Real Time Radar Quality Control for Assimilation in Numerical Models

Radar observations play an increasingly important role in numerical weather prediction (NWP) where real time forecasts of actual storms, initialized by current data are within reach. The 158 Doppler radar observation network in China, CINRAD (China Next Generation Weather Radar), is one such network that could provide much needed observations of precipitation, wind, and hail in near real time.

Radar data assimilation at the convective scale has the potential to improve the prediction of hazardous weather. At present, the three-dimensional variation (3D-Var) technique is one of the most widely used techniques in operational NWP. GRAPES (Global and Regional Assimilation and PrEdiction System) model is a system that was developed in Center for Numerical Prediction and Research (CNPR) in China, and its 3D-Var regional module arms to assimilate CINRAD data in the future. The quality of the radar data will affect the results of assimilation. Two of the issues in the use of Doppler radial velocity are the data quality control of aliased velocities and range folding. The volume of the data used requires substantial automation, and adequate accuracy. The quality control must be robust. Here we assess some of the current techniques used or proposed with data from China, representing various precipitation regimes.