22.01.2014 Views

PE EIE[R-Rg RESEARCH ON - HJ Andrews Experimental Forest

PE EIE[R-Rg RESEARCH ON - HJ Andrews Experimental Forest

PE EIE[R-Rg RESEARCH ON - HJ Andrews Experimental Forest

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

of the external quantities of S would be impossible<br />

in the context of isolated subsystems .<br />

What is needed is some translation of sub -<br />

system external variables into system externa l<br />

variables. Clymer and Bledsoe (1970) have<br />

addressed a similar question in consideratio n<br />

of "interfacing" two subsystems at the sam e<br />

level and use the term "slave model" to designate<br />

the set of equations for changing resolution.<br />

This term seems inappropriate in the<br />

context of composition of subsystem proper -<br />

ties into properties of the whole ; such a<br />

process is more master than slave . I have<br />

chosen the term "Ghost System i2 and notationally<br />

designated So as the integrating sub -<br />

system which derives the properties of th e<br />

whole from the properties of the parts . In its<br />

simplest form, So is a resolution expander of<br />

inputs and reducer of outputs . Whether it will<br />

need to assume a more complex form is yet t o<br />

be seen .<br />

In this manner we have, by statements (1 )<br />

and (2), defined the ecosystem at two organizational<br />

levels, and may elaborate lower level s<br />

by the device of decomposing subsystems i n<br />

the same manner . That is, if S may be defined<br />

at two levels, so may the elements of {Si}, an d<br />

the subsystems so defined, to any desire d<br />

degree of fineness . Ultimately, it is necessary<br />

to end with a behavioral model for all sub -<br />

systems, because it is this model that yield s<br />

explicit form to the relational expressions .<br />

Several points are of interest here . First, i t<br />

is not clear that a hierarchical decompositio n<br />

is always desirable . That is, a finer resolutio n<br />

may not best take the form of a decomposition<br />

of next most coarse form of interest .<br />

Hierarchical forms, however, have man y<br />

advantages and will be used wherever possible .<br />

Second, it is quite clear that a uniform resolution<br />

is unnecessary . That is, a very coarse<br />

resolution model of one subsystem may b e<br />

coupled with a fine resolution model of<br />

another, and such an arrangement will have<br />

many advantages . Not the least of these is th e<br />

advantage of manageability . If we want to<br />

take a closeup view of some parts of the system,<br />

there is no reason to, and many reason s<br />

not to, look at the rest of the system at th e<br />

2 After Koestler's "The Ghost in the Machine ."<br />

same degree of magnification . A hierarchical<br />

subsystem structure will be most conducive t o<br />

a variable resolution, but again, it is no t<br />

necessary .<br />

Application of the<br />

General Theory to the<br />

<strong>Forest</strong> Ecosystem<br />

With this introduction to the theoretica l<br />

background, we can now take a look at th e<br />

coarser ecosystem structures as we now perceive<br />

them . A diagrammatic convention i s<br />

established which will be recognized as simila r<br />

to Forrester's, but which differs in some<br />

important respects . Let boxes designate compartments<br />

(to be conceptualized as storage<br />

containers), let ovals or circles designate systems<br />

(or subsystems), and let diamonds designate<br />

a set of variables . Solid lines designate<br />

coupled flow variables and dotted lines represent<br />

control variables. As a convention ,<br />

diamonds are eliminated from the flow couplings,<br />

as these variables are apparent fro m<br />

the context, but identification of contro l<br />

variables is important . Occasionally these are<br />

also eliminated from a figure to reduc e<br />

clutter, and in the illustrations given here ,<br />

most control paths are eliminated for th e<br />

same reason .<br />

In figure 1 is depicted a representation of a<br />

system as a whole, with inputs and outputs . At<br />

this level, the entire system (the forest ecosystem)<br />

is defined according to (1) . In figure<br />

2, the same system is elaborated by identification<br />

of major subsystems, the terrestrial biota ,<br />

the aquatic biota and the hydrologic system .<br />

This involves definition of the system according<br />

to (2) and it should be emphasized at thi s<br />

point that there are very many ways in whic h<br />

this could be done .<br />

Implementation of this form (fig. 2) requires<br />

decisions regarding boundaries an d<br />

specifications of the nature of the coupling s<br />

between subsystems . An example will illustrate<br />

both points. Consider the uptake of<br />

water by higher plants . Question : Do we wish<br />

to consider this water as having transferred<br />

40

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!