Recent Updates to Analyses and Data Assimilation Systems for High Resolution Nowcasting and Forecasting

For the past several years CAPS has been running an efficient real-time analysis, data assimilation and forecasting system providing low latency.  The system uses 3DVAR with a cloud analysis, Incremental Analysis Updating (IAU) assimilation and the ARPS forecast model.  The system is able to complete a 2-hour forecast at 1-km grid resolution, including two cycles of data assimilation, in about 20 minutes on fewer than 200 Intel Xeon cores.  Recently the system has been running at the Collaborative-Adaptive Sensing of the Atmosphere (CASA) Dallas-Fort Worth Urban Testbed utilizing radar data from the National Weather Service 88D radars, the Federal Aviation Administration (FAA) Terminal Doppler Radars (TDWR) and X-band radars deployed in the CASA network.  The system also utilizes surface observations from diverse sources, wind profiler data, VAD wind profiles, aircraft and geostationary satellite data.  

Recently two significant upgrades were made to the system. 1) a new IAU which introduces variable- dependent timing has been designed and implemented, 2) the cloud analysis has been updated to reduce moisture in downdrafts with rain.  A third upgrade is in development, 3DVAR is combined with a Bratseth analysis scheme to better handle clustered in-situ observations, especially at the surface.  Finally, the real-time system microphysics option is updated to use the multi-moment Milbrant and Yau scheme as we complete transition to a new supercomputer.

These upgrades will be described and some sample forecasts shown, including tornadic supercells forecasted on 26 December 2015.

Figure Caption: Modified IAU Scheme with Variable Dependent Timing (VDT) to adjust wind and mass fields (red) prior to introduction of hydrometeors (blue).