Sunday, 6 January 2019
Hall 4 (Phoenix Convention Center - West and North Buildings)
Remote sensing analysis is performed on hyperspectral imagery with the intention of identifying tornado damaged swaths from the 2011 Super Outbreak in west-central Alabama. Hyperspectral imagery provides a greater number bands that are useful for calculating broadband and narrowband indices. Traditional spectral or multispectral imagery does not have the additional bands necessary to calculate the narrowband index values. The vegetation index values can be good indicators of tornado damaged regions, and their effectiveness to identify tornado damage paths will be analyzed in this study. Specifically, a number of index values were computed including, moisture stress index (MSI), normalized difference infrared index (NDII), and Normalized Difference Vegetation Index (NDVI). They are also used in a supervised classification algorithm that identifies damage within the satellite imagery. The greatest difference between damaged and undamaged regions occurred with the moisture related indices NDII and moisture stress index, although the NDII has a larger variance associated with that specific index. Overall the supervised classification worked well for more intense (EF4) tornado damaged regions that for the less intense damage (EF3 or less). The results of this work suggest that moisture related indices would be best suited for future work in evaluating tornado damaged regions through hyperspectral imagery.
![](/data/abstract/ams/2019Annual/Paper_356094_abstract_17878_0.png)
![](/data/abstract/ams/2019Annual/Paper_356094_abstract_17878_0.png)
![](images/flag.png)
![](images/blue_ribbon.png)