500m gridded surfaces for changes in climate suitability for coffee production in Risaralda, Colombia

We used 500m gridded historical and future climate surfaces for Risaralda, Colombia and coffee presences and absences to train species distribution models (suitability). Five methods were used: Generalized Boosting Model (GBM) (Friedman, 2001), Random Forest (RF) (Breiman, 2001), Maxent (Phillips et al., 2006), Generalized Linear Model (GLM) and Generalized Additive Model (GAM) (Guisan et al., 2002).我们采用哥伦比亚里萨拉尔达省的500米分辨率网格化历史与未来气候场数据，以及咖啡存在与缺失点位数据，以训练物种分布模型（适生性）。本次研究共使用五种建模方法：广义提升模型（Generalized Boosting Model, GBM）（Friedman, 2001）、随机森林（Random Forest, RF）（Breiman, 2001）、Maxent（Phillips等, 2006）、广义线性模型（Generalized Linear Model, GLM）以及广义加性模型（Generalized Additive Model, GAM）（Guisan等, 2002）。