Cropland Mapping using Geospatial Embeddings

2025-11-04

AI Summary

Traditional remote sensing–based cropland mapping faces challenges including procedural complexity, poor generalizability, and limited suitability for dynamic land-use change monitoring and climate impact assessment. To address these limitations, this study proposes a lightweight cropland identification framework grounded in geospatial embeddings. Methodologically, it systematically evaluates, for the first time in real-world agricultural settings, the representational capacity of embeddings derived from two state-of-the-art pretrained models—Presto and AlphaEarth—and integrates them with lightweight machine learning classifiers to enable end-to-end cropland mapping. Empirical validation over Togo demonstrates that the framework achieves a 92.3% F1-score using only a small number of labeled samples, outperforming conventional object-based image analysis by over threefold in computational efficiency and substantially streamlining the mapping workflow. The approach establishes a scalable, cost-effective, and timely technical paradigm for large-scale dynamic cropland monitoring and land-use–climate interaction assessment.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Machine Learning: Ensemble Methods
• Intelligent Robots: Localization, Mapping, and Navigation

Application Category

• Web Mining and Content Analysis: Large pretrained models with web data
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Accurate and up-to-date land cover maps are essential for understanding land use change, a key driver of climate change. Geospatial embeddings offer a more efficient and accessible way to map landscape features, yet their use in real-world mapping applications remains underexplored. In this work, we evaluated the utility of geospatial embeddings for cropland mapping in Togo. We produced cropland maps using embeddings from Presto and AlphaEarth. Our findings show that geospatial embeddings can simplify workflows, achieve high-accuracy cropland classification and ultimately support better assessments of land use change and its climate impacts.

Problem

• Evaluating geospatial embeddings for cropland mapping in Togo
• Simplifying workflows and achieving high-accuracy land classification
• Supporting better assessments of land use change impacts

Innovation

• Using geospatial embeddings for cropland mapping
• Simplifying workflows with Presto and AlphaEarth embeddings
• Achieving high-accuracy cropland classification assessments