19th Conf. on weather Analysis and Forecasting/15th Conf. on Numerical Weather Prediction


Implementation of a new grid-scale cloud and precipitation scheme in the NCEP Eta model

Brad S. Ferrier, NOAA/NWS/NCEP/EMC and SAIC/GSO, Camp Springs, MD; and Y. Jin, Y. Lin, T. Black, E. Rogers, and G. DiMego

A new grid-scale cloud microphysical scheme was implemented inthe 12-km Eta model on 27 November 2001, replacing the scheme of Zhao et al. (1997). The Eta Grid-scale Cloud and Precipitation scheme of 2001 (EGCP01) predicts various forms of condensate in the form of cloud water, small ice crystals, rain, and precipitation ice. Precipitation ice is determined to be in the form of snow, graupel, or sleet (frozen raindrops) based on the density of the precipitation ice, which is explicitly calculated. For computational efficiency top-down integration of precipitation fluxes is used, however, a more sophisticated algorithm was developed that partitions the downward flux of precipitation between local storage in the grid box (proportional to the layer thickness) and fall out through the bottom of the box. A single prognostic variable, total condensate, is advected in the model, where its composition in the form of individual species is maintained by storage arrays local to the microphysics. It is assumed that the fraction of total condensate in the form of ice and the fraction of liquid water in the form of rain is fixed within each column from the previous time step.

The condensation algorithm of Asai (1965) is used as in other high-resolution mesoscale models. A key closure is the assumption that the size of precipitation ice varies as a function of temperature (Ryan, 1996), which allows substantial simplification of the ice microphysics. Bulk density of precipitation ice is calculated by keeping track of the rate of growth of ice by the freezing of water versus growth by vapor deposition, in which the mass of the ice particle increases during riming and the volume is fixed by assuming that the collected liquid water filtrates into the air holes of a porous ice particle. Mixed-phase processes are considered at temperatures above -10C, but at colder temperatures all liquid is assumed to freeze to ice. The physics of freezing and melting processes, which are much more sophisticated in EGCP01 than in the Zhao scheme, will be described in more detail in the paper. Computationally efficient lookup tables store complex calculations of different moments of the distributions of rain and ice, as well as the increase in the fall speed of rimed ice as a function of rime density. Numerically stable calculations are also used for rapid processes to allow the scheme to perform well using large time steps.

The scheme was designed to be computationally efficient for use in a wide range of NWP models. Results will be presented showing improvements in the Eta quantitative precipitation forecasts using the EGCP01 scheme, resulting in increased equitable threat scores and reduced bias over a wide range of precipitation intensities. Spurious precipitation maxima ("bulls eyes") that occurred last summer in the 22-km Eta forecasts are eliminated in the EGCP01. Examples will also show how incremental adjustments in temperature associated with diabatic processes from convective and grid-scale cloud processes eliminated the spurious excitation of fast-moving gravity waves seen in Eta forecasts from last summer.

Session 10, Data Assimilation I
Thursday, 15 August 2002, 8:00 AM-9:58 AM

Next paper

Browse or search entire meeting

AMS Home Page