Convective Meteorology (Mesoscale Dynamics)

Simultaneous assimilation of radar and satellite observations into convection permitting models using an ensemble Kalman filter.

Dr. Thomas Jones

University of Oklahoma

11 December 2015, 3:00 PM

National Weather Center, Room 5600
120 David L. Boren Blvd.
University of Oklahoma
Norman, OK

This research describes the impact of assimilating both radar and satellite observations into a prototype ensemble data assimilation system for the Warn-on-Forecast (WoF) project known as the NSSL experimental WoF System for ensembles (NEWS-e). Liquid and ice water path (LWP, IWP) retrievals from the GOES Imager along with the radar reflectivity and radial velocity observations are assimilated for several tornado-producing events. Assimilating LWP and IWP retrievals improves thermodynamic conditions at the surface over the storm-scale domain through better analysis of cloud coverage in the model compared to radar-only experiments. These improvements correspond to an improved analysis of supercell storms leading to better forecasts of low-level vorticity. This positive impact was most evident for events where convection is not ongoing at the beginning of the radar and satellite data assimilation period. For more complex cases containing significant amounts of ongoing convection, only assimilating clear-sky satellite retrievals in place of clear-air reflectivity resulted in spurious regions of light precipitation compared to the radar-only experiments. The analyzed tornadic storms in these experiments are sometimes too weak and quickly diminished in intensity in the forecasts.

Several important lessons were learned as part of this study, especially those related to the techniques used to assimilate various types of remote sensing data and what thinning strategies should be applied to combined data sets. Many previous radar data ensemble assimilation studies utilized radar reflectivity calculated on the model grid by the cloud microphysics code as one of the state vectors updated by ensemble filter. Thus, gridded radar reflectivity is interpolated to the radar observation location. However, IWP and LWP are computed from the interpolated hydrometeor state and the combination of these two methods often led to poor results. To further analyze this issue, experiments were undertaken in which reflectivity is computed using the appropriate microphysics code within the forward operator from the interpolated hydrometeor state.

Finally, assimilating all available radar and satellite observations often led to examples where conflicting information was assimilated into the model, leading to a poor analysis of convection and its surrounding environment. To determine the optimal set of observations to assimilate, several experiments are conducted in which various thinning strategies were applied to MRMS reflectivity and CWP retrievals mapped to the MRMS grid. Early results indicate that removing certain observations from the assimilation cycle can improve the forecasts of supercell storms as well as improve model run performance due to a reduction in total number of assimilated observations.

Convective Meteorology (Mesoscale Dynamics) Seminar Series website