Eight months into the war in Ukraine and we continue to see widespread destruction of lives and infrastructure, more than 7.7 million Ukrainians fleeing their country, and a third of the population displaced. In addition, there now looms nuclear disaster in relation to Ukraine’s nuclear power plants, that could not only affect Ukraine, but also its EU and other neighbours. The nuclear power plant in Zaporizhzhia is the one most immediately at risk at this moment in the war.
Nuclear power plants in Ukraine
Currently, Ukraine has four nuclear power plants in operation:
- Zaporizhzhia Nuclear Power Plant
- Rivne Nuclear Power Plant
- Khmelnitsky Nuclear Power Plant
- South-Ukraine Nuclear Power Plant
After February 24, the security of these plants has been under threat. On March 4, 2022, Ukraine informed the International Atomic Energy Agency (IAEA) that Russian forces had taken control of the Zaporizhzhia nuclear power plant.
IAEA Director General Rafael Mariano Grossi expressed great concern regarding the situation and impact of the military conflict at the Chernobyl and the Zaporizhzhia nuclear plants, with respect to operating staff, physical integrity of the facilities, nuclear safety, integrity of the plant security systems, communication, and power supply. He reiterated the importance of adopting measures for ensuring nuclear safety and security during an armed conflict. The IAEA has confirmed that the Russian military actions had not yet directly impacted the Khmelnitski, Rivne, and South Ukraine nuclear power plants.
Radar remote sensing for change detection
Earlier in this series we talked about change detection using Synthetic Aperture Radar (SAR) satellite images, as a tool to detect structural changes in an area of interest.
Also read: SAR for change detection methods in Mariupol
In today’s article we use this toolset to zoom in on the Zaporizhzhia nuclear plant, looking at possible changes that occurred since February 2022, using several different methods.
Method 1: Radar change detection in RGB
The following visual combines two Sentinel-1 images of 8 January and 5 September 2022, to see a possible increase or decrease of backscattering in the area of the Zaprorizhzhia nuclear power plant. Using the RGB method, radar data is converted to red, green and blue (RGB) colours. When then ‘subtracting’ the two images, changes appear in magenta.
The resulting image shows us changes in magenta. The red block indicates the location of the nuclear power plant, where the threat of nuclear calamity has been high for months.
Please note that there are lots of magenta areas (change detected) in the fields around the city, indicating for example changes in vegetation between winter and end of summer.
According to CNN News these fears were renewed last week, after shelling intensified around the massive Zaporizhzhia plant, which is Europe’s largest, and has been under Russian control since March.
Attacks at the complex, which have ramped up as fighting flares up in Ukraine’s south, have sparked concerns about the spectre of nuclear disaster.
Method 2: Log likelihood ratio
Tech-talk: What is the Log Likelihood Ratio?
The change detections on the log likelihood decision statistic are based on the average of the backscatter ratio that may be obtained by evaluating the decision statistic in the scene, applying a threshold to the decision map and assigning values to either the changed or unchanged pixels.
The ability of a change statistic to distinguish between changed or unchanged is determined by the degree of overlap of the density functions of the decision statistic under each hypothesis which, in turn, is predominantly dependent on the average and “spread” of the density functions under each hypothesis.
The following image is obtained by again combining two Sentinel-1 images of 8 January and 5 September 2022, showing log likelihood ratio probability values between -1 (black, no change) and 1 (change detected in the period of study). We can see the westward zones and inside zone in the Zaporizhzhia Nuclear Power Plant with pixels with values of +0.6, so possible change.
Please note that also here there are lots of white areas (change detected) in the fields around the city, indicating for example changes in vegetation.
Method 3: combining RGB and log-likelihood in SNAP
Finally, we want to combine the two bands with the log ratio band, to obtain a final image. This is a method that is readily available in the Sentinel Application Platform (SNAP) software.
In this image we can see the changes more clearly, again in magenta, to the west and inside the boundaries of the nuclear power plant, confirming many of the changes that the other two images showed us earlier.
ESA SNAP Software for SAR image processing
The Sentinel Application Platform (SNAP) is a common architecture for all Sentinel Toolboxes. The software is developed by Brockmann Consult, Skywatch, Sensar and C-S.
The SNAP architecture is ideal for Earth observation (EO) processing and analysis due to the following technological innovations: extensibility, portability, modular rich client platform, generic EO data abstraction, tiled memory management, and a graph processing framework.
SNAP and the individual Sentinel Toolboxes support numerous sensors other than Sentinel sensors.
Fieldwork to study the cause of changes detected by SAR
Of course this type of remote change detection only provides a rough indication of changes, without showing the nature of these changes (we see a mix of changes in infrastructure, but also for example in vegetation), without any details of what is the cause of these detected changes. To see these details, people in the field, or observations from much closer by are required, which can be difficult in dangerous situations like we see in Zaporizhzhia.
IAEA work to safeguard nuclear power plants in Ukraine
The International Atomic Energy Agency-IAEA Director General published “Nuclear Safety, Security and Safeguards in Ukraine. 2nd Summary Report by the Director General 28 April-5 September 2022“, about the IAEA mission to Zaporizhzhia nuclear power plant. The IAEA Director General presented this report at the Security Council meeting on September 6, 2022.
The report contains seven recommendations that should be implemented by the Ukrainian and Russian authorities to avoid a nuclear accident at the Zaporizhzhia plant.