8.4 Atmosphere and surface analysis over Northern high-latitude region using the WRF model and its data assimilation system

Thursday, 5 May 2011: 4:30 PM
Rooftop Ballroom (15th Floor) (Omni Parker House )
Zhiquan Liu, NCAR, Boulder, CO; and H. C. Lin, B. Kuo, D. H. Bromwich, and L. S. Bai

Data assimilation is an optimization method to obtain accurate atmospheric and surface analysis by optimally combining information from both model forecast and observations. Harsh environmental conditions in polar region severely restrict the possibilities for in-situ measurement of the surface and atmospheric state and pose some challenge for data assimilation. In this study, the Weather Research and Forecasting (WRF) model and its data assimilation system (WRFDA) are applied to perform analysis and forecast cycles (in a 3-hour interval) over a Northern high-latitude region in a 60km horizontal resolution for a year-long period (2007). ECMWF's ERA-Interim global reanalysis data (80km resolution) are used as lateral boundary conditions for this regional analysis.

It was firstly found that WRF model forecast errors exhibit seasonal variation with largest errors in winter and minimum errors in summer. Therefore, we generated the background error covariance statistics for 4 different seasons to better characterize the WRF model error feature. Our emphasis is to optimize the exploitation of remotely sensed satellite observations such as radiances from microwave sensors (AMSU-A/B and MHS) aboard polar-orbiting platforms and GPS radio occultation (GPSRO) data from COSMIC platforms. Variational bias correction (VarBC) is applied to radiance data during the iteration procedure of assimilation. Some sensitivity experiments were conducted to find optimal configuration. We found that it needs month-long to make VarBC coefficients stable when we start cycling from no knowledge of bias coefficients. A set of “pre-trained” bias coefficients allows radiance analysis stabilized more quickly. With radiance data assimilated, the center of the low-pressure systems is more accurately positioned, which is particularly true in winter and over ocean. T2m and Q2m errors are also reduced by using radiance data. GPSRO data, which are assumed to be unbiased, was approved to be beneficial to constrain the model state and thus help radiance bias correction, particularly over ocean area where few conventional observation available. The results in this 60km resolution configuration will guide a 11-year (2000-2010) Arctic System Reanalysis in a much higher resolution, which is a collaborative effort among OSU, NCAR, CU and UIUC.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner