Ocean Responses to Typhoon Namtheun Explored with Argo Floats and Multiplatform Satellites

Abstract

Argo salinity and temperature profiles, along with other sea surface measurements, were used to explore the impacts of Typhoon Namtheun (2004) on the ocean. Namtheun took local enthalpy heat from the sea (0.39–0.7 × 10⁸ J m⁻²), cooled the sea surface water as a result of vertical mixing (maximum 3–4°C) and produced heavy precipitation over the sea (100–180 mm). During this time, the vast latent heat released (2.6–4.4 × 10⁸ J m⁻²) by the precipitation made a larger contribution to the typhoon's energy budget than the local air-sea enthalpy flux. In the upper ocean, the oceanic responses can be separated into two sub-processes, the fast spin-up accompanied by one-dimensional vertical mixing and the slow spin-down accompanied by the convergence of surface water. From Argo profiles on 28 July, it can be seen that the typhoon-induced surface mixing broke down the seasonal thermocline (approximately 20 db) within one day. In addition, the shallower (<200 db) convergence of the sea surface fresh water as a result of precipitation also made the post-typhoon water fresher (0.04 (practical salinity scale used)). In the deep ocean, the rapid upwelling at the top of the permanent thermocline suggests that the fast spin-up is a barotropic mechanism, probably gravity pressure. During the slow spin-down stage, the upwelling signal propagated downward (approximately 2 m h⁻¹) from the shallow water to the deep ocean for about 10 days; this was a baroclinic process. The baroclinic mechanism was more effective in maintaining a cyclonic eddy than in maintaining an inertial wave, and the low sea surface height anomaly and upwelling lasted much longer than the inertial oscillation (>20 days as opposed to approximately 10 days). This change in vertical structure and long-term upwelling could have impacts on the ocean environment and even on the short-term climate.

RÉSUMÉ [Traduit par la rédaction] Nous nous sommes servis des profils de salinité et de température Argo, de pair avec d'autres mesures de la surface de la mer, pour explorer les répercussions du typhon Namtheun (2004) sur l'océan. Namtheun a pris de la chaleur enthalpique locale de la mer (0.39–0.7 × 10⁸ J m⁻²), a refroidi l'eau de la surface de la mer par suite d'un mélange vertical (maximum 3°–4 °C) et a produit de fortes précipitations au-dessus de la mer (100–180 mm). Durant ce temps, l'importante quantité de chaleur latente (2.6–4.4 × 10⁸ J m⁻²) relâchée par les précipitations a apporté une plus grande contribution au bilan énergétique du typhon que le flux enthalpique air-mer local. Dans la couche supérieure de l'océan, les réponses océaniques peuvent être divisées en deux sous-processus, la surgyration rapide accompagnée d'un mélange vertical unidimensionnel et la dégyration lente accompagnée de la convergence d'eau de surface. Sur les profils Argo du 28 juillet, on peut voir que le mélange en surface produit par le typhon a brisé la thermocline saisonnière (approximativement 20 db) à l'intérieur d'une journée. De plus, la convergence moins profonde (<200 db) de l'eau douce à la surface de la mer provenant des précipitations a aussi rendu l'eau post-typhon plus douce (0.04 — en utilisant l’échelle de salinité pratique). Dans l'océan profond, la remonté d'eau rapide au sommet de la thermocline permanente suggère que la surgyration rapide est un mécanisme barotrope, probablement de pression gravitationnelle. Durant la phase de dégyration lente, le signal de remonté d'eau s'est propagé vers le bas (approximativement à 2 m h⁻¹) de la couche superficielle vers l'océan profond pendant environ 10 jours; c’était un processus barocline. Le mécanisme barocline était plus efficace à entretenir un remous cyclonique qu’à entretenir une onde d'inertie, et l'anomalie de basse hauteur de la surface de la mer ainsi que la remonté d'eau ont duré beaucoup plus longtemps que l'oscillation d'inertie (>20 jours contre environ 10 jours). Ce changement dans la structure verticale et la remontée d'eau à long terme pourrait avoir des répercussions sur l'environnement océanique et même sur le climat à court terme.

