The results were then compared with Capturing similar TRMM 3G68 rainfall data av- eraged over January–March from 1998 to 2004. Since the simulations cov- the Daily Rain Cycle ered only the 3-month period in 1998, comparisons with the 7-year TRMM observations are not straightforward; in the Tropics nevertheless, the TRMM observation provides a reference for assessing the adequacy of the simulated daily rainfall cycle in the region. ainfall over the Indonesian re- have been assumed to occur at fixed As in previous observational stud- gion has a pronounced diur- rates, and the effect of altering these ies, the rainfall over land areas in the nal signature. One test of how rates on the simulated daily rainfall TRMM data peaks anywhere from Rwell moist convection is represented in cycle has not been studied. late afternoon to midnight (Figure 1a) models is how well the diurnal rainfall At the IPRC, Yuqing Wang, Kev- depending upon the region, and then cycle is simulated in this region. In typi- in Hamilton, and doctoral student Li shifts to the ocean in the late evening cal atmospheric circulation models, the Zhou have studied how increases in to early morning hours. Over the open rainfall over the Maritime Continent the mixing and diluting rates affect the ocean there is no consistent peak rain- peaks too early in the day and the daily simulation of the daily rainfall cycle in fall time and no consistent migration. range in rainfall rate varies too much, the Maritime Continent region. They The hourly rainfall amount during the a deficiency that applies over most of used the IPRC Regional Climate Model daily cycle varies quite a bit over land, the land in the tropics. These difficul- with a 0.5°-longitude–latitude grid and but little over the ocean (Figure 2a). ties may, at least partly, have to do with 28 vertical levels. The model was run The integration with the default the way shallow convection is repre- under 3 different conditions for the parameters simulates an unrealistic sented in the models. Shallow cumulus January 1–March 31 period in 1998, rainfall pattern, but one that is typical clouds in the morning take moisture with initial and boundary conditions of most current atmospheric models: from the atmospheric boundary layer from ERA-40 reanalysis updated every rainfall peaks over most of the land ar- at the sea surface and moisten the free 6 hours and from Reynolds sea surface eas about 2–4 hours too early and over atmosphere. This sets the stage for deep temperature updated every week. In the ocean, usually during early to mid- convection later in the day. the first run, all default parameters in morning (Figure 1b); the rainfall rate An adequate representation of the mass-flux cumulus parameteriza- varies too much over land and coastal shallow convection would seem nec- tion scheme were used; in the second areas, and the proportion of large-scale essary for capturing the daily rainfall run, the mixing and diluting rates were to stratiform rainfall is too small (not cycle. The strength of shallow convec- doubled for deep convection; in the shown). tion is largely determined by the rate at third run, the rates for shallow convec- Doubling the mixing and diluting which air is mixed from the surround- tion were increased 6.7-fold, to a value rates for deep convection delays the ing environment into clouds as they similar to values inferred from large- peak rainfall by about 1–2 hours over rise and dilute the original air in the eddy simulations for shallow convec- most of the land areas and in some of cloud (a process called ‘fractional en- tion. the coastal regions, but otherwise the trainment’) and the rate at which they To obtain a local average daily cy- rainfall pattern does not become more transfer moist air inside a cloud to the cle of the hourly rainfall rate for each realistic (Figure 2c). air outside (called ‘fractional detrain- model configuration, the hourly rain- Increasing the rates for shallow ment’). In general circulation model fall rates were averaged at each grid convection, though, improves the sim- parameterizations, these two processes point and across the 3-month period. ulation in nearly all aspects. It brings

6 IPRC Climate, vol. 5, no. 2, 2005 peak rainfall times closer to observa- rate over land (Figure 2d), though it layer and slows the moistening of the tions over both the Maritime Conti- does underestimate the amplitude over middle troposphere. All this prolongs nent and the oceans and simulates the the ocean. The shallow convection ex- the onset of deep convection, delaying peak rainfall shift to offshore during periment also produces a more realis- and reducing convective rainfall dur- late evening and early morning hours tic proportion of large-scale stratiform ing daytime over land while increasing (Figure 1d). Rainfall over the western rainfall (not shown). large-scale, longer-lasting rainfall. The Pacific warm pool region, moreover, Analyses of the shallow convection change also increases the day-to-day shifts mainly westward, consistent with experiment show the reason for the im- variability in rainfall amount, some- the generally westward propagating provements: increasing the mixing and thing that has been difficult to achieve mid-sized convective systems observed diluting rates slows the destabilizing in many general circulation models. in this region. The simulation produces effect of shallow convection and verti- iprc a realistic daily amplitude in rainfall cal turbulent mixing in the boundary a) TRMM Observations b) Control Figure 1. The local time of day (see color bar) of composite peak diurnal rainfall rates from (a) 7 years of TRMM 3G68 products, and from the 3-month model simulations c) Deep Convection d) Shallow Convection under (b) control, (c) deep convection, and (d) shallow convection modifications.

a) TRMM Observations b) Control Figure 2. The daily amplitude of the rainfall rate in mm/hr from (a) 7 years of TRMM 3G68 products, and from the 3-month model simulations under (b) control, (c) deep c) Deep Convection d) Shallow Convection convection, and (d) shallow convection modifications.

International Pacific Research Center 7