Development of a GIS Tool to Assess Climate Suitability of Crops

A GIS tool was developed to assess climate suitability of minor crops using a fuzzy logic system (FLS). The FLS consists of rule statements in natural language, e.g., “temperature is suitable” and “precipitation is suitable”. The rule statements were evaluated using fuzzy logic operations to accommodate limited information on climate and crop management for assessment of climate suitability. The parameter values of the FLS were obtained from the EcoCrop database operated by Food and Agriculture Organization, which allowed to avoid an iterative calibration process. Two types of FLSs, OR_F and AND_F models, were developed using fuzzy union and intersection operations between those statements, respectively. To evaluate the FLS, yield data of annual ryegrass at sites in Australia, Belgium, and the USA were collected from literature. Because a climate suitability index would represent a potential yield rather than an actual yield, the quantile regression analysis was performed to obtain a boundary line between climate suitability index and yields for site-years. Averages of climate suitability index and yield for an extended period were also determined by sites where disease risks were relatively small. Distribution of yields for climate suitability index differed by model although the parameter values were identical for all the models. In comparison with yields near the boundary line, e.g., within 95% confidence limit, the root mean square error of the OR_F model was smaller than that of AND_F model and EcoCrop model. The OR_F model explained a greater variation of average yields for an extended period than AND_F model and EcoCrop model did at sites where disease risks were relatively small. The FLS was implemented in C++ to develop a mapping system for climate surfaces at high spatial resolution, e.g., 1 km. In particular, OpenMP was applied to accelerate calculation of climate suitability index at the global scale.