Application of support vector machines for automatic compliance monitoring of the conservation reserve program (CRP) tracts

Abstract

We study an automatic compliance monitoring approach for examining United States Department of Agriculture (USDA)'s Conservation Reserve Program (CRP) tracts. In this work, CRP compliance monitoring is aimed at checking whether each CRP tract is compliant with contract stipulations. The proposed algorithm incorporates both one-class and two-class support vector machines (SVMs) for CRP classification. Specifically, one-class SVM (OCSVM) is first used to separate minor nonCRP outliers from the majority which is assumed to be the real CRP coverage. Then OCSVM results are used to train a two-class SVM (TCSVM) to further refine the CRP classification result. We use the CRP reference data as the baseline to evaluate CRP classification results. A high consistence between the CRP classification result and the CRP reference data indicates good compliance, while a low consistency reveals possible noncompliance. Simulation results show that the proposed method provides reliable information for CRP compliance monitoring.