We have compared JMA-GSM with some observations and some other NWP models. The JMA-GSM is a hydrostatic model. It is the operational global NWP model in Japan. The comparison revealed that: the middle of the troposphere simulated by JMA-GSM was dryer than the observed by radio sonde, when deep convection was active; the dry bias of JMA-GSM was more conspicuous than those of the other NWP models; cloud occurrence evaluated by JMA-GSM was quite smaller than the observed by CloudSAT and CALIPSO in the middle of the troposphere, where deep convection was active.

We have also performed a preliminary analysis of heating and moistening rate. The greater part of the water vapor that was removed from the air appeared to be converted to rain droplet by convection process. The amount of the cloud water that was detrained in the middle of the troposphere seemed to be too small. These facts suggest that the dry bias is due to deficiencies in convection process.

We are currently comparing JMA-GSM with JMA-NHM in the tropics, aiming to improve its capability of simulating cumulus convection. The JMA-NHM is a non-hydrostatic model. It is the operational limited-area NWP model in Japan. The comparison is going to reveal deficiencies in parameterizations of JMA-GSM.

In the presentation, we show the results of the comparison between JMA-GSM and JMA-NHM, and then make some discussions of remedies for the deficiencies, after briefly introducing the dry bias. The prototype of improved parameterization will also be shown.