Journal of Geophysical Research: Atmospheres

Thermodynamic structure and evolution of the atmospheric mixed layer over the western North Pacific during the summer monsoon onset

Authors


Corresponding author: B. Geng, Research Institute for Global Change, Japan Agency for Marine-Earth Science and Technology, 2-15 Natsushima, Yokosuka, 237-0061, Japan. (bgeng@jamstec.go.jp)

Abstract

[1] This study examined the thermodynamic structure of the atmospheric mixed layer and its evolution in different periods during the onset of the western North Pacific summer monsoon by using data observed on board the research vessel Mirai at 12°N and 135°E during 6–24 June 2008. The overall mean depth of the mixed layer was 482 m, and the mixed layer tended to shrink after the monsoon onset. The mixed layer was relatively warm and moist during the early onset period. However, the mixed layer became very cool and dry during the latter onset period, when intense and sustained rainfall was observed. The surface buoyancy flux, environmental subsidence over the mixed layer, and upward motion at cloud base intensified after the monsoon onset. All of these factors were negatively correlated with the depth of the mixed layer, which implies that the upward motion at cloud base and compensating environmental subsidence act to suppress the growth of the mixed layer driven by buoyancy forcing and to shrink the mixed layer after the monsoon onset. A significant positive correlation existed between the surface buoyancy flux and the upward motion at cloud base after the monsoon onset. Such positive feedback between the sea surface and cumulus clouds facilitates the formation of the warmer and moister mixed layer during the early onset period, thereby benefiting the further development of the monsoon onset.

1 Introduction

[2] The onset of the western North Pacific summer monsoon (WNPSM) is associated with a distinct change in large-scale atmospheric circulation and sea surface temperature [Murakami and Matsumoto, 1994; Ueda et al., 1995; Wu and Wang, 2001]. It has been speculated that air-sea interaction in the tropics is important during the WNPSM onset [Ueda and Yasunari, 1996; Wu, 2002]. Needless to say, the tropical marine atmospheric boundary layer (MABL), which is the lowest part of the atmosphere over tropical oceans, plays a crucial role in the interaction between the ocean and the free atmosphere during the WNPSM onset. The tropical MABL has a well-defined vertical structure [Garstang and Betts, 1974; Gupta and Ramachandran, 1998], typically consisting of an unstable surface layer, a nearly neutral mixed layer, a stable transition layer, and a cloud layer. The mixed layer constitutes a major part of the MABL. It is characterized by vigorous turbulence tending to regulate the transport of heat and moisture from the surface aloft. As a result, the atmospheric mixed layer is an indispensable mediator of the air-sea interaction.

[3] Observational studies on the thermodynamic characteristics of the tropical atmospheric mixed layer have been conducted over the Atlantic Ocean [e.g., Malkus, 1958; Augstein et al., 1974; Fitzjarrald and Garstang, 1981], the Pacific Ocean [e.g., Bond, 1992; Serra et al., 1997; Johnson et al., 2001], and the Indian Ocean [e.g., Subrahamanyam et al., 2003; Ramana et al., 2004; Alappattu and Kunhikrishnan, 2010]. It has been found that the structures of the mixed layer over tropical oceans are similar. The depth of the mixed layer is quite variable, ranging from approximately 100–1000 m. Deeper mixed layers are usually observed under fair weather conditions with little precipitation, while mixed layers may become shallower or disappear when the MABL is disturbed by rain-induced outflow. Within the mixed layer, the potential temperature is more uniform than the specific humidity. The potential temperature and the specific humidity are regulated by heat and moisture exchanges at the surface and the transition zone, and advective and radiative effects in the mixed layer. Fitzjarrald and Garstang [1981] also found that precipitation-maintained downdrafts make the mixed layer cooler and drier or destroy the mixed layer.

[4] The depth of the atmospheric mixed layer is modulated by several factors. There are two processes that deepen the mixed layer. As indicated by Stull [1988], the growth of the mixed layer can be promoted mechanically by shears or convectively by buoyancy. Convective sources stem from heat transfer from a warm surface and radiative cooling from the top of the cloud layer. The warmth of the sea surface of tropical oceans facilitates the growth of the mixed layer driven by buoyancy forcing from below. However, there are several processes that retard the growth of the mixed layer. Environmental subsidence tends to make the mixed layer shallow [Stull, 1988]. Shallow cumulus clouds, which are frequently observed over tropical oceans, also play important roles in suppressing the mixed layer by inducing compensating subsidence between cumuli [Sarachik, 1974] and venting air out of the top of the mixed layer [Stull, 1985]. The depth of the atmospheric mixed layer is determined by a balance between deepening and retarding processes of the layer.

[5] Changes in the atmospheric mixed layer over oceans are mostly due to variations in synoptic and mesoscale environments [Stull, 1988]. Because the WNPSM onset is always associated with significant variations in atmospheric and oceanic environments, the atmospheric mixed layer during the WNPSM onset should undergo distinct modulation. Nevertheless, few studies have examined the atmospheric mixed layer during the monsoon onset over oceans. Two exceptions are the studies by Ciesielski and Johnson [2009] and Wang et al. [2010], who documented different structures of the atmospheric mixed layer and boundary layer over two domains of the South China Sea Monsoon Experiment (SCSMEX) during the monsoon onset in this region. Little is known about the evolution of the mixed layer during the WNPSM onset.

[6] Intensive observations using the research vessel Mirai were conducted in the western North Pacific in June 2008. The WNPSM onset occurred over the observational area during the observation period [Geng et al., 2011]. To our knowledge, these observations have provided the first opportunity to study detailed characteristics of the atmospheric mixed layer during the WNPSM onset in the region. The purpose of this study is to investigate the thermodynamic structure and evolution of the atmospheric mixed layer and to illustrate the processes that modulate the mixed layer during the monsoon onset. This paper is organized as follows. Section 2 describes the data and analysis procedures used in this study. In section 3, the characteristics of the atmospheric mixed layer during the monsoon onset are examined. Section 4 investigates the factors modulating the evolution of the mixed layer. Section 5 discusses the relationship between clouds and the sub-cloud layer, followed by a summary and the conclusions in section 6.

2 Data and Analysis Procedures

[7] Characteristics of the atmospheric mixed layer during the WNPSM onset were investigated mainly using data from oceanographic and atmospheric observations, including radiosondes, surface fields, a C-band Doppler radar, and a ceilometer. These observations were made on board the Mirai around a fixed site (12°N, 135°E) (Figure 1) during 6–24 June 2008. The grid point values (GPVs) of the global objective analysis data from the Japan Meteorological Agency were also used as auxiliary material.

Figure 1.

Geographic map of the observational site. The location of the research vessel Mirai is indicated by the filled circle.

[8] The radiosonde observation was conducted using Vaisala RS92 sondes at 3 h intervals; the vertical resolution of the radiosonde data was 2 s (approximately 10 m). Quality control of the radiosonde data was performed first using the Atmospheric Sounding Processing Environment (ASPEN) software developed by the National Center for Atmospheric Research (NCAR). A final low-pass filter with a wavelength of 20 s was applied when using the ASPEN software. The radiosonde data were further corrected for temperature and humidity biases. As indicated by Yoneyama et al. [2002], there were usually errors in temperature and humidity data within a few seconds after the launch due to the effects of the ship structure and the launch facility of the Mirai. Correction of the radiosonde data in the lowest layer of the atmosphere was made by utilizing the vertical gradient of the good quality data immediately above the questionable data [Yoneyama et al., 2002; Ciesielski et al., 2010]. There were also daytime dry biases due to the solar heating of the sonde humidity sensor, but correction for such biases was not performed in this study. As indicated by previous studies [Miloshevich et al., 2006; Yoneyama et al., 2008], little correction for the dry bias was needed at the low target levels of this study. The top of the mixed layer was determined from the radiosonde data using a method described in Johnson et al. [2001] that identified the mixed layer top subjectively as the level at which the potential temperature increased abruptly and the specific humidity decreased sharply with height.

