Regionalization of precipitation associated with tropical cyclones using spatial metrics and satellite precipitation

• Published in: GIScience & Remote Sensing
• Main Authors: Yao Zhou, Corene J. Matyas
• Format: Article in Journal/Newspaper
• Language: English
• Published: Taylor & Francis Group 2021
• Subjects: tropical cyclones; satellite precipitation; geographic information system; shape analysis; multivariate clustering; Mathematical geography. Cartography; GA1-1776; Environmental sciences; GE1-350; North Atlantic
• Online Access: https://doi.org/10.1080/15481603.2021.1908675; https://doaj.org/article/e0254b87fb71424ba5c940528390f11e

Precipitation associated with landfalling tropical cyclones (TCs) poses a significant flood risk to vast regions along and inland of the coasts. Quantifying spatial characteristics of tropical cyclone precipitation (TCP) and defining homogeneous rainfall regions can benefit forecasts and hazard mitigation of TCs. This work aims to evaluate the application of spatial metrics and satellite precipitation data in characterizing precipitation associated with landfalling TCs over the North Atlantic. This study applied an object-based Geographic Information System method to measure rainfall fields associated with North Atlantic landfalling TCs from the satellite-based rain rate estimates from 1998–2014. Eleven spatial metrics measuring the entire rainfall field and the largest rainfall polygon within the rain field were evaluated using a set of non-parametric tests to determine if they could distinguish rainfall patterns among four storm intensity categories. A multivariate clustering method with hotspot analysis was utilized to investigate spatial variations and regionalize the rainfall patterns into nine clusters using selected metrics. Five spatial metrics, namely area, solidity, dispersion, closure, and roundness, were selected since they meet three criteria: 1) having a relatively full range of 0– 1 (besides area); 2) having a significant increasing or decreasing trend with intensity, and 3) showing significant differences between any two storm categories. The clustering results indicate that TC rainfall patterns exhibit significant regional variations when storms are over land as well as over sub-basins, including the Caribbean Sea, Gulf of Mexico, and the North Atlantic. The clustering results reveal that intensity is one of the key factors in determining rainfall patterns and reflect the impact of other factors, such as wind shear, moisture content, and interaction with land.