Skip to content Skip to main navigation Report an accessibility issue

Danny Burrow’s Tornado Damage Research

Surveying Tornado Damage, and the Limits in Rural Areas


One of the challenges in surveying tornado damage is that our methods of tornado intensity estimation are most applicable to human-made buildings and objects. When a tornado damages rural areas where buildings are few and far between, it can be difficult to reconstruct how strong the tornado’s winds were, how wide its path was, and other characteristics. Remote sensing is a powerful tool in these situations because it allows us to quantify environmental change—or in this case, vegetation damage caused by the tornado. In this study, we used high-resolution satellite imagery to study damage caused by three tornadoes that occurred in Alabama and Georgia in March of 2019. We derived a few commonly used spectral metrics from this imagery and correlated them to estimates of tornadic wind speeds made by National Weather Service meteorologists. The metric that was best correlated with estimated wind speeds was calculated by subtracting normalized difference vegetation index (NDVI) values seven days before the tornadoes from NDVI values three days after the tornadoes. Then, we examined how these NDVI difference values varied across the width of the three tornado paths, and compared these patterns to a model, called a Rankine vortex, of how wind speeds vary across the width of a tornado. We found that NDVI difference patterns compared well to the Rankine vortex model for the strongest of the three tornadoes, but not nearly as well for the two weaker tornadoes. While it would have been great to see strong comparisons for all three of them, it is still exciting because it tells us that we can make inferences about the dynamics of strong tornadoes using freely available satellite imagery. As technology advances and super-high resolution imagery becomes more accessible to researchers, we are expecting lots of new studies similar to this one that will allow us to explore just how tornadoes behave so we can better protect people from these deadly hazards.

These are maps of spectral indices we calculated from satellite imagery along the paths of the tornadoes. The yellow and red lines indicate the width of the tornado paths, as determined by NWS meteorologists. The metric that gave us the best results was NDVI difference, shown in (b). The brown areas in (b) represent areas of vegetation damaged by the tornado.

The above is a graph of how NDVI difference varied across the path width of the strongest tornado we studied. Note the smoothing spline (red) and how it compares to the pattern of wind speeds across the width of the Rankine vortex model.

The above graph represents how maximum ground-relative wind speeds vary across the width of a tornado in the Rankine vortex model. In this graph, x = 0 represents the center of the tornado path. The strongest wind speeds are on the right side of the path, relative to the direction that the tornado moves. Wind speeds decrease quite rapidly outside the radius of maximum wind speeds (x = -1, 1 in this figure). This pattern of wind speeds compares well to the NDVI difference values in the other figure.

Helpful links:

Read the full paper here (open access):

Here’s a post from NWS Birmingham meteorologists on how tornado damage surveys are performed:

Here’s a brief report on how satellite imagery is useful for tornado damage assessment:

Find Danny Burrow Here!

UTK Geography personal page:

ORCID page:

Google Scholar: