Applying a deep learning-based approach for scaling vegetation dynamics to predict changing forest regimes under future climate and fire scenarios.