Test of a Hybrid Data Assimilation Scheme for the Warn-on-Forecast Project During the HWT Spring Experiments in 2018

A real-time, weather adaptive, hybrid dual-resolution ensemble variational analysis and forecast system with the WRF-ARW as forecast model has been developed recently for the NOAA supported Warn-on-Forecast project (WoF). The goal is to provide a deterministic physically-consistent gridded analysis and forecast products to the forecasters besides the ensemble-based products for making warning decisions in a timely manner. The storm position is determined based on NOAA/Storm Prediction Center’s convective outlook each day and it also has the on-demand capability so that end users (e.g., forecasters or scientists) can set up the location of the analysis and forecast domain in real-time based on the interested weather. The 36 ensemble members are provided with the cycled WRF-DART analysis and forecast system. Both the ensemble and the 3DEnVAR analysis systems take in available mesoscale forecasts, radar data, satellite retrieved cloud water path, and traditional observations to perform two separate 15-minute data assimilation cycles. The hybrid 3DEnVAR system incorporates the dynamic covariances from the 36 ensemble members and the static covariances based on climatological model error with dual-resolution capability. The WRF-DART ensemble analysis is performed at 3 km resolution, and the 3DEnVAR analysis is performed at 1.5 km resolution. Then 18 of the 36 ensemble forecasts based on the WRF-DART analyses and one deterministic forecast based on the hybrid 3DEnVAR analysis with 3 or 6 hours forecast length are launched every 30 minutes.

In our early implementation, raw radar data from each 88D radar are received and processed volume by volume every 4 to 10 minutes. Not all raw volume data can be received and used timely because of cut-off time window. In the new implementation, some significant changes have been made with regard to the radar data. First, radar radial velocity data are received and processed sweep by sweep within a minute/minutes as soon as observations are available. Furthermore, the reflectivity is replaced with well quality-controlled data from the NSSL Multi-Radar/Multi-Sensor system. The expectation of these changes is that the quality of the forecasts will be improved because of timely using of well quality-controlled radar data. The impact of these changes on data assimilation and 0-3 hours convective scale weather forecasts during 2018 spring experiments will be evaluated and reported in the conference.