Constrained Optimization of Spatial Sampling in a Model-based Setting using SANOS Software

摘要

As a basis for all subsequent spatial analyses, the sampling scheme is crucial to the quality of all geostatistical surveys. Spatial Simulated Annealing is an optimisation algorithm for sampling in model-based geostatistics. The algorithm can distinguish between a variety of optimisation criteria, and optimises the sampling scheme by individual placement of the observations at optimal locations. It takes into account sampling constraints such as buildings and roads. The algorithm can incorporate preliminary observations in the optimised sampling scheme. The algorithm is implemented in the SANOS software. SANOS is a windows-based, user-friendly program that allows for interactive optimisation of the sampling scheme. Implemented optimisation criteria are (ⅰ) optimising variogram estimation, (ⅱ) minimising kriging variance, (ⅲ) optimising detection of maximum contamination in multivariate studies, and (ⅳ) minimising the collocated ordinary cokriging variance. SANOS will be especially effective in optimising cost-efficiency of spatial surveys in realistic conditions. The program is freely available from the World Wide Web.