[9] The other data obtained on board the Mirai were recorded in intervals ranging from 60 s to 10 min. These data were averaged over the same time intervals as the radiosonde observation, except for periods of assessments when the mixed layer was affected by rain-induced outflow. To identify the affecting periods of outflow, the surface rain rate, surface air temperature, and sea surface temperature observed on board the Mirai were processed in 6 min intervals according to the method of Young et al. [1995] and Saxen and Rutledge [1998]. The beginning of the effect of outflow was designated when a precipitation rate higher than 2 mm h−1 or a temperature decrease greater than 1°C was observed. The effect of outflow was sustained until the surface air temperature was in equilibrium with the sea surface temperature or was terminated by the intrusion of another nearby outflow.

[10] Surface fluxes were calculated using the algorithm of Fairall et al. [2003]. Meanwhile, area mean rainfall was derived from radar reflectivity data within a range of 100 km from the radar using the reflectivity-rainfall rate (Z-R) relations suggested by Tokay and Short [1996]. C-band Doppler radar is also useful for studying the atmospheric boundary layer [Lothon et al. 2002]. The vertical velocity at cloud base was derived from Doppler radar observations through the upward integration of the divergence obtained by the velocity-azimuth display (VAD) technique [Browning and Wexler, 1968] within a 15 km radius of the Doppler radar. As in Snodgrass et al. [2009], a reflectivity of 7 dBZ was used as the rain/no-rain cutoff. To limit the effect of precipitation on vertical velocity, radar data with reflectivities greater than 7 dBZ were removed when using the VAD technique.

[11] The regression analysis of the observational data was performed. Potential outliers were detected using the Grubbs test [Grubbs, 1969] at the 5% significance level and were not used in the regression analysis.

3 Characteristics of the Mixed Layer

[12] By using data from the same observation period, Geng et al. [2011] assessed in detail the monsoon onset based on outgoing longwave radiation (OLR) data and studied large-scale circulations and precipitation features over the observational area during the pre-onset (6–11 June), early onset (12–17 June), and latter onset (18–23 June) periods of the monsoon. They discovered that the western Pacific subtropical high (WPSH) began to retreat northeastward and that the monsoon trough (MT) began to move northward during the early onset period. Meanwhile, a sudden northeastward jump of the WPSH and an abrupt northeastward advance of the MT occurred during the latter onset period. Corresponding to the evolution of large-scale circulations, the rainfall increased steadily upon the monsoon onset and underwent rapid fluctuations during the latter onset period.

[13] In this section, the structure and evolution of the mixed layer will be investigated in conjunction with the progress of the monsoon onset as indicated above.

3.1 Mixed Layer Height

[14] A time series of the mixed layer height, zi, identified from the radiosonde data during the monsoon onset is shown in Figure 2a. Three periods of the monsoon onset, as determined by Geng et al. [2011], are shown at the top of the figure. The figure shows that mixed layers could be frequently identified. The frequency of the appearance of mixed layers was 80%, close to the 83% observed in the southern domain of the SCSMEX during the monsoon onset over the South China Sea [Ciesielski and Johnson, 2009] and similar to the 72% observed during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA COARE) [Johnson et al., 2001]. It can be seen that zi was usually unidentifiable when stronger rainfall was observed at the surface (Figures 2b and 2c), which is consistent with previous studies on contamination of the mixed layer by precipitation [e.g., Fitzjarrald and Garstang, 1981; Johnson et al., 2001]. Based on the method of Young et al. [1995] and Saxen and Rutledge [1998], periods when the mixed layer was affected by rain-induced outflow are also shown in Figure 2c. By comparing Figures 2a with 2c, it can be seen that most of the shallowest mixed layers after the monsoon onset were associated with the effect of rain-induced outflow. Much more rainfall around the Mirai was observed after the monsoon onset, when echo tops higher than 4 km appeared more often (Figure 2b). This result indicates the intensification of convection with the monsoon onset. It is noted that intense and sustained rainfall associated with the deepest echo tops occurred on approximately 21 June, which made the mixed layer unidentifiable during this period.

Figure 2.