Keywords:

• Argo
• typhoon
• upwelling
• mixing
• spin-up
• spin-down
• air-sea interaction

1 Introduction

Typhoons (tropical cyclones in the northwest Pacific Ocean), when passing over the sea, remove large amounts of heat from the warm sea water. In turn, the strong winds caused by typhoons stir the sea water, transferring a vast amount of mechanical energy to the ocean. Vast energy interchanges within the atmosphere and ocean as a result of air-sea interactions have a great effect on both weather and ocean dynamics. The ocean's responses include sea surface cooling (Price, 1981), enhanced entrainment and vertical mixing (Han, Ma, & Chen, 2012; Jacob, Shay, Mariano, & Black, 2000; Lin, Chen, Pun, Liu, & Wu, 2009; Lin, Wu, Pun, & Ko, 2008), inertial oscillation (Black & Dickey, 2008; Morozov & Velarde, 2008; Price, 1981), and enhanced phytoplankton blooming (D'Asaro, 2003; Han et al., 2012; Subrahmanyam, Rao, Rao, Murty, & Sharp, 2002; Sun, Yang, Xian, Lu, & Fu, 2010; Wang & Zhao, 2008; Xian, Sun, Yang, & Fu, 2012; Yang, Fu, Sun, Liu, & Feng, 2010).

Recent investigations indicate that warm core eddies favour the enhancement of typhoons, because the eddies have a large upper ocean heat content (UOHC) (Lin et al., 2009; Wu, Lee, & Lin, 2007). It has been found that the relatively warmer upper ocean, characterized by a large D26 (depth of the 26°C isotherm), leads to a reduction in cyclone-induced ocean cooling (Lin et al., 2009). For a slow-moving Category 5 typhoon (maximum sustained winds higher than 61 m s⁻¹), the required pre-typhoon D26 and UOCH are typically 115–145 m and 1.2–1.3 × 10⁹ J m⁻², respectively (Lin et al., 2009). Although the temporal heat flux varies from 0.6 × 10³ W m⁻² (Jacob et al., 2000; Lin et al., 2008) to 2–3 × 10³ W m⁻² (Shay, Goni, & Black, 2000), the total enthalpy fluxes from the seawater to the air for typhoons are similar (0.7–1 × 10⁸ J m⁻²); this is less than the pre-typhoon UOHC.

The impacts of typhoons on the ocean are investigated using both observations (D'Asaro, 2003; Han et al., 2012; Price, 1981; Stramma, Cornillon, & Price, 1986; Sun et al., 2010; Wang & Zhao, 2008) and model simulations (Chang, 1985; Chen, Liu, Tang, & Wang, 2003; Emanuel, 1999; Liu, Wang, & Huang, 2008; Price, Sanford, & Forristall, 1994; Tsai, Chern, & Wang, 2008). From satellite imagery, it has been determined that the maximum surface cooling occurs before the passage of the eye of the typhoon (D'Asaro, 2003), or a few days after the passage of the eye (Stramma et al., 1986; Subrahmanyam et al., 2002).

It is generally recognized that the cooling of seawater by typhoons is primarily caused by upwelling or entrainment of deep cold water, in addition to the surface heat flux (Price, 1981). For example, although the maximum heat loss induced by Hurricane Opal (1995), a Category 4 storm, was 1.0 × 10⁹ J m⁻² with a heat flux of 2.0 × 10⁴ W m⁻² in 14 hours, the estimated surface heat flux was only about 2.0–3.0 × 10³ W m⁻² (Shay et al., 2000). Another possible major cooling mechanism is the horizontal advection of cold water (D'Asaro, 2003).

