Improvements to ceiling/visibility forecasts from the 13-km RAP and 3-km HRRR hourly updated forecast systems

The Rapid Refresh (RAP), an hourly-updated mesoscale atmospheric prediction system, is the primary gridded 3-d data source provided operationally by NOAA for US aviation applications requiring hourly updating. Continued development of the RAP at NOAA's Earth System Research Laboratory (ESRL) has been focused on improvements of aviation-sensitive forecasts, including those of cloud, ceiling, and visibility. The RAP replaced the previous Rapid Update Cycle (RUC) as NOAA's operational hourly updated weather model on 1 May 2012 at NOAA's National Centers for Environmental Prediction (NCEP). In this paper, we describe ceiling and visibility forecast skill from the operational RAP, comparing it with that from the previous RUC model and from the future experimental RAP version 2 currently running at NOAA/ESRL.

Consistent with the RAP focus on providing short-range “situational awareness” guidance for aviation, severe weather, renewable energy and other forecast applications, a sophisticated set of analysis procedures have been developed for initializing 3-D hydrometeor fields in cloudy and precipitating areas. Cloud/hydrometeor fields are initialized via a non-variational cloud analysis procedure, which uses METAR and satellite-derived cloud-top information to modify hourly cycled explicit 3-D cloud hydrometeor fields. Recent work has focused on 1) reducing a positive moisture bias (relative to radiosonde observations) introduced from satellite observation-based mid- and high-level cloud building while 2) still improving retention of initial cloud fields through model improvements.

We will present recent modifications to both the data assimilation and model numerics for the RAP and the experimental 3-km HRRR model to improve ceiling and visibility forecasts. We will also show work on a new 3-km analysis application of the fully 3D Gridpoint Statistical Interpolation (GSI, including the cloud, radar, and surface analysis modules) to 3-km HRRR background fields. Initial results have been encouraging and the fully 3D 3-km analysis has many possible uses, including improved HRRR forecasts.