A Separable Goal Programming Approach to Optimizing Multivariate Sampling Designs for Forest Inventory
Describes the application of a separable goal programming approach to stratified random sampling involving multiple objectives. Other attempts at solving this problem are also reviewed. The method is applied to a forest inventory problem in New Mexico involving six objectives and fourteen strata. Eight sampling allocations are presented to illustrate the sensitivity to alternate preference functions. Intercorrelations among goal criteria limit the effects of alternative preferences on resulting sampling allocations. All sampling allocations are guaranteed to be nondominated--something that goal programming does not (in general) provide. Forest Sci. 27:147-162.
No Supplementary Data
No Article Media