Towards high-resolution forecasting of tropical deforestation with deep learning
- 1École Polytechnique Fédérale de Lausanne (EPFL)
- 2EPFL
- 3University of Zurich
Abstract
Reducing tropical deforestation effectively requires spatially explicit data on deforestation dynamics. While past deforestation is now detected and mapped with high accuracy, producing reliable future forecasts remains a challenge. We present a deep learning approach for high-resolution deforestation forecasting in the Brazilian Amazon. We train a spatio-temporal ConvLSTM model on past deforestation maps and other covariates. Our model achieves an F1 score of 0.47 when forecasting one year ahead, substantially improving over comparable prior work. For multi-year predictions, we obtain promising results within a 3-year horizon, while sensitivity to shifts in deforestation dynamics suggests limitations for longer-term forecasts. We generate forecasts for years 2016–2018 across the Brazilian Amazon, and make them available through an interactive Google Earth Engine application. Our results demonstrate the potential for high-resolution, short- and medium-term forecasting that can support conservation efforts and help protect tropical forests.
Related
- FAST-ML: A Hybrid Physics-Machine Learning Framework for Tropical Cyclone Intensity Forecasting
- A deep learning global ocean forecasting model with sub-daily and eddy-resolving resolution
- Butterfly Effect Confirmed in Global AI Weather Models: Evidence from Tropical Cyclone Forecasting
- PrecipCorrDiff: A Deep Learning Framework for Generating 0.0625∘ High-Resolution IMD Gridded Precipitation Data
- Shoreline kinematics in Libreville (Gabon) derived from very high spatial resolution (VHR) imagery using machine and deep learning techniques
BibTeX
@article{pisl2026towards,
title = {{Towards high-resolution forecasting of tropical deforestation with deep learning}},
author = {Jan Pisl and Gencer Sumbul and Gaston Lenczner and Jan Dirk Wegner and Devis Tuia},
journal = {Environmental Research Letters},
year = {2026},
month = nov,
doi = {10.1088/1748-9326/aebda8},
url = {https://iopscience.iop.org/article/10.1088/1748-9326/aebda8},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Towards high-resolution forecasting of tropical deforestation with deep learning
AU - Jan Pisl
AU - Gencer Sumbul
AU - Gaston Lenczner
AU - Jan Dirk Wegner
AU - Devis Tuia
PY - 2026
DA - 2026/11/04
JO - Environmental Research Letters
DO - 10.1088/1748-9326/aebda8
UR - https://iopscience.iop.org/article/10.1088/1748-9326/aebda8
AB - Reducing tropical deforestation effectively requires spatially explicit data on deforestation dynamics. While past deforestation is now detected and mapped with high accuracy, producing reliable future forecasts remains a challenge. We present a deep learning approach for high-resolution deforestation forecasting in the Brazilian Amazon. We train a spatio-temporal ConvLSTM model on past deforestation maps and other covariates. Our model achieves an F1 score of 0.47 when forecasting one year ahead, substantially improving over comparable prior work. For multi-year predictions, we obtain promising results within a 3-year horizon, while sensitivity to shifts in deforestation dynamics suggests limitations for longer-term forecasts. We generate forecasts for years 2016–2018 across the Brazilian Amazon, and make them available through an interactive Google Earth Engine application. Our results demonstrate the potential for high-resolution, short- and medium-term forecasting that can support conservation efforts and help protect tropical forests.
SN - 1748-9326
ER -