas may not have the
(Greenback Cutthroat Trout Recovery Team 1983). Most ability to respond to this increased nitrogen loading
aquatic ecosystems in the Rocky Mountains are now in- (Nams et al. 1993). We can expect direct and indirect effluenced
by non-native brown trout from Europe and fects on ecosystem functions in forested catchments
rainbow trout from the Pacific Coast. One of the subtle, (Baron et al. 1994).
yet devastating, trends in the Rocky Mountain fishery is Water quality is a growing concern in the Rocky
loss of genetic diversity in native fishes from introduc- Mountains. The United States Geological Survey (1993)
tions of non-native fishes.
stated that all of Colorado's major drainages are affected
The Rocky Mountain goat and the moose, which are to some degree by pollution. Historic mining operations
damaging to native plant species, were deliberately in- still contribute toxic trace elements to more than 2 100 km
troduced into Colorado. Accidentally introduced mam- of rivers and streams in Colorado. Of the 50300 km of
mals in Colorado and Wyoming include the house mouse streams in Colorado, more than 900 km of streams do
and the Norway rat (Armstrong 1993). The potential ef- not meet water-quality criteria for fishing. Other Rocky
fects of these introduced mammals on Rocky Mountain' 'I\1ountain states report similar problems (United States
ecosystems are poorly understood.
Invasive exotic diseases are increasingly devastating
Geological Survey 1993). Although water developments
affect riparian zones upstream and downstream from
to native flora and fauna. White pine blister rust is caus- dams (Mills 1991), regional information on the biotic efing
up to 50% mortality of white pines in areas of Monfects of water projects and pollution is extremely limtana.
Lungworm-pneumonia complex is a bacterial disited, fragmentary, or inaccessible. Air pollution and waease
that causes spontaneous mortality in the lambs of ter pollution are immediate threats to human health,
bighorn sheep in summer (Aguirre and Starkey 1994).
Proximity to domestic sheep highly correlates with mortality
in newly reestablished bighorn sheep populations
biodiversity, and local economies.
(Singer 1995). Whirling disease, introduced from Europe,
is a parasitic infection that attacks recently hatched trout.
Potential Interactions
It is now affecting native and non-native trout popula- A vulnerability assessment requires an understanding
tions in Colorado. At first, the disease was thought to of existing and potential interactions among stresses
affect only hatchery fishes; however, the native greenback
cutthroat trout may also be susceptible.
(Fig. E.10).
For example, air pollution and fertilisation may add
While weather interacts with some of the disturbance high levels of nitrogen into rivers and streams (Baron
mechanisms such as fire, dispersal and the human redis- et al. 1994; Williams et al.1996). Nitrogen concentrations
tribution of species is the primary cause of invasion (Mack may be offset somewhat by increased precipitation and
et al. 2000). The expense and difficulties of preventing, runoff, or they could be exacerbated by drought. The
managing, and controlling invasive species may be only flow of nitrogen through the ecosystem might be fur-
weakly linked to weather relative to other factors (landther influenced by land-use change (e.g. deforestation,
use change, human-aided dispersal, pollution, trade, etc.). urbanisation, intensive grazing), invasive plant species

~sed M~tic
(e.g. nitrogen fixers or nitrogen accumulators), or by the
extirpation or re-introduction of beavers. Thus, assessing
the vulnerability of water quality in a region to rapid
change, multiple stresses and interactions must be considered.
Setting appropriate priorities for prevention,
control, mitigation, and restoration of water quality must
consider all existing and potential stresses and their interactions.
Local decisions on land use, control of invasive
species, and fire management can have substantial
Strategies to Set Priorities
Assessing the immediate and long-term needs of stakeholders,
native biodiversity, the economy, and ecosystems
requires that we understand all these stresses and
set appropriate priorities for prevention, control, mitigation
and restoration. It is common for land managers
and stakeholders to address single stresses over the
short-term (Stohlgren 1999a,b). Immediate issues related
to rapid human population growth, habitat loss
and loss of species, altered disturbance regimes, inva-
effects on important resources.
As another example, increased CO2, warmer temperatures,
and increased precipitation may favour forest growth
in several areas. However, increased forest growth may sive species, and air and water pollution often take prec-
reduce understorey plant diversity, forage for wildlife, and edence over long-term responses to these same stresses.
riparian zone width, while increasing the forest fuel build- For example, rarely is long-term change recognised as
up and the threat of catastrophic wildfire in areas of in- the top research and management concern in city plancreasing
human population growth. Meanwhile, warmer
ning documents, county budget documents, or resource
night-time or winter temperatures may facilitate insect management plans in national parks and forests. Land
outbreaks by reducing the winter-kill of insect eggs and managers are duly concerned and preoccupied over
larvae, possibly counteracting the increased forest pro- recent building permits on adjacent lands, loss of wetduction
with increased forest mortality (Robertus et al. land habitat and recent amphibian decline, invasive
1991). Changes in forest composition, age structure, and noxious weeds and diseases, air-quality violations due
biomass can greatly alter hydrology. Economic models to prescribed burning to bring forest-fuel levels under
of change must consider many possible interactions and control, and current water-pollution problems from
outcomes from a suite of forces in complex systems. air pollution or pesticide use from nearby. Land man-
Human Population
Growth, Increased
Demand for Resources
Building in fire-prone
vegetation types; fire
.suppression policies
Increased water use
and Threatened and
Distuib7 Endangered species
Invasive Species Altered Disturbance
Regimes
ttre
exotic
Soil
disturbances
Land-use change
(habitat fragmentation
and habitat loss)
Albedo; surface roughness
Fig. E.l O. Existing and potential interactions among stresses
plants Decreased water
Changes in long-term
weather patterns
N fertilization
on weeds
Forest
age class
Air and Water
Pollution
linked
to cloud cover
in riparian zones;
increased
contamination

514
CHAPTER E.5 .The Vulnerability Approach
agers are often affected most by short-term and local
issues.
Stakeholders with economic interests (e.g. local businesses,
water managers and farmers) are often concerned
with short-term economic returns and the immediate
resource supplies that maintain the status quo. Thus, there
may be significant pressure to manage for short-term
needs that are easily addressed and matched to one- and
two-year funding cycles, short-term public opinion and
management plans, and average job longevity (c. three
to five years). It follows that meagre research funds are
frequently spent on short-term studies of problems.
There are several reasons why addressing only shortterm
stresses one at a time might be very shortsighted.
First, as stated, there are many obvious interactions
among the stresses that can have compounding and confounding
effects. Second, there could be "threshold" effects,
non-linear responses, or lag effects where the effects
of single or multiple stresses go undetected until a
given level is exceeded. Subtle changes are often the most
difficult to detect and manage. Third, we do not fully
understand, or even marginally understand, the longterm
effects of multiple stresses in most ecosystems and
regions. For example, rapid, but long-term, weather
change or increases in variability or extreme events
could make the management of the other stresses that
much more difficult and expensive in the future.
Conclusion
I
The human population in the Rocky Mountain region
will probably double in the next 20-40 years, creating
increased demands for and pressures on natural resources,
especially water. National parks, forests, and wilderness
areas will probably become increasingly insular,
and habitat fragmentation will increase in the na-
ture reserves and the urban areas. Continued species
decline (e.g. prairie dogs, amphibians, deer, old-growth
and fire-dependent species), habitat loss (wetlands, riparian
zones, and old-growth forests), increased air pollution
and water developments, altered disturbance regimes,
and introduced species and diseases will continue
to affect many Rocky Mountain ecosystems. Vulnerability
assessments must recognise modern humans as being
both a primary stress factor and as stakeholders.
Combined with standardised biotic resource inventories,
predictive models, and long-term monitoring, vulnerability
assessments are a logical approach to responsible
land and water stewardship in the face of multiple
stresses.
Setting priorities for the control and mitigation of
multiple stresses requires detailed information on shortand
long-term ecological and economic effects (Pielke
1998b). Some level of effort must be reserved for long-term,
unpredictable responses. However, adequate funds must
be allocated for the primary, short -term stresses that have
an immediate effect on the region's resources and people.
Current long-term GCM sensitivity experiments, with
no known probability, may be of limited use to local land
and water managers, but they may focus some attention
on the need for a longer planning horizon and a more
complete understanding of the Earth's climate system and
potential human impacts to that system.
In the absence of adequate information and models,
setting specific "coping" priorities is tenuous and should
be an iterative process that should be improved as additional
information becomes available. The palaeoclimate
records warn us to be better prepared for rapid, abrupt
changes -in any direction. Interdisciplinary research
and monitoring of ecosystems, anticipating economic
concerns, and clearly identifying society's immediate
and long-term needs are prerequisites to vulnerability
assessments and for setting and adjusting priorities.

Case Studies
Roger A. Pielke, Sr.
In order to illustrate that environmental vulnerability
involves a spectrum of threats, several specific examples
are provided in this chapter. These examples demonstrate
why it is so important to start with the assessment
of vulnerability when studying the interaction of
humans with the rest of the environment.
E.6.1 Population and Climate
Introduction
Charles J. Vorosmarty .Jake Brunner
Carmen Revenga .Balazs Fekete. Pamela Green
Yumiko Kura .Kirsten Thompson
One aspect of the global change debate that has received
relatively little attention from the Earth systems science
community is the purposeful intervention of humans
within the terrestrial hydrological cycle as we establish
and use water resources. Recent studies have shown
that this phenomenon is indeed global (Dynesius and
Nilsson 1994; Vorosmarty et al. 1997; Postel et al. 1996)
and the growth in water use by humans is likely to continue
over the next several decades (Rijsberman 2000;
United Nations 1997; Shiklomanov 1996). Water engineering
will remain an important human endeavour
over this time frame as both economic development and
population growth will force future demands on the
natural water system. The stabilisation of water supplies
through control of terrestrial runoff and/or the translocation
of water stocks from aquifers or through interbasin
transfers will thus remain a fundamental feature
of the land-based water cycle. These activities purposely
seek to optimise water security and hence seek
to reduce the degree of vulnerability to water shortages
and/or excess.
In this contribution we document the emerging role
of direct human usage of water. We explore the value of
linking biophysical and socio-economic information
within a common framework. We also highlight key areas
of uncertainty and draw inferences about how our
current knowledge base is related to the issue of water
vulnerability. We-summarise an emerging paradigm for
successful water resources management and explain the
role of the Earth systems science community in supporting
this new strategy for water development.
Water Supply, Use and Emerging Patterns of Scarcity
Assessment of the global water supply has been an imprecise
science. If we consider the renewable resource
base to equal the long-term mean runoff we can find estimates
which vary between 33500 and 47000km3yr-1
(Kourzoun et al. 1978; L'vovich and White 1990; Gleick
2000; Shiklomanov 1996; Fekete et al. 1999). One of the
most recent such assessments, made using time series of
observed discharges over approximately 70% of the discharging
land surface of the globe and a model-based
extrapolation, offer 39600 km3yr-1 (Fekete et al. 1999).
A more classical accounting by Russian scientists cites
42750 km3yr-1 (Shiklomanov 2000). Although one could
view these differences as being of simple academic interest
only, they become amplified at regional or smaller
scales and can give substantially different views of per
capita water supplies and hence vulnerability to water
scarcity. For example, in North Mrica the former study
gives a total of 362 km3yr-1 while the latter shows
181 km3 yr-l. Assuming a population in 1995 of 150 million
we get per capita supplies varying between 1 200 and
2400 m3 per capita. This estimate highlights the difficulty
in attempting to assemble a coherent picture of
water availability in the face of declining hydrographic
monitoring (Vdrijsmarty et al. 2001). Further, in many
parts of the world, groundwater is an important source
of water supply (both renewable and non-renewable;
Shiklomanov 1996). Our understanding of the hydrography
of groundwater resources is even more limited,
since well-logged, groundwater discharge/recharge, and
aquifer property data are neither synthesised nor released
to the global change community. Noteworthy exceptions
exist for regional aquifers (e.g. High Plains in US; Weeks
and Gutentag 1988; Nativ 1992), suggesting the feasibility
of providing such information when suitable financial
and technical resources are made available. Nonetheless,
there is but one 1: 10 M scale map (Dzhamalov

--~ ~
which met strict selection criteria in terms of record
length and consistency with neighbouring gauging stations,
were selected from the GRDC archive for use in
developing the database. These stations were co-registered
to a 30-minute Simulated Topological Network
(STN-30; V6rosmarty et al. 2000).
The Center for International Earth Science Information
Network has produced a global population database
using census data for 120 000 administrative units
(CIESIN 2000). We compare conditions between 1995
and the year 2025, starting with a gridded dataset at 2.5minute
(4 km) grid increments. The 1995 cell counts were
extrapolated exponentially to 2025 and then capped using
the low growth projection for each country (United
Nations Population Division 1998). The cell population
counts were then aggregated to 30-minute grid increments
to match the runoff database. Figure E.12 shows
the population density in 1995. For this study, the lowgrowth
projection is considered to be more realistic than
either the medium or high projections (Seckler et al.
1999). But use of the low-growth projection, in conjunction
with the maximum sustainable water supply, means
that the water scarcity estimates should be considered
as conservative. For the purpose~ of this study, we do
not consider potential weather change impacts which
add an additional layer of complexity and potential vulnerability.
Future water scarcity may be more acute than
this tabulation suggests.
The runoff and population databases were summed
by river basin to calculate the renewable supply of water
per person. We estimate scarcity using a set of standard
rules (Hinrichsen et al.1998). First, populations living in
basins with water "crowding" of 590 to 1 000 people per
106 m3 yr of water supply are classified as water stressed,
and tend to experience severe shortages in drought years.
Second, people living in basins with 1000-2000 people per
106 m3 yr are classified as water scarce, and can expect
chronic shortages of freshwater that threaten food pro-
Population (thousands)
I < 1
11-10
110-100,
100-1,000
I > 1,000
No Data
~--
Fig. E.12. Population density from CIESIN (2000)
Global Estimates
E.6.1 .Population and Climate 517
duction and hinder economic development. Finally,
people living in basins with more than 2000 people per
106 m3 yr face absolute water scarcity, beyond which development
is highly constrained without access to alternative
water sources.
River Basin Estimates
Figure E.13 and Fig. E.14 show river basins under varying
degrees of water stress in 1995 and 2025. (Only basins
larger than 25000 km2 are shown.) These maps show
increased water scarcity in Southern and Eastern Africa,
Pakistan and central Asia, but stable conditions in Latin
America, South-east Asia, China and India. But this overview
hides important basin-specific trends. Table E.l shows
total water supply, population, and water crowding in
1995 and 2025 for large basins that are currently water
stressed or will be close to such stress by 2025. Together,
they hold half the world's population. Average water crowding
in these basins is projected to increase by 25%, from
546 people per 106m3yr in 1995 to 735 people per 106m3yr
by 2025. Water crowding in some basins, notably those in
China, will increase by less than 15%, but the Indus and
the Nile will suffer declines in excess of 40%. In the case
of the Indus, this will result from the growth in Pakistan's
population, from 136 million in 1995 to 246 million
in 2025. In the case of the Nile, the decline will mainly
result from a doubling of Ethiopia's population, from
55 million to 108 million. Egypt, the second most populous
country in the basin, will see its population grow
more slowly, from 52 million to 63 million.
The river basin estimates were summed to give the total
number of people living in conditions of water scar-
~
It
~ :-' ~:--~'-'~':'.-.::c~':":"'" ..,..----,
,/

81S
sa!pms as e) , 9'3 !l3.LdVH)
"W U°!lI!W I aldoad
ooo'~ -ooo'~ I
ooo'~ -009 I
009> I
BjBO ON
a6ue4::> IUa~Jad
6U!PMOJ::> JateM U!
UO!illq f"t :l1I1:{:I SMO1:{S 't'g: 3{q1l.1 'S"'tO't PU1I S"661 U! A:I!:>
-UO:> U! P3A!{ 'UO!:l1Irndod S,PPOM 3q:1}0 %It JO '3{d03d
U! (JA ,W 901 J3d 3{d03d 06S" m:>s J3:111M}0 SUO!:I!p
-ndod S,PPOM 31:{:I}0 %6t JO '3{d03d UO!illq S"'f pU1! 'S"661
J3qwnu 31:{.1 'S"'tO't Aq SUO!:I!PUO:> 3S3q:1 U! 3A!{ illM 'UO!:l1I{
A:I!:>m:>s J3:111M 3:1n{Osq1l }O SUO!:I!pUO:> SU!:>1I} 3{d03d }O
-nq 1'1 WOJ} 3S1I3J:>U! illM (JA,W 901 J3d 3{d03d 000't Y![8nb 0M.:j 3J!nb3J S:lrnS3J 3S31:{.1
U! 3U!{:>3P 1:{:>!1:{M) 1:{:lMOJS UO!:l1Irndod UO p3S1Iq A{UO 3m
:lunO:>:>1I O:lU! 3}{t1:1 :IOU Op A31:{.1 '(ppOM 31:{:I}0 S:lJ1Id M3} ~
'3SU1!1:{:> J3q:1113M 'Uo!:lnnOd }O A{ddns J3:111M uo S:I:>3JJ3 3q:1
-!:lS3J3pUn A{q1lqOJd 3JO}3J31:{:I A31:{.1 ':ju3wpunodw! JO
-:>3$ 'UOSJ3d J3d A{ddns J3:111M U! 3U!{:>3P [80:1:>11 31:{:I 3:111W
-3P J3:111M :lU1I:lSUO:> 11 S3WnSS1I S!SA[8U1I 31:{:I 3Sn1l:>3q 'pUO
:lU3S3Jd3JS!W A{q1l:l!A3U! S:I{nS3J 31:{:I 'uOSJ3d J3d PU1IW
'SU!S1Iq U!1I:lJ3:> U! A:I!:>J1I:>S J3:111M
~:~::~~~:
--"'-
~66t .11131 3~ .l0} UIsllq 3SIIU!II.lP 1q
}O UO!~:>unrUO:> 3~ 1Q p3tndwo:> SUjpMO.l:> .l3~IIM o£ L '3 °6!.i
---'-
~
P3S!UIIS.lO '1:ImQIl{!t!AII .l3~1!M pUll '3sn .l3~IIM 'UO!~IIJndod uo lI~ep p3:>U3.13}3.103S
-~
~661 °t ~lOl U! SUO!tj'pUO:J SU!lP.dwo:J pm U!~q 3S1!U!1!JP J.q P3S!U1!SlO
'UO!t1!jlldod UO 1!tep P3:JU31313103S 10 UO!t:Jun[uo:J 31{t J.q p3tndwo:J SUj'pM01:J 13t1!M U! 3Smq:) 't L '3 .6!~
uopnQ!lJUOJ l!aql PUD Salnpa:>Old uopD/nqDl)o J:>aJj3
AJU!DJla:>Un OJ
3A1!tI J1!tIJ SU!S1!q U! ~U!A!y 3yd03d )0 J3qwnu 11!JOJ 3tI.L
£w Dol 1 U1!tIJ SS3Y )0 Ayddns J3J1!M 3yq1!M3U3J 11!nuu1! U1!
SnO!A3Jd 3WOS U1!tIJ J3tI~!tI %09 S! ~661 U! UOSJ3d J3d
-In:> 3JOW tI~n°tIJTV '(9661 AOU1!WOy}{!tIS '~'3) S3J1!W!JS3
-UnJ 11!qoy~ MOY AY3A!J1!Y3J 1! PU1! 1!J1!p uo!J1!rndod JU3J
J1!tIJ S! UOS1!3J U!1!W 3tIJ 'SJOJ:>1!) ~U!Jnq!JJUo:> 3J1! ))0
SU!S1!q J3A!J ~U!Sn AJ!:>J1!:>S J3J1!M p3J1!W!JS3 ApmS S!t:p
3JOW 3Jmd1!:> S3SA11!U1! 13A3Y-U!S1!g 'S3!JJunO:> U1!tIJ J3t:p1!J
-unJ pU1! uo!J1!rndod U! AJ!Uq1!!IeA 11!!J1!ds JU3J3tIU! 3t:p)0
Y1!qoy~ 3JO)3J3tIJ PU1! (AJ!:>J1!:>S J3J1!M 3:>U3tI PU1!) ))0
pU3J I1!M SJrnS3J Y3A3Y-U!S1!q 3J1!~3J~~1! J1!tIJ S3J1!W!JS3
-J3A1! AJJuno:> uo p3S1!q 3S°tIJ U1!tIJ 3J1!Jn:>:>1! 3JOW 3q OJ
'S3~1!
--- ~
'l.tmq1!U1!A1! 13t1!M pm '3sn 13t1!M
~

~tOt pUB ~66t 'U!Seq J~A!J J..q J..~"Je:>S J~~eM .L'3 alqel
~'lO'l pu~
~66{ '1..~1:>JP.:>s 10!~~M ~qolD
"Z"3 ~Iqel
le~Ol
OOOl <
OOOl- 000 L
000 L -06S
pa~e)olleUn
06S >
LlOl
LS5
699
l60£
5995
os{t! U1!J O!M 'O!AOq1! pO!tuO!sO!Jd J..~O{OpoqtO!w O!~ ~U!s!l
-qsO!JJ O!{q1!MO!UO!J {t!qO{~ 'S!S1!q 1!t!d1!J JO!d 1! uo t1!qt MOqs
£w ooS 17 Ot S661 U! £w 0069 WOJJ {{t!J fiM J..{ddns JO!t1!M
pO!S!U~OJO!J O!q tsnw t! tnq 'doJp O!~J1!I1! S! S!qJ. .0SO'l: U!
JY!JO!ds-uo!~O!J J..{q~!q J..t!{t!O!J U! S! uo!tJnpO!J O!qt t1!~
-nq O!qt Jo }{{nq O!~ t1!qt pO!t1!W!tsO! uO!O!q S1!q t! 'PO!O!puI
Ot J..t!w!XoJd {t!JodwO!t PU1! {t!!t1!ds U! S! uo!t1!{ndod U1!W
{O!tSOd) JO!t1!MqSO!JJ Jo J..{ddns O!{q1!MO!UO!J O!qt Jo %0£ J..{UO
S! S!qJ. -J..{ddns S!~ Jo %oS O!sn O!M t1!qt PU1! (9661 .{t! to!
sno{ndod tSOW OMt S,PPOM O!qt U! tu1!tJodw! J..{{t!!JO!dsO!
UJO!qtJON i6661 JO!{}{JO!S) 1!!PUI PU1! 1!U!q::> :sO!!JtunoJ
-tS1!H lO!M J..JO!A S! 1!U!q::> UJO!~nos O!nqM 'i..Jp l..JO!A S! 1!U!q::>
O!J1! 1!!PUI uJO!qtnos PU1! tSO!M O!nqM 'to!M J..JO!A S! 1!!PUI UJO!
O!sO!~ {{t! U! o!A!{ O!ldoO!d Jo SJO!qwnu O!~Ie{ O!sn1!JO!H -J..Jp l..JO!A
-U!qt!M tu1!tJodw! S}{S1!W wO!qt ~U!t1!~O!J~~1! 'SUO!~-O!J
O!t1!tsJO!pun snqt SJO!qwnu O!t1!~O!J~~V .sO!JuO!JO!JJ!P l..JtunoJ
Ot J..{O!){!{ S! t1!qt i..t!JJ1!JS JO!t1!M {t!!tuO!tod Jo wO!{qoJd O!~
.SO!{t!JS {t!JO{ O!JOW t1! O!JO!AO!S O!JOW qJnW O!q
Lvl
-IeJS JO!t1!M O!A!t1!{O!J pO!t1!{nq1!t O!M 'UO!tJO!SS1! S!~ tSO!t OJ.
.o!.!) O!sn JO!t1!M O!A!t1!IO!J Jo S{O!AO!1 O!A!SSO!J~OJd JnoJ t1! J..t!J
61
OL
ZL
SS
V
DOL
a~eW!D pue UO!~elndod ° L °9°3
£8L l
vl9
LLOl
v6v£
96l
vLlL
OOL
61~
Su1~uno:>:>~ ~U;}J;}}J1P JnoJ Su1sn (J..lddns/It!M~JP~1M It!~o~
-;}J p~OJq 9't SSOJ:>~ SU01~~S;}JSS~ P;}J;}P1suo:> ;}M 's~1un
P1JS It!np1AWU1 pulJ 'su1s~q ;}S~Ul11JP 'S;}1J~uno:> lIt! 'SU01S
It!u01~~u Jno '~SJ1d "S~ynS;}J ;}S;}~ SMOqS ~I"H ;}mS1d "Sll;}:>
Jo }{JOM ;}~ o~ ;}:>UlJlqW;}S;}J pooS SJ~;}q U01~~S;}JSS~ I;}A;}I
Jno 'SU01S;}J ;}~1Sodwo:> 9't s1q SU1sn (9661) AOU~WOpuqS
Mol /..J;}A ~~ Il;}M SpuOdS;}JJo:> osIt! ;}~~S;}JSS~ I;}A;}I-/..J~unO:>
-UlJ~sqns SJ;}}JW ~1 qSnO~It! '(1"0 » SS;}J~S J;}~~M Jo SI;}A;}I
-P1JS ;}q~ WOJJ J..lIt!1~u~~sqns SJ;}JJ1P U01~~S;}JSS~ u1s~q
;}q~ 'uJn~ uI "SU01~~ynwn:>:>~ P1JS JO u1s~q ;}~ WOJJ J..lIt!1~
UI "SS;}J~S JO SI;}A;}I (1"0 » MOl /..J;}A ~~ Su1~uno:>:>~ P;}S~q
J..lIt!1~U~~sqns SJ;}}J1P ~1 SWJ;}~ ';}~nlosq~ ~OU ~nq ';}A1~~I;}J
-JO:> pooS S1 ;}J;}q.t "SS;}J~S JO SI;}A;}I ;}~~J;}POW ~~ Sp1JS WOJ}
-S;} P;}S~q-P1JS pU~ ;}P1M-u1s~q ;}~ U;};}M:J;}q ;}:>U;}pUOdS;}J
(01 O~ S;}1J~UnO:>-Su01S;}J JOJ ZOI-!OI) s~1un SU1~uno:>:>~ JO
J;}qwnu SU1S~;}J:>U1 UlJ ~1M "SS;}J~S JO SI;}A;}I qS1q ~~ S;}~~W1~
"S;}SS~I:>;}~ Jo U01~~J~d;}s P;}S~;}J:>u1 UlJ ;};}S ;}M '(SU1s~q JOJ
;}:>u~~JodW1 ;}q~ U1 u01~:>np;}J ;}SJ~I ~ S1 ;}J;}~ 'IP.yn:>1~IP.d uI
-J..q ~u;}J~dd~ u~ ;}J~ q:>1qM 'S;}SS~I:> ;}~~1P;}WJ;}~U1 ;}~ Jo
"U01~~WJOJU1 Jo u01~~ynwn:>:>~ It!1~~ds ;}SIP.I ~ Jo ~:>npoJd
6
sr
SL
8V
v

520
.
CHAPTER E.6 .Case Studies
Fig. E.1S.
Assessments of water scarcity
as a function of accounting
unit. Here contrasting levels of
relative water demand (withdrawal/supply)
are used on the
horizontal axis. Levels beyond
0.4 indicate severe water stress.
All aggregations are made over
the global domain
Although we can demonstrate sensitivity to the choice
of accounting unit used, we cannot identify an optimal
unit. This awaits further study by the water sciences
community and a consideration of the broader issue of
data availability and harmonisation across biophysical
and socio-economic disciplines. The choice of accounting
unit is important from the standpoint of determining
the number of people at risk from water stress. In
comparison to the regional or country-level tabulation
which yields about 400 million under severe stress, the
gridded or basin-level interpretation gives a substantially
larger population, at about 2 billion. The conventional
wisdom thus severely underestimates the degree
to which the world's population is suffering from water
scarcity.
Vulnerability and Essential Biophysical Data
The combination of biophysical and socio-economic
datasets within a common geographic framework suggests
that the world is a drier place than previously
thought, and that population growth will result in a large
and growing portion of the world's population living
under conditions of water scarcity by 2025. The World
Bank has long sought to help countries harness their
water resources for economic development through its
support to irrigation, water supply, sanitation, flood control,
and hydropower. Water has been one of the most
important areas of World Bank investment: between 1985
and 1998, it lent more than $33 billion, or 14% of project
lending (World Bank 1998a).
But these projects have encountered serious implementation
problems because they do not address the
root causes of waste and mismanagement in the water
sector. Several problems emerged. First, fragmented public
investment programmes failed to optimise water allocation
and emphasized supply development over demand
management. Second, excessive reliance on overextended
government agencies resulted in poor fmancial
accountability and cost recovery. This led to a vicious
circle of poor quality, unreliable service, low will-
ingness to pay, inadequate operating funds, and further
deterioration in service. Third, damage to freshwater
ecosystems inherent in some projects was treated, if at
all, by limited corrective actions and rarely by substantive
adjustments to project design.
In response to these problems, the World Bank issued
a water sector strategy in 1993 (World Bank 1993).
The principal goal of the strategy was to improve water
sector performance by ensuring the provision of water
services in an economically viable and environmentally
sustainable manner. The strategy recognised that to improve
the performance of the water sector, it is necessary
to help countries reform their water management
institutions, policies, and planning systems. The strategy
is being implemented through a new generation of
water resources management projects. Noteworthy examples
exist for Morocco and Mexico (World Bank 1996,
1998b).
To meet increased demand at reasonable cost, policymakers
are exploring better ways to allocate existing
water supplies and encourage users to conserve water.
Better water management requires a significant improvement
in the collection and analysis of hydrological
data yet monitoring programmes in many countries
have deteriorated sharply in recent years (Vorosmarty
ans Sahagian 2000). Paradoxically, at a time of growing
water scarcity, we know less and less about the challenges
and opportunities we face.
The constraints on hydrological monitoring are both
financial and institutional. For example, both China and
India have dense monitoring networks and tremendous
analytical capacity but the effective use of these assets
is not possible for several reasons. First, very few stations
are linked by satellite or cell phone, resulting in
lengthy delays as data are retrieved and transcribed by
hand. Second, different agencies manage different
gauges, often over different time periods, resulting in
inconsistent data quality and fragmentary records.
Third, hydrological data are not shared among agencies,
let alone with the research community. Many data are
considered state secrets. Fourth, limited data access is
exacerbated by a move by many agencies toward full cost

ecovery. Finally, because of the focus on flooding, hydrological
models are often only calibrated using data
collected during the rainy season. More accurate predictions
would be possible if the models were calibrated
using data collected throughout the year.
Hydrological monitoring capacity in many countries
is deteriorating because of inadequate cost-recovery and
reduced government support. The number of stations
reporting to the Global Runoff Data Centre peaked in
the mid-1980s and then fell sharply (Fekete et al. 1999).
The decline has been most marked in Africa, where a
recent analysis shows that the number of gauging stations
in most countries is well below WMO guidelines
(Rodda 1998). Although several international initiatives
(e.g. W -HYCOS, GRDC) have sought to compile and publish
atmospheric, oceanic, cryospheric and hydrological
data, they ultimately depend on the quality of the
national networks. Even in the United States the hydrological
monitoring network has degraded. The USGS has
been monitoring river flow and water quality for over
100 years. The network coverage expanded throughout
the 1960s and 1970S but contracted sharply during the
1990S (United States Geological Survey 1999). More than
100 river gauges with long-term records, the single most
important class of stations for environmental monitoring
and design engineering, are being lost each year
(Lanfear and Hirsch 2000). As a result, the network has
become vulnerable as collaborating agencies have cut
costs to meet their own specific needs, for example by
only monitoring high or low flows, or by limiting public
access.
Many countries reject the principle of free data access.
Some industrialised countries have introduced legislation
that would restrict the open access that the re-
search community has traditionally enjoyed. In response,
the International Union of Geodesy and Geophysics, the
International Association of Hydrological Sciences, and
the World Meteorological Organization have all passed
resolutions calling for the free and unrestricted access to
hydrological data (compare Sect. C.3.4). In less developed
countries, data access is primarily limited by high prices,
which are justified on the basis of recovering operating
costs. As a result, fewer data are available to analysts and
policy-makers in developing countries to design and
implement the new policies they need to cope with water
scarcity.
The Promise of New Technologies
Ideally, a hydrological monitoring system will generate
a real-time, continuous stream of discharge measurements
from selected locations in the river basin. In practice,
this goal is not attained: stage, not discharge, is
measured (discharge is inferred from stage using a set
E.6.1 .Population and Climate 521
of empirical relationships, which must be recalibrated
on a regular basis), measurements are taken infrequently
and irregularly, and the results are made available days
or weeks after capture.
Advances in radar gauging, telemetry, the internet, and
modelling can significantly improve monitoring performance.
Satellite and cellular communications can
transmit data in real-time from the gauging station to
the internet. GIS-based hydrological models can integrate
hydrological, meteorological, and elevation data to predict
discharge and water quality across the river basin.
For remote and inaccessible areas, Vorosmarty et al.
(1999) propose the development of a hydrologically-oriented
satellite to monitor river stage, surface velocity, and
width. Using imaging radar and/or doppler lidar sensors,
the satellite would measure the surface height of inland
water bodies with an accuracy of 5-10 cm and river surface
velocity with an accuracy of 20 cm per second, with
a frequency of three to seven days. The data would be
made freely available in near real-time to counteract the
deterioration of terrestrial monitoring networks, data
commercialisation, and long delays in proc-essing and
distribution. But however advanced the technology. an
adequate ground network is still required to calibrate
these remotely sensed measurements.
Conclusions
The future management of water must emphasize demand
management, efficiency improvements and conservation.
This requires increasingly sophisticated information
on the functioning of the hydrological system
and how it is affected by water management decisions.
To this end, the Earth systems science community
can provide valuable insight into an important component
of the analysis, namely, a quantitative description
of the terrestrial water cycle. The analysis presented
here suggests that when socio-economic and biophysical
datasets are linked in a relatively high resolution setting,
we see an even more urgent picture of potential
vulnerability in world water resources than previously
acknowledged. The challenge will be to harmonise the
information needs of the two communities, that is of
the natural sciences and the social sciences. The Earth
systems modelling community provides high-resolution
datasets on water availability which transcend political
boundaries, a reflection of the integrative nature of the
water cycle. While our knowledge of water supply continues
to require improvement, our knowledge of population
distribution and water use statistics is arguably
much more primitive. In many countries, there has been
no census for over a decade and water use statistics for
several countries are many years old. Investment in hydrological
monitoring and analysis should be balanced

