A deep learning approach to detecting volcano deformation from satellite imagery using synthetic datasets

Abstract

Satellites enable widespread, regional or global surveillance of volcanoes and can provide the first indication of volcanic unrest or eruption. Here we consider Interferometric Synthetic Aperture Radar (InSAR), which can be employed to detect surface deformation with a strong statistical link to eruption. Recent developments in technology as well as improved computational power have resulted in unprecedented quantities of monitoring data, which can no longer be inspected manually. The ability of machine learning to automatically identify signals of interest in these large InSAR datasets has already been demonstrated, but data-driven techniques, such as convolutional neutral networks (CNN) require balanced training datasets of positive and negative signals to effectively differentiate between real deformation and noise. As only a small proportion of volcanoes are deforming and atmospheric noise is ubiquitous, the use of machine learning for detecting volcanic unrest is more challenging than many other applications. In this paper, we address this problem using synthetic interferograms to train the AlexNet CNN. The synthetic interferograms are composed of 3 parts: 1) deformation patterns based on a Monte Carlo selection of parameters for analytic forward models, 2) stratified atmospheric effects derived from weather models and 3) turbulent atmospheric effects based on statistical simulations of correlated noise. The AlexNet architecture trained with synthetic data outperforms that trained using real interferograms alone, based on classification accuracy and positive predictive value (PPV). However, the models used to generate the synthetic signals are a simplification of the natural processes, so we retrain the CNN with a combined dataset consisting of synthetic models and selected real examples, achieving a final PPV of 82%. Although applying atmospheric corrections to the entire dataset is computationally expensive, it is relatively simple to apply them to the small subset of positive results. This further improves the detection performance without a significant increase in computational burden (PPV of 100%). Thus, we demonstrate that training with synthetic examples can improve the ability of CNNs to detect volcano deformation in satellite images, and propose an efficient workflow for the development of automated systems.

Towards Global Volcano Monitoring Using Multisensor Sentinel Missions and Artificial Intelligence: The MOUNTS Monitoring System

“…Also, since arbitrary amounts of data can be generated, overfitting due to insufficient data is not an issue, and the generalization capability of the final model depends primarily on the realism of the synthetic data. While the present work was under review, a study using synthetic interferograms to train neural networks was published [35]. Unlike our study, the authors generate deformation patterns based on analytic models simulating realistic deformation sources in volcanic settings (i.e., Mogi and Okada sources in particular), and model stratified atmospheric effects from weather models (i.e., GACOS).…”

LiCSAR: An Automatic InSAR Tool for Measuring and Monitoring Tectonic and Volcanic Activity

“…Because of the large number of products, it becomes impossible to visually check all of them. Therefore, the COMET team has developed several machine learning approaches for automatically detecting ground deformation signals based on blind signal separation methods [25,68] and deep learning techniques [69][70][71]. The latter algorithms can detect large ground deformation signals in wrapped interferograms, whereas the former approach can detect the onset of slow ground deformation or subtle changes in the rate of any background deformation in InSAR time series.…”

Automated Methods for Detecting Volcanic Deformation Using Sentinel‐1 InSAR Time Series Illustrated by the 2017–2018 Unrest at Agung, Indonesia

“…To be useful in volcanic crises, monitoring systems need to be available quickly, robust, and objective. Automated analysis tools are particularly useful for volcano observatories as they can be routinely applied at large number of systems and used to flag anomalous events for expert analysis (Anantrasirichai et al, , ; Gaddes et al, ).…”

“…Recent examples include the regional study of volcanic unrest in Latin America (Pritchard et al, ) and the study of specific events such as the 2014 unrest at Chiles‐Cerro Negro (Ebmeier et al, ) or the 2010 eruption at Merapi volcano (Pallister et al, ). Automated monitoring systems are being developed to routinely analyze large volumes of InSAR data, making use of artificial intelligence tools such as machine learning (Anantrasirichai et al, , ) and blind source separation (Ebmeier, ; Gaddes et al, , ).…”