ABSTRACT
This chapter provides an overview of a new remote sensing technique – called within-field anomaly detection (WFAD) – that was specifically developed for regions such as South Africa, where the extent of salt-affected areas is small. Potentially, the WFAD should also accommodate the limitations posed by the spectral ambiguity of direct and indirect indicators of salt accumulation and waterlogging. Error matrices show the accuracies achieved by the WFAD method for each irrigation scheme. WFAD was explicitly developed for regions such as South Africa where the extent of salt-affected areas are relatively small in extent. The WFAD was applied to nine different irrigation schemes in South Africa representing sufficient climatic, environmental and agricultural variability to test for model robustness. A robust WFAD approach to quantify salinity and waterlogging was applied to nine different irrigation systems in South Africa. The WFAD method uses a geographical object-based image analysis approach with a rule-based classification scheme.
