Mapping potential zones for rainwater harvesting in Gujarat’s arid frontier-Rann of Kutch

Student name: Mr Udit Saikia
Guide: Dr Priyanka Singh
Year of completion: 2025
Host Organisation: 2M Shan Engineering Solutions Pvt. Ltd.
Supervisor (Host Organisation): Dr Subash Prasad Rai

---

Abstract:

This paper presents an integrated GIS and machine learning approach for identifying potential RWH zones in the Rann of Kutch, Gujarat. Fifteen geo-environmental parameters were derived using ArcGIS and reclassified based on hydrological significance. Weights were allocated to each factor based on the Analytical Hierarchy Process (AHP). The reweighted thematic layers were subsequently analyzed using five supervised machine learning algorithms: Random Forest (RF), Gradient Boosted Trees (GBT), Boosted Regression Trees (BRT), Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP), implemented in Python. Performance of the models was assessed using statistical analysis and ROC curve. Among these, the MLP& BRT model demonstrated the highest predictive performance, effectively capturing the spatial variability and influencing parameters of suitable RWH zones. The present study underscores the potential of integrating AHP with machine learning techniques for high-resolution site suitability analysis.

KEYWORDS: Rainwater harvesting, GIS, Analytical Hierarchy Process (AHP), machine learning, Rann of Kutch.