1 INTRODUCTION

The Asian summer monsoon (ASM) is one of the most complicated components of the Earth's climate system (Wang et al., 2001; He et al., 2006, 2007). It not only directly affects the social activities and economy in Asia with the largest population in the world but also influences other area via its remote effects (Gadgil and Kumar, 2006; Trenberth et al., 2006; He et al., 2007; Wang et al., 2018). The onset process of the ASM indicates the beginning of summer monsoon. Previous studies have proposed that the ASM firstly establishes over the southeastern Bay of Bengal (BOB), then over the South China Sea (SCS), and finally over India (Wu and Zhang, 1998; Wang and LinHo, 2002; Mao and Wu, 2007; Liu et al., 2015). As the beginning of the ASM onset, many studies have paid attention to the onset process of the BOB summer monsoon (BOBSM). For instance, Mao and Wu (2007) pointed out that both the winter-tosummer seasonal transition and the ASM onset started with the northward tilting of the subtropical ridge surface towards the warmer region, presenting the emergence of positive meridional temperature gradient in the mid-upper-troposphere of the BOB. A type of low-level vortex, which always existed over the BOB in early May during the BOBSM onset (Lau et al., 1998; Liu et al., 2002), is known as the monsoon onset vortex (MOV) in the atmosphere. Vinayachandran et al. (2007) showed the detailed structure of the MOV and its association with the sea surface temperature (SST) in the BOB. They found out that the warmest SST occurred in the eastern BOB before the BOBSM onset. Mao and Wu (2011) attributed the generation of typhoon Nargis over the BOB in spring 2008 to the atmospheric barotropic instability and diabatic heating over the BOB. The BOBSM built up with the northward movement of the developed MOV (Liu et al., 2013, 2015). Case studies suggested that the formation and development of the BOB MOV were associated with the short-term spring warm pool in the BOB via the local air-sea interaction (Wu et al., 2011, 2012). In particular, the atmospheric available potential energy is generated over the BOB and then converted to the kinetic energy, thereby resulting in the formation of a MOV (Wu et al., 2012).

The evolution of the BOB MOV is closely associated with the variation of the local oceanic mixed layer, which reflects the intense turbulent mixing in the upper ocean and directly interacts with atmosphere. Wu et al.(2011, 2012) pointed out that the shallow ocean mixed layer facilitated the quickly warming of the BOB SST, while the mixed layer depth (MLD) increased rapidly with the development of the MOV to cool the local SST by the upwelling of cold water in the deeper layers. Therefore, the variation of local oceanic MLD is vital to the formation and development of the BOB MOV. Actually, the transports of mass, momentum, and energy in the mixed layer provide most of the ocean kinetic energy (de Boyer Montégut et al., 2004). On the other hand, the wind forcing also plays an important role in the oceanic energy cycle and oceanic circulation, especially the heat budget of surface mixed layer (Shi et al., 2017, 2018). Previous studies emphasized the key roles of the MLD in the distributions of the net surface heat flux (Chen et al., 1994), the near-surface acoustic propagation (Sutton et al., 1993) and even the marine ecosystems (Prasanna Kumar and Narvekar, 2005). When the spring mixed layer in the BOB is shallow, the solar radiation with a wavelength less than 700 nm is able to penetrate the mixed layer, and then directly warmed the seawater below the mixed layer. Based on the Buoy data, Sengupta et al. (2002) demonstrated that the solar radiation flux penetrating the BOB mixed layer in spring could reach about 45 W/m². Its magnitude and variations greatly affected the upper ocean dynamics and thermodynamics, as well as the marine ecological process (Kraus and Turner, 1967; Lewis et al., 1990; Ohlmann et al., 1996). The effects of the BOB MOV on the oceanic mixed layer resemble those of the tropical cyclone, which have been examined since the 1960s. In the mid-1960s, the scientists started to use one-dimensional oceanic mixed layer model to examine the variation of upper oceanic layers (Kraus and Turner, 1967). Subsequently, the two- and threedimensional ocean models were developed to investigate the oceanic responses to the tropical hurricanes (Elsberry et al., 1976; Price, 1981). It was found that the tropical cyclones brought the cold water in deeper layers into the mixed layer to decrease SST by the entrainment upwelling processes (Sakaida et al., 1998). The colder SST could change the intensity and structure of tropical cyclones as a feedback (Behera et al., 1998; McPhaden et al., 2009; Zeng and Chen, 2011). In addition, the MOV-induced precipitation increased the input of freshwater into the BOB, where the upper-oceanic salinity gradient and the pycnocline was changed (Lukas and Lindstrom, 1991; Sprintall and Tomczak, 1992). Thus, the thermocline is deeper than the pycnocline to form the barrier layer (Sprintall and Tomczak, 1992). The barrier layer thickness (BLT) was defined as the difference between the depth of the thermocline and the pycnocline. Acting as a thermal barrier for the vertical heat exchange, the barrier layer could prevent the effective heat exchange between the thermocline and the pycnocline (Godfrey and Lindstrom, 1989). It partly offset the cooling effects of the entrainment upwelling to modulate the oceanic heat budget in the mixed layer, the SST and the air-sea heat flux (Swenson and Hansen, 1999).

