Laserfiche WebLink
Simplicity is a priraarv consideration in modeling, and is generally <br />achieved if the objectives of a model are precisely kno~rn. An effective <br />model vrill not be overly complicated, will not have sophisticated data <br />requirements, and gill com7unicate easily Frith the practicing professional. <br />By the same token, an effective model cannot be expected to solve problems <br />of minute detail. <br />Perhaps the most important fzctor in model design and use has to do <br />Frith proper selection of the range over erhich the model is valid. Again, <br />suitaule boundaries or limits for the nod el are related to concise objec- <br />tives. If a model is applied in situations outside the range of conditions <br />for which it i:as designed, then the results are often misleading. <br />:Model zpplication of particular interest in this investigation is the <br />prediction of the impact produced by a chanoe imposed on the system and <br />the appror.imate extent of that impact. In order to rely on the model, one <br />must be satisfied that it is an acceptable representation of the real <br />system, and that r:odel behzvior corresponds to that of the real system. <br />i•;hile several so-called "objective" tests have been developed for <br />model vzlidation, no completely objective test exists, since all depend <br />• eventually on some underlying subjective premise (Forrester, 1969). It <br />~- should be borne in mind that the dznoer in anv ouantitative model-valida- <br />lion procedure is that it takes on an "aura of a.uthenticit~~" which may <br />• <br />lead the iner.oerienced c.odeler to for^et the underlying subjective <br />asses„otions. Although sore investigators may challenge the idea that at <br />some point "objective" model validation procedures rest on a subjective- <br />foundatien, primary confidence must depend on: (1) how acceptable or <br />plausible the model is in describing natural processes, znd (2) the <br />reasonable assumption tilat "if all the necessary components are adequztely <br />described and properly interrelated, the model system cannot do other than <br />behave as it should"(Forrester, 1969). Qecause much of the content of <br />complex natural system models is derived fror~~ nonGuantitative sources, the <br />defense of such models ultima*_ely oust rest in careful subjective evalua- - <br />tion of their performance by experienced professionals trho are familiar <br />with these syste.-.s. <br />G-22 <br />