On April 25, 2015, a magnitude - 7.8 earthquake caused widespread building damage in central Nepal. The Italian Space Agency’s COSMO-SkyMed Synthetic Aperture Radar (SAR) satellite acquired data over Kathmandu – a 50 x 50 km area – four days after the earthquake. Using the SAR information, Sang-Ho Yun and other researchers of the Advanced Rapid Imaging and Analysis (ARIA) team at NASA’s Jet Propulsion Laboratory and California Institute of Technology produced a damage proxy map showing areas of potential building damage.

Sang-Ho Yun
Sensor Technology: How were you able to create the damage proxy maps?

Sang-Ho Yun: We compared before and after images. It’s more sophisticated than just a simple comparison. We use an image called a coherence map.

Sensor Technology: What is a coherence map?

Sang-Ho Yun: A coherence map shows similarity between two radar images, pixel by pixel. High-coherence pixel value means that the ground objects under that pixel remained the same between the two timings of the radar image acquisition, whereas low coherence means that the objects changed during that time. We take the difference of the two similarity maps; therefore we can isolate a change from other naturally occurring changes. The high pixel values in the map directly indicate high probability of damage, or high degree of change.

Sensor Technology: Who used the data?

Sang-Ho Yun: The map was used by the National Geospatial-Intelligence Agency (NGA) to determine priority areas for their analysis. Six days after the earthquake, the Office of US Foreign Disaster Assistance confirmed a strong spatial correlation between the map and realities on the ground. Seven days after the earthquake, Japan Aerospace Exploration Agency’s ALOS-2 [satellite] acquired SAR data over much larger areas (70 × 180 km). Nine days after the earthquake, the DigitalGlobe [an American commercial vendor of space images] decided to use the damage proxy maps to determine where to focus their collection areas for high-resolution imagery. On the following day, we produced damage proxy maps from ALOS-2 data, which revealed potential building damage as well as landslides.

Sensor Technology: Were the maps available to the public?

Sang-Ho Yun: Damage proxy maps were made available to the public and responding agencies through the ARIA team website at http://aria-share.jpl.nasa.gov/ , and the US Geological Survey’s Hazards Data Distribution System (HDDS) website at http://hddsexplorer.usgs.gov/ . From the ARIA website alone, there were 3,198 downloads of the maps in May 2015.

Photos of the damage in Bhaktapur, Nepal, are overlaid on a damage proxy map derived from COSMO-SkyMed satellite data. Colors show increasingly significant change in terrain/building properties (including surface roughness and soil moisture). Red is most severe. (Image Credit: NASA/JPL-Caltech/Google/DigitalGlobe/CNES/Astrium/ Amy MacDonald/Thornton Tomasetti)
Sensor Technology: What are the advantages of Synthetic Aperture Radar?

Sang-Ho Yun: The imaging radar sees through clouds and can image during day and night. So it has great potential to be used for rapid disaster response.

Sensor Technology: What can be seen specifically that is so valuable for recovery teams?

Sang-Ho Yun: The expanse of the imaged area is quite large: 50 × 50 kilometers and 70 × 180 km. The comprehensive map suddenly reveals the areas of potential damage. So you immediately have an idea of what areas are heavily affected. A lot of first responders and the decision-makers can look at the map and then come up with an informed decision as to where they might want to put their resources first.

Sensor Technology: How long does it take to acquire data in a damaged area?

Sang-Ho Yun: The first latency that we have is the data acquisition latency, which is the amount of time that we have to wait until we see the first radar satellite, when you look at the sky from the affected area.

If we take into account all the existing radar satellites, the expected wait time is already within a day. That’s very encouraging. In fact, we had missed the first data acquisition opportunity; there was an Italian satellite that passed over the area six hours after the earthquake. The second [satellite passed over the area] a day after the earthquake.

This is the great potential. We can better coordinate in the future with other space agencies and produce this kind of product much faster.

Sensor Technology: What were the challenges for this type of disaster sensing?

Sang-Ho Yun: We’re using a radar sensor; that does not necessarily correspond with what we would see with our eyes on the ground. So we need to sort out the similarity between what the radar sees and what we would see. That’s where a lot of study is warranted, and we are making progress in that direction. Radar can see very subtle changes that people cannot see, like soil moisture content change. We don’t see that very well on the ground, but radar does. We want to identify it and reduce the rate of false alarm in the product.

Sensor Technology: What is the difference between what we see and what the radar sees?

Sang-Ho Yun: We see what is normal and what is damaged. Radar sees how much the object became rougher or smoother (compared to radar wavelength), and perhaps how much the ground became wetter or drier.

A damage proxy map, derived from ALOS-2 satellite data, shows images of Dhunche, Nepal. Unreinforced stone masonry buildings (shown) resulted in many collapses following the earthquake. (Image Credit: NASA/JPL-Caltech/Google/DigitalGlobe/Amy MacDonald/Thornton Tomasetti)
Sensor Technology: Where else have the maps been used?

Sang-Ho Yun: This is a change-detection map. We see changes on the ground; that can be a challenge when we don’t know exactly what caused that change. It could be a soil moisture change on the ground; it could be building damage; it could be landslide debris; it could be a newly established group of tents for a shelter for the victims. The radar sensor sees whatever is changing on the ground; it doesn’t discriminate for building damage or landslide damage.

On the flip side, it’s also an advantage that we can apply this technique to many, many different applications. In fact, we did detect a landslide, which was induced by the Nepal earthquake. We also saw great potential on detecting ash fall damage from volcanic eruption. There are many applications. If there’s ground surface change, there’s a great potential for us to apply this technique.

Sensor Technology: What are you working on now?

Sang-Ho Yun: Most recently, I applied the Synthetic Aperture Radar technique to detect and produce a flood-extent map; there was a Midwest flood occurring over [a span of] three weeks, starting toward the end of last year along the Mississippi River.

Sensor Technology: How was the SAR used to create a flood-extent map?

Sang-Ho Yun: Flooded areas often undergo surface roughness change. This change shows up as brightness change in radar images. Thus, simple comparison of before and after radar images already gives a good idea of areas of potential floods.

Sensor Technology: What are other possibilities for damage proxy maps?

Sang-Ho Yun: Other than detecting damage of artificial structure, there have been other uses of radar images, such as mapping of liquefaction damage, landslides, forest damage, storm surge damage, tropical cyclone damage, tsunami damage, and damage from volcanic ash fall and lava flow. It is also very well known that SAR images are used to precisely measure ground deformation caused by earthquakes, volcanic activities, ground water level change, and so on. Such techniques are also useful to measure the velocity of glaciers, as well as to study spatial variation of water vapor content in the troposphere and total electron content in the ionosphere. With more and more SAR missions in orbit, I think we’re reaching the capacity of where we can and should apply the technique for more direct, societal benefits. I’m happy to see an increased level of awareness of the usefulness of the technology. We’ll be continuing to improve this technique to produce a better product for better future response.

For more information, visit www.jpl.nasa.gov .

Sensor Technology Magazine

This article first appeared in the June, 2016 issue of Sensor Technology Magazine.

Read more articles from this issue here.

Read more articles from the archives here.