522
CHAPTER E.6 .Case Studies
with increased support for socio-economic data collection.
Socio-economists are well advised to adopt such a
"boundary-free" perspective by standardising, both intra
and internationally, the datasets necessary to quantify
the drivers of water withdrawal and consumption.
Finally, identification of water vulnerability will be based
on vigilance and scientific hydrology. With international
support, governments should seek to upgrade and extend
their hydrological monitoring systems using the
most cost-effective data collection, communication and
analysis technologies available.
Fig. E.16.
The Lake Erhai basin
Legend:
@ Town/City
.Scenic spot
/V River
~ Lake
~ Boundary
.Landuse
0 10 20km
Acknowledgments
The authors would like to thank Uwe Deichmann (World
Bank), who produced the population surfaces under contract
to CIESIN and advised on the extrapolation of the
data to 2025, and Greg Browder (World Bank), who provided
the World Bank water sector project data and
project papers. The authors would also like to acknowledge
the contributions made by Richard Lammers (UNH)
and Jose Simas, Danny Gunaratnam, and Nagaraja Rao
TheLMe--
EI1I8J Balin
N
W+E
S

Harshadeep (World Bank). The US Agency for International
Development and the Dutch Ministry of Foreign
Affairs provided funding for this paper.
Brad Bass Lei Liu .Gordon H. Huang
E.6.2 Water Resources in the Lake Erhai Basin, China
Introduction
One of the most critical concerns facing both the national
and local governments in China is the environmental
degradation associated with rapid economic
development. Chinese governments have been in the
process of designing region-specific environmental
regulations, but the decision-making process requires a
sound understanding of the significant drivers of regional
environmental problems and the effects of policy
changes on different sectors and decision variables. One
sector that has drawn considerable interest is water resources;
in particular, the impact of economic development
on water resources and how these impacts are
modified by other factors such as climatic variation and
change.
The Lake Erhai basin provides an illustration of the
integrated impacts of several factors, including weather,
on water resources. The basin is host to agricultural and
industrial production, a net-cage fishery, forestry and
tourism, but its water resources are also required for
municipal water and navigation. The rapid economic
development and population growth of recent years has
been accompanied by increased amounts of water pollution
due to industrial wastewater emission, eutrophication
in Lake Erhai due to nutrients from runoff and
the fishery, and soil erosion due to deforestation. In order
to address these concerns, a research project entitled
"Integrated Environmental Planning for Sustainable
Development in the Lake Erhai basin" was undertaken
between 1996-1998 with support from the United Nations
Environmental Programme.
The Lake Erhai basin is located at 25°25' -26°10' N
and 99°32' -100°27' E and covers an area of 2565 km2
within the jurisdiction of Yunan Bai National Autonomous
Prefecture, China (Fig. E.16). Lake Erhai has an
approximate area of 250 km2, an average depth of 10.2 m
and a storage volume of 28.2 x 108 m3. There are 117 rivers
and streams that drain into Lake Erhai, mostly from
the north through several smaller lakes and which drain
into the Mizu, Luoshijiang and the Yunganjiang Rivers
and from the east. The Xier River is the only river flowing
out of the lake and is the river most affected by industrial
discharge. The basin also includes mountains,
hills and alluvial plains. Land use includes forest (44.7%),
grassland (20.3%), farmland (14.5%), human habitat
(2.7%), abandoned land (1.7%), transportation (0.5%),
and water bodies occupy 9.5% of the basin.
E.6.2 .Water Resources in the lake Erhai Basin, China 523
The vulnerability assessment diagram has been
modified for this region and expanded to eight subsystems
-population, agriculture, industry, tourism, water
resources, pollution control, water quality and forestryand
their interrelations (Fig. E.l]). For example, industrial
development brings economic benefits but also consumes
raw materials from forestry and agriculture and
discharges wastewater into the Xier River. A system dynamics
(SD) approach was used to model the interactions
between the various components in the Lake Erhai
basin. The relationships were quantified using Professional
Dynamo Plus.
The governments that are involved want to consider
how limits to economic development would impact on
water quality and quantity, the sensitivity of these impacts
to variations in weather and then determine an ap-
pr

~
524
..
CHAPTER E.6 .Case Studies
""..
I do~estic sewage I
"
"
"""
"", [;;;~~J
wastewater
r pollution 1,,""" treatment
L subsys~m J
""
""",."..", COD discharge I
,~
I agricultt1ral outputs I
,.' per capita
~;;;JriCulture , income
subsystem ..
".
". """"" ". '" [ ~;~;~;~~~~J
water consumption
nutrient concentration
of agricultural runoff /
y
Fig. E. 17. Interactive relationships among different subsystems
sensitive to several population parameters. To assess the
seriousness of the problems with ICODT, planners,
policy-makers and industrial stakeholders must agree on
an acceptable uncertainty for this variable. If the model
uncertainty is much higher than the acceptable uncertainty,
then this decision variable should only be used in
a restricted manner (ICODT increases or decreases relative
to the previous time step) or not be used at all.
Vulnerability Assessment
The ErhaiSD model was run for a period of 15 years with
1996 identified as the year for generating alternative sce-
",,"'1~ater consumption
\
" '"
~
\
"
tourism
subsystem
\
" \
industrial point
sources
!
!
agricultural nonpoint
sources
f ,. I
water nonpoint
-""'"
[~~~~~
water quality
subsystem
domestic nonpoint
sources
sources
frommountains
narios. The simulation exercise included a base run (BR)
and three alternative scenarios (~I-A3), provided by
local authorities based on a previou~ planning study (Wu
et al. 1997), balancing economic and environmental objectives
(1), emphasising industrial growth (2) and emphasising
increasing water pollution control. The results
are presented for gross industrial output (PlOT), industrial
wastewater generation (ISWT), industrial COD
emission and water pollution index (Fig. E.18a-d). The
model suggests that water pollution will be reduced in
all three alternatives. Alternative 2 will result in marginally
higher amounts of economic growth than Alternative
1, yet with much larger amounts of industrial discharge.
The COD concentrations could be reduced by

'b
§;
I- 0a:
-;;;-
Q)
c
c
0
"b ~
fa
8
350
300
250
200
150
100
50
0
60
50
40
30
20
10
0
1996 2000 2004 2008 2010 1996 2000 2004 2008 2010
1996 2000 2004 2008 2010 1996 2000 2004 2008 2010
Fig. E.18. Simulation results of the ErhaiSD model with a base run (BR) and three alternative scenarios (AI-Aj). a Gross industrial output
values; b total industrial wastewater generation; c total industrial COD emissions; d water pollution index
50% through the development of regional wastewater creased surface runoff, soil eroision and in-lake turbu-
treatment systems; the best improvement occurs in Allence. In the dry scenario the N and P loadings are much
ternative 3 due to the lower amounts of COD discharge. lower. However, eutrophication could still be a problem
Alternatives 1 and 3 produce substantially lower due to intense sunlight and higher temperatures as oc-
amounts of industrial discharge, but control of this varicurred in August and September of 1996, for example,
able is not sufficient as all of the alternatives contribute when the algae count and the COD reached their high-
a similar amount of total dissolved N and P. Although est levels in ten years, most likely due to strong thermal
Alternative 3 emphasizes agriculture, the amounts of total
dissolved Nand P in the back-flowing irrigation water
(NAWUR and PAWUR) were only marginally higher
stratification and weak dispersion.
than Alternatives 1 and 2. Conclusion
Vulnerability to Weather
The Lake Erhai basin's water supplies are vulnerable to a
wide range of factors including the weather. The ErhaiSD
model suggests that choosing an alternative management
In addition to the direct discharge of industrial waste procedure, which balanced industrial growth with envi-
and nutrients, water quality is closely related to precipironmental concerns or one that favours agriculture, could
tation and runoff. To assess the vulnerability of the ba- reduce industrial discharge. This is particularly imporsin
to seasonal weather variability -and perhaps climate tant under a scenario of reduced precipitation, as the COD
change -three precipitation scenarios were considered, levels are inversely proportional to water quantity. All of
corresponding to dry, average and wet seasons in the ~ the planning alternatives were similar in terms of total
region. Table E.3a-c present the impacts of the three pre- dissolved Nand P, but this could be reduced through incipitation
scenarios on the Xier River, for the three plancreasing the forest coverage. The model suggested that
ning alternatives, before and after the installation of N and P levels were proportional to water quantity, but it
wastewater treatment. The COD concentration is high- did not reflect the potential for eutrophication under a
est under the dry scenario, even with treatment facili- scenario of reduced precipitation.
ties.
These results are problematic for planners in that any
An additional simulation was conducted using the precipitation scenario that deviates from average condi-
three precipitation scenarios under the assumption of tions will worsen water quality. The industrial discharge
increasing forest coverage. The results are presented for would have to be controlled through additional waste-
Alternative 1 only, as the impacts were similar across all water treatment and/or reduced industrial growth. If the
three alternative futures (Table E4a-c). Under the wet future weather were to favour a drier precipitation re-
scenario, the Nand P loadings are increased due to in- gime, the region's development plan could be altered to
w Q)
c
c
.9
~ ~
~
40
35
30
25
20
15
10
5
0
525

526
CHAPTER E.6 .Case Studies
Table E.3. COD (Chemical Oxygen Demand) concentration in the Xier River before and after wastewater treatment processes
a Averageseason
1998
2001
2004
2007
2010
b Dry season
1998
2001
2004
2007
2010
c Wetseason
1998
2001
2004
2007
2010
Alternative 1 Alternative 2 Alternative 3
before after before/after before after before/after before after before/after
73.02
76.24
81
87.66
91.89
97.16
101.59
108.04
117.18
123.09
61.68
64.39
68.37
73.94
77.43
Table E.4.
Simulation results for Alternative
1 for normal, dry and wet
season
63.14
59.17
56.9
56.21
55.1
83.63
78.16
74.91
73.80
72.25
53.49
50.26
48.44
47.80
47.08
1.2
1.29
1.42
1.56
1.67
1.16
1.30
1.44
1.59
1.70
1.15
1.28
1.41
1.54
1.64
a Normal season
Forest coverage (FCR) (%)
91.76
101.36
104.42
109.53
112.08
122.85
136.14
140.30
147.41
151.03
77.19
85.15
87.73
91.98
94.07
N concentration in Lake Erhai (mgl-"
P concentration in Lake Erhai (mg I-"
b Dry season
Forest coverage (FCR) (%)
N concentration in Lake Erhai (mg I-"
P concentration in Lake Erhai (mg I-"
c Wet season
Forest coverage (FCR) (%)
N concentration in Lake Erhai (mg I-"
P concentration in Lake Erhai (mg I-"
allow for more industrial growth, although it should be
noted that industry requires significant amounts of water
that might not be readily available. However, the
amount of industrial discharge is the one variable that is
highly uncertain in ErhaiSD. Without knowing the acceptable
level of uncertainty, it is difficult to use the model
as the sole basis for controlling industrial growth. Because
of the high degree of uncertainty there are risks in
accepting the model as a basis for controlling industrial
discharge. This case study did not explicitly account for
climatic uncertainties in the decision, an analysis which
is presented in Bass et al. (1997>.
78.85
77.46
70.2
66.24
60.94
105.19
103.32
93.24
87.66
80.33
66.50
65.38
59.43
56.26
51.89
1998
44.7
0.31
0.0272
44.7
0.28
0.026
44.7
0.32
0.0279
1.16
1.31
1.49
1.65
1.84
1.17
132
1.51
1.68
1.88
1.16
1.30
1.48
1.641.81
73.23
74.8
76.04
79.12
79.77
97.53
99.72
101.31
105.48
106.40
61.83
63.15
64.24
66.87
67.41
61.72
54.06
47.21
43.62
38.56
81.77
71.22
61.67
56.49
49.44
52.3
46.00
40.41
37.57
33.43
1.19
1.38
1.61
1.81
2.07
1.19
1.40
1.64
1.87
2.15
1.18
137
1.59
1.78
2.02
2001 2004 2007 2010
50.5
0.30
0.0265
50.5
0.27
0.0255
50.5
0.31
0.0272
Changming Liu
56.2
0.28
0.0254
56.2
0.26
0.0245
56.2
0.29
0.026
E.6.3 Yellow River: Recent Trends
62.0
0.27
0.0251
62.0
0.24
0.0244
62.0
0.28
0.0256
Physical and Hydrological Characteristics
67.7
0.25
0.0241
67.8
0.23
0.0235
67.7
0.26
0.0245
The Yellow River, or Huang He, the second largest river
in China, is known for its high silt content. It originates
from the Yueguzonglie basin, being about 4700 m in elevation
on the north side of the Bayankala Mountain in
the Qinghai-Tibet Plateau. It flows across nine provinces
from west to east and, finally, it enters the Bo Sea (part of

