A minimum cross-entropy approach to disaggregate agricultural data at the field level

Abstract

Agricultural policies have impacts on land use, the economy, and the environment and their analysis requires disaggregated data at the local level with geographical references. Thus, this study proposes a model for disaggregating agricultural data, which develops a supervised classification of satellite images by using a survey and empirical knowledge. To ensure the consistency with multiple sources of information, a minimum cross-entropy process was used. The proposed model was applied using two supervised classification algorithms and a more informative set of biophysical information. The results were validated and analyzed by considering various sources of information, showing that an entropy approach combined with supervised classifications may provide a reliable data disaggregation.