Satellite imagery, AI and geospatial technologies are being extensively used in conflict zone Ukraine for not just search and rescue missions, relief and rehabilitation efforts, and frontline emergency communications, but for accurate assessment of the magnitude of infrastructure damage as well.
All war is a symptom of man’s failure as a thinking animal
~John Steinbeck
When American author John Steinbeck visited the enthralling Kyiv in 1947, it was still recovering from the destruction wrought by the Second World War, and being rapidly reconstructed to match its majestic grandeur.
Decades later, Ukraine has become the turf of the biggest conflict in Europe ever since the World War, which has upended the security architecture and brought the continent to a standstill, reviving the anxieties of the Cuban Missile Crisis, and endangering the future of energy security and green transition.
The war has torn down critical infrastructure facilities, industrial enterprises, roads, highways, residential apartments, and farms, reducing large parts of Europe’s largest country and formerly vibrant cities to rubble.
Unpredictability looms about which direction the war will take. Ukraine continues to make advances in the four eastern regions annexed by Russia, while the Kremlin’s missiles are hitting as far as western part of the country.
As critical civilian infra, including power plants is being hit, accurate damage assessment is essential for post-war repair, reconstruction, and rehabilitation efforts.
According to a recent statement by Ukrainian President, Volodymyr Zelensky, “30% of Ukraine’s power stations have been destroyed, causing massive blackouts across
the country.”
As a result of the conflict, Ukraine’s economy is expected to contract by 35% this year, and more than 14 million people are estimated to have been displaced so far, says a World Bank report.
The war has also dimmed the prospects of a post-pandemic economic recovery for emerging and developing economies, exacerbating record-breaking inflation and acute food and fuel crisis.
If World Bank estimates are to be believed, recovery and reconstruction across social, productive, and infrastructure sectors would require at least $349 billion, which is more than 1.5 times the size of Ukraine’s pre-war economy in 2021.
“Russia’s invasion of Ukraine has triggered one of the biggest human displacement crises and exacted a heavy toll on human and economic life,” said Anna Bjerde, World Bank Vice President for the Europe and Central Asia region.
“Ukraine continues to need enormous financial support as the war needlessly rages on as well as for recovery and reconstruction projects that could be quickly initiated.”
ALSO READ: How Geospatial Helps in Monitoring and Eradicating Plastic Waste?
Tracking the War Toll
Since the onset of the Ukraine war in February 2022, the Kyiv School of Economics (KSE) Institute, a leading think tank and the analytical unit of the KSE, along with the Office of the President of Ukraine, Ministry of Reintegration of Temporarily Occupied Territories of Ukraine, Ministry of Infrastructure of Ukraine, Ministry for Communities and Territories Development of Ukraine has been spearheading an initiative called Russia will pay/ damaged.in.ua.
The initiative collects, evaluates, analyses, and documents information provided by government agencies, local authorities, citizens, and others, on the damage inflicted on civilian infrastructure resulting from the Russian invasion. It is effectively an opensource data project to evaluate and track the cost of war.
According to the KSE Institute, the damage inflicted on the country’s infrastructure has been estimated at a cost of $114.5bn as bombs devastated buildings, public utilities, power plants, and road networks.
“We collect the data in every possible way and then try to verify and compare these data sources. We cooperate with both local and state authorities that receive information, be it from our own reporting channels, from credible media, business associations or individual witnesses,” says Maksym Nefyodov, the KSE Institute Project Lead.
They use other digital tools which structure and facilitate the process, such as their website, the Telegram chatbot and the widely popular Kyiv Digital and Diia apps originally used for access to state services.
“We also collect the data from ortho-rectified drone images taken in liberated zones. For war zones and larger communities we cooperate with companies proficient in analyzing satellite imagery,” adds Nefyodov.
Data to Rescue
To assess the damage and procure accurate data about the war-torn infrastructure, KSE Institute has partnered with Tensorflight, a Polish/US company that uses satellite, aerial, and street-level imagery to automate commercial property inspections and claims to process helping underwriters with risk assessment and to better understand their portfolio exposures.
