Sarkar et al. (2026) A hybrid Granger TCN framework for generating climate analogues and determining the future of agricultural practices

Identification

• Journal: Scientific Reports
• Year: 2026
• Date: 2026-05-25
• Authors: Suman Saurabh Sarkar, Sushma Jain, Jatin Bedi
• DOI: 10.1038/s41598-026-47821-y

Research Groups

• Department of Computer Science & Engineering, Thapar Institute of Engineering and Technology, Patiala, Punjab, India.

Short Summary

The study proposes a hybrid framework combining Granger Causality and Temporal Convolution Networks (TCN) to identify future climate analogues for maximum temperature in Kashmir based on historical data from other Indian cities to support agricultural planning.

Objective

• To determine specific future years in which the maximum temperature of Kashmir (output region) will match the historical maximum temperatures of Coimbatore, Gwalior, and Ludhiana (input regions) to facilitate maize crop planning.

Study Configuration

• Spatial Scale: Regional (four Indian cities: Coimbatore, Gwalior, Ludhiana, and Kashmir).
• Temporal Scale: 75 years of mapping.

Methodology and Data

• Models used: Hybrid Granger TCN framework (Granger Causality Test integrated with Temporal Convolution Network).
• Data sources: Historical maximum temperature records for the four selected Indian cities.

Main Results

• The framework demonstrated significant mapping capabilities for temperature analogues.
• For the 1st rank match across the 75-year mapping period, the standard deviation of the Root Mean Square Error (RMSE) was 0.0271 with a variance of 0.0007.

Contributions

• Develops a novel hybrid deep learning and causality-based approach for generating climate analogues.
• Provides a predictive tool for agricultural automation and infrastructure setup, specifically for optimizing irrigation sources and maize crop planning.

Funding

• Not specified.

Citation

@article{Sarkar2026hybrid,
  author = {Sarkar, Suman Saurabh and Jain, Sushma and Bedi, Jatin},
  title = {A hybrid Granger TCN framework for generating climate analogues and determining the future of agricultural practices},
  journal = {Scientific Reports},
  year = {2026},
  doi = {10.1038/s41598-026-47821-y},
  url = {https://doi.org/10.1038/s41598-026-47821-y}
}