(a) Time series of the mixed layer height (zi) (top of each vertical bar with scale to the left). (b) Same as Figure 2a, but for radar-observed area-mean rainfall within 100 km of the radar (vertical bar with scale to the left) and mean 10 dBZ echo top within 15–30 km of the radar (dashed line with scale to the right). The horizontal line in Figure 2b indicates an altitude of 4 km. (c) Same as Figure 2a, but for the surface rain rate (R) (black vertical bar with scale to the left) and the surface air temperature (curve with scale to the right) at the Mirai. Note that the surface rain rate and the surface air temperature in Figure 2c have been processed at 6 min intervals and used to determine periods when the mixed layer was affected by rain-induced outflow (gray vertical bars). The intervals of the pre-onset, early onset, and latter onset periods are also shown at the top of the figure.

[15] As shown in Figure 2a, zi tended to decrease after the monsoon onset. Evidence of the decrease in zi from the pre-onset period to the latter onset period is shown in Table 1. The mean zi was 539 m during the pre-onset period, and it decreased to 474 m during the early onset period. Corresponding to the occurrence of more frequent and intense precipitation during the latter onset period (Figure 2b), the lowest mean zi of 418 m was observed during this period. The mean zi over all periods of the monsoon onset was 482 m, which is also similar to the observational results of Johnson et al. [2001] and Ciesielski and Johnson [2009]. As in these studies, the zi during the monsoon onset period fluctuated over a wide range of depths. The highest zi of 720 m was observed during the pre-onset period, while the lowest zi of 176 m was observed during the latter onset period.

Table 1. Statistics of Mixed Layer Thermodynamic Properties During Different Periods of the Monsoon Onset
Periodinline imagezimaxzimininline imageinline imageinline imageinline imageLCLCBH
 (m)(m)(m)(K)(g kg−1)(°C)(w m−2)(m)(m)
Pre-onset539720311300.91826.36.9674696
Early onset474621177301.118.526.48.6646671
Latter onset418680176300.518.22614.3622636
Total482720176300.818.226.210647668

3.2 Thermodynamic Structure

[16] The thermodynamic variables monitored in each period of the monsoon onset were scaled according to the mean value of zi observed during the respective period. Figure 3 shows scaled-mean profiles of potential temperature θ, specific humidity q, and temperature T. During each period of the monsoon onset, both the θ and q profiles showed the general structure of the tropical MABL. The surface layer existed below approximately 50 m, where rapid decreases in θ and q with height were observed. Above the surface layer and below zi, well-mixed structures of θ and q were evident. The transition layer manifested itself within a layer of approximately 100–200 m above zi, where an abrupt increase in θ and a sharp decrease in q with height could be seen. As in previous studies [e.g., Fitzjarrald and Garstang, 1981; Johnson et al., 2001], slight changes in the gradients of θ and q within the mixed layer were also evident. It can be seen that θ tended to increase slightly with height from near the center of the mixed layer, while q tended to decrease slightly with height, even in the lower portion of the mixed layer. These increase in θ and decrease in q with height within the mixed layer are likely related to the entrainment of air with higher θ and lower q from above the mixed layer [e.g., Stull, 1988].

Figure 3.

Scaled mean profiles during the pre-onset, early onset, and latter onset periods, respectively, for (a) potential temperature (θ), (b) specific humidity (q), and (c) skew temperature (T). Short horizontal lines intersecting profiles indicate the mean heights of the mixed layer.

[17] Although an incremental decrease in zi was observed, the distinct evolution of the monitored thermodynamic variables occurred during different periods after the monsoon onset (Figure 3). During the early onset period, the mixed layer became relatively warm and moist. In this period, the mean values of the mixed layer mean T and q were higher by approximately 0.1°C and 0.5 g kg−1, respectively, than those during the pre-onset period (Table 1). However, the mixed layer became cooler and drier during the latter onset period. Compared with the early onset period, the mean values of the mixed layer mean T and q decreased by approximately 0.4°C and 0.3 g kg−1, respectively, during the latter onset period (Table 1). It is noted from Figure 3 that the mixed layer was the shallowest and coolest during the latter onset period. This condition is apparently related to the intense and sustained rainfall during this period (Figure 2b).

[18] It is interesting that regions just above zi after the monsoon onset warmed and dried concomitantly (Figure 3). Compared with the pre-onset period, an approximately 0.4 K increase in θ and 1 g kg−1 drying of q occurred just above zi during the latter onset period. This concomitant warming and drying above the mixed layer may be associated with environmental subsidence from above. As will be shown in the next section, subsidence may have played an important role in modulating the mixed layer during the monsoon onset.

