Analysis of the lava flows from the La Palma 2021 volcanic eruption

To deepen the understanding of remote sensing data and its applications, an analysis of such data is part of the "Remote Sensing II" module.

Result of the pre-processing from the satellite data used
Result of the pre-processing from the satellite data used

During the summer term 2023 on of the project topics investigate the lava flows of a volcanic eruption. While a significant portion of remote sensing data falls within the visible spectrum, in this instance, radar data was employed. The selected geographical and temporal focus was the volcanic eruption on La Palma in 2021. Volcanic eruptions are often accompanied by substantial smoke, complicating analysis in the visible spectrum. Radar data, however, penetrates clouds and smoke, enabling an analysis of the lava on the surface. The analysis of the event was facilitated by the use of high-resolution radar data from the Sentinel-1 satellite, ensuring reliable detection and tracking of changes on the Earth's surface.

After performing the necessary preprocessing steps to make the data usable, various approaches were employed to localize and categorize the lava flows on La Palma. SAR data is available in two different datasets: GRD (Ground-Range Detected) and SLC (Single Look Complex). Each dataset contains two bands based on the direction of the backscattered signal compared to the outgoing signal. In our analysis, the GRD dataset was utilized.

Result of the mathematical procedure in comparison with the official eruption layer
Result of the mathematical procedure in comparison with the official eruption layer

Initially, mathematical operations were conducted on the layers and pixel values to deepen understanding of both the program and dataset. The initial idea involved dividing the pixel values of a dataset before and after the eruption. An unchanged pixel should have a value of one, while changed pixels would deviate. As the terrain in the lava field areas had undergone significant changes, a threshold was set to identify only larger changes in pixel values. This idea was constrained by noise and did not yield significant lava detection.

Result of the offset tracking
Result of the offset tracking

Next, a simple change detection algorithm was applied to the datasets before and after the eruption. The resulting image still exhibited considerable noise, but after applying a threshold, a delineation of the lava field area became visible. This could be cross-verified using a georeferenced layer of the lava field provided by Copernicus.

The final approach involved the application of offset tracking. In this function, two or more images of the same area at different times are compared, and phase differences are calculated. From this information, a change vector is derived, providing the direction and speed of movement. The results of this method were the most promising, indicating that over the entire period, partially, lava flows achieved velocities of more than 2.4 meters per day.

Authors: Robin Klein, Max Göttert, Judith Franke