Doppler Radar Information 3-d Variational Direct Assimilation

Direct assimilation of Doppler radial velocity and radar reflectivity that arenon-model variables is a challenge. With an ultimate goal of operationalapplication of Doppler radar data in numerical weather prediction (NWP), thedirect assimilation procedure of Doppler radar observations usingthree-dimensional variational (3DVAR) data assimilation method is setup forthe Weather Research and Forecasting (WRF) model. Numerical experimentswith the WRF 3DVAR system are conducted for the initialization of TyphoonRusa (2002), squall line, mesoscale rainstorm and landfalling Typhoon Matsa(2005). The Doppler radar data assimilation in WRF 3DVAR system is robustwith the data from diverse radar stations, including Jindo radar observations inKorea, IHOP₂002 (International H2O Project) WSR-88D observations in theUnited States and the Shanghai WSR-88D observations in China, respectively.The major research works and the primary conclusions are summarized asfollows:(1) To assimilate Doppler radial velocity and radar reflectivity data, thevertical velocity increments, cloud microphysics processes and observationoperators of the radar observations are introduced into the WRF 3DVARsystem. The tangent linear and adjoint programs in WRRF 3DVAR are verifiedwithin the expected accuracy. The developed Doppler radar data assimilationscheme is efficient for the WRF analyses, which are consistent with the WRFdynamical, thermodynamical and other model parameters by blending ofdifferent mesoscale observation data.(2) Doppler radial velocities from the Korean Jindo radar are directlyassimilated into the WRF 3DVAR system for Typhoon Rusa (2002) case.Assimilation and simulation results show that vertical velocity increments,tuning of scale-lengths in recursive filter, GTS (the Global TelecommunicationsSystem) data and AWS (Automatic Weather Station) data, are all to improvethe performance of Typhoon initial vortex and subsequent prediction. Thedirect assimilation of radial velocities has especially positive impact. It caneffectively extract the useful information from radar observations and producea positive impact on Typhoon vortex initialization and landfall prediction.(3) Experiments of initialization and subsequent forecast for a squall lineobserved during IHOP₂002 campaign in United States are carried out usingthe direct assimilations of the WSR-88D radar data. The results demonstratethat the analysis and simulation are improved by directly assimilating radialvelocity and reflectivity, respectively. The rainfall prediction skill withassimilation of multiple Doppler radar observations (eleven radars) is betterthan assimilation of single Doppler radar observations.(4) The direct assimilation of the Shanghai WSR-88D observations isperformed using a mesoscale rainstorm case in Shanghai. Numericalexperiments indicate that the simulated precipitation for the case is the mostreasonable when the innovation vector threshold of radial velocity is 7.5 m/s,the observation error of radial velocity is 1.5 m/s, and the innovation vectorthreshold of reflectivity is 10 dBZ, the observation error of reflectivity is 2 dBZ.The initialization of three dimensional wind fields is improved with the directassimilation of radial velocities. The direct assimilation of reflectivity canimprove the thermodynamics fields. It can adjust the dynamical fields as well.The reflectivity assimilation has more positive impact on the rainfall forecastskill than the assimilation of radial velocity. Our study also shows that thecloud radius in cloud vertical velocity parameterization is an important andsensitive parameter to the rainstorm initialization and rainfall forecast. Amongthe direct assimilations of radial velocity, reflectivity and both simultaneously,the experiment with assimilation of both radial velocity and reflectivitysimultaneously has the most positive impact on the rainstorm prediction.(5) The landfalling Typhoon Matsa (20025) is initialized and simulatedthrough directly assimilating Shanghai WSR-88D observations. Usingde-aliased radial velocity, significant improvements are achieved in theTyphoon vortex analysis, its track prediction and its precipitation forecast. Thereflectivity assimilation also improves the Typhoon rainfall prediction. The bestskill of precipitation forecast is obtained when both radial velocity andreflectivity are assimilated simultaneously.