[19] The relationship between zi and mixed layer thermodynamic properties is illustrated in Figure 4, where scatterplots of zi versus the mixed layer mean θ and q during different periods of the monsoon onset are shown. It is clear that the mixed layer mean θ generally increased, while the mixed layer mean q generally decreased with the increase in zi during each period of the monsoon onset. Linear correlations of the mixed layer mean θ and q with zi were made. All correlations, except for those between the mixed layer mean θ and zi during the pre-onset period and between the mixed layer mean q and zi during the latter onset period, were significant at the 10% level. The correlation coefficients of the mixed layer mean θ and q with zi were the highest during the early onset period, being 0.58 and −0.41, respectively. Both coefficients were significant at the 1% level. The observational results of deeper (shallower) mixed layers with a higher (lower) mixed layer mean θ and a lower (higher) mixed layer mean q are consistent with those observed by Johnson et al. [2001]. The concomitant increase in the mixed layer mean θ and the decrease in the mixed layer mean q concomitantly with the increase in zi reflect the basic developmental process of the mixed layer via buoyancy forcing, which is the formation of a deeper mixed layer through the entrainment of overlying air with higher θ and lower q [Stull, 1988]. Consequently, the observational results indicate that buoyancy may be a fundamental force that drives the formation of the mixed layer during the monsoon onset.

Figure 4.

Scatterplots of the mixed layer height (zi) versus (a–c) the mixed layer mean potential temperature (inline image) and (d–f) the mixed layer mean specific humidity (inline image) during the pre-onset, early onset, and latter onset periods, respectively. Also shown in each figure are the linear least square fit (dashed line) and the correlation coefficient (r).

4 Modulations of the Mixed Layer

[20] It has been shown that the mixed layer tended to shrink after the monsoon onset (Figure 2). On average, a cooler and drier mixed layer was observed during the latter onset period, when intensive rainfall occurred (Figures 2 and 3). This result indicates that precipitation is likely an important factor modulating the mixed layer. However, the appearance of the relatively warm and moist mixed layer during the early onset period suggests that the mixed layer could also be regulated by other factors. These factors will be investigated in this section by utilizing the regression analysis of the observational data obtained at times when they were not affected by rain-induced outflow, as shown in Figure 2c.

4.1 Surface Buoyancy Flux

[21] As suggested in Figure 4, the mixed layer during the monsoon onset period is generally buoyancy-driven. The evolution of the surface buoyancy flux during the monsoon onset period is shown in Figure 5a. The surface buoyancy flux B is defined as B = S + 0.61cpTE/L, where S is the sensible heat flux, E the latent heat flux, cp the specific heat of dry air, L the latent heat of vaporization, and T the surface air temperature. It is evident from Figure 5a that the surface buoyancy flux was positive throughout all periods of the monsoon onset. This result reinforces the importance of buoyancy forcing in promoting the development of the mixed layer during the monsoon onset period. The surface buoyancy flux generally intensified after the monsoon onset, which benefitted the warming and moistening of the mixed layer during the early onset period (Figure 3). Increases in the surface buoyancy flux were associated with increases in the sea-air temperature difference (Figure 5b) and the surface wind speed (Figure 5c) after the monsoon onset.

Figure 5.

(a) Time series of the surface buoyancy flux (B) (solid line). The dashed line indicates the pentad mean of the surface buoyancy flux. (b) Same as Figure 5a, but for the sea-air temperature difference (SST − Tair). (c) Same as Figure 5a, but for the surface wind speed (|v|).

[22] Because the surface buoyancy flux increased, one might expect an increase in zi after the monsoon onset. Nevertheless, by comparing Figures 5a with 2a, it is interesting to find that zi tended to decrease with the increase in the surface buoyancy. Such a relationship is much clearer in Figure 6, where scatterplots between these two variables are shown. A negative correlation was found between the surface buoyancy flux and zi, and it was significant at the 10% level during the pre-onset period and at the 1% level after the monsoon onset. This significant negative correlation between the surface buoyancy flux and zi suggests that other factors are more dominant in counteracting and reducing the growth of the mixed layer driven by buoyancy forcing during the monsoon onset.