(O!HSUllnH) lO!A!~ MolIO!X 0!1fI
Jo q:>IIO!l O!!PP!W pUll lO!ddfl
'6L '3 '6!~
11 '~:JUIAOJd ~UOpUlJqS 'i.1uno:) IYU~)l1lJ (lJ~S Mon~A ~q1
~~lJUreJP lJ q1IM 'N o'tv 01 't£ pUll H 0611 0196 u~~M1~q s~Il
'W)f v9v ~ JO ~U~1 ~SJno:J urew lJ pUll zW)f 000 't~L JO lJ~Jt!
nIl) PIJlJ-IW~S JO PIJlJ SI lJ~Jt! J~AI~ Mon~A ~q1 JO 1S0W
S1unO:J:JlJ UISlJq ~~lJUIlJJP S1I Jo lJ~JlJ ~qJ. '(6861 ~UlJIl SOlI
'}1!101}1!uoI1lJU ~q1 Jo %'t AluO
dn s~)[I!w JJounJ ~q1 ~nqM lJ~Jt! }1!UOI1lJU ~q1 Jo %8 JoJ
u!seq JaA!~ MOna}.. atlt Jo e~ep :!!~s!Ja~:!eJeq:! J.peaJ. aq.L 'S'3 a)qel
111.01
~:JJnos J~1lJM 1UlJ1JodwI 1S0W ~q1 sI J~AI~ Mon~A ~qJ.
J~1lJM ~q1 SlJ S~AJ~S 11 'lJUIq:) q1JOU PUlJ 1S~M-q1JOU Jo
'PUlJldOJ:J lJqW 91 'UOIillW vtI JO UOI1lJyndod lJ JOJ Alddns
UlJnJ.~uoqZ ~q1 PUlJ 'IXUlJlJqS 'rxulJqS 'lJII0~UOW J~UU1
JO ~SlJq A~J~U~ ~q1 SlJ PUlJ 'S~I1I:J P~ZIS ~IPPIW JO ~~JlJl o~
JO WlJ~J1S ~IPPIW PUlJ J~ddn ~qJ. 'spl~Y lIO Il~u~qS PUlJ
q:JIqM iW)f 000 o£L JO lJ~JlJ UlJ SJ~AO:J J~AI~ Mon~A ~q1
Il PUlJ nIl) lJ~JlJ ~~lJUIlJJP }1!101 S1I JO %L6 JOJ S1unO:J:JlJ
q~Iq PUlJ SUIlJ1UnOW JO S1SISUO:J lJ~JlJ SIq1 JO 1S0W ,( 000't
JO q1~U~llJ q1IM 1lJY ~1Inb SI WlJ~J1S J~MOI ~qJ. 'SnlJ~1lJld
'UIlJld lJUIq:) q1JON ~q1 SSOJ:JlJ ~UIMOY W)f 08L
q1IM J~AI~ Mon~A ~q1 JO M~IAJ~AO UlJ S~AI~ 61'H ~Jn~Id
'q:JlJ~J J~MOI pUll ~IPPIW 'J~ddn ,~,~ 'SUISlJq-qns ~~Jq1 S1I
SI J~AI~ Mon~A ~q1 JO JJOUnJ ~~lJJ~AlJ IlJnUUlJ ~qJ.
~q1 JO JJOUnJ }1!101 ~q1 JO %9'~~ JOJ S1unO:J:JlJ noqzUlJl
~AOqlJ lJ~JlJ ~qJ. 'lJ1lJp JO SJlJ~A 9~ UO p~SlJq EW uoInIq O'8~
uI AplJ~J~ S~~UlJq:J JJounJ ~qJ. ,( 000't Il PUlJ nIl) J~AIJ
'JlJ~AlJ UIq1IMAlu~A~un p~1nqIJ1SIP sI PUlJ SJt!~A1U~J~JJIP
-lJ~S AUIlJJ ~q1 WOJJ s1yns~J JJounJ }1!101 ~q1 Jo %OL 1noqv
SI J~AI~ MOn~A ~q1 JO :JI1SIJ~1:JlJJlJq:J 1U~UIWOJd ~qJ.
'(J~qw~1d~S 01 ~unf WOJJ) uos
-UO:J 11Is ~qJ. '"plJOI1U~WIP~s UI q:JIJ PUlJ J~1lJM JO 1JOqS"
4JeaJ Jaddn
4JeaJ aIPP!W
4JeaJ UMOa
9'8£
v'v£
l'l
l'SL
9.lLv£
9.S9L
v.9all
9.£9vS
9'~L£
9'9V~
O'L~
~'O9S
S,l!U!qJ SUOWl! :lSJ9 alp S){Ul!J JaA!~ MOna}.. alp JO :lua:l
:luawwas renUUl! aSl!JaAl! aqJ. i 6861 S~!l p~ n!l) SJaA!J
-~S alp SU!){l!:I) SUO:l UO!mq 9'1 S! (S:'a aIql!J.) aSxeq:>s!P
-uo:> :Ins aSl!JaAl! alp '(aA!:ll!:luasaJdaJ Sl! UO!:ll!:lS l!!XUaw
S! aSJl!q:>s!P :luaw!pas wnw!Xl!W aq:l £_w S){ S:£ S! :lua:l
:Ins wnw!Xl!w alp pUll (££61) Jl!a1.. Jad SUO:l UO!mq 't6'£
aSJl!q:>s!P :luaw!pas aqJ. i£-w S){ 0't6 paq:>l!aJ :lua:lUo:>
%06 sap!AoJd (JaqO:l:>O 0:1 1..IfiO UOSl!as poo{J alp SU!mp
:luawwas Jo UO!:ll!m:>n{J AJxea1.. aqJ. lunOWl! renu~ alp JO
qS!q l! U! aSJl!q:>s!P :luawwas aqJ. 'qS!q a:l!nb S! aSJl!q:>s!P
wnw!Xl!w aqJ. 'aSl!JaAl! aq:l a:>!Nl:I S! Jl!a1.. Uo!:ll!:I!d!:>aJd
U! aSJl!q:>s!P :luaw!pas aq:l 'aIdWl!xa JOd 'sanreA JaqS!q
uaAa q:>l!aJ ~:> O!:ll!J aSJl!q:>s!P :luaw!pas wnW!U!w p~
:luaw!pas pa:ll!J:lua:>uo:>Jad1..q aqJ. '(I-J1..:I UO!n!q 817'0)
8't61 Jo :ll!1p saw!:I 't'8 SlIM q:>!qM 'SUO:l UO!mq 't6'£ SlIM ££61
-uaWUOJ!AUa WOJJ pa:lIfisaJ Sl!q JaA!~ MOna}.. aq:l U! Pl!°I
'(6861 S~!l pUll n!l ~1861 nn) nl!a:l
-l!Id ssaol alp U! UO!SOJa nos aJaAaS Sl! q:>ns swaIqoJd re:l
SJOJD.J!PU/ AJ!/Dno JaJDM U! a6uD4.J
L1:~ spuaJl ~ua)aM :JaA!M MoilaA .£'9'3
-qsnu I..q JI;)S~! ;)SU~;)I:> o~ J;)A!J ;)~ JO I..~!:>~d~:> ;)~ s;):>np
-;)J q:>!qM ~U!S~;)J:>;)P S! J;)A!~ MOn;)A ;)~ JO JJOUnJ ;)q.L
MOn;)A ;)q~ JO IOJ~uo:> uo!~nnod ';)W!~ ;)W\!S ;)q~ ~V .~U!
JOJ SM~I ;)~;)Idwo:> ou ;)J~ ;)J;)~ ;):>U!S 'P;)~~~I s~q J;)A!~
~U;)W;)~~U~W p;)~!Un ~ pU~ UO!~:>;)~oJd ;):>JnOS;)J J;)~~M
-U! ;)~J~q:>S!P ;)~~M;)S pU~ ;)Sn J;)~~M ;)q~ 'WS!U~q:>;)W
J~;)I.. P~oI uo!~nnod P;)S~;)J:>U! U! P;)~ynS;)J ;)A~q S;)S~;)J:>
6'vl
0
9L
lv'l
9'LOV
l'9V~
£'SL9
6'SLv
-.
ZZ'£
ZZ'6
9v'vl
ZZ'6

Acknowlegement
This
528
CHAPTER E.6 .Case Studies
Table E.6. Statistical data of sediment yields after Chen et at. (1998) (empty spaces in the table are unmeasured values)
by year. The pollution of the Yellow River has become
worse which has been attributed to different reasons
with mining one of the main sources (Ongley 2000).
In Summer 1998, a water quality assessment was
performed, based on monitoring data of 69 segments
(26 of mainstream, 43 of tributaries) with a total length
of 7158 km in the Yellow River basin. The result illustrates
the water problem of the Yellow River, i.e. refJects
the serious degradation of the extensively used waters:
grade! I-III made up only 26.1%, grade IV made up
38.6%, grade V made up 13.0% of the total length. The
most seriously polluted water is unsuitable for any use,
and does not even get a grade as it is inferior to grade V.
22.3% of the total length belonged to this category, that
is below grade V. This means of the total length of the
river with polluted water, i.e. belonging to grade IV, V
and even worse than V, was 73.9% of the total length
assessed in June to September 1998. Comparing the water
quality of 1998 with that of 1985, the length of river with
water quality worse than grade III has increased by 58.8%.
The long periods of low flow and drying-up in the
lower reach of the Yellow River have led the wetland areas
in the delta to decrease. The groundwater table has
declined without enough fresh water recharge and the
salinisation area has increased. The water environment
of the wetland has changed and threatened the life of
hundreds of wild plants and 180 kinds of birds. It is damaging
the biological diversity of the delta.
! In China, the water quality is classified in five grades, in descend-
ing order from the highest quality (I) to the poorest one (V).
chapter was supported by the Chinese Key Proj ect:G19990436-01
Roland E. Schulze
E.6.4 Examples of Hazard Determination and Risk
Mitigation from South Africa
The Approach, the Hydrological Model and the Spatial
Databases Used
This section illustrates some of the concepts of risk, hazard
and vulnerability within a context of hydrological
risk management as described in Chapt. E.5, using examples
from South Africa, defined here as the contiguous
area of 1267 681 km2 made up of the nine provinces
of the Republic of South Africa plus the landlocked Kingdoms
of Lesotho and Swaziland. The country/regional
scale is chosen to identify regions of similar hydrological
hazard levels and thereby to distinguish between
areas of higher and lower potential hydrological risk.
Such a comparative view is important from the perspective
that risk management is a national/regional responsibility
and that this analysis may assist in identifying
target areas for potential priority attention. Many of the
hazard/risk indices presented by way of maps are in the
form of ratios, rather than as absolute values, to highlight
sensitive areas on a relative scale.

s
A regional analysis requires a hydrological model to
be operated with suitable hydrological spatial databases.
The hydrological model selected was the ACRU agrohydrological
modelling system (Schulze 1995, and see
Sect. D.2.6.6.2), developed in South Africa for catchment
scale simulation of runoff components, sediment yield,
reservoir yield, irrigation supply/demand and crop yield
analysis, and containing routines for the assessment of
land use/management as well as climate impacts on hydrology.
ACRU is a widely verified physical-conceptual
and daily time step simulator operating on a multi-layer
soil water budget described in detail in Schulze (1995).
The ACRU model has been widely verified at catchment
scale, both in South Africa under a range of hydro climatic
and physiographic conditions (e.g. Schulze 1995) and elsewhere
(e.g. in the USA, in Germany and Zimbabwe).
For the production of South Africa scale maps shown
in this chapter, quality controlled daily rainfall for the
concurrent 44-year period 1950-1993 was input for each
of the 1946 Quaternary catchments which have been
delineated for South Africa, Lesotho and Swaziland. For
each Quaternary catchment other atmospheric parameters
(e.g. daily maximum and minimum temperatures;
A-pan equivalent reference potential evaporation) were
also input, as were hydrological soil parameters. For purposes
of producing comparative hydrological hazard
maps, land cover was assumed to be grassland in fair
hydrological condition (i.e. 50-75% cover).
of General Hydrological Hazard Indices
To set the scene, Fig. E.20 (top) shows mean annual
precipitation (mm), which characterises the long-term
quantity of available water to a region, to display a general
westward decrease, with relatively low mean annual
precipitation -a first indication of a largely semi-arid
climate and potentially high risk natural environment.
A simple aridity index expressed as the ratio of mean
annual potential evaporation to mean annual precipitation
(Fig. E.20, middle) emphasizes the hazard of hydrological
semi-aridity, because it amplifies the effects
of a low mean annual precipitation when that is evaluated
in association with the region's high atmospheric
demand. The aridity index is an already high 2-3 where
mean annual precipitation is > 600 mm, increasing to
> 10 and even> 20 in the west. A consequence largely of
the high aridity index is that the conversion ratios of
rainfall to runoff over most of South Africa are exceptionally
low (Fig. E.20, bottom) with, overall, only 9% of
rainfall manifesting itself as runoff. These low runoff
ratios, simulated with the ACRU model (Schulze 1995),
result in a high hydrological vulnerability over much of
the region.
E.6.4 .Examples of Hazard Determination and Risk Mitigation from South Africa 529
Example of a "Deprivation" Event, Hydrological
Uncertainty and of Variable Sensitivity:
The 1982/3 EI Nino Event over South Africa
The 1982/1983 EI Nino was one of the most severe experienced
over South Africa. The manner in which such
an event impacts on different hydrological responses is
illustrated in Fig. E.21. Observed rainfall and simulated
runoff and recharge into the groundwater zone through
the soil profile are all expressed as ratios of their respective
long-term (1950-1993) median values. For much of
the region the EI Nino season's rainfall was 60-75% of
the median, however, with sizeable areas receiving within
the range of expected rainfalls (i.e. 75-125%) while some
others received only 20-60% of the norm (Fig. E.21, top).
The corresponding runoff responses display much more
complex patterns spatially and in the range of ratios.
Much of the region yielded only 20-60% of the longterm
runoff (Fig. E.21, middle), with considerable areas
generating < 20% of the expected runoffs. This shows
clearly once more the intensifying effects of the hydrological
cycle on rainfall perturbations, as well as the dependence
of hydrological responses not only on total
rainfall amounts, but also on individual events, rainfall
sequences and antecedent catchment wetness conditions,
i.e. on the hydrological uncertainty created by meteorological
and catchment conditions.
Some hydrological processes and responses display
higher sensitivities, and thus higher potential vulnerabilities,
than others. This is illustrated by the recharge
to groundwater during this EI Nino event, which was
impacted even more severely than runoff (Fig. E.21, bottom).
Generally only 0-20% of the expected recharge
was simulated to take place, the reason being that a
higher threshold has to be reached for recharge to commence
than for stormflow to start occurring.
Example of an "Assault" Event as an Index of
Potential Vulnerability and Stochasticity by
Quantitatively Defined Endpoints in Regard to
Depth, Duration, Frequency and Area Affected
Episodic flood generating events display considerable
stochasticity (i.e. unknowable randomness). As an index
of potential vulnerability, the flood hazard example
presented below as an "assault" event illustrates the relative
spatial differences over South Africa between the
severity of the 1: 50 year I-day flood-producing rainfall,
and consequent runoff, compared with what could be
considered the annual expected I-day values, i.e. the 1:2
year event. Ratios of 1: 50 to 1 : 2 year rainfalls are generally
between 2 and 4 (Fig. E.22, top), with lower ratios
~

of~
sa!pms as e) 0 903 ~3idVH)
-!AU3 {U:>!8010lp.l..q P!JU-!W3S .1..1
-38ml S,[!:>!l}V rpnos }O S3:>!PUI
'OZ'3"6!:!
(alPP!IU) '(ww) UO!1U1!d!:>3Jd
{Unuuu UU3W (dol) :1U3UJUOJ
UO!SJ3AUO:> 3rp (IUOJJoq) puu
UOI1U1IdI:>3Jd 01 UOI1UJOdU3
{U!1U310d ienuuu UU3W }O 0!1UJ
3rp SU p3SS3Jdx3 X3PU! A:I!P!m
(uL661 ~JnrpS WOO}) I1t!JU!UJ
01 JjOUnJ {Unuuu UU3W }O 0!1UJ
I
'
I
100"&1 < .
00"&1
0001 -00' .
I: -r _
F -C -i
LJ
WW
008 -00' 0
008 -OO~ ~
OO~ -oor
00"& -001
001 >
-0001 .
~ > .
o~ < 8
o~ -91 8
g/. -01.
O/. -9 0
(t66 t) eJ:JeUf1
: UO/l8003 3d
-t
OllBb'
9

'CfS
"ZZ"3"6!.i
sa!pms as e) , 9'3 113ldVH)
(IOO'l ;lzynq;)S WO.lJ)
5u!pooy Ot AtmqB.I;lUynA {B!t
-u;ltod Jo S;I;)!PU! SB (WOJJoq)
sJJoun.l pUB (t/OJ) SJlBJU!B.I
ABp-I .lB;lA-'l :.lB;lA oS JO SOnB~
-W3q~ ~S3J!U11W '13A3Moq 'SO!~1111{1!JU!1!1 3S3q.L '~S3M 13!lp
3tp JO S~l11d U! 01 O~ SU!S1131:JU! ~nq 'S113111 {1!1~U3:J 13AO
3q~ U! 13qS!q S3W!~ 01-t Sq~d3p pOOlJ J..11p-l S11 S3A13S
PU11 S3W!~ O~ O~ dn pUt! '11:J!lJV tpnos JO S113111 U13~S113
13AO 13qS!q
"l'l'a 'S!d) ~S3M 13!lP 3tp JO S~:J111~ ~U11:JY!US!S
3:JU311n:J:J0 ~u3nb31JU! Ut! 3q J..11W SpOOlJ 31!1{M '(monoq
3A11q spOOlJ 31111 ~11tp S3~111~Snn! 31dw11x3 S!tp '~S3M 3tp U!
'S3pm!US11W 3A!~11131 qS!q J..IP3~:J3dx3
-un 1!3q~ JO 3sn11:J3q 3S11W11P 313A3S Op O~ {1!!~U3~od 3q~
SJaSDJDQ JJoLJ5 Jo asn LJ6noJLJJ AJU!DJJa.JUn JO a/dWDX3
~q~nl!JJ J..I~u3nb3JJ S! UO!~l!U!WJ3~3p pJl!Zl!q re:)!~S!~l!~S
-l!~l!p ~Joqs ~U!Sn JO 3:)U3nb3SUO:) l! Sl! S3!~U!l!~J3:)Un 1:fI!M.
-3P JO S3nreA reAJ3~U! 3:)U3JJn:)3J q~!q 3U!WJ3~3p O~ S~3S
re:)!~OIOJPJ..q JoJ S!~ 3~l!J~Snm 0.1 °JjOUnJ JO lreJU!l!J U~!S
JjOUnJ PUl! IJ1!JU!l!J Al!p- 1 Jl!3A O~ : 1 3q~ 's3sodJnd US!S3P
P3uold S1!M. £661-'tL61 pO!J3d J1!3A-'t't ~oqs l! JOJ p3~l!W!~S3
-!~S3 JjounJ pOl! ll1!JureJ A1!p-1 J1!3A O~: 1 ~SureS1! O!~l!J l! S1!
uI '£661-0~61 SJ1!3A tt 3~ '3'! 'p0!J3d 3~ 3Iqnop JoJ p3~l!W
Sl!M UO!~nq!J~S!P 3nJ1!A 3W3JtX3 J1!WJOU SoI3~ 3Sl!:> q:>l!3
~Joqs 3q~ JI 13poW flllJV 3q~ q~!M p3~l!J3U3S JJOUnJ
JO 3Sl!:> 3~ U! 'S3!J3S wnw!Jrew J1!nUUl! 3q~ O~ P3!Iddl!
3q pynOM O!~l!J 3q~ 'S3!J3S WnW!Xl!W Il!nuul! 3q~ JO
uo!~l!yndod p3~:>3dx3 3q~ JO 3A!~l!~U3S3Jd3J 3J3M SpJO:>3J
'I pUnOJl!
3SJl!l'3Sl!:> 3q~ 3q O~ ~OU AIJl!3I:> S!q~ SMOqS £'t'a 3JnS!d
~L'o U33~3q SO!~l!J IJ1!JU!l!J Al!Ids!p l!:>!JJV ~nos Jo S~:>l!J~
SO!~l!J MOqS Sl!3JP. J3~0 3I!qM '~L'o > U3A3 PUl! ~6'o pOl!
-U!l!J US!S3P SU!~l!W!~Sa i O~'I U3A3 pOl!) ~'t'1 JO SS3:>X3 U!
3J3A3S U! ~ynS3J sn~ Al!W sq~Su3I pJO:>3J ~Joqs WOJJ sIJ1!J
-U!l!J 3q~ WOJJ p3~l!W!~S3 S! JJOUnJ US!S3P 3:>UO P3Y!Idwl!
SJOJJ3 3S3q~ ~!M 'SUO!~l!W!~S3J3AO JO SUO!~l!W!~S3J3pUn