Although the effects of spring warm pool in the BOB on the MOV and ASM onset have been shown in Wu et al. (2012), the feedback of BOB MOV on the evolution of local SST with the barrier layer is unclear. In climatology, the warming of the spring SST over the BOB is primarily controlled by the air-sea net heat flux while the role of oceanic process is limited (Santoso et al., 2010). However, the influences of spring tropical cyclone on the SST in the BOB should be quite different. Since the BOB MOV lives long after its generation in spring 2003, it implicates a potential positive feedback between MOV and SST in this year. Hence, the present study aimed to revisit how the BOB MOV influences the heat budget in the upper ocean with a barrier layer, and to show interaction between MOV and SST especially after the ASM onset.

2 DATA AND METHOD 2.1 Data

The pentad oceanic reanalysis datasets, including the oceanic temperature, salinity, and threedimensional current velocities, were investigated using the US National Centers for Environmental Prediction (NCEP) Global Ocean Data Assimilation System (GODAS; Behringer and Xue, 2004; Saha et al., 2006). This dataset had constant zonal resolution of 1° and a variable meridional grid of enhanced to 1/3° within 10° of the equator, accompanied by 40 levels with a 10-m resolution in the upper 200 m. The daily SST was obtained from the National Oceanic and Atmospheric Administration (NOAA) optimum interpolation SST, version 2 (OISSTv2; Reynolds et al., 2002), with a high horizontal resolution of 0.25°×0.25°. The daily atmospheric reanalysis datasets, including three-dimensional wind, 10-m wind, net surface shortwave radiation, net surface longwave radiation, surface latent heat flux, and surface sensible heat flux, were investigated using the National Centers for Environmental Prediction-Department of Energy (NCEP-DOE) Reanalysis 2 (Kanamitsu et al., 2002). Their surface variables are stored on a global T62 Gaussian grid, and the upperair variables are available at the 2.5°×2.5° horizontal resolution and extend from 1 000 to 10 hPa with 17 vertical pressure levels. The pentad precipitation fields were taken from the Climate Prediction Center (CPC) Merged Analysis of Precipitation (CMAP) with a resolution of 2.5°×2.5° (Xie and Arkin, 1997). All the daily variables were converted to the pentadmean values similar as the temporal resolution of the GODAS and CMAP products. The climatology of each element was defined by its arithmetic average from 1980 to 2010.

2.2 Methodology

The oceanic isothermal layer (IL) and mixed layer (ML) are defined as follows. The ML is an upper turbulent layer in the ocean, which was defined as the quasi-homogeneous surface density layer (Du et al., 2005). Kara et al.(2000a, b) have proposed the optimal definitions for determining the IL depth (ILD) from temperature profiles and the MLD from density profiles, which could be applied in all regions of the world's oceans. The ILD (MLD) was identified as the base depth of an isothermal (isopycnal) layer, where the sea temperature (density) changed by a fixed amount of , where T and S denote the sea temperature and salinity, respectively, and P=0 from the temperature (density) at a reference of 10-m depth. For the MLD, Kara et al. (2000a) compared different criteria and suggested that ΔT=0.8℃ is a proper criterion. Because of the inconsistent distributions of sea temperature and salinity, the IL was usually deeper than the ML. Their difference was generally described as a barrier layer (BL; Lukas and Lindstrom, 1991; Sprintall and Tomczak, 1992; You, 1995), and the depth from the bottom of the MLD to the top of the thermocline was defined as the barrier layer thickness (BLT).

The heat budget in the mixed layer was expressed by the following equation (Qiu, 2000; Qu, 2003; Du et al., 2005):

in which T_m indicates the mean seawater temperature in the mixed layer, h_m is the MLD, ρ and C_p are the reference density and specific heat of seawater, respectively. T_d is the temperature at 5-m below MLD to represent the cold water entrained in the ML. The indicates the zonal and meridional velocities averaged in the MLD. Each term in Eq.1 is averaged in the mixed layer. The term in the left-hand side of Eq.1 is for the temperature tendency, while the four terms in the right-hand side represented the effect of surface thermal forcing, horizontal advection, vertical entrainment, and the residual term (R), respectively.

In Eq.1, Q₀ denotes the net surface heat flux, which was the sum of the net downward shortwave radiation (Q_sw), net downward surface longwave radiation, and surface latent and sensible heat fluxes. Here a positive value indicates the downward heat flux into the ocean. The penetration of solar radiation below the ML (q_d) is estimated as (Pacanowski and Griffies, 1999)

As to the entrainment rate (W_ent) in Eq.1, it was calculated as per Qu (2003):

in which , and are for the tendency of MLD, the velocity of water parcel on the base of the ML, and the horizontal advection of water parcels below the ML, respectively.

3 RESULT AND DISCUSSION 3.1 Onset process of the BOBSM in 2003

The MOV is actually a tropical cyclone during the BOB summer monsoon onset. These tropical cyclones show different genesis and complicated tracks, and their corresponding lifetimes are inconsistent. In 2003, the BOBSM built up on May 12. Before the BOBSM onset, the MOV formed in the southeastern BOB on May 8. It then developed and moved northwestward. The MOV peaked on May 13 with the minimum of sea level pressure existing on the BOB. Subsequently, the MOV gradually weakened and moved northeastward. It finally faded away over the western Burma on May 19. Therefore, the 2003 case is one of the most classical examples for investigating the effects of MOV on the BOB mixed layer heat budget because of its typical track and lifetime. The lifespan of the BOB MOV can be verified by the pentad evolution of the low-level wind and precipitation in spring 2003 (Fig. 1). In Pentad 24 (April 26–30), the subtropical high was continuously extending from the Arabian sea to the northern SCS, presenting two centers of the low-level anticyclone over the northern Indian peninsula and the northern Indo-China peninsula, respectively. Its ridgeline, which was the 700-hPa zonal wind shear line (u=0 and (∂u/∂y)>0), was located at 20°N (Fig. 1a). At this moment, the precipitation took place in the equatorial region (to south of Indian peninsula). Subsequently, a closed cyclonic circulation was formed over the northern Sumatra, along with the enhancement of local precipitation (Fig. 1b). In Pentad 26 (May 6–10), this cyclone strengthened rapidly to be an MOV, and the associated rainfall over the southern BOB increased rapidly (exceeding 30 mm/d). Meanwhile, a weak low-level Indo-Burma trough was observed over the northern BOB (Fig. 1c). The MOV continued to develop and moved quickly northward in Pentad 27, along with the maximum of precipitation shifting northward to the northwestern BOB. When the MOV moved northward to meet the Indo-Burma trough, the original zonally-oriented ridge of the subtropical high started to split over the western Indo-China peninsula, corresponding to the onset of the BOBSM (Fig. 1d). In Pentad 28 when the MOV completely merged with the Indo-Burma trough, the MOV diminished and disappeared in a short time. Simultaneously, the continuous subtropical high belt located near 20°N thoroughly split and the precipitation was centered over the northeastern BOB (Fig. 1e). After the BOBSM commenced in Pentad 27, the ASM onset started to propagate towards the SCS and western Pacific, but it was blocked on the eastern coast of India where was influenced by the northwesterly wind behind the deep Indo-Burma trough (Fig. 1f).

