P1.6 High Resolution Assimilation of Casa Radar Data from a Tornadic Convective System

In late 2006, the NSF Engineering Research Center (ERC) for Collaborative and Adaptive Sensing of the Atmosphere (CASA, McLaughlin et al. 2005) began its first integrated project (IP1, Brotzge et al. 2007). CASA-IP1 consists of a network of four x-band dual-polarization Doppler radars in southwest Oklahoma. During the spring of 2007, this radar network collected data from a number of severe convective events. As part of the CASA spring Experiment 2007 (CSET-2007, Brewster et al. 2007), quasi real-time forecasts were performed using the CAPS ADAS analysis system (Brewster 2006). Preliminary results from these “real-time” forecasts showed success from assimilating combined reflectivity from WSR-88D and CASA radars. This paper is concerned with the assimilation of CASA radar data into the three-dimensional variational data assimilation (3DVAR) package of a non-hydrostatic numerical model, the Advanced Regional Prediction System (ARPS). The ARPS 3DVAR minimizes a cost function that includes the departure of the analysis from the background, the departure of the observations from the analysis and a penalty term. The penalty term imposes a weak anelastic mass continuity constraint on the analyzed wind field. The ARPS 3DVAR is described in detail in Gao et al. 2004 and Hu et al 2006b. Previous work found that 3DVAR reflectivity assimilation is most effective when combined with a cloud analysis (Hu et al. 2006a, Hu et al. 2006b). By expediting the “spin-up” of important convective cells and associated clouds, a cloud analysis improves the accuracy of the overall analysis (Hu et al. 2006a, Hu and Xue 2007). The cloud analysis used in this study evolved from the Local Analysis and Prediction System (LAPS , Albers et al. 1996) with modifications by Zhang (1998) and Brewster (2002).