**4. The control experiment: C**

Model domain mean surface rain rate starts on 3 June 2008 with the magnitude of about 1 mm h-1 (Fig. 3), which corresponds to the weak upward motions with a maximum of 2 cm s-1 at 6-8 km (Fig. 1a). The rain rate increases to 2 mm h-1 as the upward motions increase up to over 6 cm s-1 on 4 June. When the upward motions weaken in the evening of 4 June and a weak downward motion occurs near the surface, the mean rainfall vanishes. As upward motions pick their strengths on 5 June, the mean rain rate intensifies (over 2 mm h-1). The mean rainfall reaches its peak on 6 June (over 4 mm h-1) as the upward motions have a maximum of over 20 cm s-1. The upward motions rapidly weaken on 7 June, which leads to the significant reduction in the mean rainfall. Thus, four days (4, 5, 6, and 7 June) are defined as the onset, development, mature, and decay phases of the rainfall event, respectively. During 3-6 June, the mean rainfall is mainly associated with the mean water vapor convergence (*QWVF*>0) in water vapor related surface rainfall budget and the mean heat divergence (*SHF*>0) in thermally related surface rainfall budget. Local atmospheric drying (*QWVT*>0) and moistening (*QWVT*<0) occur while the mean local atmospheric cooling (*SHT*<0) prevails. The mean hydrometeor loss/convergence (*QCM*) has small hourly fluctuations. The mean radiative heating during the daytime and mean radiative cooling during the nighttime have the much smaller magnitudes than the mean heat divergence and the mean local heat change do in thermally related surface rainfall budget. On 7 June, the mean water vapor related surface rainfall budget shows that the rainfall is associated with local atmospheric drying while water vapor divergence prevails. The mean thermally related surface rainfall budget reveals that the rainfall is related to heat divergence while the heat divergence cools local atmosphere.

The calculation of precipitation efficiency using model domain mean model simulation data shows that *PEWV* generally is higher than *PWH* (Fig. 4a) because the rainfall source from the mean water vapor convergence in water vapor related surface rainfall budget is weaker than the rainfall source from the mean heat divergence in thermally related surface rainfall budget (Fig. 3). This suggests that the precipitation system is more efficient in the consumption of rainfall source from water vapor than in the consumption of the rainfall source from heat. The root-mean-squared (RMS) difference between *PEWV* and *PEH* is 24.4%. Both *PEWV* and *PEH* generally increase as surface rain rate increases (Fig. 5a). This indicates that the precipitation system generally is more efficient for high surface rain rates than for low surface rain rates. The ranges of *PEWV* and *PEH* are smaller when surface rain rate is higher than 3 mm h-1 (70-100%) than when surface rain rate is lower than 3 mm h-1 (0-100%).

Thermodynamic Aspects of Precipitation Efficiency 83

Fig. 6. Time series of contributions of model domain mean of (a) *PSWV* (dark solid), *QWVT* (light solid), *QWVF* (short dash), *QWVE* (dot), *QCM* (dot dash), and (b) *PSH* (dark solid), *SHT* (light solid), *SHF* (short dash), *SHS* (dot), *SLHLF* (long short dash), *SRAD* (long dash), and *QCM*

Model domain mean surface rainfall consists of convective and stratiform rainfall. Convective rainfall differs from stratiform rainfall in four ways. First, convective rain rates are higher than stratiform rain rates. Second, convective rainfall is associated with stronger horizontal reflectivity gradients than stratiform rainfall. Third, upward motions associated with convective rainfall are much stronger than those associated with stratiform rainfall. Fourth, the accretion of cloud water by raindrops via collisions in strong updraft cores and the vapor deposition on ice particles are primary microphysical processes that are responsible for the development of convective and stratiform rainfall,respectively (Houghton 1968). The convective-stratiform rainfall partitioning scheme used in this study is developed by Tao and Simpson (1993) and modified by Sui et al. (1994). This scheme partitions each vertical column containing clouds in 2-D x-z framework into convective or stratiform based on the following criterion. Model grid point is identified as convective if it has a rain rate twice as large as the average taken over the surrounding four grid points, the one grid point on either side of this grid point, and any grid point with a rain rate of 20 mm h-1 or more. All non-convective cloudy points are considered as stratiform. In addition, grid points over stratiform regions are further checked and identified as convective when following conditions are met. Over raining stratiform regions, cloud water below the melting level is greater than 0.5 g kg-1 or the maximum updraft above 600 hPa exceeds 5 m s-1, or over non-raining stratiform regions, cloud water mixing ratio of 0.025 g kg-1 or more exists or the maximum updraft

