The yield strikes back: enhancing the transferability of field scale wheat and barley yield models by leveraging Sentinel-1/2
Accurate and transferable crop monitoring from remote sensing remains challenging because vegetation signals are strongly affected by phenological asynchrony, climatic variability, and sensor-specific responses. Existing approaches rely on local calibrated relationships , limiting their effectivenes...
| Autores: | , , , , , , , , , , , , , |
|---|---|
| Tipo de documento: | artigo |
| Estado: | Versão publicada |
| Data de publicação: | 2026 |
| País: | España |
| Recursos: | Universitat de Lleida (UdL) |
| Repositório: | Repositori Obert UdL |
| OAI Identifier: | oai:dnet:.___________::6990ec5d5c8373b73a39e7188528804f |
| Acesso em linha: | https://doi.org/10.1016/j.jag.2026.105140 https://hdl.handle.net/10459.1/469917 |
| Access Level: | Acceso aberto |
| Palavra-chave: | Yield Wheat Barley Sentinel-1 Sentinel-2 |
| Resumo: | Accurate and transferable crop monitoring from remote sensing remains challenging because vegetation signals are strongly affected by phenological asynchrony, climatic variability, and sensor-specific responses. Existing approaches rely on local calibrated relationships , limiting their effectiveness in data-sparse regions. This study investigates whether models calibrated on high-quality localized reference data can generalize to other regions by stabilizing sensor–biophysical relationships. The proposed methodology integrates two components (i) thermal time normalization based on growing degree days (GDD) to reduce phenology-driven variability, and (ii) physically motivated optical and optical–SAR fusion indices designed within this normalized framework to enhance the consistency of learned relationships across contrasting environments. The approach was evaluated through within-region cross-season, and cross-region experiments. Results show that GDD normalization improves performance relative to calendar-based approaches by up to 35%. In cross-season validation, fusion-based linear models achieved R2 > 0.60 for wheat and 0.65 in barley. Cross-region validation shows reduced but meaningful transferability, with both crops reaching R2 ≈ 0.45, and fusion reducing RMSE by ∼ 200 kg ha-1 compared to optical-only models. Machine-learning models did not improve generalization over simple parametric fits. These findings confirm that stabilizing phenological and multi-sensor relationships is critical for transferring models from data-rich to data-limited areas, providing a foundation for scalable, global agricultural monitoring. |
|---|