However, the responses of the ocean beneath the thermocline are seldom known because of a lack of observations. From buoy data (Price, 1981) and mooring data (Morozov & Velarde, 2008; Shay & Elsberry, 1987), it has been found that the responses to typhoons in the deep ocean are near-inertial oscillations. The mooring data suggest that the pressure jump beneath 1000 m depth caused by a typhoon occurs within hours of the typhoon's passage (Morozov & Velarde, 2008; Shay & Elsberry, 1987). Afterwards, it takes about 10 days for the signals generated by typhoon winds to propagate downward to the deep ocean (Morozov & Velarde, 2008; Price, 1981). The downward propagation speed (1–10 m h⁻¹) has an inverse relationship with the stratification (i.e., the weaker the stratification, the faster the propagation; Morozov & Velarde, 2008). The group velocity associated with the first and second baroclinic modes was 5.4 m h⁻¹ and 1 m h⁻¹, respectively (Shay & Elsberry, 1987). Although the buoy and mooring data have very fine spatial resolutions, the sampling layers are restricted to a few depths (120, 400, 1200 and 4700 m). Han et al. (2012) showed the drastic change in temperature, salinity and density in the entire water column at a shelf location after the passage of Hurricane Igor.

Vertical profiles from Argo floats 2900139 and 2900141 provide an opportunity to investigate the ocean's responses to typhoons. Float 2900139, which moved slowly from 28 July to 2 August 2004, provides two useful profiles for this study. Another float, 2900141, lingered around 142.680°E, 30.497°N and moved very little from 23 July to 12 August 2004 (Table 1) when a Category 4 typhoon (maximum sustained winds of 59–69 m s⁻¹), Namtheun, passed near the float (approximately 40 km away) at a speed of 4.5 m s⁻¹ (Fig. 1a). The average propagation speed of the typhoon was about 4.0 m s⁻¹ over its lifetime. The typhoon induced upper ocean responses (e.g., decreasing sea level and sea surface cooling over a large region). It also contributed to a large Kuroshio meander (Miyazawa, Kagimoto, Guo, & Sakuma, 2008; Sun, Yang, & Fu, 2009; Yang, Sun, Liu, Xian, & Fu, 2010).

Table 1. Dates, positions, MLT/ULT (average mixed layer/upper ocean temperatures within 90/200 m), MLS/ULS (average mixed layer/upper ocean salinities within 90/200 m), and SSHAs (sea surface height anomalies) for Argo floats 2900139 and 2900141.

Date (float)	Position	MLT/ULT (°C)	MLS/ULS	SSHA (mm)
07/28 (139)	141.103°E, 32.525°N	22.95/20.030	34.834/34.819	62.4
08/02 (139)	140.877°E, 32.720°N	22.67/20.054	34.754/34.777	40.2
07/23 (141)	142.680°E, 30.497°N	22.62/19.806	34.796/34.800	−35.2
07/28 (141)	142.589°E, 30.388°N	22.56/19.807	34.836/34.816	−90.5
08/02 (141)	142.676°E, 30.555°N	21.60/19.303	34.779/34.796	−108.8
08/07 (141)	142.670°E, 30.553°N	23.20/20.099	34.769/34.789	−103.0
08/12 (141)	142.674°E, 30.554°N	23.28/20.121	34.749/34.776	−95.6

Fig. 1 (a) The SSHA prior to Typhoon Namtheun's passage (between 17 July and 23 July). The track and the intensity of the typhoon are shown as open circles (maximum sustained wind speeds (MSW, m s⁻¹), central pressure (hPa)). The boxes (A, B and C) represent the regions of the anticyclonic eddy, the weak cyclonic eddy and the large cyclonic eddy, respectively. The positions of the Argo floats before the passage of Namtheun are indicated by black triangles. (b) The temperature (T) and the salinity (S) profiles of Argo float 2900141 on 23 July. The surface mixed layer, the seasonal thermocline, the mode water and the permanent thermocline are separated by dotted lines. (c) The potential density (σ, kg m⁻³) and the Brunt-Väisälä frequency (N ², s⁻²) vertical profiles. (d) The T-S diagrams.