(dot dash) from convective regions in C. Unit is mm h-1.

exceeds 5 m s-1 below the melting level.

Fig. 4. Time series of *PEWV* (solid) and *PEH* (dash) calculated using (a) model domain mean data and data from (b) convective and (c) raining stratiform regions in C. Unit is %.

Fig. 5. *PEWV* versus *PS* (cross) and *PEH* versus *PS* (open circle) calculated using (a) model domain mean data and data from (b) convective and (c) raining stratiform regions in C. Units are % for *PEWV* and *PEH* and mm h-1 for *PS*.

Fig. 4. Time series of *PEWV* (solid) and *PEH* (dash) calculated using (a) model domain mean

Fig. 5. *PEWV* versus *PS* (cross) and *PEH* versus *PS* (open circle) calculated using (a) model domain mean data and data from (b) convective and (c) raining stratiform regions in C.

Units are % for *PEWV* and *PEH* and mm h-1 for *PS*.

data and data from (b) convective and (c) raining stratiform regions in C. Unit is %.

Fig. 6. Time series of contributions of model domain mean of (a) *PSWV* (dark solid), *QWVT* (light solid), *QWVF* (short dash), *QWVE* (dot), *QCM* (dot dash), and (b) *PSH* (dark solid), *SHT* (light solid), *SHF* (short dash), *SHS* (dot), *SLHLF* (long short dash), *SRAD* (long dash), and *QCM* (dot dash) from convective regions in C. Unit is mm h-1.

Model domain mean surface rainfall consists of convective and stratiform rainfall. Convective rainfall differs from stratiform rainfall in four ways. First, convective rain rates are higher than stratiform rain rates. Second, convective rainfall is associated with stronger horizontal reflectivity gradients than stratiform rainfall. Third, upward motions associated with convective rainfall are much stronger than those associated with stratiform rainfall. Fourth, the accretion of cloud water by raindrops via collisions in strong updraft cores and the vapor deposition on ice particles are primary microphysical processes that are responsible for the development of convective and stratiform rainfall,respectively (Houghton 1968). The convective-stratiform rainfall partitioning scheme used in this study is developed by Tao and Simpson (1993) and modified by Sui et al. (1994). This scheme partitions each vertical column containing clouds in 2-D x-z framework into convective or stratiform based on the following criterion. Model grid point is identified as convective if it has a rain rate twice as large as the average taken over the surrounding four grid points, the one grid point on either side of this grid point, and any grid point with a rain rate of 20 mm h-1 or more. All non-convective cloudy points are considered as stratiform. In addition, grid points over stratiform regions are further checked and identified as convective when following conditions are met. Over raining stratiform regions, cloud water below the melting level is greater than 0.5 g kg-1 or the maximum updraft above 600 hPa exceeds 5 m s-1, or over non-raining stratiform regions, cloud water mixing ratio of 0.025 g kg-1 or more exists or the maximum updraft exceeds 5 m s-1 below the melting level.

Thermodynamic Aspects of Precipitation Efficiency 85

Over raining stratiform regions, rainfall is primarily associated with the transport of hydrometeor concentration from convective regions to raining stratiform regions because water vapor divergence dries local atmosphere on 3 and 7 June and water vapor convergence moistens local atmosphere on 4-6 June and heat divergence cools local atmosphere (Fig. 7). *PWH* is generally higher than *PEWV* on 3, 4, and 7 June whereas it is generally lower than *PEWV* on 5-6 June (Fig. 4c). The RMS difference between *PEWV* and *PEH* is 23.8%, which largely accounts for the RMS difference in the model domain mean calculations. The ranges of precipitation efficiencies for the surface rain rate of lower than 1 mm h-1 (0-100%) are larger than those for the surface rain rate of higher than 1 mm h-1

The calculations of model domain mean simulation data show that *PEWV* is insensitive to radiative effects of ice clouds on 4 June, whereas the exclusion of radiative effects of ice clouds decreases *PEH* (Table 1). The removal of radiative effects of ice clouds increases *PEWV,* but it barely affects *PEH* on 5 June. The elimination of radiative effects of ice clouds decreases *PEWV* and *PEH* on 6 June. The exclusion of radiative effects of ice clouds increases *PEWV* but it decreases *PEH* on 7 June. On 4 June, the water vapor realted surface rainfall budgets reveal that all rainfall processes contribute to rain rate in C and CNIR (Table 2), which leads to 100% of *PEWV* in the two experiments. The thermally related surface rainfall budgets show that rainfall is associated with heat divergence and radiative cooling in the two experiments (Table 3). Thus, local atmospheric cooling

(a) Model domain mean Convective regions Raining stratiform

(b) Model domain mean Convective regions Raining stratiform

 C CNIR CNIM C CNIR CNIM C CNIR CNIM 4 June 77.0 72.0 68.5 73.5 64.1 77.0 65.2 67.5 47.0 5 June 63.3 63.8 57.0 52.1 51.5 56.5 47.3 47.1 34.4 6 June 71.9 69.3 68.7 70.6 69.4 77.6 49.1 43.7 34.3 7 June 36.7 35.1 34.2 48.6 49.2 49.4 33.4 28.2 29.1 Table 1. (a) *PEWV* and (b) *PEH* calculated data averaged daily and over model domain, convective regions, and raining stratiform regions in C, CNIR, and CNIM. Unit is %.

 C CNIR CNIM C CNIR CNIM C CNIR CNIM 4 June 100.0 100.0 91.5 77.8 68.8 83.0 82.3 66.7 58.3 5 June 92.0 98.4 81.0 71.5 70.3 80.6 100.0 90.8 100.0 6 June 99.8 96.3 91.3 81.9 80.5 87.3 100.0 82.2 100.0 7 June 62.8 64.6 60.4 55.0 67.2 77.2 28.1 36.8 51.5

regions

regions

**5. Radiative effects of ice clouds: CNIR versus C** 

(*SHT*<0) makes PEH less than 100% in the two experiments.

(45-100%).

Fig. 7. Time series of contributions of model domain mean of (a) *PSWV* (dark solid), *QWVT* (light solid), *QWVF* (short dash), *QWVE* (dot), *QCM* (dot dash), and (b) *PSH* (dark solid), *SHT* (light solid), *SHF* (short dash), *SHS* (dot), *SLHLF* (long short dash), *SRAD* (long dash), and *QCM* (dot dash) from raining stratiform regions in C. Unit is mm h-1.

