32nd Conference on Broadcast Meteorology/31st Conference on Radar Meteorology/Fifth Conference on Coastal Atmospheric and Oceanic Prediction and Processes

Thursday, 7 August 2003: 5:45 PM
Physically-based "Down-to-Earth" modelling of surface precipitation using synergetic radar and multisource information
D. B. Michelson, Swedish Meteorological and Hydrological Institute, Norrköping, Sweden; and C. G. Jones, T. Landelius, C. G. Collier, G. Haase, and M. Heen
Poster PDF (104.1 kB)
Radar measurements are made at increasing height and with an increasing measurement volume with increasing range, making them decreasingly representative for surface conditions. Yet many users of radar data are interested in considering a radar image as containing surface information, an example being the use of radar-derived precipitation information as input to a hydrological runoff model. This necessitates the systematic correction of radar reflectivities aloft to be valid at the surface prior to such quantitative application. A family of methods which utilize the analyzed vertical profile of reflectivity (VPR) as the basis for such a correction has shown that VPR-based corrections are successful in both reducing scatter and minimizing bias in comparisons with independent precipitation gauge measurements. Parallel to these developments, we have attempted to determine whether the synergetic use of radar, NWP and analysis fields can be the basis for a physically-based VPR-like correction which could potentially circumvent inherent deficiencies in convensional VPR correction techniques.

Several previous attempts at formulating "Down-to-Earth" (DTE) over the last few years have led to disappointing results. We have found that the information required to perform this kind of VRP-like correction is not directly available in the forecast fields produced by the current generation of operational NWP models. The experiences gained from these previous studies have led to a final attempt at formulating DTE. Instead of using forecast and analyzed fields at face value, a robust wet adiabatic profile is instead reconstructed from forecast temperature and pressure profiles, and this profile is modified and used as the basis for the application of a cloud physics package (Rasch-Kristjansson (RK) scheme) which has been broken out of the HIRLAM-5 forecast model and modified for DTE use. The initial precipitation rate is given by the converted radar reflectivity to precipitation rate at the echo's height. Optionally, the precipitation phase can be diagnosed and appropriate Z-R relations applied in an attempt to resolve and treat the melting layer (bright band), and improve the accuracy of the DTE procedure. From the echo height to the cloud base, this precipitation rate is subject to the RK scheme at each model layer using the forecast variables from HIRLAM, depleting the parameterized cloud water profile. From the analyzed cloud base height to the surface, a seperate evaporation scheme acts upon the generated precipitation at each model layer.

Evaluation of this model has been conducted by applying DTE on radar data from the BALTRAD network during the wet summer of 2000, and comparing daily precipitation accumulations with around 1600 corresponding gauge measurements from Norway, Sweden, and Finland. The results show that DTE is successful at minimizing the range bias beyond 120 km, but that it also introduces an underestimation by radar at shorter ranges as a result of the evaporation scheme. Scatter is reduced only slightly at all ranges, indicating the impact of other error sources. The bright band treatment does not lead to significant improvements. DTE will be presented and discussed at the conference.

Supplementary URL: