Explainable Earth Surface Forecasting Under Extreme Events

Abstract

With climate change-related extreme events on the rise, high-dimensional Earth observation data present a unique opportunity for forecasting and understanding impacts on ecosystems. This is, however, impeded by the complexity of processing, visualizing, modeling, and explaining this data. We train a convolutional long short-term memory-based architecture on the novel DeepExtremeCubes data set to showcase how this challenge can be met. DeepExtremeCubes includes around 40,000 long-term Sentinel-2 minicubes (January 2016–October 2022) worldwide, along with labeled extreme events, meteorological data, vegetation land cover, and a topography map, sampled from locations affected by extreme climate events and surrounding areas. When predicting future reflectances and vegetation impacts through the kernel normalized difference vegetation index, the model achieved an $R^2$ score of 0.9055 in the test set. Explainable artificial intelligence was used to analyze the model's predictions during the October 2020 Central South America compound heatwave and drought event. We chose the same area exactly 1 year before the event as a counterfactual, finding that the average temperature and surface pressure are generally the most important predictors. In contrast, minimum evaporation anomalies play a leading role during the event. We also found the anomalies of the reflectances in the timestep before the extreme event to be critical predictors of its impact on vegetation. The code to replicate all experiments and figures in this paper is publicly available at https://github.com/DeepExtremes/txyXAI.