Monitoring Aquatic Weeds in Indian Wetlands Using Multitemporal Remote Sensing Data with Machine Learning Techniques
Vahid Akbari, Morgan Simpson, Savi Maharaj, Armando Marino, Deepayan Bhowmik, G. Nagendra Prabhu, Srikanth Rupavatharam, Aviraj Datta, A. Kleczkowski, J. Alice R. P. Sujeetha
Abstract
The main objective of this paper to show the potential of multitemporal Sentinel-1 (S-1) and Sentinel-2 (S-2) for detection of water hyacinth in Indian wetlands. Water hyacinth (Pontederia crassipes, also called Eichhornia crassipes) is one of the most destructive invasive weed species in many lakes and river systems worldwide, causing significant adverse economic and ecological impacts. We use the expectation maximization (EM) as a benchmark machine learning algorithm and compare its results with three supervised machine learning classifiers, Support Vector Machine (SVM), Random Forest (RF), and k-Nearest Neighbour (kNN), using both synthetic aperture radar (SAR) and optical data to distinguish between clean and infested waters.