Comparative Study of Fuzzy Clustering Algorithms for Mapping Sugarcane Ratoon Fields

Remote sensing is widely used in agriculture to monitor crop health, map harvested areas, and estimate soil properties like moisture, texture, and fertility. High temporal resolution satellite data, combined with fuzzy-based machine learning, enables precise crop mapping. This study focuses on detecting the harvesting period of sugarcane and ratoon crops at the Sugarcane Research Center, Pyinmana, using the Modified Possibilistic C-Mean (MPCM) classifier and the MPCM with Spatial Constraints (MPCM-S) approach. Classifier accuracy was assessed using the Mean Membership Difference (MMD) method. Results show that MMD values ranged from 0.00980 to 0.04117 for MPCM and 0.00196 to 0.03725 for MPCM-S. The highest variance within mapped ratoon fields was 0.000847 for MPCM and 0.0006937 for MPCM-S.