Predictive Modelling of Contagious Deforestation in the Brazilian Amazon

摘要: Tropical forests are diminishing in extent due primarily to the rapid expansion of agriculture, but future magnitude and geographical distribution tropical deforestation is uncertain. Here, we introduce a dynamic spatially-explicit model that predicts potential spatial pattern Amazon deforestation. Our differs from previous models three ways: (1) it probabilistic quantifies uncertainty around predictions parameters; (2) overall rate emerges “bottom up”, as sum local-scale driven by local processes; (3) contagious, such increases through time if adjacent locations deforested. For scenarios evaluated–pre- post-PPCDAM (“Plano de Acao para Protecao e Controle do Desmatamento na Amazonia”)–the parameter estimates confirmed near roads already deforested areas significantly more likely be less protected areas. Validation tests showed our correctly predicted accumulates over time, there very high surrounding exact sequence which pixels The under pre-PPCDAM (assuming no change values to, for example, changes government policy), annual rates would halve between 2050 compared 2002, although this partly reflects reliance on static map road network. Consistent with other models, scenario, states south east Brazilian have probability losing nearly all forest outside 2050. This strong scenario. Contagious spread along lacking formal protection could allow reach core, currently experiencing low its isolation.