ANNOUNCEMENTS
Water hyacinth (Pontederia crassipes) infestation poses a significant threat to urban lakes, impacting water quality, biodiversity, and ecosystem services. This study investigates the distribution of water hyacinth across lakes within major Indian urban agglomerations such as Hyderabad, Bengaluru, Chennai, Mumbai, and Kolkata over the periods from 2021 to 2022 and from 2023 to 2024, using Sentinel 2 Level 2A satellite imagery combined with advanced geospatial techniques.
A comprehensive analysis of 211 lakes identified 40 lakes significantly affected by water hyacinth, with Kolkata exhibiting the highest infestation rate. To accurately map water hyacinth coverage, multiple machine learning classifiers which includes Random Forest (RF), Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), Convolutional Neural Network (CNN), and K- Nearest Neighbors (KNN) were evaluated. The Random Forest model, configured with 100 decision trees (n_estimators equal to 100), achieved the highest classification accuracy of 0.925, outperforming XGBoost with 0.883, CNN with 0.829, SVM with 0.781, and KNN with 0.758.
Technical validation was conducted by comparing RF classified outputs with manually digitized water hyacinth extents for two lakes, namely Fox Sagar Lake and Ana Sagar Lake, yielding close agreement and minimal percentage differences of 3.91 percent and 0.73 percent, respectively. These results demonstrate the robustness and reliability of the RF algorithm for large scale, automated detection of water hyacinth.
The findings emphasize the critical need for integrated management strategies to control invasive species and preserve the ecological health of urban water bodies. This research contributes valuable insights and methodological approaches for effective monitoring and sustainable management of aquatic vegetation in rapidly urbanizing landscapes.
Key words: Water Hyacinth, Random Forest, Urban Agglomeration.
