Large-scale demonstration of machine learning for the detection of volcanic deformation in Sentinel-1 satellite imagery.

Radar (SAR) satellites systematically acquire imagery that can be used for volcano monitoring, characterising magmatic systems and potentially forecasting eruptions on a global scale. However, exploiting the large dataset is limited by the need for manual inspection, meaning timely dissemination of information is challenging. Here we automatically process ~ 600,000 images of > 1000 volcanoes acquired by the Sentinel-1 satellite in a 5-year period (2015-2020) and use the dataset to demonstrate the applicability and limitations of machine learning for detecting deformation signals. Of the 16 volcanoes flagged most often, 5 experienced eruptions, 6 showed slow deformation, 2 had non-volcanic deformation and 3 had atmospheric artefacts. The detection threshold for the whole dataset is 5.9 cm, equivalent to a rate of 1.2 cm/year over the 5-year study period. We then use the large testing dataset to explore the effects of atmospheric conditions, land cover and signal characteristics on detectability and find that the performance of the machine learning algorithm is primarily limited by the quality of the available data, with poor coherence and slow signals being particularly challenging. The expanding dataset of systematically acquired, processed and flagged images will enable the quantitative analysis of volcanic monitoring signals on an unprecedented scale, but tailored processing will be needed for routine monitoring applications. Supplementary Information: The online version contains supplementary material available at 10.1007/s00445-022-01608-x.

Use of InSAR data for measuring land subsidence induced by groundwater withdrawal and climate change in Ardabil Plain, Iran.

The Ardabil plain, with an approximate area of 1097.2 km2 in northwestern Iran, has experienced land subsidence due to intensive groundwater withdrawal and long seasons of drought in recent years. Different techniques have been used to investigate and evaluate subsidence in this region including: Global Positioning Systems (GPS), Levelling, and Geotechnical methods. These methods are typically expensive, time-consuming, and identify only a small fraction of the areas prone to subsidence. This study employs an Interferometric Synthetic Aperture Radar (InSAR) technique to measure the long-term subsidence of the plain. An open-source SAR interferometry time series analysis package, LiCSBAS, that integrates with the automated Sentinel-1 InSAR processor (COMET-LiCSAR) is used to analyze Sentinel-1 satellite images from October 2014 to January 2021. Processing of Sentinel-1 images shows that the Ardabil plain has been facing rapid subsidence due to groundwater pumping and reduced rainfall, especially between May 2018 to January 2019. The maximum subsidence rate was 45 mm/yr, measured at the southeastern part of the plain. While providing significant advantages (less processing time and disk space) over other InSAR processing packages, implementation of the LiCSBAS processing package and its accuracy for land subsidence measurements at different scales needs further evaluation. This study provides a procedure for evaluating its efficiency and accuracy for land subsidence measurements by comparing its measurements with the results of the GMTSAR and geotechnical numerical modeling. The results of geotechnical numerical modeling showed land subsidence with an average annual rate of 38 mm between 2006 and 2020, which was close to measurements using the InSAR technique. Comparison of the subsidence measurements of the Ardabil plain using the LiCSBAS package with results obtained from other techniques shows that LiCSBAS is able to accurately detect land deformation at large scales (~ km). However, they may not be optimized for more local deformations such as infrastructure monitoring.

Extreme subsidence in a populated city (Mashhad) detected by PSInSAR considering groundwater withdrawal and geotechnical properties.

Ground deformation can cause serious environmental issues such as infrastructure damage, ground compaction, and reducing the ground capacity to store water. Mashhad, as one of the largest and most populated cities in the Middle East, has been suffering from extreme subsidence. In the last decade, some researchers have been interested in measuring land subsidence rates in the Mashhad valley by InSAR techniques. However, most of those studies were based on inaccurate measurements introducing uncertainties in the resulting subsidence rates. These researches used a small number of EnviSat data with long perpendicular and inhomogeneous temporal baseline. This paper seeks to determine the subsidence rate in urban areas of Mashhad in recent years, the threat that was neglected by the city managers and decision-makers. For this purpose, the Persistent Scatterer InSAR technique was applied in the study area using two time-series of descending and ascending Sentinel-1A acquisitions between 2014 and 2017. The results demonstrated the maximum line-of-sight deformation rate of 14.6 cm/year and maximum vertical deformation (subsidence) rate about 19.1 cm/year which could have irreversible consequences. The results were assessed and validated using piezometric data, GPS stations, and geotechnical properties. This assessment confirms that the main reason for subsidence in the interested area is groundwater over-extraction. Also, investigation of geotechnical properties shows that thick fine-grained layers in the northwest of the city could strongly affect the results. At the end of this paper, a new simplified method was proposed to estimate specific storage in special cases to predict the subsidence rate.