Assessing hurricane impact on vegetation and endangered deer habitat using airborne lidar and multispectral images

Authors: Jiyeon Kim, Sorin C. Popescu, Roel R. Lopez, X. Ben Wu, Nova J. Silvy

In the lower Florida Keys, the endangered Florida Key deer and numerous other wildlife species inhabit a vulnerable island environment susceptible to storm surges and rising seawater due to low elevation and flat terrain. Timely and reliable assessment of vegetation damage from natural disasters, such as Hurricane Irma, is crucial for effective habitat management. The study’s overall objective is to examine Hurricane Irma’s impact on vegetation on No Name Key, Florida, using remote sensing. The study relates the area change in vegetation obtained from remote sensing analysis to Florida Key deer population changes following the storm. The methodology involved performing a thematic change detection analysis using the following data sources: (1) aerial multispectral images (for pre- and post-Hurricane), (2) airborne lidar data (for pre- and post-Hurricane), (3) an existing vegetation map, and (4) soil data. A Support Vector Machine (SVM) image classification algorithm was applied to pre- and post-storm input image stacks to create pre- and post-Hurricane Irma vegetation maps. We were then able to obtain the area change information (for various vegetation categories) by performing the change detection analysis of the 2 SVM-classified images. The differences in areas following the storm were calculated for 7 affected vegetation types. Using the area change information following Hurricane Irma, we estimated the number of deer supported by the storm-affected vegetation. These estimated deer numbers, based on the area differences in post-Hurricane Irma vegetation types, were compared to observed deer numbers collected during the post-Hurricane Irma Texas A&M Natural Resources Institute (NRI) deer field survey. The results showed the following: mangroves had the largest negative area changes (area loss), followed by pinelands, hardwoods/hammocks, developed areas, and buttonwoods. Freshwater marshes had the largest positive area changes (area gain). The deer's preferred vegetation areas had decreased post-Hurricane Irma, resulting in a reduced deer population compared to pre-storm numbers. The predicted number of the Key deer post-Hurricane Irma fell within a 95% confidence interval of the observed deer population from the post-storm field survey. The study findings and techniques could be applied to study climate change impact, especially sea level rise. This methodology can be valuable in assessing the impact of storms on other wildlife species in similar environments. The applications and methodology are especially relevant considering the increasing frequency and intensity of storm surges and the accelerating rate of sea level rise.

Suggested Citation