Guide to Hydrological Practices, 6th edition, Volume I - Hydrology.nl

I.2-20
GUIDE TO HYDROLOGICAL PRACTICES
hydrological uncertainty. In other words, the
probabilistic descriptions of uncertainty, based on
statistics of finite samples of hydrological
data, are uncertain in themselves. Reduction of
uncertainty about uncertainty is a key aspect of
taking full advantage of the information
contained in the data that the network will
generate.
The column in the middle of the structure, labelled
optimization theory, is often included taxonomically
as a part of socio-economic analysis. However,
even in the absence of socio-economics, the optimization
theory is often used in hydrological
network design. Thus, it is included here as a separate
component of the structure. A suite of
mathematical programmes, each with its own utility
and shortcomings, comprises optimization
theory, which is often referred to as operations
research. The context of the network-design problem
determines which, if any, of the mathematical
programmes can be used in a given situation. Often,
the choice between two or more network designs
must be made on the basis of judgement because
appropriate optimization tools either do not exist
or are too consuming of computer resources to be
efficient.
Atop the pyramid is decision theory, which is a
formal mechanism for integrating all of the underlying
components. The application of decision
theory in network design is not required – it is not
even possible in most circumstances. However, an
understanding of its pretexts and premises can
make a network designer more cognizant of the
impacts of his or her final decisions.
The left-hand side of the pyramid represents a
rather amorphous group of technologies under the
heading of socio-economic analysis. In addition to
social sciences and economics, this part of the
network-design structure also encompasses policy
science and even politics. The latter plays a very
important role in the realization of the potential
benefits of water and, thus, also in the ultimate
value of the data from the network. The left-hand
side of the structure is the part that usually receives
little rigorous consideration in the design of the
data network. This is probably attributable to two
causes: the subject matter is difficult to treat in an
objective, mathematical way; and to do so in a
substantive manner requires the synthesis of
inputs from many disciplines beyond those of
hydrology and water resources engineering. Thus,
a network design that includes a significant socioeconomic
analysis will probably be both expensive
and time-consuming.
Nevertheless, hydrological data-collection sites
are often installed to meet pressing social needs
and economic constraints with relatively little
thought to meeting long-term hydrological
information needs. Aside from meeting scientific
needs, data-collection sites may be installed
to assist water mangers in responding to extreme
events such as floods or droughts, allocating
water supplies among competing uses, or meeting
regulatory requirements. Sites operated for
these latter purposes may also lead to increased
hydrological understanding, but the resulting
network is by no means optimized for that
purpose.
2.4.1.2 Surrogate approaches
Since full-scale and complete network design is
either impossible or impractical in today’s world,
approaches that substitute surrogate measures,
objectives, or criteria are actually used to answer
the questions that comprise network design. For
example, a common substitution is to maximize
information content from a network in lieu of
optimizing the economic value of the data. Studies
have shown that, if information is used properly,
it can be expected to contribute to the economic
worth resulting from a decision. The more information,
the better the decision. However, the
economic impact of information is not linearly
related to its magnitude and the marginal worth of
additional information decreases with the amount
of information that is available. Thus, the use of
this surrogate criterion can lead a Hydrological
Service in the right direction if only sparse hydrological
information is available, but its use can
cause the collection of excess data if the region of
interest already has a reasonably adequate information
base.
Among the basic analytical techniques that take
advantage of surrogates in the design of networks
are cartographic analysis, correlation and
regression methods, probabilistic modelling,
deterministic modelling and regionalization
techniques. Each method has particular
applications and the choice depends on the limitations
of available data and the type of problem
under consideration. Quite often the different
techniques are combined in certain applications.
The Casebook on Hydrological Network Design
Practice (WMO-No. 324) presents applications of
these techniques as a means of determining
network requirements. Further examples are
contained in other publications (WMO/IHD
Project Report No. 12; WMO-Nos. 433, 580,
806).