Impact of bulk microphysics scheme assumptions on simulated mesoscale convective system precipitation biases

Several cloud-resolving model and limited area model simulations of a large mesoscale convective system (MCS) during the Tropical Warm Pool-International Cloud Experiment produce consistent biases with respect to observations that are modulated by the assumptions used in the bulk microphysics schemes. Simulated convective area and reflectivity aloft are high biased, while simulated stratiform rainfall is low biased. Several studies of mid-latitude continental squall lines have shown that two-moment bulk microphysics schemes decrease the low bias in stratiform rainfall produced by excessive evaporation in one-moment schemes. For this tropical monsoonal MCS, two-moment schemes produce larger stratiform raindrops with Doppler velocities close to those retrieved by an S-band vertically profiling radar, but one-moment schemes produce larger rainwater contents that are closer to those retrieved by disdrometer and vertically profiling radars. The greater fall speeds and lesser rainwater content in two-moment simulations cancel out to produce similar rain rates to those in one-moment simulations. Because relative humidity is high in this case, excessive evaporation in one-moment schemes is not an issue. Instead, the low bias in simulated stratiform precipitation results from a combination of ECMWF model forcing dry biases relative to a variational analysis using a three-hourly five sounding array and overly intense convection relative to a dual-Doppler retrieval. This overly intense convection leads to incorrect ice properties aloft and detrainment that is too high in the troposphere. The greatest updraft vertical wind speeds in the upper troposphere (> 35 m/s) result from lofting and freezing of large rainwater contents, and thus, one of the keys to obtaining proper convective and stratiform structural properties is properly simulating low-mid level convective dynamics and microphysics for a given a large-scale environment. Despite these convective dynamics issues, there are some simple changes to the bulk microphysics scheme that are more realistic and produce better comparison with observations. Among these are changing the rain gamma shape parameter to a value greater than 0 or one that varies as a function of the slope of the gamma size distribution. This reduces convective area and improves the vertical profile of stratiform raindrop size, sedimentation, and rainwater content. Another is using a snow mass-diameter relationship that assumes snow mass is proportional to its area rather than its volume, which reduces the high bias in convective reflectivity aloft in some schemes.