Radar reflectivity-based initialization of precipitation systems using a diabatic digital filter within the Rapid Update Cycle
Stephen S. Weygandt, NOAA Earth System Research Laboratory, Boulder, CO; and S. G. Benjamin
A new procedure for initializing ongoing precipitation systems using a national mosaic of radar reflectivity data has been developed and is currently being tested for implementation within the Rapid Update Cycle (RUC) assimilation and prediction system. The RUC utilizes a 1-h update cycle to advance a three-dimensional mesoscale analysis used to initialize a short range forecast. Cycled fields include all atmospheric variables, cloud and precipitation variables, and land surface model variables. These fields are updated via the RUC analysis system, which is compossed of a 3DVAR solver for the atmospheric fields and a non-variational cloud analysis procedure. Within the cloud analysis, surface METAR, radar, and satellite data are used to update cloud and hydrometeor variables. A digital filter initialization (DFI) procedure is used to control inertial-gravity waves excited by the assimilation. In June of 2006, the forward model portion of the digital filter was upgraded in the operational RUC from adiabatic to diabatic, yielding improved moisture forecasts and also facilitating the development of a radar data-based diabatic initialization procedure.
In the new radar data-based initialization procedure, latent heating is forced to be consistent with the hydrometeor field derived from the National Severe Storms Laboratory (NSSL) 3D radar mosaic data. The resultant latent heat based temperature tendency field is then applied during the forward-model integration portion of the diabatic DFI. This is accomplished by replacing the temperature tendency from both the parameterized convection and explicit microphysics with that derived from the radar data. Application of this latent heat derived temperature tendency induces an associated vertical circulation, with low-level convergence and upper-level divergence. In addition to this radar-based latent heat specification during the diabatic DFI, the relative humidity is increased in the radar reflectivity echo regions.
Initial test results from a January 2007 squall line case confirm that the radar-based latent heat specification during the DDFI does induce a vertical circulation (low-level convergence, upward motion, upper-level divergence) associated with the radar-observed precipitation systems. During the subsequent model integration, this circulation projects onto both the parameterized and explicit precipitation processes yielding a much improved prediction of the squall-line compared to a run without the radar assimilation. Based on the encouraging test case results, the new radar data assimilation procedure has been implemented in a real-time parallel version of the RUC run at the NOAA Earth System Research Laboratory (ESRL) in Boulder, CO. Qualitative assessment of the radar data assimilation cycle indicates impressive performance, with improved initialization of ongoing precipitation systems that are often missed or displaced otherwise. In addition, mesoscale details of larger-scale precipitation areas often appear more realistic in the first few hours. The impact of the radar-based latent heat specification diminishes after the first 6 hours of the model integration; however we have only evaluated it for strongly forced wintertime events. Consistent with the qualitative assessment, very preliminary assessment of 0-6h precipitation verification scores indicates substantial improvement for moderate precipitation thresholds.
We are currently examining the sensitivity of the procedure to: 1) strength of the latent heat temperature tendency, 2) inclusion of hydrometeors in addition to the latent heating, 3) specification of a low-level cold pool, and 4) degree of moistening in the near-storm environment. In addition, we are further evaluating the impact of the assimilation procedure on the skill of precipitation forecasts and other measures. We anticipate the possible inclusion of this procedure in a summer 2007 upgrade to the operational RUC at NCEP and this work will serve as the basis for the inclusion of a similar technique in the WRF-based Rapid Refresh system, which will replace the RUC in 2009. At the conference, we will present quantitative verification and case study results from the real-time cycle tests, as well as more detailed analysis of selected off-line test cases.
Extended Abstract (2.1M)
Session 1B, Radar Data Assimilation
Tuesday, 26 June 2007, 8:00 AM-10:00 AM, Summit B
Previous paper Next paper
Browse or search entire meeting
AMS Home Page