Spatiotemporal Tracking of the Volcanic Cloud Dispersion Using ConvLSTM Models and SEVIRI Imagery

Explosive volcanic eruptions inject a variety of particles and gases into the atmosphere, forming volcanic clouds that significantly impact human health, climate, and aviation safety. Accurately capturing the temporal evolution of these clouds is essential for understanding their dynamics and improving predictive capabilities. Due to the rapid and unpredictable nature of explosive eruptions, volcanic clouds can form, expand, and disperse in short timeframes. For this reason, high-temporal-resolution geostationary satellite data are indispensable for near-real-time monitoring. SEVIRI (Spinning Enhanced Visible and InfraRed Imager), onboard the Meteosat Second Generation (MSG) geostationary satellite, provides high-frequency radiometric data essential for tracking volcanic clouds on a global scale. SEVIRI's ability to acquire images at intervals of 5–15 minutes enables the identification of patterns in cloud formation and dispersion, supporting timely warnings and informed decision-making during crises. Here, we propose a novel approach using a convolutional long short-term memory (ConvLSTM) model, a type of recurrent neural network designed to handle spatiotemporal data, for effectively tracking the spread of volcanic clouds using satellite imagery. By training the model on a dataset of Ash RGB images derived from SEVIRI data, we analyze volcanic events at Mt. Etna (Italy) to demonstrate the model's capability to capture both spatial and temporal dynamics. Our findings show that ConvLSTM models excel in addressing complex spatiotemporal challenges, providing robust segmentation and reliable tracking of volcanic clouds over time. This approach delivers timely information that enhances aviation safety, emergency response, and public health monitoring, contributing to more effective management of volcanic crises.