Fig. E.23.
Ratios of "short" (1972-1993)
to "longer" (1950-1993) design
rainfall and runoff in South
Africa for the 50-year return
period I-day event (from
Schulze 2001)
fall (Fig. E.23, bottom). The importance of record length
can therefore not be overemphasized, particularly in
light of the worldwide trend, certainly in developing
countries and evident also in South Africa, of declining
hydrometeorological recording networks.
Example of Secondary Hazard Modification Through
Land-use Practices: The Case of Grazing Management
Land-use practices have already been shown to playa
significant role in long-term average hydrological responses
(cf. Mgeni Case Study, Chapt. D.n, but perhaps
even more dramatically so at the extremities of frequency
distributions (e.g. Schulze 1989, 2000 ).An example of secondary
hazard modification by manipulating land-management
practices is given below.
E.6.4 .Examples of Hazard Determination and Risk Mitigation from South Africa 533
Much of South Africa's natural grassland (veld) has
recently been shown to be heavily over-utilised (Hoffman
et al.1999). Hydrologically, the degradation through
overgrazing of veld from good to poor condition, with
its reduction in vegetal cover from> 75% to < 50%,
implies enhanced stormflows through reduced interception,
evaporation and transpiration potentials as
well as infiltrability, shortened catchment lag times
which increased peak discharges and greater exposure
to soil erodibility through removal of mulch and shorter
drop fall heights. These variables were changed for each
of the 1946 Quaternary catchments covering South Africa
in ACRU model simulations of stormflows, peak
discharges and sediment yield to reflect veld in good v.
poor management condition. Figure E.24 (top) shows
that annual stormflows from veld in degraded condition
are generally 1.5-2.5 times as high as those from

534
CHAPTER E.6 .Case Studies
Fig. E.24.
Ratios of annual stormflows
(top) and sediment yields (bottom)
in South Africa from veld
in poor v. good hydrological
condition (from Schulze 2001)
Fig. E.25.
Example of simple benefit
analysis of seasonal runoff
forecasts over South Africa
(after Schulze et al. 1998)
~Ro/n~.~
Forecasted c/ima!9
."WIn" f8sulrs in situation
.No dffef9ncs
I ."lose" situation I
0 No data

Fig. E.26.
Schematic illustration of the
reduction of uncertainty in a
reservoir operation through
application of forecasting
techniques
535
With forecast! Uncertainty
Without forecast: Reduction
veld under good management. When converted to sedi- wrong, i.e. 1981/1982, 1987/1988 and 1990/1991, most of
ment yields, however, the factor difference becomes southern Africa scores more "wins" than "losses". The
2.5-7.5 times, and even> 7.5 times (Fig. E.24, bottom), impacts of short term climate perturbations such as the
clearly illustrating how a hazard, in this case stormflow EI Nino phenomenon have already been illustrated. This
and especially sediment yield, can be modified positively forecast analysis indicates that even at the current level
by good grazing management and/or rehabilitation of of seasonal forecast accuracy (around 62% if the 3 worst
overgrazed lands.
forecasts in 15 are omitted) these can potentially be
"translated" into an operational tool for water resources
managers which could prove statistically more accurate
Example of Vulnerability Modification through
than the current practice of forecasting based on his-
Seasonal Forecasting of Runoff
Vulnerability modification is a form of risk mitigation
which includes, inter alia, assessing the benefits of foretorical
expected, i.e. median, runoffs with wide uncertainty
bands, as shown in Fig. E.26.
casting streamflows for the rainy season ahead. Statisti- Are Certain Areas in South Africa Hydrologically
cally derived categorical seasonal rainfall forecasts four More Sensitive than Others to the Individual Forcing
months ahead are made for eight regions of South Africa
by the South African Weather Service, for three cat-
Variables of Climate Change?
egories, i.e. "above average", "near average" and "below Figure E.2? illustrates the relative sensitivities on mean
average" seasonal rainfalls. If seasonal rainfall forecasts annual runoff of dT (assumed to be a uniform increase
were a random process, such three-category forecasts of 2 °C over southern Africa) and M (changed through
would be correct 33% of the time. If seasonal categori- -10% to +10% of the present). In each case the other
cal rainfall forecasts are "translated" into seasonal run- two variables are held constant at present levels when
off forecasts, these could become very valuable reser- running the daily ACRU model. The hydrological sysvoir
operations and irrigation application planning tools tem is relatively insensitive to temperature changes that
for water resources managers. Seasonal categorical rain- affect evaporation and hence runoff. The increase of 2 °C
fall forecasts for the eight forecast regions in South Af- reduces mean annual runoff over most of summer
rica were downscaled to daily rainfall values using tech- rainfall areas in South Africa by only 5% (Fig. E.2], top).
niques described in Schulze et al. (1998b) for applica- However, in the south-west winter rainfall region, the
tion with the ACRU modelling system to over 1500 Qua- response to temperature becomes more dramatic, with
ternary catchments in South Africa. A simple benefit a 2 °C increase by itself producing a simulated reduc-
analysis of forecasting skill was undertaken, in which a tion in mean annual runoff in excess of 50%. The rea-
"win" was recorded if, for the historical seasonal rainsons for this are that under present climatic conditions
fall forecast, the simulated seasonal runoff was closer to evaporation losses there are relatively low from the moist
the runoff simulated with actual historical rainfall than soils in winter, but that with warming, faster drying soils
the median seasonal runoff, while a "loss" was recorded between rainfall events significantly reduce runoff. The
when median runoff was closer to the actual than the most significant sensitivity to climate change, however,
forecast runoff. "No difference" implies forecasted and remains that due to rainfall, with changes by one unit
median runoffs within 5% of one another. Figure E.25 manifesting themselves as runoff changes by a factor of
illustrates that, when excluding three seasons out of 15 2 to 5 (Fig. E.2?, bottom), with the sensitivity more domi-
for which the rainfall forecast accuracy proved 100% nant in the extreme south-west.

536
CHAPTER E.6 .Case Studies
Fig. E.27.
Sensitivities to selected atmospheric
variables: Temperature
(top) and rainfall (bottom)
change impacts on mean annual
runoff over South Africa
(from Schulze and Perks 2000)
Conclusions
This chapter has provided examples of hydrological risk
from South Africa. The examples bring to the fore two
overarching issues in hydrological risk management. The
first is the question of uncertainty in risk-related hydrological
studies -uncertainties regarding meteorological
and catchment conditions now and in the future, and uncertainties
around input data and uncertainties emanating
from the models used in hydrological risk management.
The second revolves around the recurring theme
of the hydrological system's potential for amplifying
perturbations of the climate drivers, predominately of
rainfall. It is the amplification and the uncertainty issues
Ratio
0 0.86 -1.00
0 O. 80 -D. 96
~ 0.16 -O. 80
0 O. 76 -D. 86
~ 0.50 -O. 76
.> O. 60
Mcxte/ : ACRU
Ratio
.5
Mode/:ACRU
which will need to be stressed to practitioners and managers
of hydrological risk and vulnerability time and
again and to which researchers will need to focus more
of their attention.
Acknowledgements
Results shown and discussed in this chapter derive from
research funded by the Water Research Commission of South
Africa and the US Country Studies for C1in1ate Change Programme.
They are thanked for their support, as are the
Computing Centre for Water Research, Lucille Perks (climate
change), Mark Horan (GIS) and Jason Hallowes (forecasting),
all of the University of KwaZulu-Natal.

suo!snl:>uo)
'lS ';}lIl;}ld 'V l;}~O'M
11!.I3U39 .£"3 ~Iqel
!UU:ln'J :lp OABlH sl..{:l1
-dt! OM~ u33M~3q ~St!l~UO:> ){It!~S t! P3P!AOld 3At!q 3M
0!lt!U3:>S 3q~ 'q:>t!olddt! ~SlY 3qJ. .3~Ut!q:> re~U3WUOl!A
-U3 3ln~n) )0 S~:>3JJ3 re!~u3~od 3q~ ~U!SS3SSt! o~ s3q:>t!Old
)0 ~3S t! do13A3P o~ 's13poW UO!~t!lD:>l!:J rel3U39 mOl)
Alre:>!dJ..~ 's~nd~no 13poW l3~ndwo:> uo S3!13l 'q:>t!olddt!
.~U3WUOl!AU3 s,q~lt!H 3~ )0 3~t!~S 3ln~n) 3~ lO) SO!reU3:>S
S! l3~t!3M ~t!~ S3WnSSt! Alrensn q:>t!olddt! 0!lt!U3:>S S!qJ.
-UO:»O lO~:>t!) ~U!:>lO) 'AluO S3W!~3WOS pUt! '~Ut!U!WOP 3~
-0) '~St!l~UO:> U! 'q:>t!olddt! A:J!{!qt!l3UlDA 3qJ. .S3re:>S It!q
-ol~ ~t! 3~Ut!q:> wl3~-~uol uo A{!reW!ld S3Sn:>0) pUt! 'Ul3:>
S3re:>S reUO!~3l pUt! re:>°l ~t! S3SS3l~S 31d!~lDW uo S3Sn:>
.s~:>t!dw! Wl3~-l3~lOqS '3~t!!P3WW! 3l0W pUt! (A{!lt!W!ld)
'p3Sn A13P!M ~sow 3~ S! q:>!qM 'q:>t!olddt! 0!reU3:>S 3q~ ~nq
S3q:>t!Olddt! q~oq U! S3SS3U}[B3M pUt! S~~U3l~S 3lt! 3l3qJ.
3M 3:>UO AluO .S~U3WSS3SSt! A:J!1!qt!l3ulDA q~!M ~lt!~S ~snw
ssaJ~S ~ueu!wcp pawnsslf
p3sn J..111!:>!dJ..~ s~ s3q:>~0.Idd~ J..~mq~.I3UJnA pU~ 0!.I~U3:>S 3tp }O S:>!~S!.I3~:>~:mq:>
)1a' SIOSOJae 'saJmeJadwa1 uea)o 'aJa4dsoW1e
41° 1 SUo!ss!wa se6 asn04UaaJ61Ua)aJ 'a1ew!D
aj1:l!j S! aJa4~ 'Ja/laMOH "sanb!U4)a~ 6uHeJSuMOp
ap!/loJd slapow ~uasaJd ~e4~ ~sa66ns o~a)uap!/la
~e SO!Jeua)s a~ew!p aspaJd JO 'a~eJro)e ')!~sHeaJ
'sale)s leUo!6a1 JO le)Oj
"~lroYJ!P f.1~uasa1d) pa~e~sun a/le4 aJn~nJ ~Ue~S!p d4~ JO
'f.1!I!qeqoJd 10 f.1U!eua) JO Sja/lal (alqeMOU)(Un
~U~WUOJ!AU~ JnO O~ S~lJ~Jq~ :>lJ!:>~ds ~q~ ~JlJ ~lJqM MOU){
o~ q~nou~ ~~JlJI (Alq!ssod Jo) AI~}{!I S! ~~UlJq:> re~U~WUOJ
-!AU~ J~q~~qM (SO!JlJU~:>S ~U!Sn) ~~lJW!~S~ O~ p~~U ~M Op
lJ ~AlJq O~ SlJ Il~M SlJ 'p~S!re~J ~q o~ S~lJ~Jq~ ~s~~ ~snlJ:>
:>!Iqnd ~~ ~JlJd~Jd J~~~~q illM q:>lJOJddlJ A~!nqlJJ~UJnA v
"SO!JlJU~:>S ~S~~ O~ p~q:>lJ~~lJ ~U!lJ~J~:>un ~q~ JO ~JnSlJ~W
S~J!qs :>!J~qdsow~lJ ~dnJqlJ 'J~q~lJ~M ~Jn~nJ ~IqlJ!JlJA JoJ
s~:>u~J~JJ!P ~q~ S~s!JlJwwns L"H ~IqlJ.L "S~SS~J~S re~u~w
-UOJ!Au~ ApSO:> PUlJ ~U~~Jn J~~O PUlJ 'UO!~:>~J!P AUlJ U!
-oJd ~U~WSS~SSlJ ~!nqlJJ~UJnA ~q~ JO S~IdwlJx~ :>lJ!:>~dS
"S~q:>lJOJddlJ o~ ~q~ u~~~~q
p~~u~s~Jd ~J~M 's~:>JnOS~J J~~lJM O~ ~:>~ds~J ~!M '~mp~:>
~lJq~ p~MOqS s~n~lJ~Il°:> PUlJ ~JlJwsQJ9.A "J~~dlJq:> s~ U!
-:>~p M~J 1X~U ~~ U! uo!~lJJndod ulJwnq U! ~SlJ~J:>U! ~~
JO A~!nqlJnlJAlJ ~~ U~~lJ~J~ O~ AI~}{!I ~JOW q:>nw S! S~plJ
')~a'SSOI ~e~!qe4
'aJn~nJ
"4)JeasaJ wJa~-6uOI pue (sJeaf. Of 0~O)UJJa~.~J04S a4~ ul sJeaf. OOL o~ Of '1 OJ pajqnop 'wJa~.6uol UJa)UO) JO aweJJaw!~ lensn
'Sale)S leqol6 pue leUo!~eu 1euo!6aJ 1e)01 spaau ale)s le)Ol leuo!6aJ saw!~aWOs 1eqOj9 uJa)UO) JO ale)s lensn
sJa~aweJed a!dj~lnW U! sanle/l aWaJ1Xa te!~ua~Od saJn~eJadwa~ ueaw U! sa6ue4) pa6eJa/le f.lIe!~eds uJa)UO) JO sJa~aweJed Jo[ew
pue (sf.ep aaJJ.1S0JJ!uo!~e~!dpaJd 'aJn~eJadlUa~) aWOS4~!M slla) p!J6 a61el f.IJ!eJ U! uo!~e~!dpaJd pue
'saJY 'spOOIJ)s~ua/la awaJ~xa UO sroOJ !eUo!~!ppe '~46n01p JOJ SO!leUa)S leuo!6aJ
'f.1u!e:\Ja)unJo saJnseaw:(')~a.'s~4 6n01p
ald!~lnW uo e~ep paj!e~ap saJ!nbaJ 4)e01ddy Ja4~0 JOJ ssaupaJedaJd S~!w1.l ssaJ~s al6u!s uo sroo:! -Ia/lap 10J suo!~ei!W1.l10(ew
1euo!6aJ 'le)OI ~esuOt~)e1a~U! Jja4~ pue sassaJ~s 'sassaJ~s saI6a~eJ~s 6Uldo) 6uldo
)()el seaJe f.uewpue -saje)$ leqol6pue jeUOj~eu a~ewH) JO 6u!dweJ lenpe16 M04s uaijo S~lnsa~
'Uo!~ewJoJu! a~enbape 's~ua/la awaJ1Xa JOJ ssa~pa1edaJ~6ul~!W!I-a6ue4)
-e!Je/la~ewtl.J te)j10~s!4)a~ew1.l) :sassaJ~satd!1tnw
a)ueqJn~s!p paJa~le 'asn Ja~eMpUe asn puej '(f.1!I!q
's~ue~nuod/s~ueu!we~uo) 'sapadSd/l!Se/lU! 'saw!6aJ
-aJd ~!WH f.ew sanss! WJa~-U04s UO s!Se4dw3 la)(!1 lie JO ~asqns pa~!W!1 e f.luo ~uasaJdaJ s1jnSa~
'wJa~-6UOj10 -U04s a4~ U! aw!~ awos a~ew1.l) ,
Uj sa6ue4) .PI04sa14~" ~dnJqeJoJ ssaupa1ed f. spua1~leuo!~)aJ!P!un f.llensn -saWo)~no aJn~nJ
, " 'e1pawa4~ 's~s!~uaps f.uew f.q pa~da))e aJe s~lnsa~
f.ue U! sa6ue4) ~dnJqe6u!pnpu!'$aJn~nJ ajq!ssod ,"
JO UOI~eIJe/l Ja~eaJ6 JeJ e JOJ ssaupa1eda1d sdJmba~ .suO!P!paJd. lenpe se )!lqnd a4~ pue
~af. ')!~s!leaJ dJOW f.lqeqoJd St St4~- UO1~)dJ!P siapow lie 1e4~ S! S~lnsa1 JO UO!~eISueJ~ a4~ Uj ~SOl

