Prediction of Water Resources in Marathwada Region using Spatial Data Analysis

Abstract

The Marathwada region of Maharashtra, India, is chronically afflicted by hydrological droughts and severe water scarcity, posing significant challenges to its agro-based economy. Effective water resource management requires accurate, timely, and accessible data. This research addresses the critical gap between complex geospatial data and an actionable understanding of water availability for local stakeholders. This study integrates Remote Sensing (RS), Geographical Information Systems (GIS), and spatial data mining (SDM) techniques to identify, analyze, and predict standing water resources across the Marathwada region. Primary data was acquired from Sentinel-2 satellite imagery, while secondary data included digitized maps and meteorological records. Water bodies were extracted using the Normalized Difference Water Index (NDWI) and subsequently vectorized to create a comprehensive spatial database. Data mining algorithms, including Decision Tree (DT), Naive Bayes (NB), Support Vector Machine (SVM), and Artificial Neural Networks (ANN), were employed for classification and predictive modeling. The results yielded a detailed, district-wise inventory of standing water resources, quantifying their number and surface area. A comparative analysis of the algorithms demonstrated that the Artificial Neural Network model achieved the highest classification accuracy (94.2%) in predicting water body persistence. The generated non-spatial tabular data provides an accessible resource for farmers and policymakers, facilitating informed decision-making for sustainable agriculture and water resource allocation. This work demonstrates the efficacy of a hybrid SDM-GIS approach in transforming raw satellite data into a valuable knowledge base for regional water management.