As one of the main low-level atmospheric heat sources, the surface sensible heat flux is an important component of the surface heat balance, and its inhomogeneous distribution could affect the establishment and maintenance of the monsoon circulation. Figures 2 and 3 depict the pentad evolution of the SST, the surface sensible heat flux and 10-m winds during the lifespan of the BOB MOV, respectively. The high SST with its maximum greater than 30℃ emerged in the central and eastern BOB to form the local spring warm pool from late April to early May (Fig. 2a & b). At this time, the low-level anticyclone was settled over the northern BOB, along with the relatively weak surface wind speed, corresponding to the small surface sensible heat flux (Fig. 3a & b). In Pentad 26, in spite of a decrease of SST, the spring warm pool still maintained in the BOB (Fig. 2c). Due to the strengthened surface easterly wind over the southern BOB, the surface sensible heat flux greatly increased with its maximum exceeding 25 W/m². Based on the atmospheric thermal adaptation theory (Wu and Liu 2000), a remarkable surface cyclonic circulation (i.e., MOV) was formed as response to the strong surface sensible heating (Fig. 3c). When the MOV moved northward to the northeastern BOB in Pentad 27, the local SST still exceeded 30℃ to support the further development of the MOV (Fig. 2d). The northeastern BOB was thus controlled by the southerly wind during the arrival of the MOV (Fig. 3d). The corresponding onshore current accumulated the warm water in the northeastern BOB to maintain the higher SST and the stronger surface sensible heating over the northeastern BOB. In contrast, over the northwestern BOB, the offshore current induced by the surface northerly wind could induce the upwelling of cold water, which resulted in the decrease of local SST and surface sensible heat flux (Figs. 2d & 3d). As the MOV died out in Pentad 28, the southwesterly wind over the northern BOB was strengthened remarkably with the drastically decrease of the local SST (Figs. 2e & 3e). The stronger offshore current could further weaken the surface sensible heat flux over the western BOB (Fig. 3e). In Pentad 29, the BOB spring warm pool disappeared completely, and the surface sensible heat became negative over the entire BOB (Figs. 2f & 3f).

Figure 4 displays the pentad evolution of the seawater temperature tendency averaged in the mixed layer to indicate the SST variation in the BOB. Prior to the formation of the BOB MOV, the SST tended to rise in the northern Indian ocean with its maximum in the BOB (Fig. 4a & b), corresponding to the generation of the BOB spring warm pool. When the MOV formed in Pentad 26, the SST tendency began to decrease in the southern BOB, where the surface wind speed and sensible heat flux got enhanced (Fig. 4c). As a result of the northward movement of the MOV, the cooling tendency of SST occurred in the northeastern BOB (Fig. 4d), suggesting the weakening of the spring warm pool in the BOB. As the MOV gradually decayed in Pentad 28, the strong low-level northwesterly wind was evident over the BOB, followed by the continuous decrease tendency of the local SST (Fig. 4e). This decrease tendency could persist until Pentad 29 when the BOB spring warm pool disappeared completely (Fig. 4f). Since the BOB MOV could bring on large amount of rainfall, the barrier layer would come up due to the mass of fresh water input during the ASM onset. In the next section, we will show how the barrier layer modifies the role of air-sea heat flux in the SST evolution during the onset of the BOBSM.

3.2 Heat budget analysis 3.2.1 The net surface heat flux

Figure 5 shows the pentad evolution of the temperature tendency induced by the surface heat flux. Before the formation of the BOB MOV, the net surface heat flux could increase the SST in the northern Indian ocean. The warming center with a maximum greater than +0.4 K/pentad was positioned in the central and eastern BOB (Fig. 5a & b), which was close to the center of the total SST tendency. It indicated that the formation of BOB spring warm pool was attributed primarily to the net surface heat flux. As the MOV formed over the southern BOB in Pentad 26, the SST tendency due to the net surface heat flux changed from positive to negative (Fig. 5c), implying the loss of heat from the sea surface. In the developing stage of the MOV, the net surface heat flux was negative with its center over the central and eastern BOB (Fig. 5d). Mass of heat lost from the upper ocean to the atmosphere via the air-sea interaction to decrease the local SST rapidly. The BOB spring warm pool began to attenuate in this stage. When the MOV weakened and disappeared in Pentad 28, the cooling center related to the net surface heat flux was confined to the northeastern BOB (Fig. 5e) because of the strong low-level southwesterly wind. The SST thus further decreased to break the spring warm pool in the BOB. In Pentad 29, the southwesterly wind over the BOB was weakened to some extent (Fig. 5f), corresponding to the disappearance of the cooling tendency associated with the net surface heat flux.

