Identifying Local Deforestation Patterns Using Geographically Weighted Regression Models

This study aimed at identifying drivers and patterns of deforestation in Mexico by applying Geographically Weighted Regression (GWR) models to cartographic and statistical data. We constucted a nation-wide multidate GIS database incorporating digital data about deforestation from the Global Forest Change database (2000–2013); along with ancillary data (topography, road network, settlements and population disribution, socio-economical indices and government policies). We computed the rate of deforestation during the period 2008–2011 at the municipal level. Local linear models were fitted using the rate of deforestation as dependent variable. In comparison with the global model, the use of GWR increased the goodness-of-fit (adjusted R2) from 0.20 (global model) to 0.63. The mapping of GWR models’ parameters and its significance, anables us to highlight the spatial variation of the relationship between the rate of deforestation and its drivers. Factors identified as having a major impact on deforestation were related to topography, accessibility, cattle ranching and marginalization. Results indicate that the effect of these drivers varies over space, and that the same driver can even exhibit opposite effects depending on the region.