Radiative Effects of Water Clouds on Heat , Cloud Microphysical and Surface Rainfall Budgets Associated with Pre-Summer Torrential Rainfall

This study investigates thermal, cloud microphysical and surface-rainfall responses to the radiative effects of water clouds by analyzing two pairs of two-dimensional cloud-resolving model sensitivity experiments of a pre-summer heavy rainfall event. In the presence of the radiative effects of ice clouds, exclusion of the radiative effects of water clouds reduces the model domain mean rain rate through the mean hydrometeor increase, which is associated with the decreases in the melting of graupel and cloud ice caused by enhanced local atmospheric cooling. In the absence of the radiative effects of ice clouds, removal of the radiative effects of water clouds increases model domain mean rain rate via the enhancements in the mean net condensation and the mean hydrometeor loss. The enhanced mean net condensation and increased mean latent heat are related to the strengthened mean infrared radiative cooling in the lower troposphere. The increased mean hydrometeor loss associated with the reduction in the melting of graupel is caused by the enhanced local atmospheric cooling.


INTRODUCTION
Cloud radiative processes play an important role in the development of precipitation systems.The cloud-radiation interaction could lead to the destabilization of the environment (e.g., Dudhia 1989), the unstable thermal stratification of stratiform clouds in the upper troposphere (e.g., Lilly 1988), and the development of second circulation by the different radiative heating between cloudy and clear-sky regions (e.g., Gray and Jacobson 1977).Tao et al. (1993) found that infrared radiative cooling enhances surface rainfall in their cloud-resolving model simulations of precipitation systems in the tropics and mid-latitudes.Fu et al. (1995) revealed that clear-sky infrared radiative cooling enhanced rainfall while weakened infrared radiative cooling due to stratiform clouds reduces rainfall.The nocturnal rainfall peaks in the diurnal variations of surface rainfall result from the decreased saturation mixing ratio due to the falling temperature caused by the infrared radiative cooling (e.g., Sui et al. 1997Sui et al. , 1998;;Gao and Li 2010).Wang et al. (2010) and Shen et al. (2011a, b) studied cloud radiative effects of pre-summer heavy rainfall processes and found that the rainfall responses to cloud radiative processes depend on cloud type and stage of convective development.The exclusion of cloud radiative effects increases the mean rainfall during the onset and decay phases, but it reduces the mean rainfall during the mature phase (Shen et al. 2011b).The removal of the radiative effects of ice clouds reduces the mean rainfall during the onset phase, whereas it enhances the mean rainfall during the mature and decay phases.The elimination of the radiative effects of water clouds weakens the mean rainfall during the mature phase, but increases the mean rainfall during the decay phase (Shen et al. 2011a).Although Shen et al. (2011a) examined the radiative effects of water clouds on rainfall through the analysis of vertical structures of heat budget, they did not provide explanations of why and how the radiative processes of water clouds affect hydrometeor change/convergence, which may be an important rainfall generating process.
In this study, the radiative effects of water clouds on heat, cloud microphysical and surface rainfall budgets associated with pre-summer torrential rainfall event are investigated by revisiting the sensitivity simulation data from Shen et al. (2011a).Five-day mean analysis is conducted to examine the radiative effects of water clouds on a pre-summer heavy rainfall event during 3 -8 June 2008.Since ice clouds may reduce the incoming solar radiative flux that reaches water clouds and the outgoing infrared radiative flux emitted from water clouds and surface, the radiative processes of ice clouds may alter the radiative effects of water clouds on rainfall.Thus, the radiative effects of water clouds on heat, cloud microphysical and surface rainfall budgets will be respectively discussed in the presence and absence of the radiative effects of ice clouds.Section 2 briefly describes model, large-scale forcing, and sensitivity experiments.Section 3 discusses the analysis, and section 4 summarizes the findings.

MODEL AND EXPERIMENTS
The data used in this study come from Shen et al. (2011a), which are obtained from the four sensitivity experiments with a two-dimensional cloud-resolving model.Model setups used in this study are summarized in Table 1, and model microphysical schemes can be seen in Table 2.A longitudinally oriented rectangular area of 108 -116°E, 21 -22°N, within which torrential rainfall occurred, is chosen as the model domain for the calculation of large-scale forcing Fig. 1.The model is integrated from 0200 Local Standard Time (LST) 3 June to 0200 LST 8 June 2008, during the pre-summer heavy rainfall event.Model simulations in the control experiment (CTL) are compared with available observations in terms of rain rate, and temperature and water vapor profiles (Wang et al. 2010;Shen et al. 2011b), which show good agreement.The three sensitivity experiments are identical to the control experiment, except that the mixing ratios of water clouds, ice clouds, and both water and ice clouds are set to zero in the calculations of radiation, in the sensitivity experiments NWR, NIR, and NCR respectively.The comparison between CTL and NWR is conducted to study the radiative effects of water clouds on rainfall in the presence of radiative effects of ice clouds.NIR and NCR are compared to examine the radiative effects of water clouds on rainfall in the absence of radiative effects of ice clouds.

Radiative Effects of Water Clouds in the Presence of Radiative Effects of Ice Clouds
In the presence of radiative effects of ice clouds, the exclusion of radiative effects of water clouds reduces the mean rain rate from 1.36 mm h -1 in CTL to 1.32 mm h -1 in NWR (Table 3).The mass-integrated cloud budget shows that the rain rate (P S ) is associated with the net condensation (Q NC ) and hydrometeor change/convergence (Q CM ), i.e., q 5 = q c + q r + q i + q s + q g , q c , q r , q i , q s , and q g are the mixing ratios of cloud water, raindrops, cloud ice, snow, and graupel, respectively; , z t and z b are the heights of the top and bottom of the model atmosphere respectively;

Prognostic equations
Potential temperature, specific humidity, five hydrometeor species, and perturbation zonal wind and vertical velocity.
Basic model parameters Model domain of 768 km, grid mash of 1.5 km, time step of 12 s, and 33 vertical layers.

Lateral boundary conditions Cyclic
Model reference Gao and Li (2008) and Li and Gao (2011) Model integration 0200 LST 3 June -0200 LST 8 June 2008 Large-scale forcing Vertical velocity and zonal wind (see Fig. 1), and horizontal advections (not shown).

