Fathalla Rihan, United Arab Emirates

A Basis for Improving Numerical Weather Prediction by Assimilating Doppler Radar Radial Winds

M. N. Anwar, H. Al-Sakaji


Numerical Weather Prediction (NWP) is considered as an initial-boundary value problem: given an estimate of the present state of the atmosphere, the model simulates (forecasts) its evolution. Specification of proper initial conditions and boundary conditions for numerical dynamical models is essential in order to have a well-posed problem and subsequently a good forecast model (A well-posed initial/boundary problem has a unique solution that depends continuously on the initial/boundary conditions). The goal of data assimilation is to construct the best possible initial and boundary conditions, known as the analysis, from which to integrate the NWP model forward in time. In this paper, we describe an approach to assimilate Doppler radar radial winds into a high resolution NWP model using 3D-Var system. We discuss the types of errors that occur in radar radial winds. Some related problems such as nonlinearity and sensitivity of the forecast to possible small errors in initial conditions, random observation errors, and the background states are also considered. The technique can be used to improve the model forecasts, in the Gulf area, at the local scale and under high aerosol (dust/sand/pollution) conditions.

File available here