Specifically, in this study, we investigate the physical responses of the ocean to Typhoon Namtheun using data from Argo float profiles and satellites. Because of the low temporal resolution of the observations, the inertial oscillations in the deep ocean cannot be detected directly in this investigation, though it is a very important response (Morozov & Velarde, 2008; Price, 1981; Shay & Elsberry, 1987; Wang, Zhou, & Qin, 1997). In the next section, we briefly describe the air-sea fluxes observed by satellites. Then two sub-processes in the upper ocean responses are investigated using heat and salt budgets. In Section 3, both the barotropic and baroclinic responses of the deep ocean are explored using the vertical profiles. The impacts of a typhoon can reach lower than the 200 m depth immediately after the typhoon's passage. Several related problems are discussed, including the assumption of entrainment free depth, the upwelling height in the deep ocean and the energy budget for the maintenance of typhoon intensity. Finally, a discussion and conclusions are presented.

2 Data and method

a Observation data

Typhoon track data, taken every 6 h, including the location of the centre, central pressure, and maximum sustained wind speeds (MSW), were obtained from the Shanghai Typhoon Institute (STI) of the China Meteorological Administration (CMA). Merged sea surface temperature (SST) data were derived from the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and the Advanced Microwave Scanning Radiometer-Earth observing system (AMSR-E) onboard the Aqua satellite. The spatial resolution of the SST data is 0.25° × 0.25°. The TMI/AMSR-E SST data are available at www.remss.com. In contrast to infrared radiation, microwaves can penetrate clouds with little attenuation, giving an uninterrupted view of the ocean surface accompanying a typhoon (Wang, Fu, Liu, Liu, & Sun, 2009; Wentz, Gentemann, Smith, & Chelton, 2000). The One-Degree Daily (1DD) air-sea enthalpy flux data are produced by the Objectively Analyzed air-sea Fluxes (OAFlux) project of the Woods Hole Oceanographic Institution (WHOI). This product integrates satellite observations with surface moorings, ship reports, and reanalyzed surface meteorological data based on an atmospheric model (Yu, Jin, & Weller, 2008; Yu & Weller, 2007). The 1DD precipitation data were obtained from the Global Precipitation Climatology Project (GPCP). The sea surface height anomaly (SSHA) data were produced and distributed by Archiving, Validation and Interpretation of Satellite Oceanographic data (Aviso). The Argo float profiles were extracted from the real-time quality-controlled Argo database of the China Argo Real-time Data Center. The Argo data were collected and made freely available by the International Argo Project and the national programs that contribute to it.

b Method

The UOHC is calculated using the temperature profile T(z) as follows:

where ρ and c_p are the seawater density (1.026 × 10³ kg m⁻³) and heat capacity (4.18 × 10³ J kg⁻¹), respectively. To investigate the oceanic responses to a typhoon, it is necessary to know the heat and salinity budgets. They are derived from the conservation of bulk heat Q and bulk salinity S:

where U is the horizontal velocity; Q_e is the air-sea heat flux; E_s and P_s are the evaporation from and precipitation over the sea surface, respectively; and Q_s and S_e are, respectively, the heat and salt fluxes resulting from entrainment. If the bulk water is deeper than the entrainment depth (e.g., 90 db below the seasonal thermocline in Fig. 1b), both Q_s and S_e can be assumed negligible. Moreover, the local air-sea fluxes can be calculated from local heat flux and precipitation data, and the changes in Q and S can be calculated from the vertical profiles. Thus, we can use Q, S, and the local air-sea fluxes to determine the heat and salt budget within the upper ocean. The assumption of the negligible Q_s and S_e is discussed in Section 4.

In addition, we need to calculate the typhoon-induced upwelling (or displacement of the thermocline). There are two empirical formulas to calculate the upwelling height h after a typhoon. The first uses atmospheric conditions alone (i.e., sea surface wind stress (τ) and translation speed (u)) (Price et al., 1994):

The second uses oceanic conditions alone:

where η, g and g′ are sea surface height, the acceleration due to gravity, and reduced gravity, respectively (Shay et al., 2000). In this study, the upwelling heights of the water were calculated according to the density profiles pre- and post-Namtheun (except for 2 August when some measurements below 200 db were lost), and values below 800 db were omitted because of the coarse vertical resolution. For example, the density profiles pre- and post-typhoon are ρ ₁(d) and ρ ₂(d), respectively, where d is the depth below the sea surface. If ρ ₁(d) = ρ ₂(d − h), then the water at depth d has an upwelling height h. This method is similar to, but more accurate than, Black and Dickey's (2008) which requires the derivative of the density profile.

