Arai et al. (2026) Crop Yield Estimation with MODIS Derived Normalized Difference Vegetation Index and Comparative Study on Crop Yield Prediction Among Linear Regression, Random Forest and Gradient Boosting as Well as CatBoost

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

Identification

• Journal: Remote Sensing
• Year: 2026
• Date: 2026-09-10
• Authors: Kohei Arai, Sara Sanwal
• DOI: 10.3390/rs18183107

Research Groups

• Indian Institute of Technology (IIT) Delhi
• National Centre for Medium Range Weather Forecasting (NCMRWF)
• University of Agricultural Sciences (UAS)

Short Summary

This paper presents a machine-learning system to forecast agricultural crop yields in India between 2000 and 2026, with a complementary verification method using MODIS-derived NDVI. The study finds that random forest outperforms other models in predicting crop yields.

Objective

• To develop a machine-learning system for forecasting agricultural crop yields across Indian states.
• To propose a complementary aggregate-level verification method for predicted crop yield using MODIS-derived NDVI.

Study Configuration

• Spatial Scale: National scale, with focus on Indian states.
• Temporal Scale: 2000-2026 period, with a walk-forward validation procedure and a final holdout period (2024-2026).

Methodology and Data

• Models used: Linear regression, random forest, gradient boosting, CatBoost.
• Data sources: MODIS-derived NDVI time series (MOD13A3.061 Vegetation Indices Monthly L3 Global 1 km SIN Grid), crop yield data.

Main Results

• Random forest produced the most reliable and consistent forecasts, with a mean absolute percentage error (MAPE) of 11.4% and R2 = 0.982 on the final holdout.
• The inclusion of MODIS-derived NDVI significantly improved prediction accuracy.
• Prediction error varies considerably by crop.

Contributions

• This study proposes a complementary aggregate-level verification method for predicted crop yield using MODIS-derived NDVI.
• It also contributes an NDVI-based estimation approach for total foodgrain output.

Funding

• The research was funded by the Ministry of Earth Sciences (MoES), Government of India, under the project code MoES/ICPB/2020-21.

Citation

@article{Arai2026Crop,
  author = {Arai, Kohei and Sanwal, Sara},
  title = {Crop Yield Estimation with MODIS Derived Normalized Difference Vegetation Index and Comparative Study on Crop Yield Prediction Among Linear Regression, Random Forest and Gradient Boosting as Well as CatBoost},
  journal = {Remote Sensing},
  year = {2026},
  doi = {10.3390/rs18183107},
  url = {https://doi.org/10.3390/rs18183107}
}