Improving a spectral bin microphysical scheme using TRMM satellite observations

TRMM-observed mature-stage squall lines during late spring and early summer in the central USA over a 9-year period are compiled and compared with a case simulation by the Goddard Cumulus Ensemble (GCE) model with a spectral bin microphysical scheme. During the quasi-steady state of the simulation, a forward radiative transfer model calculates TRMM Precipitation Radar (PR) reflectivity and 85 GHz brightness temperatures from simulated particle size distributions. Comparisons between model and TRIVIM observations using radar Contoured Frequency with Altitude Diagrams (CFADs) and 85 GHz brightness temperature probability density distributions are performed, in addition to CFADs from a surface C-band radar for the same case. Radar CFADs comparisons reveal that the model overestimates sizes of snow/aggregates in the stratiform region. Three sets of sensitivity tests are carried out in order to improve the simulated radar reflectivity profiles: increase of aggregates' density and terminal fall velocity; changing temperature dependency of collection efficiency between ice-phase particles, particularly those of the plate-type; and adding a break-up scheme for large aggregates. While all three approaches mitigate the discrepancies, changing collection efficiency produces the best match in magnitudes and characteristics of radar CFADs. In addition, interactions between ice- and water-phase particles also need to be adjusted in order to have good comparisons in both radar CFADs and 85 GHz brightness temperature distributions. This study shows that long-term satellite observations, especially those with multiple sensors, can be very useful in constraining model microphysics. Copyright (C) 2010 Royal Meteorological Society