Detecting Hail Damage Using the GOES Advanced Baseline Imager

In rural areas, verification of damage caused by severe wind/hail events can be difficult because of the lack of observer reports. Satellite data provides a complementary method to observer reports for documentation of damage from hail swaths. Previous research has shown that hail damage is generally associated with more negative changes in the Normalized Differential Vegetation Index (NDVI) calculated from satellite observations over a 2 to 3-week period bracketing the hail storm. However, the value of NDVI change associated with hail damage varies by time of year and the type of vegetation. Therefore, the threshold for the maximum value for NDVI change associated with hail damage must be recalculated with each hail event making automation of hail detection more difficult.

In 2014, storm surveys of hail damage to corn and soybeans were made across portions of Nebraska, Iowa, and South Dakota for 3 events. A fourth hail swath of corn and soybean damage was surveyed in Iowa in 2019. The degree of damage to crops was determined for over one hundred points for each storm. Then the NDVI change was calculated for each survey point providing a relationship between observed hail damage and NDVI change. Using information from the 2014 storm surveys, several methods to find thresholds for NDVI change separating hail damage for no damage will be tested. The Critical Success Index will be calculated for each method in order to determine which has the best skill on the 2014 events. The best performing methods will be tested on the independent data set from the 2019 survey.