Activity Number:
|
562
- Statistical Methods for Multivariate Spatial and Spatio-Temporal Models with Application to the Environment
|
Type:
|
Invited
|
Date/Time:
|
Thursday, August 6, 2020 : 3:00 PM to 4:50 PM
|
Sponsor:
|
Section on Statistics and the Environment
|
Abstract #309308
|
|
Title:
|
A Stochastic Tropical Cyclone Precipitation Field Generator
|
Author(s):
|
William Kleiber* and Steve Sain
|
Companies:
|
University of Colorado and Jupiter Intelligence
|
Keywords:
|
Hurricane;
Lagrangian;
Nonstationary
|
Abstract:
|
Tropical cyclones (TCs) pose serious risks to coastal communities by endangering human safety and property. Coastal flooding is due to combined risk from wind-driven storm surge and intense precipitation driven by warm and wet atmospheric conditions. Only recently have models for TC rainfall been developed in meteorology, with most work focusing on wind-driven storm surge. However, due to the limited nature of available TC data, statistical modeling is beckoned. We introduce a new approach for simulating TC precipitation fields given a small number of TC track predictors such as storm center and central pressure deficit. The model is trained on numerical weather prediction output, and validated by cross-validation. The approach captures TC asymmetry as well as realistic space-time evolution of precipitation fields.
|
Authors who are presenting talks have a * after their name.