Spatial decision support for selecting tropical crops and forages in uncertain environments

Abstract

Farmers in the developing world frequently find themselves in uncertain and risky environments: often having to make decisions based on very little information. Functional models are needed to support farmers tactical decisions. In order to develop an appropriate model, a comparison is carried out of potential modelling approaches to address the question of what to grow where. A probabilistic GlS model is identified in this research as a suitable model for this purpose. This model is implemented as the stand-alone Spatial Decision Support System (SDSS) CaNaSTA, based on trial data and expert knowledge available for Central America and forage crops. The processes and methods used address many of the problems encountered with other agricultural DSS and SDSS. CaNaSTA shows significant overlap with recommendations from other sources. In addition, CaNaSTA provides details on the likely adaptation distribution of each species at each location, as well as measures of sensitivity and certainty. The combination of data and expert knowledge in a spatial environment allows spatial and aspatial uncertainty to be explicitly modelled. This is an original approach to the problem of helping farmers decide what to plant where.