Desertification Forecasting in Semi-Arid Regions: Using Deep Learning to Probe the Observability Limits of Land Degradation Dynamics
ConvLSTM forecasts regional desertification trends in Ceará but not cell-level land degradation.
Researchers train a ConvLSTM on ten annual Earth Observation variables (2000-2023) over Ceará, Brazil, on a 256x206 grid at about 2.2 km, to forecast a MapBiomas-derived degradation index. Statewide mean IDV rose from 0.238 to 0.350, but a gradient-boosting model explains only 4.7% of cell-level variance. The ConvLSTM cuts RMSE versus persistence by 1.6% one year ahead and 4.4% four years ahead and reproduces 53-69% of regional drift. Under climatological forcing, mean IDV is projected to rise from 0.350 in 2023 to 0.370 in 2030, so forecasts are reliable for the regional trend, not individual cells.
- ConvLSTM trained on 2000-2023 Earth Observation data over Ceará, Brazil.
- Mean degradation index rose from 0.238 to 0.350; cell-level variance is mostly noise.
- RMSE beats persistence by 1.6% at one year and 4.4% at four years.
- Under climatological forcing, mean IDV is projected from 0.350 in 2023 to 0.370 in 2030.
From the original article
1 min read at arxiv.orgForecasting desertification has received far less attention than mapping it. We use a convolutional LSTM (ConvLSTM) trained on ten annual Earth Observation (EO) variables (2000-2023) over the semi-arid state of Ceará, Brazil, on a 256x206 grid at roughly 2. 2 km resolution, to measure how much of the year-to-year evolution of a MapBiomas-derived degradation index (IDV) can be predicted from satellite data.
By Filipe Moura Da Silva · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.