Near Real-Time Monitoring and Time Series Analysis of Water Bodies in Coimbatore District Using Google Earth Engine and Machine Learning Algorithms
Sakthivel R
*
Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, India.
Geethakarthi A
Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, India.
Vaishnava Devi S
Department of Civil Engineering, Kumaraguru College of Technology, Coimbatore, India.
Janani N
Agricultural Engineering College & Research Institute, Tamil Nadu Agricultural University, Coimbatore, India.
Maheshkumar P
Agricultural Engineering College & Research Institute, Tamil Nadu Agricultural University, Coimbatore, India.
Vasanthakumar M
Agricultural Engineering College & Research Institute, Tamil Nadu Agricultural University, Coimbatore, India.
*Author to whom correspondence should be addressed.
Abstract
This study presents a Google Earth Engine (GEE)-based web application for near real-time monitoring and time series analysis of water bodies in Coimbatore district using Sentinel-1A SAR data. The application generates dynamic water area maps from 2015 to the present and enables users to select individual lakes and tanks from a dropdown menu for focused analysis. Validation was carried out using five major lakes, where historical water spread data derived from Sentinel-2 NDVI and NDWI indices were compared with SAR-derived results, showing high accuracy. A case study on Kadamparai Dam involved extracting time series data from the platform and applying multiple machine learning models for prediction; the Random Forest Regressor (R2=0.728, MSE=0.005, MAE=0.059) achieved the highest accuracy among other ML algorithms. The tool supports climate change impact assessment and hydrological analysis, and also aids agricultural planning by offering timely and critical insights into water availability trends.
Keywords: Water body monitoring, time series analysis, random forest, GEE