Avaliação do modelo de mudança do uso do solo CLUE-S na presença de dependência espacial.

Land use changes impacts global environmental changes worldwide. Land use and cover changes (LUCC) are complex dynamical processes. In general, LUCC depends on geographical, environmental, social, economical and political factors with effects in diverse temporal and spatial scales. Great efforts are being made to create LUCC models, both theoretically and computationality. An important step in LUCC models is to quantify local influence of driving forces over the probability of occurrence of a specific land use type. This probability is usually quantified by a Logistic Regression Model. This paper compares a Logistic approach with a Spatial Lag Regression Model (SLRM). Generally, both approaches select different driving forces but SLRM incorporates spatial neighborhood information which is ignored by a logistic regression. Based on a lineal trend scenario for land use demand, simulations for land use changes for Coxim micro basin, Mato Grosso do Sul, were generated using CLUE-S program under a logistic and spatial regression models for the period 2001-2011. Simulation results for both approaches revealed elevated concordance indexes, high global accuracy and Kappa index, with respect to land use for the reference year of 2004. Differences were observed for spatial distribution for land use simulations for year 2011. The spatial regression approach simulation for 2011 reached less discrepancy to the land use demand for that year