A Collaborative Platform for Soil Organic Carbon Inference Based on Spatiotemporal Remote Sensing Data
🤖 AI Summary
To address challenges in large-scale, dynamic soil organic carbon (SOC) monitoring—including strong spatiotemporal heterogeneity, difficulty in fusing multi-source data, and limited model interpretability—this study develops WALGREEN, a cloud-based platform. Methodologically, WALGREEN introduces a novel collaborative reasoning architecture explicitly designed for SOC’s spatiotemporal heterogeneity, integrating multi-temporal remote sensing data (Sentinel-2 via Google Earth Engine) with heterogeneous in-situ soil samples and establishing a closed-loop public–private data updating mechanism. It enables cross-scale, interpretable SOC stock forecasting using machine learning models (XGBoost, Random Forest), coupled with OpenLayers-based geospatial visualization and a Thymeleaf-MVC web framework. Evaluation yields an R² of 0.89 and significantly improved spatial coverage. The platform has been deployed to support annual assessments across three national agricultural carbon sink pilot zones, reducing response time to within 72 hours.
Application Category
📝 Abstract
Soil organic carbon (SOC) is a key indicator of soil health, fertility, and carbon sequestration, making it essential for sustainable land management and climate change mitigation. However, large-scale SOC monitoring remains challenging due to spatial variability, temporal dynamics, and multiple influencing factors. We present WALGREEN, a platform that enhances SOC inference by overcoming limitations of current applications. Leveraging machine learning and diverse soil samples, WALGREEN generates predictive models using historical public and private data. Built on cloud-based technologies, it offers a user-friendly interface for researchers, policymakers, and land managers to access carbon data, analyze trends, and support evidence-based decision-making. Implemented in Python, Java, and JavaScript, WALGREEN integrates Google Earth Engine and Sentinel Copernicus via scripting, OpenLayers, and Thymeleaf in a Model-View-Controller framework. This paper aims to advance soil science, promote sustainable agriculture, and drive critical ecosystem responses to climate change.
Problem
Research questions and friction points this paper is trying to address.
Enables large-scale soil organic carbon monitoring despite spatial variability
Overcomes limitations in SOC inference using machine learning and diverse data
Provides accessible platform for SOC analysis to support sustainable land management
Innovation
Methods, ideas, or system contributions that make the work stand out.
Uses machine learning for SOC predictive models
Cloud-based platform with user-friendly interface
Integrates Google Earth Engine and Sentinel data
🔎 Similar Papers
No similar papers found.
