Mapping invasive plant Prosopis juliflora in arid land using high resolution remote sensing data and biophysical parameters

In this study, high resolution remote sensing data is used to extract Prosopis juliflora (P.juliflora), which is a major invader in the study area. Support Vector Machine (SVM) classification is applied to map this invader with Normalized Difference Vegetation Index (NDVI) as an additional parameter. Optimal kernel selection has been done for SVM classification, and a polynomial kernel. has been selected for the analysis. SVM polynomial kernel generated the overall accuracy of 70% and Kappa of 0.63. Classification results were compared with the results of conventional maximum likelihood classification (MLC). It was observed that the classification accuracy is improved from 68% to 74% when NDVI was used in MLC. But, when the SVM approach was used with NDVI, the accuracy dramatically increased to 93%. This is because the NDVI is a ratio based index, which introduces information about biophysical properties, thereby helping in better separation of P.juliflora.