The NASA Earth Science Disasters Program and NASA SPoRT has been working with the National Weather Service and the Damage Assessment Toolkit (DAT) team for a number of years on the production of hi-resolution satellite imagery for inclusion in the DAT, and training on the usage of the imagery. Imagery from the Landsat satellite at 30 m resolution is available regularly in the DAT as swaths become available, but higher resolution imagery from sources such as Sentinel (~10 m) and Worldview (<1 m) are available upon request through the USGS Hazards Data Distribution System. These images are then processed by members of SPoRT and the Disasters Program for inclusion in the DAT.
Several tornado events earlier this year have helped to illustrate the effectiveness of the imagery for post-storm analysis and damage assessment. This is especially true when damage is of sufficient magnitude and spatial extent to be resolved by the particular satellite instrument and cloud-free (or near cloud-free) conditions exist at the time of the satellite pass. On April 12-13, a number of tornadoes developed and moved across the Southeast region of the U.S., some of which produced damage up to EF-4 scale in parts of southeastern Mississippi. Meanwhile, closer to the Tennessee Valley, EF-2 tornado damage occurred in portions of the Huntsville and Birmingham, AL County Warning and Forecast Areas (CWFAs), with a tornado up to EF-3 strength damaging portions of Chattanooga in the Morristown, TN NWS CWFA.
This first image of damage is along a small portion of Interstate-65 in Cullman County, AL, and just south of the city of Cullman (Image 1 below). Notice the area of damaged vegetation, which consisted of downed trees to the north of the preliminary tornado path and just east of Interstate 65 (red circle). Although some of this area was shaded by clouds, damaged trees were clearly evident. The location of damaged vegetation allowed for a relocation of the damage indicator (green triangle) farther north, which was originally closer to the green line.
A zoomed-in look at the damage also shows that trees were not just downed or uprooted, but snapped. Imagery from sources such as Worldview is of sufficiently high resolution to see snapped tree trunks, which will often be indicated by small, bright dots against the darker green/brown background, due to the higher reflectivity of wood on the inside of the tree.
Farther to the east, along the same damage path in Cullman County, a damage swath was previously undetected due to its distance from the road network that made visual ground inspection impossible (without a trek through the woods!). However, the advantage of satellite imagery allowed for assessment of the tree damage (Image 3 below). In Image 3, the blue triangle denotes the damage point that was added thanks to the presence of tree damage indicated by the satellite image. Notice the lack of small bright dots as in Images 1 and 2 (in the red circle), suggesting that most if not all of these trees in this area were uprooted rather than snapped.
Another advantage of the DAT is the ability to underlay maps or other baseline imagery below the near real-time hi-res imagery. Not only does this allow for proper geographic referencing, but the ability to assess vegetation type and state at a previous time. Image 4 (below) is one such representative image in the DAT taken from the cold season. Notice the lack of leaves and long shadows extending from the hardwood trees in the image. However, the softwood trees, likely consisting of cedar and pine species, remain green. Notice that the blue triangle was located within a patch of softwood trees, allowing for a better identification of tree types and thus the proper Damage Indicator to assess the wind rating for the uprooted trees.
Lastly for this post, we’ll take a look at another area of damage in Cullman County, just downstream and farther east along the same damage path. This was also an area of previously unknown damage due to the lack of a specific report and the inability to view the damage from a nearby roadway. The tree damage in this location was similar to the damage shown in images 1 and 2, with numerous bright dots indicating trees were likely snapped.
The background reference image (Image 6 below, same imagery type as used in Image 4) from the same location however, showed that much of this scene was dominated by hardwood trees.
This allowed for the proper Damage Indicator again to be used suggesting a higher wind speed rating than the damage from the previous scene (Image 3). Due to the presence of many snapped hardwood trees, the damage here was rated EF1.
So, in these cases, we demonstrated several uses of the high-resolution imagery for damage assessment: the ability to better affix damage points, locate damage otherwise hidden from typical roadway viewing, detect uprooted versus snapped trees, and identify primary tree types that were damaged in a location and assign proper damage indicators.
We would like to thank members of NASA MSFC, and the USGS who helped make this post possible, along with Digital Globe for permission to use and distribute the imagery in these examples.
-Kris and Lori
You must be logged in to post a comment.