Stratified Sampling Frame in GIS: A Computer Code for Latin Hypercube Sampling (LHS) Technique in Environmental Modelling

Abstract

Field sampling is a widely applied data collection method in environmental engineering. Sampling, which is supposed to provide adequate and unbiased representation of a study interest, is often faced with challenges such as choice of the minimum cost-effective sampling size, accurate location of sampling sites and statistical justification of both location of the sites and accuracy of the sampling process. Although there are many statistical sampling protocols in the literature, they largely emphasize on accuracy tests for unbiased sampling size. There is still a lack of statistical routines for unbiased sample locations. This has led to a lot of errors in spatial modeling of many environmental processes. In spite of the spatial modeling opportunities in GIS, there is currently no implementation of efficient statistical sampling frameworks within the GIS to improve sampling accuracy during primary data collection. This study developed a computer program for implementing a Latin Hypercube Sampling (LHS) protocol within a GIS environment. LHS is a robust and efficient statistical framework that has been widely used to guide the determination of sample sizes given a set of constraints. When combined with GIS software such as ArcView, LHS can improve sample size selection and subsequent unbiased strategic location of the samples in the landscape. This study developed a computer code for implementing LHS in ArcView. ArcView scripts were written in avenue language for various applications in environmental engineering. The scripts were then integrated into one unit to produce the LHS extension in ArcView GIS