Méthodes mathématiques pour estimer le champ de concentration d'un polluant gazeux à partir des valeurs mesurées aux points dispersés

Abstract

In order to evaluate a pollutant concentration field, when only few scattered values of pollutant concentration are available on a given territory, it is necessary to estimate the corresponding values in many other points. An approximation method to quantify the concentrations at the nodes of a rectangular mesh associated to the given territory was applied. These values were then used to generate concentration contours, in order to obtain an image of the concentration spatial distribution. This paper is concerned with the analysis of two types of approximation methods : statistical and analytical ones, which were applied to the AIRPARIF Network data. The Kriging statistical method, based on ordinary semivariograms can not be used in the particular case of AIRPARIF data. Among the analytical methods studied here, the thin plate interpolation spline offers the best results.