HRRR Assimilation of Zdr Column Information from Polarimetric Radar Data and GOES Satellite-Derived Cloud-Top-Cooling Rate Information

The High Resolution Rapid Refresh (HRRR) assimilation system includes a one-hour “pre-forecast” period in which the initial state is downscaled from the parent RAP forecast and latent heating (LH) is specified based on observed radar reflectivity and lightning (converted to proxy reflectivity via an empirical relationship). This LH procedure is essential for providing the HRRR with initial fields that include information about ongoing convection, thereby significantly reducing model spin-up for the subsequent HRRR forecast. We have expanded this LH-based assimilation procedure to include assimilation of two novel indicators of convection: differential reflectivity (Zdr) columns detected in polarimetric radar observations and GOES-satellite-derived cloud-top-cooling rate (CTCR) data.

The detection by polarimetric radar of Zdr columns, indicative of supercooled liquid hydrometeors lofted above the environmental freezing level (and associated with this, strong updrafts) provides practical information that can be used in our latent-heating-based assimilation procedure. Within the HRRR LH assimilation, we are conducting tests with assimilation of an LH field derived from a combination of standard radar reflectivity data and polarimetric radar detected Zdr columns. Our hypothesis is that use of the Zdr data will allow the latent heating to be focused in the strong updraft regions, where the largest amount of condensation and latent heating occurs.

The detection of areas of rapid cloud-top cooling rates within GOES satellite data (accounting for cloud movement between successive images) provides practical information on nascent or developing convection that can also be used in the HRRR LH-based assimilation procedure. We are continuing work in this area to evaluate the effectiveness of assimilating this information in the HRRR. It is hypothesized that assimilation of these data may help with the challenging problem of model prediction of weakly forced, scattered convection. Results so far have indicated modest improvement for very short-range forecasts (1-3 hours) of developing scattered thunderstorms. As part of the categorical forecast verification for this work, we are examining “miss” to “hit” changes relative to “correct rejection” to “false alarm” changes from the CTCR assimilation.

At the conference, we will present the latest results for both the Zdr and CTCR assimilation.