W3 Seminar: Variations in moisture surrounding hurricanes

Details

Speaker:
Corene Matyas, Ph.D., Professor, Geography Department, UF

Abstract:
Understanding the spatial patterns of moisture surrounding hurricanes is important to forecasting storm structure and rainfall production. This talk will highlight two different methods our research group used to assess to examine total column water vapor (TCWV) patterns around North Atlantic hurricanes during 2000-2021. One approach uses a convolutional autoencoder (CAE) with k-means cluster analysis to process over 4600 images. The second uses a principal components analysis on a smaller sample of cases to avoid autocorrelation and with high-shear cases removed. Both approaches identified four distinct spatial distributions of TCWV in terms of amount, symmetry, and areal extent, and three of the four patterns are very closely-matched between the two methods, suggesting these patterns are robust. Hurricane intensity, size, and precipitation features vary among the four clusters, characteristics which are closely associated with the moisture and environmental conditions of each cluster.