Figure 6.

Scatterplots of the surface buoyancy flux (B) versus the mixed layer height (zi) during the (a) pre-onset, (b) early onset, and (c) latter onset periods. Also shown in each figure are the linear least square fit (dashed line) and the correlation coefficient (r).

4.2 Subsidence

[23] Environmental subsidence can shallow the mixed layer [Stull, 1988]. Although subsidence was not measured directly, its effect on the evolution of the mixed layer during the monsoon onset can be indirectly inferred from the sounding data.

[24] Figure 7a shows a time-height section of θ and a time series of zi during the monsoon onset period. Above the mixed layer, air with higher θ frequently extended downward. The downward-sloping isentropes were consistent with increased drying over the mixed layer (Figure 7b). Because this concomitant warming and drying occurred most prominently below cloud bases (Figures 7a and 7b), cumulus heating and drying would be less important to its formation. The concomitant warming and drying over the mixed layer could be related to the existence of subsidence there. The existence of subsidence over the mixed layer can be seen more clearly in Figure 7c, which shows a time series of vertical p velocity over the Mirai at 925 hPa from the objective analysis data. By comparing Figures 7a with 7c, it is interesting to note that a proxy for the subsidence of air could be well outlined by a contour of 301.5 K θ. Although the time intervals of the objective analysis data were 6 h (i.e., twice as long as those of the sounding data), the intensification of the downward motion from the objective analysis data generally corresponded to the downward extension of the contour of 301.5 K θ, with their correlation coefficients being −0.20, −0.31, and −0.23, respectively, during the pre-onset, early onset, and latter onset periods.

Figure 7.

(a) Time-height section of potential temperature (shaded). The 301.5 K contour is outlined by the solid line. Also shown are time series of the mixed layer height (open circle), the low-level cloud base height (CBH) (<1 km) measured by the ceilometer (dashed line), and the lifted condensation level (LCL) (dotted line). For the sake of clarity, the CBH and LCL have been filtered with a daily running mean. (b) Time-height section of specific humidity (shaded) superimposed by the 301.5 K contour of potential temperature (solid line). (c) Time series of vertical p velocity (ω) over the Mirai at 925 hPa from the objective analysis data.

[25] It can be seen from Figure 7a that there was a tendency for zi to decrease when the contour of 301.5 K θ extended downward. Scatterplots of the height of 301.5 K θ versus zi are shown in Figure 8 to illustrate this relationship more clearly. A positive correlation existed between the height of 301.5 K θ and zi, which was significant at the 5% level during the pre-onset period and at the 1% level after the monsoon onset. The value of the correlation coefficient increased during the early onset period and was the highest during the latter onset period. Figure 8 indicates that the mixed layer generally shallows with the intensification of the warming over the mixed layer. As shown in Figure 7, the warming over the mixed layer could be related to the subsidence of air from above, although their relationship deserves further investigation by analyzing the budgets of heat and moisture. Consequently, Figure 8 implies that environmental subsidence could suppress the depth of the mixed layer during the monsoon onset.

Figure 8.

Same as Figure 6, but for the 301.5 K potential temperature height (Hθ = 301.5) versus the mixed layer height (zi).

[26] Figure 7a also shows the evolution of the low-level cloud base height (CBH) (<1 km) measured by the ceilometer on board the Mirai and the lifted condensation level (LCL) calculated from the mixed layer mean temperature and moisture. Subsidence over the mixed layer may be closely related to cumulus activity during the monsoon onset period, which could be inferred from several observational facts. First, it is evident from Figure 7a that the concomitant warming and drying occurred near the CBH, indicating that the air subsided markedly around the CBH. It is also noted that stronger downward motions were observed during the early onset and latter onset periods (Figure 7c), which is consistent with the intensification of convection after the monsoon onset (Figure 2b). Furthermore, most echo tops around the radar during the monsoon onset period were lower than 4 km (Figure 2b) and belonged to the category of shallow cumulus convection as defined by Johnson et al. [1999]. Consequently, the compensating downward motion induced by shallow cumulus [e.g., Sarachik, 1974] would greatly affect the subsidence motion near the base of clouds during the monsoon onset.

