SAR Remote Sensing for Volcanic Hazards: Ground Deformation Monitoring and InSAR
Catching signs of an upcoming eruption using satellite imagery in Reykjanes Peninsula, Iceland.
Radar remote sensing is a powerful tool which can be used alongside more popular optical remote sensing strategies. In this post we will have a quick look at how free SAR data can be utilized to provide deeper insight into volcanic hazards through the use of radar interferometry (InSAR). The data will focus on the Reyjkanes Peninsula of Iceland. Scientists have theorized that this peninsula will feature volcanic activity for decades to come.
What is SAR?
SAR is an active sensor thus radar signals are continuously transmitted to the ground then the backscatter is received by the sensor. The brightness of the pixels correlate to the strength of the backscatter received from the terrain. Ground deformation can be monitored by analyzing phase differences between 2 radar imagery.
InSAR is the unique selling point of radar remote sensing. Using phase differences between radar imagery we can monitor ground deformation. Unfortunately this is something that cannot be done using Google Earth Engine since their Sentinel-1 data is only GRD data which lacks phase information.
SAR is widely used for volcanic hazard monitoring because immense pressure from the magma underneath the surface can cause the surface to deform and are often used as one of the indicators of when a volcano is about to erupt.
In the image below we see timelapse animation of the peninsula region using Google Earth Engine and Sentinel-1 SAR GRD imagery. This timelapse data is based on the backscatter strength of the signal. We can see the bright areas are slopes that are facing towards the sensor and urban areas.
Unfortunately there are some distortions are present in large areas of the peninsula. It appears that some images in the Google Earth Engine catalog are terrain corrected and some images are not terrain corrected. I’m not very familiar with the area so I can only assume that it’s a processing error or maybe signal delay due to atmospheric effects.
This type of data cannot be used for ground deformation monitoring since it is missing phase data. I usually like to check the study area on Google Earth Engine since the Sentinel-1 dataset is easily accessible using that service.
Ground Deformation through InSAR
Data was manually downloaded and processed to minimize the geometric distortions seen in the timelapse animations. This would ensure that images are aligned for proper timeseries analysis.
I download images ranging from August 9 2019 to March 19 2021. I also used a descending pass for the images. March 19 is the exact date that the volcano erupted so the data right up to the point of the eruption can be observed.
The software I’m using for this analysis is MintPy which uses a small baseline subset technique.
Point 1: Fagradalsfjall Volcano
As of writing Fagradalsfjall Volcano is the main area where the eruption is occuring.
In the plot above we can see the large deformations around the volcano area which is marked by a red triangle. Large areas around the volcano do not pass the coherence threshold so it is masked out but we can see the large negative and positive ground deformations surrounding the volcano. The black square is the reference point for all the displacement calculations.
For the graph below a coherent pixel is selected to observe ground deformation.
The area around the point features a noticeable uplift starting late 2021 before eruption and has a linear velocity of 5.15 ± 0.46 cm per year. Starting near the end of 2021 there is a lot of movement as well as a lot of earthquakes. Iceland raised the yellow alert on February 2, 2021 due to increased ground deformation and earthquakes.
Towards the last few weeks of the data there is a sharp upward velocity. This is also right around the time that there were numerous earthquakes which led researchers to estimate that an eruption was coming very soon.
Point 2: Swelling around Grindavik
For the second point we will look at the area around Grindavik which also has swelling. The Iceland Met office released this displacement map based on data in January 2020:
If we compare it with the data from MintPy we can observe that there was also a large deformation event in the beginning of January 2020 which matches their data. There is another large deformation event near the end of the timeseries which was when the volcano of Fagradalsfjall was about to erupt. In this sampling point there was a linear velocity of 7.90 ± 0.48 cm per year.
That the quick glimpse of the activity in Iceland. I hope it showed some interesting insight into how radar remote sensing can be used in our everyday world. This post is related to my previous post about InSAR validation based on GPS data in Iceland.
If you like what you see you can follow me on Medium and give me a clap.
Please contact me on LinkedIn if you want to get in touch.