The Gsd Cloud Analysis: Rapidly-Updated Realtime Hydrometeor Initialization for Short-Range NWP

An accurate short-range cloud and precipitation forecast is a fundamental component of rapidly updating data assimilation/short-range model forecast systems such as the NOAA 3-km High-Resolution Rapid Refresh or the 13-km Rapid Refresh (RAP). To reduce cloud and precipitation spin-up problems, a cloud/hydrometeor non-variational assimilation technique for stratiform clouds was developed within the Gridpoint Statistical Interpolation (GSI) data assimilation system. The goal of this technique was retention into the subsequent model forecast, as appropriate, of observed stratiform cloud and observed clear 3-d volumes.

The cloud observations available for this assimilation include satellite brightness temperatures, radar reflectivity, surface-based ceilometer data, and surface visibility. Quality control, expansion into full information content, and forward operators are described for each of the observation types. This transformation of these observation types, mostly remote, into a full observation information 3-d field is accomplished via identification of cloudy, clear and unknown-cloud-content 3-dimensional volumes. Out of this 3-dimensional cloud/hydrometeor information at a given time, updating of forecast background fields is accomplished through clearing and building of hydrometeors and associated modifications to water vapor, temperature, and water droplet or ice particle concentrations.

The impact of the cloud/hydrometeor assimilation on the short-range forecast is assessed with a set of retrospective experiments using the RAP model with GSI assimilation, showing an improvement on short-range forecasts primarily for cloud ceiling, but also with slight improvement for relative humidity and wind. The 3-d cloud information content pre-processing described here is also applicable to an ensemble/variational assimilation technique which is in development for use in the Experimental HRRR Ensemble (HRRRE).