4.3 Cumulus Clouds Atop the Mixed Layer

[27] Cumulus clouds atop the mixed layer can also reduce the depth of the mixed layer due to the removal of the sub-cloud layer air by an upward cloud-base mass flux [Stull, 1985]. As shown in Figure 7, the base height of the low-level clouds measured by the ceilometer was close to the LCL. The mean difference between the CBH and zi was 184 m, suggesting that the CBH was close to the top of the transition layer, as can be inferred from Figure 3. Consequently, there were clouds with bases frequently atop the mixed layer during the monsoon onset period.

[28] Upward cloud-base mass flux is proportional to upward motion at cloud base. Using the method described in section 2, the vertical motion at cloud base was calculated from Doppler radar observations. Figure 9 shows the evolution of the upward motion at cloud base. Similar to the surface buoyancy flux, a general increase in the intensity of the upward motion at cloud base occurred after the monsoon onset, reinforcing the argument that the intensification of cumulus convection promotes compensating subsidence over the mixed layer during the monsoon onset period (Figure 7).

Figure 9.

Time series of upward motion at cloud base (wup) (vertical bar). The dashed line indicates the pentad mean of the upward motion at cloud base.

[29] A regression analysis on the upward motion at cloud base and zi is shown in Figure 10. A negative correlation generally existed between the upward motion at cloud base and zi. The correlation coefficient became higher after the monsoon onset. It was significant at the 5% level during the early onset period and at the 1% level during the latter onset period. This result indicates that with the intensification of convection, cumulus clouds atop the mixed layer also contribute to the suppression of the mixed layer after the monsoon onset.

Figure 10.

Same as Figure 6, but for upward motion at cloud base (wup) versus the mixed layer height (zi).

5 Relationship Between Clouds and the Sub-Cloud Layer

[30] By comparing Figures 9 with 5a, it can be seen that the evolution of the upward motion at cloud base bore a resemblance to that of the surface buoyancy flux. It was found that peaks of the upward motion at cloud base coincided well with those of the surface buoyancy flux. The relationship between the upward motion at cloud base and the surface buoyancy flux is shown in Figure 11. A positive correlation generally existed in each period of the monsoon onset. Much better correlations appeared after the monsoon onset when both the upward motion at cloud base and the surface buoyancy flux began to intensify. The correlation coefficient was 0.49 (significant at the 5% level) during the early onset period and increased to 0.53 (significant at the 1% level) during the latter onset period.

Figure 11.

Same as Figure 6, but for upward motion at cloud base (wup) versus the surface buoyancy flux (B).

[31] This positive correlation between the upward motion at cloud base and the surface buoyancy flux is consistent with the result of Nicholls and LeMone [1980]. They analyzed turbulence measurements during the Global Atmospheric Research Program Atlantic Tropical Experiment (GATE) and found a close connection between the cloud-base mass flux and the surface buoyancy flux. The much better correlation between the upward motion at cloud base and the surface buoyancy flux after the monsoon onset is also similar to the observational result of LeMone and Pennell [1976] who found a stronger coupling between the cloud and sub-cloud layers under more enhanced cumulus convection.

[32] Previous observational and numerical studies have indicated the importance of the interaction between the cloud and sub-cloud layers. It was found that cumulus convection is enhanced with an increase in surface turbulent fluxes [Ogura and Cho, 1974; Ghate et al., 2011]. However, cumulus clouds can also modify and increase sub-cloud-layer turbulent fluxes [Emmitt, 1978; Tiedtke et al., 1988; Grant, 2001; Chandra et al., 2010]. Consequently, Figure 11 leads to the conclusion that there would be positive feedback between the upward motion at cloud base and the surface buoyancy flux during the monsoon onset.

[33] Such positive feedback after the monsoon onset is conducive to the enhancement of the monsoon onset. The intensification of the surface buoyancy flux warms and moistens the mixed layer from below, while the enhancement of the upward motion at cloud base strengthens the compensating subsidence and shrinks the mixed layer, thereby reducing the dilution of the mixed layer air via the mix-out effect from above. As a result, the positive feedback between the upward motion at cloud base and the surface buoyancy flux facilitates the formation of the warmer and moister mixed layer, as observed during the early onset period of the monsoon (Figure 3). Consequently, this process would modulate the mixed layer into a state favorable for the further development of the monsoon convection.

