The grid-point atmospheric model of IAP LASG (GAMIL) was developed in and has been evaluated since early 2004. Although the model shows its ability in simulating the global climate, it suffers from some problems in simulating precipitation in the tropics. These biases seem to result mainly from the treatment of the subgrid scale convection, which is parameterized with Tiedtke’s massflux scheme (or the Zhang-McFarlane scheme, as an option) in the model. In order to reduce the systematic biases, several modifications were made to the Tiedtke scheme used in GAMIL, including (1) an increase in lateral convective entrainment/detrainment rate for shallow convection, (2) inclusion of a relative humidity threshold for the triggering of deep convection, and (3) a reduced efficiency for the conversion of cloud water to rainwater in the convection scheme.

Two experiments, one with the original Tiedtke scheme used in GAMIL and the other with the modified scheme, were conducted to evaluate the performance of the modified scheme in this study. The results show that both the climatological mean state, such as precipitation, temperature and specific humidity, and interannual variability in the model simulation are improved with the use of this modified scheme. Results from several additional experiments show that the improvements in the model performance in different regions mainly result from either the introduction of the relative humidity threshold for triggering of the deep convection or the suppressed shallow convection due to enhanced lateral convective entrainment/detrainment rates.