Spatial Extreme Precipitation Modeling for The Pacific Northwest United States Using Satellite Information and Max-Stable Processes
Abstract
Extreme precipitation models are crucial for a wide range of engineering applications, including planning and designing hydrologic infrastructure. The conventional approach for modeling extreme precipitation is based on the regional frequency analysis methods together with the ground-based observations. There are a few restrictive assumptions underlying the regional methods. As a result, the sparsity of the rain gauges, especially in remote and high-elevation areas, tends to undermine the application of the conventional approaches. Max-stable processes, as the extension of the extreme value theory to infinite dimensional scale, can model the marginal behavior and the dependence structure of the spatial extreme rainfall data, simultaneously. In this study, PERSIANN-CDR, a long-term satellite-based precipitation dataset has been used together with max-stable processes to model extreme precipitation in the Pacific Northwest (PNW) United States. We first obtain the extreme rainfall estimates from PERSIANN-CDR, which are bias-adjusted with the ground-based observations. Then, various parametric max-stable models including Schlather, Brown-Resnick, and Extremal-t processes are fitted to the data. Various geographical covariates are considered in modeling of location, scale, and shape parameters of the GEV distribution. Finally, the models are evaluated and compared based on their abilities to model the marginal behavior and the dependence structure of the spatial extreme data. Results of this study show that there exists an apparent spatial structure in the extreme precipitation data. Also, it is observed that the Brown-Resnick and the Extremal-t processes provide similar and satisfying performances in estimating the marginal parameters and the spatial dependence structure of the extreme data. Results of this study also demonstrate that the PERSIANN-CDR dataset is a valuable tool for spatial extreme precipitation modeling in the poorly gauged regions.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFM.H52A..08F
- Keywords:
-
- 1807 Climate impacts;
- HYDROLOGY;
- 1817 Extreme events;
- HYDROLOGY;
- 1821 Floods;
- HYDROLOGY;
- 1854 Precipitation;
- HYDROLOGY