3R.7
Ensemble generator of radar-based QPE
Grzegorz J. Ciach, IIHR Hydroscience & Engineering, Iowa City, IA; and W. F. Krajewski, G. Villarini, D. Kitzmiller, and R. A. Fulton
Large uncertainties associated with operational precipitation estimates produced by U.S. national network of WSR-88D radars are well-acknowledged. These errors are due to numerous sources, including both systematic and random effects. The propagation of these uncertainties in all those models where rainfall is used as input can cause large errors in the model output. However, a comprehensive quantitative evaluation of these uncertainties has not yet been achieved. In order to fill this need, the NOAA National Weather Service (NWS) is supporting the development of a model for the quantitative description of radar-rainfall errors. This model is based on the empirical characterization of the statistical error structure in the operational WSR-88D precipitation products for different conditions, such as different spatio-temporal scales and rain regimes. To achieve a realistic parameterization of the relation between true rainfall (RA) and radar-rainfall (RR), the authors have envisioned a model characterized by two elements: a deterministic distortion function and a random component, which represents all the sources of uncertainty. For their characterization, a non-parametric framework is used and rain gauge estimates are considered as an approximation of the true rainfall. The proposed results are based on a six-year sample of Level II data from the Oklahoma City radar site (KTLX) and processed through Build 4 of the Open Radar Product Generator Precipitation Processing System (PPS). The radar data are complemented with the corresponding rain gauge observations from the Oklahoma Mesonet, Agricultural Research Service Micronet, and the Environmental Verification and Analysis Center PicoNet. These nested gauge networks provide the flexibility to span a range of spatial scales. In this paper the authors presents early results of error model development. They also show an implementation structure of a generator of rainfall realizations that are consistent with the actual distribution of the uncertainties in radar-rainfall products. The results are illustrated with real data based examples.
Session 3R, Quantitative precipitation estimation and validation
Tuesday, 25 October 2005, 10:30 AM-12:15 PM, Alvarado ABCD
Previous paper