Detecting Rainfall Extreme Fields and Their Scaling Using Weather Radar Data

Thursday, 18 December 2014
Ali Hamidi1,2, Naresh Devineni1,2, Ali Zahraei1,2 and Reza Khanbilvardi1,2, (1)CUNY City College, New York, NY, United States, (2)CUNY-NOAA CREST, New York, NY, United States
Information on the probability of extreme rainfall events of various durations is required for hydraulic design in order to control storm runoff. Such information is usually expressed as a relationship between Intensity-Duration-Frequency (IDF) of extreme rainfall. The general IDF curve approach assumes a stationary climate and typically is regionalized based on small number of gauges. However, with the ongoing accumulation of weather radar records, radar-rainfall data represent an alternative to gauging data providing much needed spatial resolution. A clear understanding of the space-time rainfall patterns for events or for a season will enable in assessing the spatial distribution of areas likely to have a high/low inundation potential for each type of rainfall forcing. The Next Generation Weather Radar system (NEXRAD) comprises of 160 Weather Surveillance Radar-1988 Doppler (WSR-88D) sites throughout the United States and at selected overseas locations. Stage IV is a national multi-sensor radar product from NCEP, mosaicked from the regional multi-sensor analyses with 4km×4km and 1h resolution of space and time respectively. In the current study, 11 years of HRAP (Hydrologic Rainfall Analysis Project) gridded Stage IV radar data is employed to generate a relationship between intensity, duration, frequency and the storm exposed area of New York Metropolitan area covering almost 30,000 km2 of the most populous cities at the east part of United States. We investigate the statistical properties of the spatial manifestation of the rainfall exceedances and present the scaling phenomena of contiguous flooded areas as a result of large scale organization of storms. This can be used for spatially distributed flood risk assessment conditional on a particular rainfall scenario. Statistical models for spatio-temporal loss simulation including model uncertainty to support regional analysis can be developed. In this project, we explore a non-parametric multivariate approach and a parametric Bayesian approach to develop stochastic scenarios of spatial extreme rainfall fields for regional risk assessment.