Fuzzy machine learning based algorithms for mapping chickpea agricultural crop fields using sentinel-2 satellite data

Abstract

This study examined the NC and PCM classifiers, which classified and mapped the chickpea agricultural fields in Rajasthan, Nagaur district using two distinct techniques for selecting training samples. Two training parameter approaches are "mean" and "individual sample as mean" which have been tested in this study area. In processing temporal data, two approaches have been attempted to decrease spectral information. One is the Modified Soil Adjusted Vegetation Index 2 (MSAVI2), and the other is Class-Based Sensor Independent Modified Soil Adjusted Vegetation Index-2 (CBSI-MSAVI2).The MMD (Mean Membership Difference) and RMSE (Root Mean Square Error) approaches were employed to measure accuracy, In order to demonstrate that the classifier successfully identifies classes, cluster validity (SSE) was also carried out, and the variance parameter was computed to handle heterogeneity among chickpea crop fields. To obtain RMSE results, Sentinel-2 satellite data was classified, whereas Planet scope satellite data was used as reference data set.NC classifier applying ‘individual sample as mean’ on CBSI-MSAVI2 temporal indices gives the best result. RMSE, MMD, Variance, and SSE values for NC classifier using ‘individual sample as mean’ on CBSI-MSAVI-2 temporal indices were 0.01802, 0.00046, 0.13077, and 5.10003,respectively for the m=1.1.