An explainable deep learning framework for sub-seasonal to seasonal agricultural drought prediction over India
- 1Centre of Studies in Resources Engineering, Indian Institute of Technology Bombay, Mumbai, India
- 2Centre for Climate Studies, Indian Institute of Technology Bombay, Mumbai, India
- 3Department of Civil Engineering, Indian Institute of Technology Bombay, Mumbai, India
Abstract
Agricultural drought (AGD) prediction is critical for agricultural water conservation in India, where ∼60% of the population depends on farming. This study presents an explainable data-driven framework to predict AGD at sub-seasonal-to-seasonal (S2S) scales (1–3-month lead) using a Stacked Long Short-Term Memory (SLSTM) model at 0.25° grid resolution across croplands in India. The framework integrates local antecedent land-atmosphere variables and non-local climate drivers, represented by Sea Surface Temperature (SST) anomalies, with Soil Moisture Percentile (SMP) as the drought indicator. A novel extreme-aware custom loss function combining weighted mean squared error and a time-shift penalty is used to train SLSTM model specifically to capture drought events. The model is trained from June 1981–May 2015 and tested from June 2015–May 2022. SLSTM model outperformed benchmark deep learning models (XGBoost and LSTM) and the SEAS5 dynamical forecast system in predicting major AGD events, including the 2015–16 event. At a 1-month lead time, SLSTM model demonstrated strong performance, with percentage correct and probability of detection between 0.6–1 and false alarm rate between 0–0.3, particularly in Northern, Southern, and Eastern India. Although performance declined with increasing lead time, the model remains conservative while minimizing overprediction. Predicted drought area correlates reasonably well with that of observations across all three lead times. We further analyzed shifts in grid-wise predictor importance between local and non-local features in response to changes in lead time at seasonal and climatological scales. This research supports the development of early warning systems for AGD risk management in India.
Related
- Reservoir eutrophication prediction with explainable framework integrating temporal convolutional networks and bidirectional sequence modeling
- Machine learning models with explainable AI for PM2.5 prediction for Urban air quality assessment
- Seasonal hydropower forecast framework using a combination of deep learning and physics-based methods
- Skillful Seasonal Tropical Cyclone Prediction with AI Weather--Climate Models Forced by Dynamically Predicted SSTs
- A cross-regional machine learning transfer prediction framework for heat risk based on environmental characteristics of mature new towns
BibTeX
@article{verma2026explainable,
title = {{An explainable deep learning framework for sub-seasonal to seasonal agricultural drought prediction over India}},
author = {Saurabh Verma and L. Karthikeyan and Subimal Ghosh},
journal = {Agricultural Water Management},
year = {2026},
month = jul,
doi = {10.1016/j.agwat.2026.110597},
url = {https://doi.org/10.1016/j.agwat.2026.110597},
note = {CC BY-NC},
}
RIS
TY - JOUR
TI - An explainable deep learning framework for sub-seasonal to seasonal agricultural drought prediction over India
AU - Saurabh Verma
AU - L. Karthikeyan
AU - Subimal Ghosh
PY - 2026
DA - 2026/07/07
JO - Agricultural Water Management
DO - 10.1016/j.agwat.2026.110597
UR - https://doi.org/10.1016/j.agwat.2026.110597
AB - Agricultural drought (AGD) prediction is critical for agricultural water conservation in India, where ∼60% of the population depends on farming. This study presents an explainable data-driven framework to predict AGD at sub-seasonal-to-seasonal (S2S) scales (1–3-month lead) using a Stacked Long Short-Term Memory (SLSTM) model at 0.25° grid resolution across croplands in India. The framework integrates local antecedent land-atmosphere variables and non-local climate drivers, represented by Sea Surface Temperature (SST) anomalies, with Soil Moisture Percentile (SMP) as the drought indicator. A novel extreme-aware custom loss function combining weighted mean squared error and a time-shift penalty is used to train SLSTM model specifically to capture drought events. The model is trained from June 1981–May 2015 and tested from June 2015–May 2022. SLSTM model outperformed benchmark deep learning models (XGBoost and LSTM) and the SEAS5 dynamical forecast system in predicting major AGD events, including the 2015–16 event. At a 1-month lead time, SLSTM model demonstrated strong performance, with percentage correct and probability of detection between 0.6–1 and false alarm rate between 0–0.3, particularly in Northern, Southern, and Eastern India. Although performance declined with increasing lead time, the model remains conservative while minimizing overprediction. Predicted drought area correlates reasonably well with that of observations across all three lead times. We further analyzed shifts in grid-wise predictor importance between local and non-local features in response to changes in lead time at seasonal and climatological scales. This research supports the development of early warning systems for AGD risk management in India.
SN - 0378-3774
SN - 1873-2283
ER -