Optimal spatial design for air quality measurement surveys: what criteria ?

摘要

In this work, we present a spatial statistical methodology to designrnbenzene air concentration measurement surveys at the urban scale. In a frst step, werndefne an a priori modeling based on an analysis of data coming from previous cam-rnpaigns on two diu000berent agglomerations. More precisely, we retain a modeling with anrnexternal drift which consists of a drift plus a spatially correlated residual. The statis-rntical analysis performed leads us to choose the most relevant auxiliary variables and torndetermine an a priori variogram model for the residual. An a priori distribution is alsorndefned for the variogram parameters, whose values appear to vary from a campaignrnto another. In a second step, we optimize the positioning of the measuring devices onrna third agglomeration according to a Bayesian criterion. Practically, we aim at fndingrnthe design that minimizes the mean over the urban domain of the universal krigingrnvariance, whose parameters are based on the a priori modeling, while accounting forrnthe prior distribution over the variogram parameters. Two optimization algorithms arernthen compared: simulated annealing and a particle flter based algorithm.