To quantitatively evaluate the relative contribution of each component in the net surface heat flux to the SST tendency, Fig. 6 presents the temporal evolution of these terms in the BOB (85°E–95°E, 5°N–15°N, shown in Fig. 2a) during the lifespan of the MOV. Although some studies indicated that great uncertainties existed on the radiation flux between observation and different reanalysis datasets (Wang et al., 2017; Zhang et al., 2018), the surface radiation flux from the NCEP/DOE was used to ensure the dynamic consistency among the datasets. Generally, the net surface heat flux depended on the downward solar radiation flux and the upward surface latent heat flux. In contrast, the contributions of the solar radiation penetrating the mixed layer (q_d) and the surface longwave radiation flux were limited, and the effect of surface sensible heat flux was almost negligible. Before Pentad 26, the apparent low-level anticyclone over the BOB produced the tropospheric sinking and the clear-sky condition. Thus, strong solar radiation arrived at the sea surface (exceeding 250 W/m²). Meanwhile, the surface latent heat flux became weaker due to the small surface wind speed in situ. As the weak mixing effect in the upper ocean under the influences of small surface wind speed, a number of heating absorbed by the shallow mixed layer gave rise to the spring warm pool in the BOB. This favored the formation of the MOV and the onset of the BOBSM. However, the q_d was strong with its value over 50 W/m² and was quite close to the buoy observation (Sengupta et al., 2002). As a result, the heating effects of the net surface heat flux on the ML was reduced before the MOV formation.

When the MOV formed and developed in Pentads 26–27, more thicker clouds appeared over the BOB to remarkably reduce the net solar shortwave radiation (Q_sw) and the net upward longwave radiation (Q_lw), accompanied by the decrease of the solar radiation penetrating the MLD (q_d). In the meantime, the enhanced surface southwesterly wind induced by the MOV could increase the upward surface latent heat flux (Q_lw) rapidly via the wind-evaporation mechanism. Thus, mass of energy lost from ocean to atmosphere via the air-sea interface, presenting the negative-to-positive transition of the downward net surface heat flux over the BOB, where the SST was cooled evidently. As the MOV faded away in Pentad 28, the surface wind speed gradually weakened to attenuate the upward latent heat flux. The less transport of heat from ocean to atmosphere resulted in a weaker cooling tendency of SST over the BOB. Subsequently, the local net surface heat flux was below 50 W/m² in early June, when the solar radiation penetrating the MLD (q_d) also gradually decreased to almost zero.

Therefore, the evolution of the net surface heat flux during the formation and development of the MOV was primarily attributed to the changes of the downward solar radiation and the upward latent heat flux. Their joint effect caused the loss of upperoceanic heat to decrease SST in the BOB. The upward latent heat flux played a relatively more important role. In addition, the change of the solar radiation penetrating the MLD (q_d) followed the variation of the downward solar radiation reaching the sea surface. Although the magnitude of q_d was much smaller, suggesting its little contribution to the local SST after the generation of the MOV, we cannot ignore its effects on the formation of spring warm pool in the BOB before the formation of the MOV. Actually, the q_d could warm the seawater temperature below the mixed layer due to the shallower MLD, and then indirectly affected SST through the mixed layer process after the MOV formation. We will further investigate the role of the oceanic processes in the SST tendency over the BOB during the lifespan of the MOV.

3.2.2 Oceanic processes in the mixed layer

As shown by the Eq.1, the large-scale oceanic processes associated with the SST tendency included the horizontal advection of seawater temperature and the effect of vertical entrainment. Fig. 7 depicted the temporal evolution of the SST tendency induced by the horizontal advection in the BOB. The horizontal advection was quite weak in the whole BOB region before the formation of the MOV (Fig. 7a & b). Since the MOV formed over the southern BOB in Pentads 26–27, the horizontal temperature advection was strong in the southern but still weak in the northern BOB (Fig. 7c & d). In Pentad 28 when the MOV began to decay, the positive center of horizontal temperature advection started to occur in the northern BOB (Fig. 7e). The horizontal temperature advection in the mixed layer of the BOB completely vanished in Pentad 29 (Fig. 7f). It was obviously noted that the warm and cold horizontal temperature advections were respectively found in the western and eastern of the southern BOB region in Pentads 27–28 (Fig. 7d & e), which suggested the horizontal temperature advection was somehow important to the mixed layer temperature tendency. However, based on the results of the area averaged horizontal temperature advection, the total horizontal temperature advection in the mixed layer of the BOB contributed little to the SST cooling tendency during the formation and development processes of the MOV.