Convective rain rate (Fig. 6) is much higher than stratiform rain rate (Fig. 7) and mainly accounts for model domain mean surface rain rate (Fig. 3). Over convective regions, rainfall is associated with water vapor convergence in water vapor related surface rainfall budget (Fig. 6a) and heat divergence in thermally related surface rainfall budget (Fig. 6b). *QCM* is negative over convective regions whereas *QCM* is positive over raining stratiform regions (Fig. 8), which indicates the transport of hydrometeor concentration from convective regions (*QCM*<0) to raining stratiform regions (*QCM*>0). The hydrometeor transport is associated with the local atmospheric drying (*QWVT*>0) over convective regions because *QWVT* and *QCM* have similar magnitudes but opposite signs. The water vapor convergence, local atmospheric drying, and heat divergence are the rainfall sources whereas the local atmospheric cooling and the transport of hydrometeor concentration from convective regions to raining stratiform regions are the rainfall sinks. *PEWV* generally is higher than *PEH* (Fig. 4b) because the rainfall source from water vapor convergence in water vapor related surface rainfall budget is weaker than the rainfall source from heat divergence in thermally related surface rainfall budget (Fig. 6). *PEWV* is lower than *PEH* on 4 and 7 June because the rainfall source from water vapor convergence is stronger than the rainfall source from heat divergence. The RMS difference between *PEWV* and *PEH* is 12.5%. Both *PWEV* and *PEH* increase as convective rainfall intensifies (Fig. 5b). Precipitation efficiency ranges from 60% to 90% as convective rain rate is higher than 2 mm h-1, whereas it ranges from 0 to 100% as convective rain rate is lower than 2 mm h-1.

Fig. 7. Time series of contributions of model domain mean of (a) *PSWV* (dark solid), *QWVT* (light solid), *QWVF* (short dash), *QWVE* (dot), *QCM* (dot dash), and (b) *PSH* (dark solid), *SHT* (light solid), *SHF* (short dash), *SHS* (dot), *SLHLF* (long short dash), *SRAD* (long dash), and *QCM*

Convective rain rate (Fig. 6) is much higher than stratiform rain rate (Fig. 7) and mainly accounts for model domain mean surface rain rate (Fig. 3). Over convective regions, rainfall is associated with water vapor convergence in water vapor related surface rainfall budget (Fig. 6a) and heat divergence in thermally related surface rainfall budget (Fig. 6b). *QCM* is negative over convective regions whereas *QCM* is positive over raining stratiform regions (Fig. 8), which indicates the transport of hydrometeor concentration from convective regions (*QCM*<0) to raining stratiform regions (*QCM*>0). The hydrometeor transport is associated with the local atmospheric drying (*QWVT*>0) over convective regions because *QWVT* and *QCM* have similar magnitudes but opposite signs. The water vapor convergence, local atmospheric drying, and heat divergence are the rainfall sources whereas the local atmospheric cooling and the transport of hydrometeor concentration from convective regions to raining stratiform regions are the rainfall sinks. *PEWV* generally is higher than *PEH* (Fig. 4b) because the rainfall source from water vapor convergence in water vapor related surface rainfall budget is weaker than the rainfall source from heat divergence in thermally related surface rainfall budget (Fig. 6). *PEWV* is lower than *PEH* on 4 and 7 June because the rainfall source from water vapor convergence is stronger than the rainfall source from heat divergence. The RMS difference between *PEWV* and *PEH* is 12.5%. Both *PWEV* and *PEH* increase as convective rainfall intensifies (Fig. 5b). Precipitation efficiency ranges from 60% to 90% as convective rain rate is higher than 2 mm h-1, whereas it ranges from 0 to 100% as

(dot dash) from raining stratiform regions in C. Unit is mm h-1.

convective rain rate is lower than 2 mm h-1.

Over raining stratiform regions, rainfall is primarily associated with the transport of hydrometeor concentration from convective regions to raining stratiform regions because water vapor divergence dries local atmosphere on 3 and 7 June and water vapor convergence moistens local atmosphere on 4-6 June and heat divergence cools local atmosphere (Fig. 7). *PWH* is generally higher than *PEWV* on 3, 4, and 7 June whereas it is generally lower than *PEWV* on 5-6 June (Fig. 4c). The RMS difference between *PEWV* and *PEH* is 23.8%, which largely accounts for the RMS difference in the model domain mean calculations. The ranges of precipitation efficiencies for the surface rain rate of lower than 1 mm h-1 (0-100%) are larger than those for the surface rain rate of higher than 1 mm h-1 (45-100%).
