Identification of areas in probability of being deforested, through logistic regression. study in Chihuahua (Mexico) for period 2007-2013

The objectives of this work were to identify areas in probability of being deforested in the west of Chihuahua, Mexico, and their causal variables, using a logistic regression model. The work was developed in a Geographic Information System, through the deployment and statistical regression between ten layers of geographic information (biophysical and socioeconomic) related to deforestation and deforested areas of four municipalities of the state during the period 2007-2013. The results indicate a good predictive power of the technique (McFadden's Pseudo-R2 of 0.26), and identify the degree of marginalization, land slope, land tenure and soil agrological aptitude as the variables with the biggest weight in the loss of forests. The findings facilitate the environmental management of forested areas, and the mitigation of causes of deforestation in already degraded areas.