Yeşilköy et al. (2026) Linking drought indicators and crop yields through causality and information transfer: a phenology-based analysis

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-01-03
• Authors: Serhan Yeşilköy, Özlem Baydaroğlu, İbrahim Demir
• DOI: 10.1038/s41598-025-32185-6

Research Groups

• İstanbul Provincial Directorate of Agriculture and Forestry, Ministry of Agriculture and Forestry, İstanbul, Türkiye
• Oak Ridge Institute for Science and Education, Oak Ridge, TN, USA
• USDA-ARS, Adaptive Cropping Systems Laboratory, Beltsville, MD, USA
• University of Colorado-Boulder, Civil, Environmental and Architectural Engineering, Boulder, CO, USA
• NOAA, Global Systems Laboratory, Boulder, CO, USA
• National Academies of Sciences, Engineering, and Medicine, Washington, DC, USA
• River-Coastal Science and Engineering, Tulane University, New Orleans, LA, USA
• ByWater Institute, Tulane University, New Orleans, LA, USA

Short Summary

This study employs causal inference and information theory to identify the most representative drought indicators and meteorological parameters for corn yield in Iowa, revealing that specific indices like SPEI-9m and SPI-6m during silking, and SPI-9m and SPI-6m during doughing, have the strongest causal links to crop production.

Objective

• To identify the most representative drought indicators and meteorological parameters that causally influence corn yield, considering crop phenological stages, for improved crop yield prediction models.

Study Configuration

• Spatial Scale: Rainfed agricultural lands in Iowa, USA, across 9 distinct agricultural districts, with data aggregated to a state-level assessment.
• Temporal Scale: 19-year time series (2005 to 2023), focusing on corn silking and doughing phenological periods (July and August).

Methodology and Data

• Models used: Cross Convergent Mapping (CCM) for causal connection, Transfer Entropy (TE) for information transfer, Simplex projection for optimal embedding dimension and time delay, S-map for nonlinearity determination.
• Data sources:
  • Daily maximum air temperature (Tmax) and precipitation data from DAYMET (1-kilometer resolution).
  • Gridded drought indices (Evaporative Demand Drought Index (EDDI), self-calibrated Palmer Drought Severity Index (scPDSI), Standardized Precipitation and Evapotranspiration Index (SPEI), Standardized Precipitation Index (SPI) with 3-, 6-, 9-, 12-month scales) from gridMET (4-kilometer spatial resolution).
  • Crop (corn and soybean) phenological data from USDA-NASS database for Iowa.
  • County-level yield data aggregated to agricultural districts.

Main Results

• The analysis focused on corn yield due to the lack of nonlinearity in soybean datasets, which rendered CCM inappropriate for soybean causality.
• For corn yield during the silking period, maximum air temperature (Tmax) was the most influential factor (CCM: 0.35), while SPEI-9m (CCM: 0.34) and SPI-6m (CCM: 0.33) were the drought indices with the strongest causal relationships.
• During the doughing period, SPI-9m (CCM: 0.39) showed the strongest causal relationship with corn yield, followed by SPI-6m (CCM: 0.37) and scPDSI (CCM: 0.36).
• Transfer Entropy (TE) analysis indicated that SPI-9m exhibited the highest information transfer (0.12 bits) for corn yield in both silking and doughing periods.
• Phenological periods significantly influence the causal relationships between meteorological variables/drought indices and corn yield.
• Precipitation is identified as the primary meteorological factor influencing corn production, with simpler, precipitation-based indices (SPI) being more effective than complex multivariate drought indices.

Contributions

• First study to apply both Cross Convergent Mapping (CCM) and Transfer Entropy (TE) to link drought indicators and crop yields in Iowa, specifically considering crop phenological stages.
• Identified key drought indices (SPEI-9m, SPI-6m, SPI-9m, scPDSI) and meteorological parameters (Tmax, precipitation) that exhibit strong causal relationships with corn yield during critical growth stages.
• Demonstrated the crucial importance of incorporating phenological periods into crop yield prediction models for enhanced accuracy.
• Suggested that simpler, precipitation-only drought indices (like SPI) are more effective in capturing the underlying processes influencing corn yield compared to more intricate multivariate indices.
• Provided foundational insights for the development of more accurate corn yield forecast models for the US Corn Belt.

Funding

• University of Iowa Interdisciplinary Scalable Solutions for a Sustainable Future Project (Grant title: “Watershed-Level Multicriteria Quantification of Agricultural Sustainability for Iowa”).
• Agricultural Research Service (ARS) Research Participation Program administered by the Oak Ridge Institute for Science and Education (ORISE) (for Serhan Yeşilköy).
• NRC Research Associateship Program at the National Oceanic and Atmospheric Administration - Global System Laboratory (NOAA-GSL), administered by the Fellowships Office of the National Academies of Sciences, Engineering, and Medicine (for Özlem Baydaroğlu).

Citation

@article{Yeşilköy2026Linking,
  author = {Yeşilköy, Serhan and Baydaroğlu, Özlem and Demir, İbrahim},
  title = {Linking drought indicators and crop yields through causality and information transfer: a phenology-based analysis},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-025-32185-6},
  url = {https://doi.org/10.1038/s41598-025-32185-6}
}