IrrMap: A Large-Scale Comprehensive Dataset for Irrigation Method Mapping

2025-05-13

AI Summary

This work addresses the lack of large-scale, cross-regional, and long-term benchmark datasets for fine-grained irrigation method identification in agricultural remote sensing. To this end, we introduce IrrMap—the first open-source irrigation mapping dataset—covering 1.68 million farms and 14.1 million acres of cropland across the U.S. West from 2013 to 2023. IrrMap comprises 1.1 million multi-source georeferenced image patches (224×224 GeoTIFFs), integrating multi-temporal Landsat/Sentinel imagery, crop type, land use, and vegetation indices. It features the first cross-regional, decadal-scale manual annotations of fine-grained irrigation types (e.g., center-pivot sprinkler, drip, flood). We propose a geographically aligned quality control framework and a scalable data generation pipeline to support regional transfer and task generalization. Accompanying resources—including PyTorch-ready splits, baseline models, and fully documented code—are released on GitHub and Hugging Face, significantly advancing irrigation pattern distribution modeling and deep learning–based semantic interpretation.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data

Abstract

We introduce IrrMap, the first large-scale dataset (1.1 million patches) for irrigation method mapping across regions. IrrMap consists of multi-resolution satellite imagery from LandSat and Sentinel, along with key auxiliary data such as crop type, land use, and vegetation indices. The dataset spans 1,687,899 farms and 14,117,330 acres across multiple western U.S. states from 2013 to 2023, providing a rich and diverse foundation for irrigation analysis and ensuring geospatial alignment and quality control. The dataset is ML-ready, with standardized 224x224 GeoTIFF patches, the multiple input modalities, carefully chosen train-test-split data, and accompanying dataloaders for seamless deep learning model training andbenchmarking in irrigation mapping. The dataset is also accompanied by a complete pipeline for dataset generation, enabling researchers to extend IrrMap to new regions for irrigation data collection or adapt it with minimal effort for other similar applications in agricultural and geospatial analysis. We also analyze the irrigation method distribution across crop groups, spatial irrigation patterns (using Shannon diversity indices), and irrigated area variations for both LandSat and Sentinel, providing insights into regional and resolution-based differences. To promote further exploration, we openly release IrrMap, along with the derived datasets, benchmark models, and pipeline code, through a GitHub repository: https://github.com/Nibir088/IrrMap and Data repository: https://huggingface.co/Nibir/IrrMap, providing comprehensive documentation and implementation details.

Problem

• Large-scale dataset for irrigation method mapping
• Multi-source satellite imagery with auxiliary agricultural data
• Standardized ML-ready geospatial data for deep learning

Innovation

• Large-scale multi-resolution satellite imagery dataset
• Standardized ML-ready GeoTIFF patches with dataloaders
• Complete pipeline for dataset generation and extension