Improving disaster response from space

From September 28 to October 10 of this year, Hurricane Matthew swept through Haiti, Florida and the Carolinas, leaving communities scrambling to beat back the onslaught of floodwaters. First responders, government agencies and insurers needed to know which areas required immediate attention. However, in many of the hardest-hit locations, damage to infrastructure meant that there was no viable way to gather that data on the ground.

FirstLook, DigitalGlobe’s online subscription service for emergency management, offers fast web-based access to pre-event and post-event imagery, plus updates as our constellation continues to collect data on a priority basis.

And when you’re looking for more than a picture, GBDX, DigitalGlobe’s geospatial big data platform, has the tools to turn satellite images into actionable data. Using GBDX, you can integrate DigitalGlobe’s trained neural network algorithms with open-source data sets like OpenStreetMap. In the case of Hurricane Matthew, we found this layering particularly useful in identifying underwater and at-risk infrastructure.

Here, we performed a water extraction on images collected before and after the hurricane.

Lake growth in Cavaillon, Haiti

The lake’s area has grown significantly in the hurricane’s wake. Now, you can see that structures (represented by red dots) that were previously safe are now at risk of flooding.

River changes course, Les Anglais, Haiti

In some cases, a body of water doesn’t just grow—it completely changes course. In the pre-storm image above, you can see that the river in Les Anglais, Haiti had a wide bow that bent away from the structures in the left side of the frame. Now, after the storm, the blue lines of our water mask show that the river’s course has changed completely.

Pre- and post-event imagery

Water extraction and water extraction with OSM data

The ability to determine exactly what infrastructure is underwater is one of the most powerful applications of GBDX for disasters like Hurricane Matthew. Analyzing this information at scale is no trivial task and automating the steps of taking pre-event imagery all the way to post-event plus meaningful insight is not easy. In the above four images of an area of Goldsboro, North Carolina, we step through the process of turning the imagery into something actionable.

How many structures have been impacted—and to what extent is the road damaged, too? Using the GBDX-ArcMap add-in, we can pull in water vectors and query for OSM features that are within the water. With this extraction, we get an output of the specific impact within the AOI:

Using GBDX-ArcMap add-in we can pull in water vectors and query for OSM features that are within the water.

From counting the number of impacted structures to showing change over time, there’s so much more that GBDX can do with these FirstLook images. If you’re working in the affected areas, let us know what else you’re looking for and we’ll do our best to engineer accurate, cost-effective solutions.