Sensitivity of Tropical Pacific Convection to Dry Layers at Mid- to Upper Levels: Simulation and Parameterization Tests

Abstract

Numerical forecast experiments are carried out to investigate the implications of observed moisture variability in the tropical Pacific for deep convection. The study uses a series of quasi-cloud-resolving model forecasts with a triple-nested version of the Naval Research Laboratory's Coupled Ocean?Atmosphere Mesoscale Prediction System (COAMPS). The forecasts are carried out for a dry tongue episode in the tropical western Pacific in November 1998. During a 24-h forecast, a number of convective cells develop in an area of deep convection near the edge of the simulated dry tongue in the COAMPS 3-km mesh inner grid, which is located in the southern portion of the domain of the Tropical Ocean Global Atmosphere Coupled Ocean?Atmosphere Response Experiment (TOGA COARE) field program. The forecast is repeated multiple times using a moisture profile from a TOGA COARE dry tongue episode to modify the lateral boundary conditions for the 3-km mesh grid over layers of varying depth and altitude. The entrainment of drier air results in a 37% reduction in rainfall when the dry layer extends from the 600-hPa level to the model top. Tests show that entrainment of dry air at mid levels has a major impact in suppressing precipitation production due to cold rain processes. It should be noted that conditions where warm rain processes dominate are not addressed by the present experiments. Associated with the suppression of deep convection by dry layers in these forecasts, there is a large increase in stored buoyant energy. The presence of dry midtropospheric air may thus serve to help permit the buildup of buoyant energy for subsequent episodes of deep convection, such as associated with the onset of the Madden?Julian oscillation. In order to investigate the ability of convective parameterizations to represent the moisture sensitivity observed in the COAMPS experiments, data from the COAMPS 3-km mesh grid were used for tests of convective parameterizations in both semiprognostic model (SPM) and single-column model (SCM) mode. Both the Emanuel and the Kain?Fritsch convective parameterizations are unable to account in the SPM tests for the reduction in rainfall and cloud-base mass flux obtained on the COAMPS 3-km mesh grid when dry lateral boundary conditions are imposed above the 600-hPa level. The results are consistent with the interpretation that inadequacies in the convective closure assumptions play a signficant role. Comparisons of updraft mass flux profiles in the SPM tests point to considerable sensitivity to the details of the implementation of the stochastic buoyancy-sorting model in these schemes. In the SCM tests, the parameterized rainfall amounts are better than in semiprognostic mode, but apparently too low in one case by a substantial amount. In addition, large errors are observed in the simulated stored buoyant energy. The convective schemes are shown to perform best for greater horizontal resolutions of the forcing data.