From Level-1 to Level-3: Integrating Time Series Analysis Into the Sentinel-1-based SARVIEWS Volcano Monitoring Service

The recent occurrence of a number of destructive earthquakes and volcanic eruptions has reinforced the need for timely disaster management information. Synthetic Aperture Radar (SAR), with its all-weather, day-and-night capabilities, has long been a useful tool in hazard monitoring and disaster management, but only recently, with the launch of the Sentinel-1 mission, has SAR finally become a relevant data source for operational response.

Here, we present recent technical advances to the Sentinel-1-based SARVIEWS processing system, which was developed by the University of Alaska Fairbanks in collaboration with the Alaska Satellite Facility, to generate SAR-derived hazard products for volcano hazard monitoring. Previous versions of SARVIEWS enabled automated production of a series of Level-2 SAR products and provided a means for their free-and-open distribution. Level-2 products included fully geocoded and radiometrically terrain corrected image time series as well as unwrapped line-of-sight deformation maps. Distribution of these products is facilitated through the SARVIEWS Hazard Portal (http://sarviews-hazards.alaska.edu/), which has found extensive use since its initiation.

We will highlight recent enhancements to the SARVIEWS service, which now enables generation of Level-3 products such as inverted deformation time series and amplitude change detection maps from existing Level-2 SARVIEWS data. Level-3 product generation is enabled through an interactive and expandable Jupyter notebook lab. This lab comes pre-loaded with ready-to-use Level-3 algorithms such as SBAS-type InSAR time series analysis and time series change detection. It also allows development of additional algorithms by including a range of pre-installed Python libraries, public domain remote sensing software, and deep learning tools.

To demonstrate the benefits of the new Level-3 components of SARVIEWS, we will showcase their use for analysis of volcanic activity. We highlight the production of deformation time series for the recent eruptions of Fernandina and Sierra Negra on the Galapagos Islands, and demonstrate deformation measurements at Alaska's Okmok volcano. We also demonstrate amplitude change detection by analyzing the chronology of the 2016-2017 volcanic eruption at Bogoslof Island, Alaska.