RSCM: A Bayesian remote sensing-integrated crop model software framework for yield estimation

Abstract

RSCM is an open-source, process-based crop simulation framework that integrates satellite-derived vegetation indices directly into parameter estimation via Bayesian Maximum A Posteriori (MAP) optimization. This approach automates estimation of leaf area index, aboveground dry matter, and grain yield without extensive ground-based calibration. The system couples a Python data interface with a high-performance C simulation engine, enabling efficient regional-scale processing. Validation using independent datasets for rice, wheat, and maize demonstrated robust performance: yield Model Efficiency reached 0.99, with a minimum ME of 0.67 for wheat. The Bayesian prior regularization constrained parameter estimates while maintaining predictive accuracy. Regional applications in South Korea, North Korea, and the U.S. Corn Belt captured spatial yield gradients and inter-annual variability across millions of pixels. RSCM provides a computationally efficient tool bridging process-based modeling and remote sensing for precision agriculture and food security monitoring.

Plant phenotyping relevance

衛星データと作物モデルを統合し、LAI・地上部乾物量・収量という植物形質を推定するソフトウェア手法を開発・検証しており、形質取得・推定法が研究の中心である。

Code and data availability

公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。

No evidence-backed public reproduction asset is currently recorded.