538
-
CHAPTER E.7 .Conclusions
potable water than any of the climate change scenarios assessment procedure is a more inclusive evaluation
created by the general circulation models. Brad Bass and method than relying on scenarios at the start of an im-
colleagues, and Changming Liu, documented the similar pact assessment. The scenario approach is only required
importance (and threat) associated with increasing popu- once specific environmental threats are recognised. Once
lation pressures in China. Roland Schulze shows hydro- we identify environmental risk, the scenario approach
logical risks exemplified for South Africa with special can be used to determine if there is a possibility that the
consideration of aridity and difficulties in management threat could be realised and, if possible, quantify the prob-
practices due to uncertainties in the prediction of seaability of the risk. Hutchinson shows how the strong toposonal
weather such as ENSO events. Tom Stohlgren and graphic coherence of weather can be used to reduce some
colleagues document the range of threats to a pristine of the uncertainty in constructing fine scale scenarios
park area, in which invasive exotic plant species and air and assessing vulnerability. Even without quantification
pollution are major threats. Other integrated assessments and the ability to predict the future, however, the vulner-
are also taking place (Becker et al. 1999; Stohlgren 1999b, ability perspective provides useful information for
and others). As shown in this chapter, the vulnerability policy-makers (Sarewitz et al. 2000).

References Part E
Adams RM, Crocker TD, Katz RW (1984) Assessing the adequacy
of natural science information: a bayesian approach. Rev Econ
Stat LXVI:568-575
Aguirre AA, Starkey EE (1994) Wildlife disease in US national
parks: historical and coevolutionary perspectives. Conserv Bioi
8:654-661
Alcamo J (ed)(1994) IMAGE 2.0: integrated modeling of global
climate change. Kluwer Academic Publishers, Dordrecht
Alcamo J, Kreilemans E (1996) Emission scenarios and global climate
protection. Global Environ Chang 6:305-334
Alcamo J et al. (1998) Baseline scenarios of global environmental
change. In: Alcamo J, Leemans R, Kreileman E (eds) Global
change scenarios of the 21'1 century. Results from the IMAGE 2.1
model. Elseviers Science, London, pp 97-139
Alcamo J, Henrichs T, Rosch T (2000) World water in 2015: global
modelling and scenario analysis for the world commision on
water for the 21'1 century. University of Kassel, Kassel
Alexander RR (1985) Major habitat types, community types and plant
communities in the Rocky Mountains. USDA Forest Service, Rocky
Mountain Forest and Range Experiment Station, Fort Collins, Co
Alexandratos N (ed) (1995) Agriculture: towards 2010. John Wiley
& Sons, Chichester
Alley RB, Meese DA, Shuman AJ, Gow AJ, Taylor KC, Grootes PM,
White JWC, Ram M, Waddington ED, Mayewski PA, Zielinski
GA (1993) Abrupt accumulation increase in the Younger-Dryas
termination in the GISP2 ice core. Nature 362:527-529
Anthes RA, Warner TT (1978) Development of hydrodynamic models
suitable for air pollution and other mesometeorological studies.
Mon Weather Rev 106:1045-1078
ArmstrongDM (1993) lions, ferrets and bears. A guide to the mammals
of Colorado. Colorado Division of Wildlife and University
of Colorado Museum, Denver. Co
Arnell NW (1999) Climate change and global water resources. Global
Environ Chang 9:31-49
Aubin J-p, Cellina A (1984) Differential inclusions. Springer. Berlin
Bankes SC (1993) Exploratory modeling for policy analysis. Oper
Res 41:435-449
Bantelmann A, Hoffmann D, Menke B (1984) Veranderungen des
Kiistenverlaufs, Ursachen und Auswirkungen. In: Kossack G,
Behre KE, Schmid P (eds) Archaologische und naturwissenschaftliche
Untersuchungen an landlichen und friihstadtischen
Siedlungen im deutschen Kiistengebiet vom 5. Jahrhundert v.
Chr. bis zum 11. Jahrhundert n. Chr.. Band 1: Landliche Siedlungen.
Deutsche Forschungsgemeinschaft, Weinheim, pp 51-68
Bardossy A, Plate EJ (1992) Space-time model for daily rainfall using
atmospheric circulation patterns. Water Resour Res 28:1247-1259
Baron JS. Ojima DS, Holland EA, Parton WJ (1994) Analysis of nitrogen
saturation potential in Rocky Mountain tundra and forest:
implications for aquatic systems. Biogeochelnistry 17:61-82
Baron JS, Rueth HM, Wolfe AP, Nydick K, AlIstott EJ. Minear JT,
Moraska B (2000) Ecosystem responses to nitrogen deposition
in the Colorado Front Range. Ecosystems 3:352-368
Barrow CJ (1991) Land degradation. Cambridge University Press.
Cambridge
Bass B, Huang G, Russo J (1997) Incorporating climate change into
risk assessment using grey mathematical programming. J Environ
Manage 49:107-123
Becker A, Wenzel V,Krysanova V,LahrnerW (1999) Regional analysis
of global change impacts: concepts, tools and first results.
Environ Model Assess 4:243-257
Behnke RJ (1992) Native trout of Western North America. American
Fisheries Society, Bethesda, Maryland
Bellone E, Hughes JP, Guttorp P (2000) A hidden Markov model
for relating synoptic atmospheric patterns to precipitation
amounts. Climate Res 15:1-12
Bernabo IC, Webb TI (1977) Changing patterns in the Holocene
pollen record of northeastern North America: a mapped summary.
Quaternary Res 8:64-96
Blaikie P, Cannon T, Davis I, Wisner B (1994) At risk: natural
hazards, people's vulnerability and disasters. Routledge,
London
Bloh Wvon, Block A, Schellnhuber H-J (1997> Self-stabilization of
biosphere under global change: a tutorial geophysiological
approach. Tellus 49B:249-262
Bloh W von, Block A, Parade M, Schellnhuber H-J (1999) Tutorial
modeling of geosphere-biosphere interactions; the effect of
percolation-type habitat fragmentation. Physica A 266:186-196
Bonan GB (1997> Effects of land use on the climate of the United
States. Climatic Change 37:449-486
Bounoua 1, GJ Collatz, Los SO, Sellers PI, Dazlich DA, Tucker CJ,
Randall DA (2000) Sensitivity of climate to changes in NDVI.
J Climate 13:2277-2292
Braswell BH, Schimel Ds, Linder E, Moore III B (1997> The response
of global terrestrial ecosystems to interannual temperature variability.
Science 278:870-872
Brovkin V, Ganopolski A, Claussen M, Kubatzki C, Petoukhov V
(1999) Modelling climate response to historical land-cover
ch.ange. Global Ecol Biogeogr 8:509-517
Brown RA, Rosenberg NJ (1999) Climate change impacts on the
potential productivity of corn and winter wheat in their
primary United States growing regions. Climatic Change 41:
73-107
Buchholtz CW (1983) Rocky Mountain National Park: a history.
Colorado Associated University Press, Boulder, Co
Bury RB, Corn PS, Dodd CK, McDiarmid RW Jr (1995) Amphibians.
In: LaRoe ET (ed) Our living resources. National Biological
Service, Washington DC
Canad~ll P (2000) Non-linear responses and surprises: a new Earth
system science initiative. Global Change Newsletter 43:1-2
Casella G, George EI (1992) Explaining the Gibbs sampler. Am Stat
46:167-174
Chadde SW,Kay CE (1991) TalI-wilIowcornrnunities on Yellowstone's
northern range: a test of the natural regulation paradigm. In:
Keiter RB, Boyce MS (eds) Greater Yellowstone ecosystem:
redifining America's wilderness heritage. Yale University Press,
New Haven, Conn., pp 231-262
Changnon SA (1998) Comments on "Secular trends of precipitation
amount, frequency, and intensity in the United States." B
Am Meteorol Soc 79:2550-2552
Chase TN, Pielke RA, Kittel TGF, Nemani R, Running SW (1996)
The sensitivity of a general circulation model to global changes
in leaf area index. J Geophys Res-Atmos 101:7393-7408
Chen J et al. (1998) Yellow River harnessing and water resources
development. Yellow River Conservancy Press, 44 p

540
References Part E
CIESIN (Center for International Earth Science Information Network)
(2000) Gridded population of the world: provisional release
of updated database of 1990 and 1995 estimates. Columbia
University's Center for International Earth Science Information
Network
Claussen M (2001) Earth system models. In: Ehlers EKT (ed) Understanding
the Earth system: compartments, processes and
interactions. Springer-Verlag, Heidelberg, pp 145-162
Claussen M,GaylerV (1997> The greening of Sahara during the mid-
Holocene: results of an interactive atmosphere-biome model.
Global Ecol Biogeogr 6:369-377
Claussen M, Brovkin V, Ganopolski A, Kubatzki C, Petoukhov V
(1998) Modeling global terrestrial vegetation-climate interaction.
Philos TRoy Soc B 353:53-63
Claussen M, Brovkin V, Petoukhov V, Ganopolski A (2001) Biogeophysical
versus biogeochemical feedbacks of large-scale landcover
change. Geophys Res Lett 28:1011-1014
Cohen J (1996) How many people can earth support? WW Norton,
New York
Colorado Division of Wildlife (1986) Wildlife in danger: the status
of Colorado's threatened or endangered fish, amphibians,
birds, and mammals, Fort Collins, Colo
Conway D, Jones PD (1998) The use of weather types and air flow
indices for GCM downscaling. J Hydrol 213:348-361
Corn PS, Fogelman JC (1984) Extinction of montane populations
of the northern leopard frog (Rana pipiens) in Colorado. J Her-
petoI18:147-152
Cubasch U,Santer BD,Hellbach A, Hegerl G, HiSck H,Maier-Reimer
E, Mikolajewicz U, StiSssel A, Voss R (1994) Monte Carlo climate
change forecasts with a global coupled ocean-atmosphere
model. Clim Dynam 10:1-19
Cubasch U, Storch H von, Waszkewitz J, Zorita E (1996) Estimates
of climate changes in southern Europe using differnt downscaling
techniques. Climate Res 7:129-149
de Vries HJM, Bollen J, Bouwman L, den Elzen MGJ, Janssen M,
Kreilemann E (2000) Greenhouse-gas emissions in equity,
environment- and service-oriented world: an IMAGE-based
scenario for the next century. Technol Forecast Soc 63:137-174
deKoning GHJ, Veldkamp A, Fresco LO (1998) Land use in Ecuador:
a statistical analysis at different aggregation levels. Agr
Ecosyst Environ 70:231-247
DiSll P, Siebert S (2002) Global modeling of irrigation water requirements.
Water Resources Research 38(4):10.102912001
WRoO0355,2002
DiSiSs BR (1994) Environmental degradation, global food production,
and risk for large-scale migrations. Ambio 23:124-130
Dowlatabadi H (1999) Integrated assessment: implications of uncertainty.
In: Encylopedia of Life Support Systems. Oxford University
Press, Oxford
Dowlatabadi H, Morgan G (1995) Integrated assessment climate
assement model 2.0, technical documentation. Department of
Engineering and Public Policy, Carnegie Mellon University,
Philadelphia
Downing TE, Olsthoorn AJ, Tol RSJ (eds) (1999) Climate change
and risk. Routledge, London
Dynesius M, Nilsson C (1994) Fragmentation and flow regulation
of river systems in the northern third of the world. Science
266:753-762
Dzhamalov RG, Zekster IS (eds) (1999) World map of hydrogeological
conditions and groundwater flow and digital maps of
hydrogeological conditions and groundwater flow. HydroScience
Press
Elias SA, Short SK, Clark PU (1986) Paleoenvironment interpretations
of the late Holocene, RoCky Mountain National Park, Colorado,
USA. Review De Paleobiologie 5:127-142
Fairman R, Mead CD, Williams WR (1998) Environmental risk assessment:
approaches, experiences and information sources.
European Environment Agency, Copenhagen, 252 p
Falkenmark M (1998) Dilemma when entering the 21't century -
rapid change but lack of a sense of urgency. Water Policy 1:
421-436
Fall PL (1988) Vegetation dynamics in the southern Rocky Mountains:
Late Pleistocene and Holocene timber-line fluctuations.
University of Arizona, 303 p
FAO (1998) FAO statistics on diskette. Food and Agriculture Organization
of the United Nations, Rome
FAO (1999) FAOSTAT database collections, vol. 1999. Food and
Agriculture Organization of the United Nations
Fekete BM, Vorosmarty Q, Grabs W (1999) Global, composite runoff
fields based on observed river discharge and simulated
water balance. Global Runoff Data Centre, Koblenz
Fekete BM, Vorosmarty Q, Grabs W (2002) High resolution fields
of global runoff combining observed river discharge and simulated
water balances. Global Biogeochem Cy 16(3):10.10291
1999GBoO1254,2002
Flather CH, Joyce LA, Bloomgarten CA (1984) Species endangerment
patterns in the United States. USDA Forest Service, Rocky Mountain
Forest and Range Experiment Station, Fort Collins, Co, 42 P
Flint EP (1994) Changes in land use in south and South-east Asia
from 1880 to 1980: a data base prepared as part of a coordinated
research program on carbon fluxes in the tropics. Chemosphere
29:1015-1062
Foster DR, Motzkin G, Slater B (1998) Land-use history as longterm
broad-scale disturbance: regional forest dynamics in central
New England. Ecosystems 1:96-119
Frenzel B (1994) Moore als Umweltindikatoren in klimatischen
Grenzgebieten. Hohenheimer Umwelttagung 26:155-167
Frolking S,Xiao X,ZhuangY,Salas W,Li C (1999) Agricultural landuse
in China: a comparison of area estimates from ground-based
census and satellite-borne remote sensing. Global Ecol Biogeogr
8:407-416
Ganopolski A, Rahmstorf S (2001) Rapid changes of glacial climate
simulated in a coupled climate model. Nature 409:153-158
Ghil M, Childress S (1987) Topics in geophysical fluid dynamics:
atmospheric dynamics, dynamo theory, and climate dynamics.
Springer-Verlag, Heidelberg
Giorg!,»larinucci MR, Visconti G (1992) A 2 X CO2 climate change
scenario over Europe generated using a limited area model
nested in a general circulation model. 2. Climate change scenario.
J Geophys Res 97:10011-10028
Glantz MH (2001) Currents of change: impacts of El Nino and La
Nina on climate and society, 2nd edn. Cambridge University Press,
Cambridge, New York
Gleick P (2000) Water futures: a review of global water resources
projections. In: Rijsberman FR (ed) World water scenarios:
analyses. Earthscan Publications, London, pp 27-37
Gliemeroth AK (1995) Palaookologische Untersuchungen fiber die
letzten 22000 Jahre in Europa. S. Fischer, Stuttgart
Godwin H (1952) Recurrence-surfaces. Danmarks Geol Unders,
II. raekke, nr. 80, pp 22-30
Gossow H (1976) Wildokologie. Begriffe, Methoden, Ergebnisse,
Konsequenzen. BLV; Miinchen
Graumlich LJ (1993) A 1000-year old record of temperature and
precipitation in the Sierra Nevada. Quaternary Res 39:249-255
Greenback Cutthroat Trout Recovery Team (1983) Greenback cutthroat
trout recovery plan revision. US Fish and Wildlife Service,
Colorado Division of Wildlife, Fort Collins, Colo
Groats PM, Stuiver M, White JWC, Johnson SJ, Jouzel J (1993) Comparison
of oxygen isotope records from the GISP2 and GRIP
Greenland ice cores. Nature 366:552-554
Guenni 1 et al. (2001) A biogeophysical approach to the vulnerability
concept: extreme rainfall and population distribution
in South America during 1999. Global Change Open Science
Conference, Amsterdam, The Netherlands, (poster paper)
Hall CAS, Tian H, Qi Y, Pontius G, Cornell J (1995) Modelling spatial
and temporal patterns of tropical land-use change. J Biogeogr
22:753-757
Henry D, Henry M (1991) Greenbacks are back. Colorado Outdoors
40:23-25
Hilderink HBM et al. (1999) TARGETS -The CD-Rom
Hinrichsen D, Robey B, Upadhyay UD (1998) Solutions for a water-short
world. Johns Hopkins University School of Public
Health, Baltimore, Md
Hoekstra AY (1996) AQUA: a framework for integrated water policy
analysis. Dutch National Institute of Public Health and the Environment
(RIVM), Bilthoven, The Netherlands, 70 p
Hoekstra AY (1998) Perspectives on water. An integrated modelbased
exploration of the future. Technical University, Delft