Compared to the horizontal temperature advection in the mixed layer of the BOB, the upwelling of cold water caused by the vertical entrainment (Fig. 8) played more important role in the SST tendency. Before the formation of the MOV, the oceanic mixing was relatively small due to the weak surface wind, along with little vertical entrainment in the northern Indian Ocean (Fig. 8a & b). As the MOV formed over the southeastern BOB, the local surface wind was strengthened to increase the upper ocean mixing to the west of the MOV center, which enhanced the insitu vertical entrainment (Fig. 8c). The SST thus decreased remarkably. With the northward movement of the MOV, the center of cooling due to vertical entrainment extended into the northern BOB in Pentad 27 (Fig. 8d), resulting in a rapid decrease of the SST in the BOB. Since the northern BOB was controlled by the southwesterly wind over the northern BOB even after the MOV vanished, this cooling effect could still maintain in Pentad 28 (Fig. 8e).

Figure 9 provides more details about the vertical entrainment in the mixed layer of the BOB during the MOV lifespan. The entrainment rate depended on the vertical temperature gradient at the base of the mixed layer, the MLD and the BLT. Before Pentad 26, the upper ocean mixing in the BOB was rather weak due to the relatively small surface wind speed, corresponding to little entrainment in the mixed layer of the BOB (Fig. 9a). Meanwhile, the average MLD in the BOB was quite shallow with its value of about 25 m (Fig. 9b), but the vertical temperature gradient at the base of the mixed layer was quite large (Fig. 9a). The thin mixed layer facilitated the penetrating of the solar radiation to the base of mixed layer where the barrier layer is thin in this stage (Fig. 9b). When the MOV formed over the eastern BOB in Pentad 26, the local surface wind was strengthened rapidly (Fig. 3c), leading to the stronger ocean mixing and the entrainment rate. Since the vertical temperature gradient changed little, the cold water was taken into the upper ocean by the entrainment at the base of the mixed layer. Afterwards, the intensified ocean mixing could rapidly deepen the local MLD and enhance the entrainment rate in the upper ocean in Pentad 27, accompanied by the northward movement of the BOB MOV. The MOV-induced precipitation brought more freshwater into the BOB, where the salinity stratification was changed to significantly thicken the BLT. The vertical temperature gradient at the base of the mixed layer began to decrease remarkably (Fig. 9a). As the MOV decayed over the northern BOB in Pentad 28, the strong surface southwesterly wind maintained the stirring and mixing in the upper ocean, followed by the continuous increase of the MLD to about 45 m (Fig. 9b). On the other hand, the monsoonal precipitation could further stabilize the upper ocean salinity stratification, causing that the BLT gradually increased to about 12 m, accompanied by the continuous decrease of the vertical temperature gradient (Fig. 9). Thus, the upwelling of cold water was jointly attributed to both the vertical entrainment rate and the vertical temperature gradient at the base of the mixed layer during the development and decay phases of the BOB MOV.

With the enhancement of the surface wind and the monsoonal precipitation over the BOB, both the local MLD and BLT increased in the development and decay stages of the MOV. The stronger vertical entrainment was jointly induced by the deepened MLD and the enhanced upwelling motion at the base of the mixed layer, whereas the vertical temperature gradient at the base of the mixed layer was decreased due to the variations of the q_d and the BLT. Prior to the formation of the MOV with shallow MLD in the BOB, the strong q_d could penetrate the mixed layer and directly heat the deeper water. When the MOV formed and moved onto the northern BOB, its resultant precipitation brought more fresh water into the BOB to change the salinity stratification in situ, leading to a rapid increase of the BLT to 12–15 m. The vertical temperature gradient at the base of the mixed layer thus gradually decreased as the BLT increased. This result indicated that the thermal barrier effect of the BL could prevent the cooling effect of the upwelling cold water on the SST. On the other hand, the seawater below the mixed layer heated by the q_d might further rise the seawater temperature in the barrier layer, which enhanced the thermal barrier effect of the BL in advance. Therefore, the oceanic process associated with the changes of q_d and BLT could slow down the SST decrease after the formation of the BOB MOV to support the development of the MOV and the enhancement of the monsoonal convection.

Consequently, the changes of SST in the BOB were jointly modulated by the net surface heat flux and the upwelling of cold water during the lifespan of the MOV in 2003. Their relative contributions were quantitatively presented by Fig. 10. Before the formation of the MOV in Pentad 26, the downward net surface heat flux could rise the SST to produce a spring warm pool in the BOB. At this moment, the oceanic process exhibited little impact on the SST changes. When the MOV formed over the eastern BOB in Pentad 26, the local SST started to decrease rapidly due to the upwelling of cold water induced by the stronger surface wind speed, along with little effects of the net surface heat flux. Whereas the atmospheric process became dominant in Pentad 27 and 28 when the MOV developed and moved northward. The evident reduction of the surface solar radiation and increase of the surface latent heat flux jointly produced the mass of heat loss from ocean to atmosphere, which rapidly decreased the SST in the BOB. In this episode, the cooling effects due to the oceanic process was weakened because of the thickening of BLT and the reduction of vertical temperature gradient at the base of the mixed layer, indicating the potential influences of barrier layer on slowing the SST decrease after the MOV formation. In addition, the horizontal ocean advection in the mixed layer always had little effect on the BOB SST change during the formation and development of the MOV.

