Anomaly detection in satellite imagery through temporal inpainting

TL;DR

This paper introduces a deep learning-based method for anomaly detection in satellite imagery by predicting expected surface appearance and identifying discrepancies, enabling more sensitive and automated environmental monitoring.

Contribution

The authors develop a novel temporal inpainting model based on SATLAS that leverages global satellite data to detect surface anomalies with higher sensitivity than traditional methods.

Findings

Detects surface anomalies with three times lower thresholds than baselines.

Successfully identified earthquake-induced surface ruptures in Turkey-Syria.

Outperforms traditional change detection methods in sensitivity and specificity.

Abstract

Detecting surface changes from satellite imagery is critical for rapid disaster response and environmental monitoring, yet remains challenging due to the complex interplay between atmospheric noise, seasonal variations, and sensor artifacts. Here we show that deep learning can leverage the temporal redundancy of satellite time series to detect anomalies at unprecedented sensitivity, by learning to predict what the surface should look like in the absence of change. We train an inpainting model built upon the SATLAS foundation model to reconstruct the last frame of a Sentinel-2 time series from preceding acquisitions, using globally distributed training data spanning diverse climate zones and land cover types. When applied to regions affected by sudden surface changes, the discrepancy between prediction and observation reveals anomalies that traditional change detection methods miss. We validate our approach on earthquake-triggered surface ruptures from the 2023 Turkey-Syria earthquake sequence, demonstrating detection of a rift feature in Tepehan with higher sensitivity and specificity than temporal median or Reed-Xiaoli anomaly detectors. Our method reaches detection thresholds approximately three times lower than baseline approaches, providing a path towards automated, global-scale monitoring of surface changes from freely available multi-spectral satellite data.