Evaluation of cloud‐resolving model intercomparison simulations using...
The Tropical Warm Pool–International Cloud Experiment (TWP‐ICE) provided extensive observational data sets designed to initialize, force, and constrain atmospheric model simulations. In this first of a two‐part study, precipitation and cloud structures within nine cloud‐resolving model simulations are compared with scanning radar reflectivity and satellite infrared brightness temperature observations during an active monsoon period from 19 to 25 January 2006. Seven of nine simulations overestimate convective area by 20% or more leading to general overestimation of convective rainfall. This is balanced by underestimation of stratiform rainfall by 5% to 50% despite overestimation of stratiform area by up to 65% because of a preponderance of very low stratiform rain rates in all simulations. All simulations fail to reproduce observed radar reflectivity distributions above the melting level in convective regions and throughout the troposphere in stratiform regions. Observed precipitation‐sized ice reaches higher altitudes than simulated precipitation‐sized ice despite some simulations that predict lower than observed top‐of‐atmosphere infrared brightness temperatures. For the simulations that overestimate radar reflectivity aloft, graupel is the cause with one‐moment microphysics schemes whereas snow is the cause with two‐moment microphysics schemes. Differences in simulated radar reflectivity are more highly correlated with differences in mass mean melted diameter (Dm) than differences in ice water content. Dm is largely dependent on the mass‐dimension relationship and gamma size distribution parameters such as size intercept (N0) and shape parameter (m). Having variable density, variable N0, or m greater than zero produces radar reflectivities closest to those observed.