Physics informed Transformer-VAE for biophysical parameter estimation: PROSAIL model inversion in Sentinel-2 imagery

The article discusses the emergence of Earth observation (EO) foundation models as effective tools for deriving latent representations of the Earth system from remote sensing data. These models facilitate ecosystem dynamics modeling without extensive preprocessing. However, existing models are limited by fixed spatial or temporal scales. The proposed framework integrates various EO modalities into a unified feature space, achieving high spatio-temporal resolution using Sentinel-1 and Sentinel-2 data, thus enhancing ecological analysis capabilities.

Monitoring the spatiotemporal dynamics of forest CO2 uptake, known as Gross Primary Production (GPP), poses significant challenges in terrestrial ecosystem research. While Eddy Covariance towers provide high-frequency estimates, their spatial limitations hinder large-scale assessments. Remote sensing offers a scalable alternative, yet many methods rely on single-sensor spectral indices and statistical models that struggle to capture GPP's complex temporal dynamics. This study evaluates the performance of GPT-2, a transformer model, against LSTM, a recurrent neural network, for GPP prediction u…