Remote Sensing Reveals Adoption of Sustainable Rice Farming Practices Across Punjab, India

2025-07-11

AI Summary

Rice irrigation consumes 24–30% of global freshwater, intensifying water stress and undermining progress toward Zero Hunger. While sustainable practices—such as direct-seeded rice (DSR) and alternate wetting and drying (AWD)—can reduce water use by 20–40% without yield loss, their spatial distribution remains poorly mapped, hindering evidence-based policy. To address this, we develop the first DSR mapping system that does not rely on prior knowledge of planting dates, integrating Sentinel-1 SAR imagery with PRANA field survey data and employing machine learning for high-accuracy classification (F1-score = 78%). Applied across Punjab, India, the system maps DSR cultivation on ~3 million agricultural fields. District-level predictions strongly align with official statistics (Pearson correlation = 0.77; Rank-Biased Overlap = 0.77), substantially closing a critical data gap and enabling integrated, evidence-driven governance of water–food security trade-offs.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Distributed Machine Learning & Federated Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics

Abstract

Rice cultivation consumes 24-30% of global freshwater, creating critical water management challenges in major rice-producing regions. Sustainable irrigation practices like direct seeded rice (DSR) and alternate wetting and drying (AWD) can reduce water use by 20-40% while maintaining yields, helping secure long-term agricultural productivity as water scarcity intensifies - a key component of the Zero Hunger Sustainable Development Goal. However, limited data on adoption rates of these practices prevents evidence-based policymaking and targeted resource allocation. We developed a novel remote sensing framework to monitor sustainable water management practices at scale in Punjab, India - a region facing severe groundwater depletion of 41.6 cm/year. To collect essential ground truth data, we partnered with the Nature Conservancy's Promoting Regenerative and No-burn Agriculture (PRANA) program, which trained approximately 1,400 farmers on water-saving techniques while documenting their field-level practices. Using this data, we created a classification system with Sentinel-1 satellite imagery that separates water management along sowing and irrigation dimensions. Our approach achieved a 78% F1-score in distinguishing DSR from traditional puddled transplanted rice without requiring prior knowledge of planting dates. We demonstrated scalability by mapping DSR adoption across approximately 3 million agricultural plots in Punjab, with district-level predictions showing strong correlation (Pearson=0.77, RBO= 0.77) with government records. This study provides policymakers with a powerful tool to track sustainable water management adoption, target interventions, and measure program impacts at scale.

Problem

• Monitoring sustainable rice farming practices adoption remotely
• Addressing groundwater depletion in Punjab using satellite data
• Lack of data for evidence-based water management policies

Innovation

• Remote sensing monitors sustainable rice farming
• Sentinel-1 imagery classifies water management practices
• Scalable framework maps DSR adoption in Punjab