Model surface boundary conditions
The surface temperature and specific humidity from NCEP/GDAS are also imposed in the model to calculate surface sensible heat flux and evaporation flux.P GFR Growth of graupel by the freezing of raindrops LFO Table 2. List of microphysical processes and their parameterization schemes [Lin et al. (1983, LFO), Rutledge and Hobbs (1983;1984;RH83, RH84), Tao et al. (1989, TSM), and Krueger et al. (1995, KFLC)].
(a) (b)  Note that the model domain mean hydrometeor convergence is zero due to the cyclic lateral boundaries furnished in the model.Since the radiative effects of ice clouds are the term in the heat budget, the mean heat budget can be analyzed.Following Li et al. (1999), the model domain mean heat and specific humidity budgets can be expressed by Here, ( / ) ; R is the gas constant; c p is the specific heat of dry air at constant pressure p, and p o = 1000 mb; t is height-dependent air density; T is the air temperature, θ is the potential temperature; u and w are the zonal and vertical components of winds; Q cn is the latent heat due to the phase change between water vapor and five cloud species; Q R is the radiative heating rate due to convergence of the net flux of solar and infrared radiative fluxes; overbar denotes model domain mean, and prime is perturbation from the domain mean; subscript "o" is a value imposed on the model, which is constructed from National Centers for Environmental Prediction/Global Data Assimilation System (NCEP/GDAS) The heat budget (2) states that the local change of model domain mean temperature is determined by condensational heating, radiative heating, convergence of vertical heat flux, vertical temperature advection, and imposed horizontal temperature advection.
The removal of the radiative effects of water clouds weakens the mean infrared radiative cooling from CTL to NWR from altitudes of 2 to 13 km (Fig. 2b).The reduced mean latent heat between altitudes of 2.5 and 6.5 km corresponds to the weakened mean infrared radiative cooling (Fig. 2a).The suppressed mean latent heat tends to cool the local atmosphere at altitudes between 2.5 and 4.5 km, which causes an unstable vertical stratification near the surface.In addition, the reduced net condensation associated with the decreased mean latent heat decreases the consumption of water vapor, which is available for the enhancement of convection near the surface.As a result, the mean net condensation and associated mean latent heat increase near the surface.The enhanced mean latent heat near the surface is largely canceled by the weakened mean latent heat in the mid troposphere, but the magnitude of the former is larger than that of the latter, which leads to a slight increase in the mean net condensation from CTL to NWR (Table 3).
To examine the difference in Q CM between CTL and NWR, the mean cloud microphysical budgets are analyzed.
The comparison in the mean cloud microphysical budget between CTL and NWR reveals similar Q CMR (-0.01 mm h -1 ) in the two experiments (Fig. 3).The difference in Q CM is associated with the differences in Q CMC (0.0 in CTL versus -0.02 mm h -1 in NWR), Q CMI (0.02 mm h -1 in CTL versus 0.0 in NWR), and Q CMG (0.02 mm h -1 in CTL versus 0.0 in NWR).The vapor condensation rate (P CND ) is larger in NWR than in CTL forming a source for cloud water in NWR, while the collection rate of cloud water by rain (P RACW ) and the accretion rate of cloud water by graupel [P GACW (T < T 0 )] are larger in CTL than in NWR as a result of the melting of cloud ice to cloud water (P IMLT ).Compared to those in CTL, more vapor condensation in NWR may be related to more water vapor due to less consumption of water vapor as indicated by less rainfall.Compared to those in NWR, the melting of cloud ice to cloud water is the microphysical process that is responsible for the sink for cloud ice in CTL.The melting of graupel to rain (P GMLT ) is larger in CTL than in NWR, which leads to a sink for graupel in CTL.The melting of cloud ice to cloud water in CTL and the enhancement in the melting of graupel to rain from NWR to CTL may be associated with the reduced atmospheric cooling from 2.5 to 4.5 km, which corresponds to the increased convergence of vertical heat flux from NWR to CTL at altitudes around 4 km and the suppressed heat divergence and enhanced latent heat NWR to CTL between 2.5 and 4 km (Fig. 2a).
The reduction in mean rain rate caused by the exclusion of the radiative effects of water clouds can also be examined by analyzing surface rainfall budget.In the surface rainfall budget (Gao et al. 2005;Cui and Li 2006), the rain rate is associated with local water-vapor change (Q WVT ), water vapor convergence (Q WVF ), surface evaporation (Q WVE ), and hydrometeor change/convergence (Q CM ), i.e., where q v is the specific humidity; u and w are the zonal and vertical components of wind, respectively; E S is the surface evaporation rate; overbar denotes a spatial mean; prime is a perturbation from spatial mean, and the superscript ° defines imposed NCEP/GDAS data.Shen et al. (2010) analyzed grid-scale simulation data of tropical rainfall by partitioning them into eight rainfall types based on the surface rainfall budget.Shen et al. (2012) applied this rainfall separation scheme to the evaluation of an existing convective-stratiform rainfall separation scheme based on the rain intensity (e.g., Tao et al. 1993) and found that the convective rainfall contains a considerable amount of rainfall associated with water vapor divergence.Since convective and stratiform rainfalls are associated with water vapor convergence and divergence, respectively, we separated the grid-scale rainfall simulation data into rain types associated with water vapor convergence and divergence.Table 4 reveals that the reduction in the mean rain rate from CTL to NWR results from the weakened rain rate associated with water vapor divergence, which corresponds to the slowdown in hydrometeor loss/convergence while the reduced water vapor divergence decreases local atmospheric drying.

