Timeliness in recognizing damaged buildings after an earthquake can save lives. Researchers in Japan recently took advantage of the unusual occurrence of two strong earthquakes that jolted the city of Kumamoto about a day apart to investigate the prospects for using airborne lidar surveys to hasten rescues from compromised buildings.
Lidar is an acronym for light detection and ranging. The technology uses pulsed laser light to measure distances to a target. When a laser mounted on an aircraft scans terrain below, the resulting data can feed the construction of high-resolution, three-dimensional digital maps. These maps include accurate renderings of the elevations of structures and topographic features within the scanned area.
The study indicates that if there are prequake and postquake lidar surveys of a stricken city, lidar may outperform aerial photography in providing accurate detection of damaged buildings, and the comparison could be automated. Lidar also works over a wider range of visibility and weather conditions than aerial photography or satellite imaging does.
“The faster we [pinpoint] the collapsed buildings, the faster we can send for help.”
The new results suggest that earthquake-prone cities may benefit from regular lidar surveys that would enable them to have an up-to-date basis for comparison when the next temblor strikes. If lidar methods can automatically detect collapsed buildings after an extreme event, they can better inform responders where people might be trapped under rubble or where supplies are needed, said Luis Moya, an engineer at Tohoku University in Japan. He is the lead author of a Natural Hazards and Earth System Sciences paper on the research published last month. “The faster we [pinpoint] the collapsed buildings, the faster we can send for help,” Moya said.
One-Two Punch
On 14 April 2016, a magnitude 6.2 earthquake struck Kumamoto Prefecture on the island of Kyushu in Japan. A second quake of magnitude 7 struck 28 hours later. Given the magnitude of the first quake—now considered the foreshock—the stronger main shock on 16 April took seismologists and emergency planners alike by surprise. The two events directly caused 50 fatalities.
Most remote sensing data are collected only after an event, and because the foreshock was initially assumed to be the main event, an airplane surveyed the area on 15 April using lidar. In this unusual case, however, the airborne survey was repeated 8 days later, after the main shock had occurred, ultimately providing Moya and his colleagues an unexpected chance to test lidar’s usefulness for earthquake damage assessment.
Assessing the Potential
Although the resolution of aerial photographs has been improving, using them to identify individual compromised buildings remains difficult.
In most cases, postevent remote sensing information takes the form of aerial photographs. Although the resolution of those images has been improving, using them to identify individual compromised buildings remains difficult, explained Moya. In addition, cloud cover and darkness limit collection of optical imagery, whereas lidar works day or night, with or without clouds.
The Kumamoto main shock permanently shifted the ground in some areas. To make certain that they could still compare identical points on each building despite the shifting of entire structures, the researchers corrected the postquake data set using a method that they describe in a 2017 paper.
Now that the researchers have figured out how to make the adjustments, remapping the postquake survey to its prequake cousin could go much more quickly, Moya told Eos, for instance, by using speedier software or other modifications that so far remain to be tested. In cases where the target area is distant from the earthquake source or the magnitude of the quake isn’t too large, the permanent deformation may be small enough to neglect, he noted.
Follow the Footprints
With the surveys realigned and using building footprint data from the Geospatial Information Authority of Japan, Moya and his coauthors examined horizontal and vertical changes between the surveys for buildings with a variety of damage patterns. After analyzing simple parameters that could be used for automatic detection, the research team found that one parameter—the average height difference between the two surveys within the building footprint—had detection accuracy similar to that of all the parameters combined.
When they evaluated aerial lidar’s success at collapsed building detection, Moya’s team determined that the technique achieved its greatest accuracy (93%) for structures that had lost 0.5 meter or more in height.
Not all other detection methods consider the vertical displacements of structures, according to Chris Renschler, a geographer and investigator with the Multidisciplinary Center for Earthquake Engineering Research at the University at Buffalo in New York who was not involved with the study. But he thinks it’s important. When comparing before and after optical imagery, even with high resolution, it’s difficult to detect a damaged building that dropped down a story with only minimal horizontal change.
When Moya’s team evaluated aerial lidar’s success at collapsed building detection, it determined that the technique achieved its greatest accuracy (93%) for structures that had lost 0.5 meter or more in height. To come up with that accuracy, the researchers compared their lidar results with the findings from a field assessment of damage conducted by another research group (that included one of Moya’s coauthors) in Japan that was studying the impacts of the same pair of quakes, Moya said.
The Future of Damage Assessment
Rapid response to an extreme event is possible if accurate information is continuously gathered before the event occurs and can be followed by collecting the same kind of data after the event for quick comparisons, according to Renschler. “What is often forgotten is that emergency managers may not have the opportunity of gathering such data, but once something happens, they are really in need of that information,” he said.
Decision makers in earthquake-prone communities should take note of the potential of the lidar methods demonstrated in this study, he added. “With the technology getting cheaper, communities may want to do this assessment on a continuous basis, so that they are updated,” he said.
In the meantime, Moya isn’t waiting for decision makers. His next step is to try to combine lidar data with other sources of data to explore the possibility of identifying damaged structures with only postevent information.