Hoffman MT, Todd SW, Ntshona ZM, Turner SD (1999) Land degradation
in South Africa. Pretoria, RSA, 245 P
Hubbard KG, Flores-Mendoza FJ (1995) Relating United States crop
land use to natural resources and climate change. J Climate 8:
229-335
Hughes MK, Graumlich LJ (1996) Multimillenial dendroclimate
records from the western United States. In: Bradley RS, Jones
PD, Jouzel J (eds) Climatic variation and forcing mechanisms
of the last 2000 years. Springer-Verlag, New York, pp 109-124
Hulme M, Carter TR (1999) Representing uncertainty in climate
change scenarios and impact studies. ECLAT -2:12-54
Huntley B,Birks HJB (1983) An atlas of past and present pollen maps
of Europe, 0-13 000 years ago. Cambridge University Press,Cam-
bridge
IPCC (Intergovernmental Panel on Climate Change) (1996) Climate
change 1995, the science of climate change. Contribution
of Working Group I of the Second Assessment Report of the
IPCC. Cambridge University Press, Cambridge
IPCC (Intergovernmental Panel on Climate Change) (2001) Summaryfor
policymakers.A Report of Working Group I of the IPCC
Jamieson D, Herrick C (1995) The social construction of acid rain:
some implications for science/policy assessment. Global Environ
Chang 5:105-112
Jeton EE, Dettinger MD, Smith JL (1999) Potential effects of climate
change on streamflow, eastern and western slopes of the
Sierra Nevada, California and Nevada. USGS Water Resource
Investigations, 44 P
Jones RG, Murphy JM, Noguer M (1995) Simulation of climate change
over Europe using a nested regional climate model I. Assessment
of control climate, including sensitivity to location of lateral
boundaries. Q J Roy Meteor Soc 121:1413-1449
Karl TR, Knight RW (1998a) Secular trend of precipitation amount,
frequency, and intensity in the United States. B Am Meteorol
Soc 79:231-241
Karl TR, Knight RW (1998b) Reply. B Am Meteorol Sac 79:2552-2554
Karl TR, Knight RW, Plummer N (1995) Trends in high-frequency
climate variability in the twentieth century. Nature 377:217-220
Katz RW (1999) Techniques for estimating uncertainty in climate
change scenarios and impact studies. ECLAT -2:40-55
Kite GW (1993) Application of a land class hydrological model to
climate change. Water Resour Res 29:2377-2384
Klein Goldewijk CGM, Battjes JJ (1997> A hundred year (1890-1990)
database for integrated environmental assessments. National
Institute of Public Health and the Environment, Bilthoven
Knight DH (1994) Mountains and plains: the ecology of Wyoming
landscapes. Yale University, Thomson-Shore, Dexter, Mich
Kolstad CD, Kelly D (1999) Bayesian Learning, pollution, and growth.
J Econ Dyn Control 23:491-518
Koruba V, Jabbar MA, Akinwumi JA (1996) Crop-livestock competition
in the West African derived savannah: application of
a multi-objective programming model. Agr Syst 52
Kossack G,Behre K-E, Schmid P (1984) Archiologische und naturwissenschaftliche
Untersuchungen an lindlichen und friihstidtischen
Siedlungen im deutschen Kiistengebiet vom 5. Jahrhundert
v. Chr. bis zum 11. Jahrhundert n. Chr. Deutsche Forschungsgemeinschaft,
Weinheim
Kourzoun VI, Sokolov AA, Budyko MI, Voskrensensky KP, Kalinin
GP, Konoplyantsev AA, Korotkevich ES, L'vovich MI (1978) Atlas
of the world water balance. UNESCO, Paris
Krzysztofowicz R (1999) Bayesian theory of probalistic forecasting
via deterministic hydrologic model. Water Resour Res 35:
2739-2750
Kunkel KE, Pielke RA Jr, Changnon SA (1999) Temporal fluctuations
in weather and climate extremes that cause economic and human
health impacts: a review. B Am Meteorol Sac 80(6):1077-1098
Landsea CW,KnaffJA (2000) How much skil1 was there in forecasting
the very strong 1997-1998 El Nifio? B Am Meteorol Soc 81:
2107-2119
Lanfear KJ,Hirsch RM (2000) USGS study reveals a decline in longrecord
streamgauges. EOS Trans Am Geophys Union 80
Langner LL, Flather CH (1994) Biological diversity: status and
trends in the United States. Rocky Mountain Forest and Range
Experiment Station, Fort Collins, Colo
Lavender D (1975) The Rockies. Harper and Row, New York
References Part E 541
Lee X, Black TA (1993) Atmospheric turbulence within and above
a Douglas-fir stand. Part II: Eddy fluxes of sensible heat and
water vapour. Bound-Lay Meteorol 64:369-389
Leemans R (1999) Modelling for species and habitats: new opportunities
for problem solving. Sci Total Environ 240:51-73
Leemans R, Agrawala S, Edmonds JA, MacCracken MC, Moss RM,
Ramakrishnan PS (1996) Mitigation: cross-sectoral and other
issues. In: Watson RT, Zinyowera Mc, Moss RH (eds) Climate
change 1995. Impacts, adaptations and mitigation of climate
change: scientific-technical analysis. Cambridge University
Press, Cambridge, pp 799-797
Lins H, Slack JR (1999) Streamflow trends in the United States.
Geophys Res Lett 26:227-230
Liu C (1981) The influence of forest cover upon annual runoff in the
loess plateau of China. In: Ma LJC, Noble AG (eds) Water transfer:
its environment: Chinese and American views. Methuen, New
York, pp 131-142
Liu c, Li C (2000) Analysis on runoff series with special reference
to drying up courses. Acta Geogr Sinica 55:257-265
Liu c, Liang J (1989) Some problems of flooding and its prevention
in the lower Yellow River. In: Liu c, Liang J (eds) Taming
the Yellow River: silt and floods. Kluwer Academic Publishers,
Dordrecht, pp 223-242
Liu KB,Sun Sc, Jiang XH (1992) Environmental change in the Yangtze
River delta since 12000 years B.P. Quaternary Res 38:32-45
Lorenz EN (1968) Climatic determinism. In: Causes of climatic
change, vol 30, 8. AMS, pp 1-3
Lorenz EN (1975) Climate predictability. In: Committee W-IJO (ed)
The physical basis of climate and climate modelling, vol 16.
GARU Publication Series
Lorenz EN (1979) Forced and free variations of weather and climate.
J Atmos Sci 36:1367-1376
Loveland TR,BelwardAS (1997> The IGBP-DIS global 1 krn land-cover
dataset, DISCover: first results.lnt J Remote Sens 18:3291-3295
Ludeke MKB, Moldenhauer 0, Petschel-Held G (1999) Rural poverty
driven soil degradation of climate change: the sensitivity
of the disposition towards the Sahel syndrome with respect to
climate. Environ Model Assess 4:315-326
L'vovich MI, White GF (1990) Use and transformation of terrestrial
water systems. In: Thrner BL, Clark Wc, Kates RW, Richards JF,
Mathews JT, Meyer WB (eds) The Earth as transformed by human
action. Cambridge University Press, Cambridge, pp 235-252
Lynch AH, McIlwaine S, Beringer J, Bonan GB (2001) An investigation
of the sensitivity of a land-surface model to climate change
using a reduced form model. Clim Dynam 17=643-652
MacCracken MC (2002) Do the uncertainty ranges in the IPCC
and US national assessments account adequately for possibly
overlooked climatic influences? Climatic Change 52:13-23
Mack RN, Simberloff D, Lonsdale WM, Evans H, Clout M, Bazzaz
FA (2000) Biotic invasions: causes, epidemiology, global consequences,
and control. Ecol Appll0:689-710
Matsuoka Y, Kainuma M, Morita T (1994) Scenario analysis of global
warming using the Asian-Pacific Integrated Model (AIM).
Energ Policy 23:1-15
Matthews E, Rotmans J, Ruffing K, Waller-Hunter J, Zhu J (1997>
Critical trends: global change and sustainable development.
UN Publication Department for Policy Coordination and Sustainable
Development, New York
McCabe G, Wolock DM Jr (1997> Climate change and the detection
of trends in runoff. Climate Res 8:129-134
Mearns 10, Rosenzweig c, Goldberg R (1997> Mean and variance
change in climate scenarios: methods, agricultural applications,
and measures of uncertainty. Climatic Change 35:367-396
Meszaros E, Palvolgyi T (1990) Daisyworld with an atmosphere.
Idojaras 94:339-345
Mills JD (1991) Wyoming's Jackson Hole darn, horizontal channel
stability, and floodplain vegetation dynamics. University of
Wyoming, Laramie, 54 p
Morozova TD (1981) Razvitie pocvennogo pokrova Evropy v
pozdnem pleistocene. Nauka, Moskva (in Russian: The development
of European soils during the Late Pleistocene)
Moxey AP, White B, O'Callaghan JR (1995) CAP reform: an application
of the NELUP economic model. J Environ Plan MaiJage 38:
117-123

542
References Part E
Nams VO, Folkard NFG, Smith JNM (1993) Effects of nitrogen fertilization
on several woody and non-woody boreal forest species.
Can J Bot 71:93-97
Nativ R (1992) Recharge into southern high plains aquifer -possible
mechanisms, unresolved questions. Environ Geol Water S
19(1):21-32
Nevison C, Gupta VK, Klinger L (1999) Self-sustained oscillations
on Daisyworld. Tellus 51B:806-814
Nichols H (1982) Review of late Quaternary history of vegetation
and climate in the mountains of Colorado. In: Halfpenny JC
(ed) Ecological studies in the Colorado alpine: a festschrift for
John W. Marr. Inst. Arctic and Alpine Res., occasional paper 37.
University of Colorado, Boulder, Co, pp 27-33
Nordstrom K, Gupta V, Chase T (2004) Salvaging the Daisyworld
parable under the dynamic area fraction framework. Scientist
on Gaia-II. Cambridge, Ma, MIT Press (in press)
Olson JS (1994a) Global ecosystem framework-definitions. USGS
EROS Data Center, Sioux Falls, SD
Olson JS (1994b) Global ecosystem framework-translation strategy.
USGS EROS Data Center, Sioux Falls, SD
Ongley ED (2000) The Yellow River: managing the unmanageable.
Water Int 25:227-231
Oreskes N, Shrader-Frechette K, Belitz K (1994) Verification, validation,
and confirmation of numerical models in the earth sciences.
Science 263:641-646
Pachur HJ, Hoelzmann P (1991) Paleoclimatic implications of Late
Quaternary lacustrine sediments in Western Nubia, Sudan.
Quaternary Res 36:257-276
Pate-Cornell E (1996) Uncertainties in global climate change estimates.
Climatic Change 33:145-149
Peer T (2000) The highland steppes of the Hindukush Range as
indicators of centuries old pasture farming. Marburger Geogr
Schriften 135:312-325
Penning de Vries FWT, Rabbinge R, Groot JJR (1997> Potential and
attainable food production and food security in different regions.
Philos TRoy Soc B 352:917-928
Peters EF, Bunting SC (1994) Fire conditions pre- and post-occurrence
of annual grasses on the Snake River plain. In: Monsen
SB, Kitchen SG (eds) Proceedings -ecology and management
of annual rangelands. USDA Forest Service Inter. Forest and
Range Experiment Station, Ogden, Utah, pp 31-36
Peters RL, Lovejoy TE (eds) (1992) Global warming and biological
diversity. Yale University Press, New Haven
Petschel-Held G, Bruckner T,Schellnhuber H-J, T6th F, Hasselmann
K (1999a) The tolerable windows approach. Theoretical and
methodological foundations. Climatic Change 41:303-331
Petschel-Held G, Block A, Cassel-Gintz M, Kropp J, LUdeke MKB,
Moldenhauer 0, Reusswig F, Schellnhuber H-J (1999b) Syndromes
of global change: a qualitative modelling approach to
assist global environmental management. Environ Model Assess
4:295-314
Pielke RA Sr (1998a) Climate predictions as an initial value problem.
B Am Meteorol Soc 79:2743-2746
Pielke RA Jr (1998b) Rethinking the role of adaptation in climate
policy. Global Environ Chang 8:159-170
Pielke RA Sr (2001) Earth system modeling -an integrated assessment
tool for environmental studies. In: Matsuno T, Kida
H (eds) Present and future of modeling global environmental
change: toward integrated modeling. Terrapub, pp 311-337
Pielke RA Sr (2002) Overlooked issues in the US National Climate
and IPCC assessments. Climatic Change 52:1-11
Pielke RA Jr, Downton MW (2000) Precipitation and damaging
floods: trends in the United States, 1932-1997. J Climate 13:
3625-3637
Pielke RA Sr, Guenni L (1999) Vulnerability assessment of water
resources to changing environmental conditions. IGBP Global
Change Newsletter 39:22-24
Pielke RA Sr, Uliasz M (1998) Use of meteorological models as input
to regional and mesoscale air quality models -limitations
and strengths. Atmos Environ 32:1455-1466
Pielke RA Jr, Sarewitz D, Byerly R, Jamieson D (1999) Prediction
in the earth sciences: Use and misuse in policy making. EOS
Trans Am Geophys Union 80:309ff
Pimentel D, Lach L, Zuniga R, Morrison D (1999) Environmental
and economic costs associated with non-indigenous species
in the United States. College of Agriculture and Life Sciences,
Cornell University, Ithaca, New York
Pitman A, Pielke RA Sr, Avissar R, Claussen M, Gash J, Dolman H
(1999) The role of the land surface in weather and climate: does
the land surface matter. IGPB Global Change Newsletter 39:
4-11
Plambeck EL, Hope C (1996) PAGE95: an updated valuation of the
impacts of global warming. Energ Policy 24:783-793
Plate EJ (1996) Risk management for hydraulic systems under hydrologicalloads.
In: 3rd IHP/IAHS Kovacs Colloquium on "Risk,
Reliability, Uncertainty and Robustness of Water Resources Systems".
UNESCO, Paris, 18 p
Postel SL, Daily GC, Ehrlich PR (1996) Human appropriation of renewable
fresh water. Science 271:785-788
Rabbinge R, van Oijen M (1997> Scenario studies for future agriculture
and crop protection. Eur J Plant Patholl03:197-201
Ramankutty N, Foley JA (1998) Estimating historical changes in globalland
cover: Croplands from 1700 to 1992. Global Biogeochem
Cy 13:997-1027
Rayner S,Malone EL (1997> Zen and the art of climate maintenance.
Nature 390:332-334
Riebsame W (1997> Atlas of the New West. W. W. Norton, New York
Rijsberman FRe (2000) World water scenarios: analyses. Earthscan
Publications, London
Risbey J,Stone P (1996) A case study of the adequacyofGCM simulations
for input to regional climate change. J Climate 9:1441-1448
Robertus AJ, Burns BR, Veblen TT (1991) Stand dynamics of Pinus
flexilis-dominated subalpine forests in the Colorado Front
Range. J Veg Sci 2:445-458
Rocky Mo_ain National Park Resources Management Plan
(1999) 1JS National Park Service, Estes Park, Colorado, USA
Rodda JC (1998) Hydrological networks need improving! In: Water:
a looming crisis. UNESCO, International Hydrological Programme,
Paris
Rosegrant MW (1997> Water resources in the twenty-first century:
challenges and implications for actions. International Food
Policy Research Institute, Washington DC
Rosegrant MW,Agcaoili-Saombilla M, Perez ND (1995) Global food
projections to 2020: implications for investment. International
Food Policy Research Institute (IFPRI), Washington DC
Rosentreter R (1994) Displacement of rare plants by exotic grasses.
In: Monsen SB, Kitchen SG (eds) Proceedings -ecology and
management of annual rangelands. USDA Forest Service
Inter. Forest and Range Experiment Station, Ogden, Utah,
pp 170-175
Rosenzweig C, Hillel D (1998) Climate change and the global harvest.
Oxford University Press, Oxford
Rosenzweig C, Phillip I, Goldberg R, Carroll I, Hodges T (1996) Potential
impacts of climate change on citrus and potato production
in the US. Agr Syst 52:455-479
Rotmans J, de Vries B (1997> Perspectives on global change: The
TARGETS approach. Cambridge University Press, Cambridge
Sala OE, Chapin FS III, Armesto JJ, Berlow E, Bloomfield J, Dirzo R,
Huber-Sanwaid E, Huenneke LF, Jackson RB, Kinzig A, Leemans
R, Lodge DM, Mooney HA, Oesterheld M, Poff NL, Sykes MT,
Walker BH, Walker M, Wall DH (2000) Global biodiversity scenarios
for the year 2100. Science 287:1770-1774
Samjaa R, Zophel U, Peterson J (2000) The impact of the vole Microtus
brandti on Mongolian steppe ecosystems. Marburger
Geogr Schriften 135:346-360
Sans6 B, Guenni L (1999) A stochastic model for tropical rainfall
at a single location. J Hydrol 214:64-73
Sarewitz D, Pielke RA Jr (1999) Prediction in science and policy.
Technol Soc 21:121-133
Sarewitz D, Pielke RA Jr, Byerly R (2000) Prediction, science, decision
making and the future of nature. Island Press, Washington
Schaller GB (1998) Wildlife of the Tibetan steppe. University of
Chicago Press, Chicago
Schellnhuber H-J (1998) Earth system management: the scope of
the challenge. In: Schellnhuber HJ, Wenzel V (eds) Earth system
analysis. Springer-Verlag, Berlin, pp 1-209