Radiative Effects of Water Clouds in the Absence of Radiative Effects of Ice Clouds
In the absence of the radiative effects of ice clouds,  exclusion of the radiative effects of water clouds increases the mean rain rate through the increased mean net condensation from NIR to NCR and the mean hydrometeor loss in NCR (Table 3).A comparison of the mean heat budget between the two experiments shows that the removal of radiative effects of water clouds weakens the mean infrared radiative cooling and the mean solar radiative heating at altitudes above 4.5 and 3 km respectively, but enhances the mean infrared radiative cooling and increases the mean solar radiative heating at altitudes below 4 and 3 km, respectively (Fig. 4).As a result, the suppressed mean latent heat above 3.5 km altitude corresponds to the weakened mean infrared radiative cooling, whereas the enhanced mean latent heat below 3.5 km corresponds to the strengthened mean infrared radiative cooling (Fig. 4a).The weakened mean latent heat in the mid and upper troposphere is largely offset by the intensified mean latent heat in the lower troposphere, which causes a slight increase in the net condensation in a mass integration from NIR to NCR.The analysis of cloud microphysical budgets shown in Fig. 5 reveals that the difference in Q CM is associated with the differences in Q CMR (0.01 mm h -1 in NIR versus 0.02 mm h -1 in NCR), Q CMS (-0.01 mm h -1 in NIR versus 0.0 in NCR), and Q CMG (0.01 mm h -1 in NIR versus 0.0 in NCR) (Fig. 5).Q CMR is slightly larger in NCR than NIR because the melting rate of graupel to rain is slightly lower in NCR than NIR.The reduction in the melting from NIR to NCR may be associated with enhanced local atmospheric cooling above 4 km (Fig. 4a), which is caused by the suppressed mean latent heat and weakened convergence of vertical heat flux.Q CMS + Q CMG = 0 in NIR mainly due to the fact the accretion rate of snow by graupel (P GACS ) is slightly lower in NIR than NCR.
The enhanced rainfall from NIR to NCR is caused by the increased rainfall associated with water vapor divergence, which is associated with the strengthened hydrome-teor loss/convergence while the intensified water vapor divergence increases the local atmospheric drying.

SUMMARY
The radiative effects of water clouds on heat, cloud microphysical and surface rainfall budgets during pre-summer rainfall over southern China were examined through the analysis of sensitivity experiment data for a heavy rainfall event that occurred from 3 -8 June 2008.The control experiment and sensitivity experiment without the radiative effects of water clouds were compared to study the rainfall responses to the radiative effects of water clouds when the radiative effects of ice clouds were turned on.The sensitivity experiments without the radiative effects of ice clouds and cloud-radiation interaction were also compared to study the rainfall responses to water cloud-radiation interaction when the radiative effects of ice clouds were turned off.The main conclusions are:

Fig. 2 .
Fig. 2. Vertical profiles of differences between NWR and CTL (NWR-CTL) for (a) local temperature changes (black), latent heat (red), convergence of vertical heat flux (green), vertical temperature advection (blue), and radiation (orange), and (b) radiation (orange) and its solar heating (red) and infrared cooling (blue) components averaged for 5 days and model domain.Unit is °C d -1 .

Fig. 3 .
Fig. 3. Time mean cloud microphysical budgets in (a) CTL and (b) NWR.Units for cloud hydrometeors and conversions are mm and mm h -1 , respectively.Cloud microphysical conversion terms and their schemes can be found in Table2.T 0 = 0°C.
Fig. 4. Vertical profiles of differences between NCR and NIR (NCR-NIR) for (a) local temperature changes (black), latent heat (red), convergence of vertical heat flux (green), vertical temperature advection (blue), and radiation (orange), and (b) radiation (orange) and its solar heating (red) and infrared cooling (blue) components averaged for 5 days and model domain.Unit is °C d -1 .
GACS Growth of graupel by the accretion of snow RH84P GACW Growth of graupel by the accretion of cloud water RH84P WACS Growth of graupel by the riming of snow RH84P GDEP Growth of graupel by the deposition of vapor RH84 CND ] + [P DEP ] + [P SDEP ] +[P GDEP ]) represents the cloud source term that consists of vapor condensation rate for the growth of cloud water ([P CND ]), vapor deposition rates for the growth of cloud ice ([P DEP ]), snow ([P SDEP ]) and graupel ([P GDEP ]); -([P REVP ] + [P MLTG ] + [P MLTS ]) denotes the cloud sink term that includes growth of vapor by evaporation of raindrops ([P ([P REVP ]), evaporation of liquid from graupel surfaces ([P MLTG ]), and evaporation of melting snow ([P MLTS ]).Comparison of the mean cloud microphysical budget between CTL and NWR reveals that the weakened mean rain rate from CTL to NWR is associated with the mean hydrometeor change from a loss in CTL to a gain in NWR while the mean net condensation increases from CTL to NWR.

Table 3 .
Cloud microphysical budgets (P S , Q NC , and Q CM ) averaged for 5 days over the model domain in CTL, NWR, NIR, and NCR, and their differences between NWR and CTL (NWR-CTL) and NCR and NIR (NCR-NIR).Unit is mm h -1 .