Day 4

Detailed paper information

Back to list

Paper title MONITORING VOLCANIC ERUPTIONS WITH SENTINEL-1 GRD DATA - THE 2014/2015 FOGO VOLCANO CASE-STUDY (FOGO ISLAND, CABO VERDE)
Authors
  1. Rafaela Tiengo Speaker
  2. José Pacheco IVAR - University of the Azores
  3. Jéssica Uchôa IVAR - University of the Azores
  4. Artur Gil IVAR - University of the Azores
Form of presentation Poster
Topics
  • D1. Managing Risks
    • D1.01 Satellite EO for Geohazard Risks
Abstract text The last eruption in the Fogo Volcano (Archipelago of Cabo Verde, Africa), which began in November 2014, was the first eruptive event captured by the Sentinel-1 mission. The GRD data from the Sentinel-1 mission was used in this study to identify the progress of the lava flow and measure the affected area, in order to assess its potential to monitor and assess eruptive scenarios in near-real-time, which is fundamental to mitigate risks and to better support crisis management. The present work sought to complement previous research and explore the potential of utilizing data from the Synthetic Aperture Radar (SAR) Sentinel-1 mission to better monitor active volcanic areas. Sentinel-1 Ground Range Detected (GRD) data was used to analyze the changes that occurred in the area before, during, and after the eruptive event and was able to identify the progress of the lava flow and measure the affected area (3.89 km² in total). After processing the GRD data using the standard SNAP workflow, the raster calculation tool of Arcmap 10.4 GIS software was used to compute an Image Differencing Change Detection. In this procedure, each image after the start of the event is subtracted from a pre-event image. For this purpose, the value of an image referring to the last hours of the eruption was subtracted from an image prior to the beginning of the event. Very high (“change”) and very low (“no change”) values were thresholded in order to obtain the change detection map. To assess the accuracy and validate each change detection procedure, the Overall Accuracy was computed with independent validation datasets with 50 change/no change sampling points. The successive change detection procedures showed Overall Accuracies ranging between 0.70 and 0.90. The identification and mapping of the affected area (3.89 km² in total) are in relative agreement with other authors' results when applying different techniques to different SAR datasets, including high-resolution commercial data (from 4.53 to 5.42 km2). Nevertheless, in the attached figure, it is possible to note that some of the areas previously observed as affected by the 2014/15 lava flow were not identified in the change detection procedures with GRD data. It might be explained by the fact that there were no substantial roughness changes in the overlap area of 2014/15 lava flow with that of 1995 which occurred at the "Chã das Caldeiras" place. Monitoring surface changes during eruptive events using Sentinel-1 GRD data proved cost-effective in terms of data processing and analysis, with lower computational cost, and results consistent and coherent with those previously obtained with Sentinel-1 SLC data or other types of SAR data. Therefore, this approach is pertinent and suitable for research but is especially valuable to integrate low-cost monitoring systems of active volcanic areas in near-real-time. The systematic use of GRD products can thus serve as the basis for event monitoring that confers greater agility in computation and analysis time for decision support.