Toulouse School of Economics, Université Toulouse 1 Capitole
Land use predictions on a regular grid at different scales and with easily accessible covariates. Application to the Teruti-Lucas survey
It is widely accepted that land use is among the main human pressures on the environment, including climate change, biodiversity loss and pollution of water, soil and air. In this context, it is much needed to develop econometric and statistical tools that help to predict the possible land use patterns in order to improve our understanding of the causes and consequences of these phenomena. In this presentation, we consider the problem of land use modeling using point level data such as the Teruti-Lucas survey and some easily accessible explanatory variables. We analyze the components of the prediction error at different spatial scales using a synthetic data set constructed from the Teruti-Lucas points in the Midi-Pyrénées region and a five categories land use classification. We explore the link between the prediction errors and the Gini-Simpson impurity index of the vector of probabilities of each category.
The study first shows that the number of points in the Teruti-Lucas survey is quite enough for estimating the probabilities of each land use category with a good quality. Furthermore it reveals that, contrary to usual practice, when the objective is to predict land use at aggregated levels, land use probabilities should be estimated at more locations where explanatory variables are available rather than restricting to the initial Teruti-Lucas points. Indeed this strategy borrows strength from the knowledge of the explanatory variables which may be heterogeneous in a given Teruti-Lucas segment. Finally, guidelines for constructing the grid of locations for estimation are given from the analysis of the heterogeneity of each explanatory variable.