Experimental Design Matrix of Realizations for Optimal Sensitivity Analysis

Experimental design principles can be used in natural resource management for greater efficiency in sensitivity analysis. Traditionally, practical application of these designs have been limited due to the availability of very specific designs. An approach to determine a design matrix for sensitivity analysis for any case is proposed. The methodology uses an objective function to minimize the difference between first and second order sensitivity terms, that is, the optimized design permits the most reliable inference of first and second order sensitivity terms. Any number of input variables, response variables and case values are permitted. The output design consists of specific settings to run each realization. Some examples are shown and a framework for future research is presented.