4 CONCLUSION

Based on the pentad mean NCEP GODAS reanalysis datasets, this study diagnosed the heat budget in the mixed layer of the BOB during the MOV lifespan from April to May 2003, and further investigated the potential influences of the solar radiation penetrating the MLD and the BLT on the evolution of spring warm pool. Figure 11 is a schematic diagram illustrating the major air-sea interaction process associated with the MOV. Before the formation of the MOV, the clear-sky condition induced by the anticyclone over the BOB led to mass of solar radiation reaching the sea surface. It could produce the BOB spring warm pool with shallow MLD in a short time. Part of solar radiation with about 50 W/m² could penetrate the shallow mixed layer and heat the deeper water. The direct effect of the solar radiation penetrating the MLD on the simultaneous SST was very weak, but it showed potential influences on the SST via the oceanic process after the MOV formation. When the MOV formed over the BOB in Pentad 26, the local surface wind speed strengthened rapidly, resulting in the stronger oceanic mixing and the deeper MLD. Therefore, the upwelling of cold water decreased the SST tendency evidently. During the development of the MOV with continuous strengthening surface southwesterlies, the net solar radiation reduced remarkably owing to the cloud-radiation effect, but the upward surface latent heat flux increased rapidly via the wind-evaporation mechanism. Thus, more energy lost from oceanic surface to atmosphere to further decrease SST in the BOB. In particular, the increased upward surface latent heat flux was the primary factor for the oceanic heat loss in this stage.

In the ocean, the cooling effect of entrainment upwelling was determined by the entrainment rate and the vertical temperature gradient at the base of the mixed layer. During the formation and development of the MOV, the entrainment rate increased rapidly along with the deepening of MLD, whereas the vertical temperature gradient at the base of the mixed layer decreased remarkably. This weaker vertical temperature gradient was attributable to the heating effect of the solar radiation penetrating the MLD and the thickening of BLT. On one hand, the solar radiations penetrating the MLD directly heat the water below the mixed layer before the formation of the MOV. On the other hand, as the MOV formed and developed, the monsoonal precipitation provided more fresh water into the BOB to change the salinity stratification, resulting in a thicker BLT. The warmer seawater in the barrier layer could weaken the cooling effect of entrainment upwelling, which prolonged the life of spring warm pool in the BOB and facilitated the MOV development and the ASM onset. These results suggested that, besides the local air-sea interaction, other oceanic elements, such as the solar radiation flux penetrating the ML and the existence of barrier layer, might also play a role in modulating the effects of MOV on the evolution of SST in the BOB.

To ensure the dynamic consistency among the datasets, the surface radiation flux from the NCEP/DOE was used to emphasize the influences of the solar radiation penetrating the MLD and the BLT on the SST tendency during the MOV lifespan. However, recent studies indicated that great uncertainties existed on the radiation flux between observation and different reanalysis datasets. Although the solar radiation penetrating the MLD estimated from the NCEP/DOE dataset was close to the result from the buoy observation, more reanalysis datasets might be needed to inspect the reliability of the solar radiation penetrating the MLD in the future work. In this study, we mainly focused on the influences of the air-sea heat flux and the cold water upwelling on the SST change, but did not considered other oceanic processes, such as the lateral diffusion and the mesoscale eddy, which might be the main source to the residual term. In addition, the variation of the BLT is controlled by the fresh water flux injecting into the upper ocean, however, the MOV-induced precipitation over the BOB exhibited large year-by-year differences. Therefore, more cases are needed to reveal the influences of the solar radiation penetrating the MLD and the BLT on the BOB SST changes on interannual timescale.

5 DATA AVAILABILITY STATEMENT

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

6 ACKNOWLEDGMENT

The authors gratefully acknowledge the use of the HPCC for all numeric simulations and data analysis at the South China Sea Institute of Oceanology, Chinese Academy of Sciences.