Since April 2022, Tensorflight has been providing and analyzing imagery from state-of-the-art satellites to assess the damage caused to infrastructure within Bucha, Irpin and Mariupol, on a pro bono basis, to the KSE Institute. These cities had a combined pre-invasion population of almost 570,000 and span an area of approximately 387 square kilometres.
This partnership is enabling Ukraine to access additional expertise in satellite, aerial, and street level imagery to eliminate inaccuracies within data, and to develop Machine Learningmodels to speed up imagery processing.
With the help of high-resolution satellite imagery, combined with geolocation, and AI, infrastructure damage is being determined.
Need for Accurate Assessment
Till date, 61188 buildings in Mariupol, 3927 in Irpin, and 2510 in Bucha, have been identified and analysed using Tensorflight’s Deep Learning model.
In addition to supporting the KSE Institute to track the cost of war, the project is also helping the Ukrainian government assess the reconstruction requirements of the country.
For example, the intelligence will help inform whether to priorities the rebuilding of a power station or to prepare for the winter months ahead or a hospital or school based on the needs of residents in affected areas.
“The total amount of infrastructure damage since the start of war has reached hundreds of billions of dollars and is increasing on a daily basis. Tracking the damage resulting from the Russian invasion is an immense undertaking, and we are providing our expertise pro bono, to support the KSE Institute,” said Jakub Dryjas, CEO of Tensorflight.
According to Nefyodov, as access to information is improving, the project seeks to go down to the most granular level of analysis.
Ukraine’s Ministry of Communal Development is building a GIS-based Geoportal that will provide data-based insights and analytics, allowing accurate planning of reconstruction efforts employing the latest urban development concepts and the build-back better principle, which are usually considered out of bounds for embattled communities on the verge of destruction.
“We anticipate to analyse even more images of Ukraine’s infrastructure to ensure data-driven decisions during the recovery process of impacted areas,” Nefyodov added.
“We also are a part of the RISE.org.ua coalition that aims to design transparent and accountable ways of managing these recovery projects once they start, as well as proper civil oversight and reporting to the donor community.”
Identifying Damaged Areas
War-torn areas are difficult to access and so is getting a clear picture of the infrastructure. Scheduling an optical satellite requires a number of conditions to be met: daylight, cloudless weather, appropriate position of the satellite and free time in the satellite imaging schedule.
“The technology used to detect damaged buildings in areas not accessible by drones or planes is our standard product modified and optimized for the detection of damage caused by shelling”, said Władysław Surała, Head of Partnerships, Tensorflight.
“We rely on high resolution satellite imagery analyzed by neural networks trained specifically to detect building damage. The images are taken from the top (~0 degrees from the axis) and oblique (30-50 degrees from the axis) so as to not omit damaged buildings with roofs still intact. As we collect more imagery of urban war zones and have more examples to re-train the model, our models will incrementally improve,” he added.
Surala explains that after the imagery is delivered to their servers, they pre-process to sharpen it. If there is auxiliary data available, such as footprints of the buildings, this is the step when they prepare it – they load the coordinates of the footprints to help detect the buildings.
ALSO READ: Cloud, 5G to be Decisive Technology Trends in 2023: Study
Next, the image is loaded into the neural networks, designed and trained to detect the buildings, and then they are classified as damaged or intact.
Tensorflight’s Deep Learning models then estimate the level of the damage to the buildings. This is further categorized into classes indicating the degree of damage, e.g. from completely demolished, to suitable for repair.
On top of that, the engine can load historical data about the damaged building and estimate the cost of reconstruction – this is also part of their standard product.
“Another area of our project’s expansion is accounting for more complex types of damage and losses – ecological, de-mining, lost harvest, movable property, and many more.
A very sad example is damage to the properties of deceased owners. The next stages of our work will gradually cover these areas,” Nefyodov added.
