EVALUATING THE COMPATABILITY OF SIMULATION MODEL OUTPUT AND OBSERVED DATA

Abstract

With the increasing popularity of computer simulation models as tools for environmental management and assessment, there is often a need to consider how well the output from such models represent reality. Statisticians evaluate the basic assumptions of least‐squares regression models by either computing statistics from or graphically evaluating residuals (observed minus predicted values). Even though there are usually a priori assumptions about residuals from simulation models we have found that their graphical or statistical evaluation can suggest model deficiencies. In this paper residuals from three simulation models were examined with the objective of suggesting limitations or desirable changes. The ability of simulation models to mimic observed data were used to provide support for visual appraisals.

Citation

THOMAS, J.M., EBERHARDT, L.L. and HANSON, W.C. (1982), "EVALUATING THE COMPATABILITY OF SIMULATION MODEL OUTPUT AND OBSERVED DATA", Kybernetes, Vol. 11 No. 3, pp. 195-203. https://doi.org/10.1108/eb005621