An Input-Frugal Deep Learning Framework for Weather-Driven National Crop-Yield Forecasting: A Case Study of Brazilian Soybean
Abstract
Reliable, timely crop-yield forecasts are essential for market stability and risk management, yet many approaches rely on costly or hard-to-scale inputs. We present a frugal, transferable, and architecture-agnostic deep learning framework that uses routine weather as the only time-varying input plus two lightweight static context inputs (crop year and an agro-environmental label) to capture long-run change and regional heterogeneity, while supporting multiple sequence encoders under identical data requirements. Using a 20-season Brazilian soybean case study (2001/02-2020/21) with leave-one-year-out cross-validation, we benchmark MLP, CNN, LSTM, CNN-LSTM, a Transformer encoder and the Mamba state-space model against linear ridge regression and a five-year moving-average "farmer" baseline. All deep learning variants outperform ridge, and all sequential encoders surpass the non-sequential MLP. The Transformer achieves the best national accuracy (RMSE 149 kg ha^-1; rRMSE 5.3%; R^2 = 0.784), reducing error by 47.6% relative to the farmer baseline. In-season forecasts improve monotonically from early- to late-season issuance, reaching approximately 50% lower error than the baseline at the latest forecast point. Ablations indicate that the agro-environmental label and spatial instance expansion (multiple grid-node weather sequences per municipality-year) contribute positively without increasing input complexity. SHAP diagnostics suggest crop year explains most of the long-run trajectory, whereas within-season weather and agro-environmental context primarily drive interannual deviations, with moisture/cloud and thermal-demand variables dominating. Overall, the framework is straightforward to deploy across other crops and geographic regions and is naturally compatible with operational weather forecasts for routine monitoring.
Related
- A hybrid car-following behavior prediction framework for snowy weather based on stacking ensemble learning
- SDECast: Probabilistic Weather Forecasting in Continuous Time with Neural SDEs
- M2Weather: A Benchmark for Joint Multi-Station and Multi-Variable Weather Forecasting
- Combining Weather Forecast Aggregation and State-Space Models for Adaptive Probabilistic Electricity Load Forecasting
- Learning Kolmogorov Backward Operators for Spectral Analysis and Forecasting: A Data-Driven Variational Galerkin Framework
BibTeX
@misc{mello2026inputfrugal,
title = {{An Input-Frugal Deep Learning Framework for Weather-Driven National Crop-Yield Forecasting: A Case Study of Brazilian Soybean}},
author = {Fernando Dupin da Cunha Mello and Prashant Kumar and Erick G. Sperandio Nascimento},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.38447},
eprint = {2609.38447},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.38447},
}
RIS
TY - UNPB
TI - An Input-Frugal Deep Learning Framework for Weather-Driven National Crop-Yield Forecasting: A Case Study of Brazilian Soybean
AU - Fernando Dupin da Cunha Mello
AU - Prashant Kumar
AU - Erick G. Sperandio Nascimento
PY - 2026
DA - 2026/09/29
PB - arXiv
DO - 10.48550/arxiv.2609.38447
UR - https://arxiv.org/abs/2609.38447
AB - Reliable, timely crop-yield forecasts are essential for market stability and risk management, yet many approaches rely on costly or hard-to-scale inputs. We present a frugal, transferable, and architecture-agnostic deep learning framework that uses routine weather as the only time-varying input plus two lightweight static context inputs (crop year and an agro-environmental label) to capture long-run change and regional heterogeneity, while supporting multiple sequence encoders under identical data requirements. Using a 20-season Brazilian soybean case study (2001/02-2020/21) with leave-one-year-out cross-validation, we benchmark MLP, CNN, LSTM, CNN-LSTM, a Transformer encoder and the Mamba state-space model against linear ridge regression and a five-year moving-average "farmer" baseline. All deep learning variants outperform ridge, and all sequential encoders surpass the non-sequential MLP. The Transformer achieves the best national accuracy (RMSE 149 kg ha^-1; rRMSE 5.3%; R^2 = 0.784), reducing error by 47.6% relative to the farmer baseline. In-season forecasts improve monotonically from early- to late-season issuance, reaching approximately 50% lower error than the baseline at the latest forecast point. Ablations indicate that the agro-environmental label and spatial instance expansion (multiple grid-node weather sequences per municipality-year) contribute positively without increasing input complexity. SHAP diagnostics suggest crop year explains most of the long-run trajectory, whereas within-season weather and agro-environmental context primarily drive interannual deviations, with moisture/cloud and thermal-demand variables dominating. Overall, the framework is straightforward to deploy across other crops and geographic regions and is naturally compatible with operational weather forecasts for routine monitoring.
ER -