Schellnhuber H-J, Yohe G (1998) Comprehending the economic
and social dimensions of climate change by integrated assessment.
WCRP-Conference: Achievements, Benefits and Challenges,
Geneva, 179 p
Schellnhuber H -J, Block A, Cassel-Gintz M, Kropp J, Lammel G, Lass
W, Lienenkamp R, Loose C, Liideke M, Moldenhauer 0, Petschel-
Held G,PlBchl M,Reusswig F (1997) Syndromes of global change.
GAIA 60:19-34
Schulze RE (1989) Non-stationary catchment responses and other
problems in determining flood series: a case for a simulation
modelling approach. In: 4th South African National Hydrological
Symposium. Water Research Commission, Pretoria, RSA,
pp 135-157
Schulze RE (1995) Hydrology and agrohydrology: a text to accompany
the ACRU 3.00 Agrohydrological Modelling System. Water
Research Commission, Pretoria, South Africa, 552 p
Schulze RE (1997a) South African atlas of agrohydrology and climatology.
Water Research Commission, Pretoria, South Africa,
273 p
Schulze RE (1997b) Impacts of global climate change in hydrologically
vulnerable region: challenges to South African hydrologists.
Prog Phys Geog 21:113-136
Schulze RE (2000) Modelling hydrological responses to land use
and climate change: a southern African perspective. Ambio 29:
12-22
Schulze RE (2001) Risk, uncertainty and risk management in hydrology:
a conceptual framework and a South African perspective.
School of Bioresources Engineering and Environmental
Hydrology, University of Natal, Pietermaritzburg, RSA., 52 P
Schulze RE, Perks LA (2000) Assessment of the impact of climate
change on hydrology and water resources in South Africa. School
of Bioresources Engineering and Environmental Hydrology,
University of Natal, Pietermaritzburg, RSA, 118 P
Schulze RE, Hallowes JS, Lynch SD, Perks LA, Horan MJC (1998b)
Forecasting seasonal runoff in South Africa: A preliminary investigation.
Witwatersrand University, Johannesburg, RSA., P 19
Schuster PF, White DE, Naftz DL, DeWayne L (2000) Chronological
refmement of an ice core record at upper Fremont Glacier
in south central North America. J Geophys Res 105:4657-4666
Seckler D (1999) Revisiting the IWMI paradigm: increasing the efficiency
and productivity of water use. Water Brief NO.2, IWWI
Seckler D, Barker R, Anarasunghe U (1999) Water scarcity in the
twenty-first century. Int J Water Res Dev 15:29-42
ShackelyS, YoungP,Parkinson S, Wynne B (1998) Uncertainty,complexity
and concepts of good science in climate change modelling:
are GCMs the best tools? Climatic Change 38:159-205
Shiklomanov IA (1996) Assessment of water resources and water
availability in the world. Stockholm Environment Institute,
St. Petersburg, State Hydrological Institute
Shiklomanov I (2000) Water resouces and water use: present assessment
and outlook for 2025. In: Rijsberman FR (ed) World water
scenarios: analyses. Earthscan Publications, London, pp 160-203
Shukla J (1998) Predictability in the midst of chaos: a scientific
basis for climate forecasting. Science 282:728-731
Singer F (1995) Bighorn sheep in the Rocky Mountain National
Parks. In: La Roe Tea (ed) Our living resources: a report to the
nation on the distribution, abundance, and health of US plants,
animals, and ecosystems. USDI National Biological Service,
Washington DC, pp 332-336
Sivillo JK,AhlquistJE, TothZ (1997) An ensemble forecasting primer.
Weather Forecast 12:809-818
Skole D, Tucker C (1993) Tropical deforestation and habitat fragmentation
in the Amazon: satellite data from 1978 to 1988. Science
260:1905-1910
Slack JR, Landwehr JM (1992) Hydro-climatic data network: a US
Geological Survey streamflow dataset for the United States for
the study of climate variations, 1974-1988. United States Geological
Survey, Reston, VA 20192, 193 P
Smith K (1996) Environmental hazards. Routledge, London
Stahle DW, Cook ER, Cleaveland MK, Therrell MD, Meko DM,
Grissino-Mayer HD, Watson E, Luckman BH (2000) Tree-ring
data document 16th century megadrought over North America.
EOS Trans Am Geophys Union 81:121-125
References Part E 543
Stohlgren TJ (1999a) The Rocky Mountains. In: Mac MJ, Opler PA,
Puckett Haecker CE, Doran PD (eds) Status and trends of the
nation's biological resources, vol. 2. Biological Resources Division,
US Geological Survey, Reston, VA, pp 473-504
Stohlgren TJ (1999b) Global change impacts in nature reserves in
the United States. In: Aguirre-Bravo C, Franco CR (eds) North
American symposium toward a unified framework for inventory
and monitoring forest ecosystem resources, RMRS-P-12.
USDA Forest Service Rocky Mountain Forest and Range Experiment
Station, Fort Collins, CO, pp 5-9
Stohlgren TJ, Coughenour MB, Chong GW,Binkley D, Kalkhan MA,
Schell LD, Buckley DJ, Berry JK (1997) Landscape-scale gap
analysis: a complementary geographic approach for land managers.
Landscape Ecol12:155-170
Stohlgren TJ, Chase TN, Pielke RA Jr, Kittel TGF, Baron J (1998)
Evidence that local land-use practise influences regional climate
and vegetation patterns in adjacent natural areas. Global
Change Bioi 4:495-504
Stohlgren TJ, Owen AJ, Lee M (2000) Monitoring shifts in plant
diversity in response to climate change: a method for landscapes.
Biodivers Conserv 9:65-86
Storch H von (1994) Inconsistencies at the interface of climate
impact studies and global climate research. Max Planck Institute
for Meteorology Report 122:12
Storch H von, Zorita E, Cubasch U (1993) Downscaling of global
change estimates to regional scales: an application to Iberian
rainfall in wintertime. J Climate 6:1161-1171
Suter GW (1993) Ecological risk assessment. Lewis Publishers,
Michigan
Thompson DQ, Stuckey RL, Thompson EB (1987) Spread, impact,
and control of purple loosestrife (Lythrum salicaria) in North
American wetlands. US Fish and Wildlife Service ResearCh Report
2, Washington D.C., 55 P
Tobolski K (1982) Anthropogenic changes in vegetation of the
Gardno- Leba lowland, North Poland. Preliminary report. Acta
Palaeobot XXII:131-139
Tol RSJ, de Vos AF (1998) A Bayesian statistical analysis of the enhanced
greenhouse effect. Climatic Change 38:87-112
T6th F, Bruckner T, Filssel HM, Leimbach M, Petschel-Held G,
Schellnhuber H-J (1997) From impact to emission: the tolerable
windows approach. In: IPCC (ed) climate change and integrated
assessment models -bridging the gaps. CGER-Report-
1029-97. Center for Global Environmental Research, Tsukuba,
Japan
Townshend JR, Justic GC, Li W, Gurney C, McManus J (1991) Global
land-cover classification by remote sensing: present capabilities
and future possibilities. Remote Sens Environ 35:243-355
Trotter PC (1987) Cutthroat: native trout of the west. Colorado Associated
University Press, Boulder
Thrk JT, Campbell DH, Ingersoll GP, Clow DA (1992) Initial findings
of synoptic snowpack sampling in the Colorado RoCky Mountains.
US Geological Survey, Denver, CO
Thrner C (1975) Der EinfluB groBer Mammalier auf die interglaziale
Vegetation. Quartarpalaontologie 1:13-19
Turner BL, Skole DL, Sanderson S, Fischer G, Fresco L, Leemans R
(1995) Land-use and Land-cover change: science/research plan.
International Geosphere-Biosphere Programme, Human Dimensions
of Global Environmental Change Programme, Stockholm,
132 p
UNEP (2000) Assessing human vulnerability due to environmental
change; concepts, issues, methods and case studies. United
Nations Environmental Programme, Nairobi, Kenya
United Nations (1997) Comprehensive assessment of the freshwater
resources of the World. World Meteorological Organization,
Geneva
United Nations Population Division (1998) World population prospects:
the 1998 Revision. United Nations, New York
United States Geological Survey (1990) National water summary
1987 -Hydrologic events and water supply and use. In: USGS
National Water Summary Series WSP 2350, Denver, Colorado
United States Geological Survey (1993) National water summary
1990-1991- Hydrologic events and stream water quality. In: USGS
National Water Summary Series WSP 2400, Washington D.C.

544
References Part E
United States Geological Survey (1999) Streamflow information
for the next century: a plan for the national streamflow information
program of the USGS. USGS, pp 99-456
United States Geological Survey -EROS Data Center (1998) Globalland-cover
characterization based on l-km AVHRR data
for 1992/93. US Geological Survey Earth Resources Data Center,
Sioux Falls, SD
van Noortijk J, Kok M, Cooke RM (1997) Optimal Decisions that
Reduce Flood Damage along the Meuse: an Uncertainty Analysis.
In: French S,Smith JQ (eds) The practice of Bayesian Analysis.
Arnold, London, pp 151-172
Veblen TT, Lorenz DC (1991) The Colorado Front Range: A century
of ecological change. University of Utah Press, Salt Lake City
Veblen TT, Kitzberger T, Donnegan J (2000) Climatic and human
influences on fire regimes in ponderosa pine forests in the Colorado
Front Range. Ecol Appll0:U78-1195
Vitousek PM, Mooney HA, Lubchenco J, Melillo JM (1997) Human
domination of earth's ecosystems. Science 277:494-499
Vogel C (1998) Vulnerability and global environmental change.
LUCC Newsletter 3:15-19
V"6rosmarty CJ, Sahagian D (2000) Anthropogenic disturbance of
the terrestrial water cycle. Bioscience 50:753-765
V"6rosmarty CJ, Sharma KP, Fekete BM, Copeland AH, Holden J,
Marble J, Lough JA (1997) Storage and aging of continental
runoff in large reservoir systems of the world. Ambio 26:
210-219
V"6rosmarty CJ, Birkett C, Dingman L, Lettenmaier DP, Kim Y, Rodriguez
E,Emmitt GD (1999) HYDRA-SAT: hydrological applications
satellite. A report from the NASA POst-2002 Land Surface
Hydrology Planning Workshop, Irvine, California, April 12-14,
1999
Vorosmarty CJ (lead), Askew A, Grabs W, Barry RG, Birkett C, Doll
P, Grabs W, Hall A, Jenne R, Kitaev L, Landwehr J, Keeler M,
Leavesley J, Schaake J, Strzepek K, Sundarvel S, Takeuchi K,
Webster F (IAHS Ad Hoc Group on Global Water Data Sets)
(2001) Global water data: a newly endangered species. An opinion
editorial to AGU Eos Transactions 82:54-58
V"6rosmartyCJ, Fekete BM,MeybeckM, Lammers R (2000) A simulated
topological network representing the global system of rivers
at30-minute spatial resolution (STN-30). Global Biogeochem
Cy 14:599-621
Wang G, Eltahir EAB (2000) Biosphere-atmosphere interactions
over West Africa. 2. Multiple Equilibria. Q J Roy Meteor Soc 126:
1261-1280
Washburn AL (1973) Geocryology. A survey of periglacial processes
and environments. E. Arnold, London
Watson A, Lovelock J (1983) Biological homeostasis of the global
environment: the parable of Daisyworld. Tellus 49B:249-262
WBGU (German Advisory Council on Global Change) (1994)
World in transition: the threat to the soils. Economica, Bonn
WBGU (German Advisory Council on Global Change) (1999)
World in transition: ways towards sustainable management of
freshwater resources. Springer-Verlag, Berlin
Weber SL (2001) On homeostasis in Daisyworld. Climatic Change
48:465-485
Weeks J, Gutentag E (1988) Region 17: high plains. In: Back W, Rosenshein
J, Seaber P (eds) Hydrogeology. Geological Society of
America, Boulder, CO, 524 P
Weisberg PI, Baker WL (1995) Spatial variation in tree seedling and
Krummholz growth in the forest-tundra ecotone of Rocky Mountain
Nati9nal Park, Colorado, USA Arctic Alpine Res 27:116-129
Weyant J, Davidson O. Dowlabathi H, Edmonds J, Grubb M, Parson
EA. Richels R, Rotmans J, Shukla R, Tol RSJ, Cline W (1996)
Integrated assessment of climate change: an overview and comparison
of approaches and results. In: Bruce JP, Lee H, Haites
EF (eds) Climate change 1995. Economic and social dimensions
of climate change. Cambridge University Press, Cambridge
White R, Engelen G, Uljee I (1997) The use of constrained cellular
automata for high-resolution modelling of urban land-use dynamics.
Environ Plann B 24:323-343
Whitlock C (1993) Post-glacial vegetation and climate of Grand Teton
and southern Yellowstone National Parks. Ecol Monogr 63:173-198
Wikle CK, Berliner LM, Cressie N (1998) Hierarchical Bayesian
space-time models. Environ Ecol Stat 5:117-154
Wilhite DA (ed) (2000) Drought: a global assessment. Routledge
Press, London
Williams MW, Baron JS, Caine N, Sommerfeld RA, Sanford RL
(1996) Nitrogen saturation in the Rocky Mountains. Environ
Sci Technol 30:640-646
World Bank (1993) Water resources management: a World Bank
policy paper (prep. by Easter KW, Feder G, Le Moigne G, Duda
AM). The World Bank, Washington DC, 141 P
~ Bank (1996) Mexico -Water resources management project:
staff appraisal report. The World Bank, Water Resources Management
Project No. 15435, May 1996, 182 P
World Bank (1998a) Evaluating the Bank's water resources management
policy: an approach paper (prep. by Bruce-Konuah A).
The World Bank, Water Resources Management Program, October
1998, 6 P
World Bank (1998b) Morocco -Water resources management project:
staff appraisal report. The World Bank, Water Resources
Management Project No. 15760, January 1998, 76 P
World Resources Institute (1996) World resources: 1996-1997. Oxford
University Press, New York
Wu SM, Huang GH, Guo HC (1997) An interactive inexact-fuzzy
approach for multiobjective planning of water resources systems.
Water Sci Technol 36:235-242
Zhou HM (1995) Towards an operational risk assessment in flood
alleviation -theory, operationalization and application. Technical
University, Delft, Netherlands, 204 p