Researchers at Ohio State University have adapted algorithms originally developed to detect meteor craters on the moon to instead search for bomb craters to help find Vietnam War-era unexploded bombs in Cambodia.
Erin Lin, assistant professor of political science at Ohio State, says demining—the process of finding and safety removing unexploded bombs and landmines—has been ineffective in Cambodia. A number of factors play into this, including laborious, inefficient means to find high-density areas, dangerous and inadequate removal practices, and the Cambodian National Clearance Agency’s focus on low-risk areas.
For their study, Lin and co-author Rongjun Qin chose a 100-square-kilometer area near the town of Kampong Trabaek. The area was the target of carpet bombing by the U.S. Air Force from May 1970 to August 1973, and is an ideal site given the array of terrains surrounding bomb craters, including rice paddies, peri-urban development and riverplains.
“This area also represents a ‘most likely’ case for finding a high ratio of undetonated to detonated bombs,” Lin explained to Laboratory Equipment. “Only 119,857 m2 (or 0.12% of the image) have been cleared by professional deminers, despite the Cambodian Mine Action Center labeling this region as a high priority. Within that cleared space, deminers found two general purpose bombs and hundreds of scrap metal pieces. This ratio highlights the recurring inefficiency in the clearance process, particularly the difficulty that deminers face in distinguishing bombs from leftover metals.”
Cambodian farmers working in contaminated fields told Lin and her team that rusty, unexploded bombs resemble other objects commonly found on the ground, such as rocks, scrap metal and even edible fungus.
The study, which was recently published in PLOS One, was conducted in two stages. In the first stage, Lin and her team used algorithms developed to detect meteor craters on the moon and planets. While the algorithm was effective, it wasn’t enough given key differences between bomb and meteor craters. For example, due to erosion levels and vegetal overgrowth, bomb craters highly vary in appearance, whereas meteor craters are relatively stable over time. Bomb craters are also much smaller, especially from a remote sensing perspective—bomb craters measure 3 to 12 meters in diameter while meteor craters can be up to 3,000 meters. Thus, for the second stage, the researchers further adapted the algorithm to consider the novel features of bomb craters.
The first stage of the model identified 89 percent of true craters (157 of 177), but also identified 1,142 false positives. The second stage eliminated 96 percent of the false positives, while losing only five of the real bomb craters. The resulting accuracy rate was about 86 percent, identifying 152 of 177 craters. This proposed method increased true bomb detection by more than 160 percent.
“We’re hoping our method can help operational teams more safely traverse conflict-affected regions,” Lin said. “Beyond logistical support, this method can also help guide policy to set the foundations for long-term growth in areas that still suffer from the threat of violence. For instance, since the process of demining is an expensive and time-intensive one, this framework helps identify the most vulnerable areas that should be demined first. We also hope that our study can inspire more academic work that examines the effects of unexploded ordnance exposure to poverty, agricultural productivity and farmer risk tolerance.”
Photo: Bomb craters on Earth have similarities to meteor craters on the moon. Credit: Ohio State University.