[34] Although the observed modulation processes can contribute to the formation of the warmer and moister mixed layer during the early onset period of the monsoon, it is noted that the role of advection and radiation in the mixed layer during the monsoon onset is unresolved in this paper. Johnson et al. [2001] found that advective and radiative effects are important in the thermodynamic budget of the mixed layer over the warm pool prior to and during the late stages of the strong westerly wind burst associated with a Madden-Julian oscillation event. The rapid buildup of convective activity during the monsoon onset may enhance the local advection of temperature and moisture and alter the local radiation balance in the mixed layer. Further research is needed to determine whether the physical processes of advection and radiation during the monsoon onset can modulate the mixed layer to facilitate the further development of the monsoon convection and whether their contributions are more significant than the processes illustrated in this study.

[35] In addition, the positive feedback between the upward motion at cloud base and the surface buoyancy flux observed here supports the previous speculation about the significance of the air-sea interaction during the monsoon onset [e.g., Ueda and Yasunari, 1996; Wu, 2002]. The results of this study have indicated a distinct coupling between the sea surface and cumulus clouds through atmospheric mixed layer processes during the monsoon onset. Physical processes such as the evolution of the oceanic mixed layer and its relation to the coupling between the sea surface and cumulus clouds during the monsoon onset also deserve further study.

6 Summary and Conclusions

[36] The thermodynamic structure and evolution of the atmospheric mixed layer during the onset of the western North Pacific summer monsoon were analyzed using data collected from the research vessel Mirai at 12°N and 135°E during 6–24 June 2008. The mixed layer was identified from three-hourly upper-air soundings and investigated in conjunction with the progress of the monsoon onset. Meanwhile, oceanographic and atmospheric observations including the surface, a C-band Doppler radar, and a ceilometer have been analyzed together with the upper-air soundings to examine important factors modulating the evolution of the mixed layer during the monsoon onset period.

[37] The mixed layer could be identified in 80% of the sounding data. The depth of the mixed layer ranged from 176 to 720 m and tended to shrink after the monsoon onset. The mean depth of the mixed layer was 539 m during the pre-onset period, 474 m during the early onset period, and 418 m during the latter onset period of the monsoon. The overall mean depth of the mixed layer during the monsoon onset period was 482 m. The mixed layer became relatively warm and moist during the early onset period. However, a very cool and dry mixed layer was observed during the latter onset period when intense and sustained rainfall occurred.

[38] The intensification of precipitation greatly affected the formation of shallow mixed layers after the monsoon onset. The modulation of the mixed layer through factors other than precipitation was also investigated. Although the surface buoyancy flux intensified with the monsoon onset, a significant negative correlation existed between the surface buoyancy flux and the depth of the mixed layer. This correlation implies that that upward motion at cloud base and compensating environmental subsidence over the mixed layer, both of which intensified after the monsoon onset and were significantly negatively correlated with the depth of the mixed layer, act to suppress the growth of the mixed layer driven by buoyancy forcing and shrink the mixed layer after the monsoon onset.

[39] A significant positive correlation existed between the surface buoyancy flux and the upward motion at cloud base after the monsoon onset, when both began to intensify. This correlation suggests positive feedback between the surface buoyancy flux and the upward motion at cloud base during the monsoon onset period. Such positive feedback would promote the warming and moistening of the mixed layer from below and reduce the dilution of the mixed layer via entrainment from above, thereby facilitating the formation of the warmer and moister mixed layer during the early onset period. This result implies that active coupling between the sea surface and cumulus clouds after the monsoon onset could modulate the atmospheric mixed layer into a state favorable for the further development of the monsoon onset.

Acknowledgments

[40] The authors would like to express their deep appreciation to the entire crew of the research vessel Mirai and to the technical staffs of Global Ocean Development Inc. and Marine Works Japan, Ltd., for their great support in obtaining the intensive observation data.