Assessing the Potential Impact of Assimilating Nationwide Mesoscale Profile Network Data on Short Range Weather Forecasting Using OSSE

Filling the gaps in the current U.S. observation network with a mesoscale network has been identified as a key priority in the 2021 NOAA Science Advisory Board report for improving nation-scale weather prediction. Beyond surface observations, accurate sampling of the boundary layer in the form of profile measurements has the potential for further improving numerical weather prediction as boundary layer processes are important for initiating and supporting convection and other forms of weather. Ground-based profiling instruments, such as Doppler Wind Lidar (DWL) and Microwave Radiometer (MWR) deployed at the New York State Mesonet (NYSM) are viable candidates for this purpose. These instruments offer a cost-effective alternative to rawinsondes or airborne observations, providing continuous high-resolution vertical profiles of wind, temperature, and moisture, which are essential for identifying key environmental conditions such as vertical wind shear, atmospheric instability, and moisture content for storm development.

Aligned with this national priority and as part of the National Mesonet program goals, this study investigates the potential benefits of assimilating data from a nationwide DWL and MWR network for short-range weather forecasting utilizing Observing System Simulation Experiments (OSSEs). The JEDI LETKF system is used to assimilate simulated DWL and MWR observations across a 3-km domain covering about 2/3 of CONUS (east of the Rockies), with a spatial density similar to that of the NYSM profile network. To account for model error, a truth simulation is generated using the WRF model initialized with GFS analysis while the data assimilation (DA) experiments are performed using the FV3 limited-area model. Initial ensemble comes from the 40-member GDAS EnKF analyses. An hourly cycling DA configuration, similar to that used in the operational RRFS, is applied over a 12-hour DA window following free ensemble forecasts up to 24 hours for evaluation. A single case involving active tornado-producing storms is chosen for the study.

A series of sensitivity experiments are first conducted to determine the optimal vertical localization radii for assimilating profiler data. Results show that using a 0.2 scalar height (in logarithmic pressure levels) out of the tested 0.1, 0.2, and 0.4 leads to best forecasts of surface fields (10-m winds, 2-m temperature and humidity) and the upper-level atmospheric states when verified against the truth. The impacts of the two different types of profile data are then examined by assimilating the DWL (u-, v-winds) and MWR (temperature and humidity) separately or together, in addition to conventional data; their forecasts are compared to a baseline experiment that assimilates only conventional data. Overall, the positive impact of assimilating profile data is most significant within the first 12-h forecast range. The benefit of assimilating DWL alone is least pronounced, possibly due to its relatively limited observing range (below 2.5 km AGL). The best wind prediction and the highest ensemble-averaged ETS for moderate composite reflectivity (at a 30-dBZ threshold) forecasts are achieved by the experiment assimilating both DWL and MWR data. More detailed analyses on the impact of assimilating DWL and MWR on analyzed and forecast storm structures, will be presented at the conference.