Dávid et al. (2025) Modeling the Start of Season Date of Hungarian Grasslands Using Remote Sensing Data and 10 Process-Based Models

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Atmosphere
• Year: 2025
• Date: 2025-12-30
• Authors: Réka Ágnes Dávid, Zoltán Barcza, Roland Hollós, Anikó Kern
• DOI: 10.3390/atmos17010049

Research Groups

Not explicitly stated in the provided text.

Short Summary

This study investigated the start of the growing season (SOS) for Hungarian grasslands (2000-2023) using MODIS NDVI and ten process-based models, demonstrating that pixel-level calibration and integration of local climate and soil information significantly enhance prediction accuracy.

Objective

• To investigate the start of the growing season (SOS) of Hungarian grasslands and evaluate the performance of ten process-based phenology models under different calibration strategies for accurate SOS prediction.

Study Configuration

• Spatial Scale: Hungarian grasslands (country-level, pixel-level analysis).
• Temporal Scale: 2000 to 2023.

Methodology and Data

• Models used: Ten process-based models of varying complexity (e.g., AGSI, AHSGSI, AGSIwSW, MGDDwPP, GDD). Model parameters were optimized using the differential evolution algorithm under generic (GEN), grassland-type (GEN GRASS), and pixel-level (PIX) calibration strategies.
• Data sources: MODIS NDVI data.

Main Results

• Under pixel-level (PIX) calibration, AGSI and AHSGSI models (driven by temperature, vapor pressure deficit, and photoperiod) achieved the lowest root mean square error (RMSE) of 3.3 days.
• Under generic (GEN) and grassland-type (GEN GRASS) calibration strategies, AGSIwSW (driven by temperature, soil water content, and photoperiod) performed best with RMSEs of 7.6 days and 6.3 days, respectively. MGDDwPP (driven by temperature and photoperiod) also showed an RMSE of 7.6 days under GEN.
• Considering the Akaike Information Criteria (AIC), the simplest GDD model (temperature-driven) was proposed for PIX, while MGDDwPP was identified as the best model for both GEN and GEN GRASS strategies.
• Residual analysis revealed strong co-variation between model errors and climate anomalies (most notably spring temperature and soil water content), enabling statistical corrections that reduced bias close to zero across all models.
• Integrating local climate and soil information into phenology models significantly enhances their accuracy for grassland SOS estimation in Central Europe.

Contributions

• Comprehensive evaluation of ten process-based phenology models and three distinct calibration strategies (generic, grassland-type, pixel-level) for grassland SOS prediction in Central Europe.
• Demonstration of the superior performance of pixel-level calibration, achieving significantly lower RMSE values compared to generic or grassland-type approaches.
• Identification of specific best-performing models (e.g., AGSI, AHSGSI, AGSIwSW, MGDDwPP) under different calibration strategies and complexity considerations (AIC).
• Highlighting the importance and effectiveness of integrating local climate and soil information for improving model accuracy and reducing bias through statistical corrections.

Funding

Not explicitly stated in the provided text.

Citation

@article{Dávid2025Modeling,
  author = {Dávid, Réka Ágnes and Barcza, Zoltán and Hollós, Roland and Kern, Anikó},
  title = {Modeling the Start of Season Date of Hungarian Grasslands Using Remote Sensing Data and 10 Process-Based Models},
  journal = {Atmosphere},
  year = {2025},
  doi = {10.3390/atmos17010049},
  url = {https://doi.org/10.3390/atmos17010049}
}