Seasonal and Intraseasonal Variation of Tropical Climate in NCAR CCM2 General Circulation Model : Sensitivity to Cumulus and Surface Processes

Abstract

In this thesis, we have undertaken a detailed study of the role of surface processes and the effect of changes in cumulus parameterization on the simulation of tropical climate by the CCM2 model of NCAR. We have examined the simulation of the mean tropical climate as well as its seasonal and intraseasonal variability. Two sets of climate simulations were analyzed to study the effect of surface hydrology feedback (a version with interactive surface hydrology, VAR.HYD, and with hydrology fixed to annual climatological values, FIX-HYD) and the effect of changes in the cumulus scheme (with mass flux scheme, HACK, and moist convective adjustment scheme, MCA). In addition to the climate simulations, a set of sensitivity simulations was carried out to understand the impact of the specification of surface wetness, albedo, and orography on the simulation of the Indian summer monsoon. Since the focus of this study was on the seasonal and intraseasonal variation, the simulations were forced with seasonally varying climatological Sea Surface Temperature (SST).

Simulated seasonal and intraseasonal variation have been studied using harmonic analysis in both spatial and temporal dimensions. The ability of the model to simulate the intraseasonal variation has been explained by identifying dominant periodicities through power spectrum analysis. The simulation of the MJO propagation characteristics has been assessed using OLR and 200 hPa circulation fields after applying a 20-70 day Lanczos band-pass filter, and their spectral characteristics using the space-time spectral technique. The time-lagged correlation analysis has been used to identify the simulation of meridional propagation of convective zones.

This study showed that the impact of surface hydrology specification was not entirely local in the model and appeared to have a remote impact on other parts of the tropics. This teleconnection between changes in the convective regions over the Indian region and West Pacific region was shown to be associated with changes in the Rossby wave pattern over these regions. It was also observed that the impact of surface hydrology on the mean simulation was larger during the boreal summer than in winter, whereas the effect of the cumulus scheme was found to be on both seasons, with maximum improvement over the Indian summer monsoon region with the MCA scheme. This improved Indian summer monsoon simulation was found to be due to the more realistic mean moist static energy in the lower (1000-400 hPa) layers of the troposphere.

The sensitivity studies of surface processes showed that the impact of soil wetness specification is large over the Indian monsoon region with a high coefficient of variation compared to other continental regions. It was found that there exists an optimum combination of surface temperature (~27-28°C) and soil moisture fraction (0.5 to 0.7) for a realistic simulation of Indian summer monsoon rainfall.

The model with the MCA scheme was able to simulate subseasonal variation of convection such as meridional propagation, equatorial eastward propagation, and off-equatorial westward propagation with reasonable accuracy. The interactive hydrology was found to be necessary for the simulation of meridional propagations of convective zones over the Indian region. This improvement in the simulation of northward propagations was illustrated as the northward tilting of the positive correlation area in the time-lagged correlation pattern of precipitation and found to be associated with a more realistic latitudinal variation of vertically integrated moist static energy over 1000 - 400 hPa. However, the change in the cumulus parameterization did not significantly modify the simulation of the meridional propagations of convective zones over this region. Our study revealed that the manner in which surface hydrology is specified does not affect the simulation of the MJO.

The cumulus parameterization was found to affect the spectral and propagation characteristics of the MJO. The reason for this improvement was explained through a realistic simulation of mean moist static energy in the lower troposphere (1000-400 hPa) by the accumulation of moisture in the lower layers, which resulted in a more realistic heating profile. The period of propagation for wavenumber 1 was found to be higher for the MCA version and lower for the HACK version compared to the observation. This study showed that the simulation of the monsoon by using the MCA scheme was better than HACK in the Indian region, but this was not true for all other tropical regions.