3 Results

a Environment before Namtheun

The study area is depicted in Fig. 1a. Contour lines representing the SSHA indicate that there were many mesoscale eddies in this region. An anticyclonic eddy (marked A), a weak cyclonic eddy (marked B) and a strong cyclonic eddy (marked C) were located along the typhoon track. The pre-typhoon environment, especially the eddies, would interact strongly with the typhoon.

In Fig. 1a, the positions of the pre-Namtheun Argo floats are marked with black triangles. The vertical profiles measured by Argo float 2900141 (Fig. 1a, Table 1) are shown in Fig. 1b. The calculated potential density profile is depicted in Fig. 1c. The deeper the water, the colder and the denser it is. There was a shallow mixed layer near the surface, where both temperature and salinity were homogeneous because of vertical mixing. A seasonal thermocline (40–90 m depth in the temperature profile), a halocline (in the salinity profile) and a pycnocline (in the density profile) were present just below the mixed layer. The seasonal thermocline separated the mode water from the well-mixed surface water. The mode water, located between approximately 100 and 300 m depths, is homogeneous with very weak stratification (Masuzawa, 1969). Its formation may be a result of the “stability gap” (low potential vorticity water) in the ocean (Pan & Liu, 2005). A permanent thermocline existed below the mode water.

Typhoon Namtheun formed over the subtropical Pacific Ocean on 24 July 2004. On 26 July, after passing over anticyclonic eddy A, it intensified into a Category 4 typhoon with a high propagation speed (approximately 6 m s⁻¹). Between 28 July and 30 July (Fig. 1), Namtheun weakened to Category 2 over the cyclonic eddy in region B and lingered westward along 31.5°N at a slow speed (approximately 2 m s⁻¹). After that, Namtheun moved quickly (approximately 6 m s⁻¹) to the northwest along the pre-existing large cyclonic eddy C and finally made landfall on Shikoku Island, Japan, and weakened gradually. This typhoon brought heavy rain to Shikoku, setting a new Japanese daily precipitation record (1317 mm).

b Sea Surface Responses to Namtheun

The strong winds of Typhoon Namtheun removed heat from the sea (Fig. 2). During the passage of the typhoon, the temporal heat flux power was about 500–800 W m⁻², which is consistent with past investigations (Jacob et al., 2000). However, the average heat flux power was only about 100–180 W m⁻² from 28 July to 1 August (Fig. 2b), and the enthalpy flux (integrating from the heat flux) was about 0.39–0.7 × 10⁸ J m⁻² from the sea to the air. Comparing the enthalpy flux from 28 July to 2 August with that from 23 to 27 July, the typhoon induced an increase in average heat flux power of 60–90 W m⁻². The increase in heat flux had a rightward bias (i.e., the increase in heat flux was larger to the right of the typhoon track). After the passage of the typhoon the heat flux returned to its pre-typhoon value (Figs 2c and 2d). In addition, the typhoon also produced a large amount of precipitation along its track (Fig. 2). During the passage of the typhoon, the average precipitation was about 100–180 mm. The heavy precipitation produced a vast latent heat release of about 2.6–4.4 × 10⁸ J m⁻² to the air. The large latent heat release (about 6.5 times the local heat flux) provided most of the energy for the typhoon, as discussed below.

Fig. 2 The enthalpy fluxes (shaded, W m⁻²) and the precipitation (contour, mm d⁻¹) for the periods (a) 07/23–07/27, (b) 07/28–08/01, (c) 08/02–08/06 and (d) 08/07–08/11.

The warm (cold) core eddy has a deeper (shallower) D26 and a larger (smaller) UOHC. For a slow-moving typhoon, a deeper subsurface layer is needed (Lin et al., 2009). On the one hand, the pre-typhoon D26 was shallow (20–30 m) and the UOHC was about 1.6 × 10⁸ J m⁻² in the vicinity of the Argo floats, which was smaller than the required D26 depth (>45 m) and UOHC (6.4 × 10⁸ J m⁻²) for a Category 3 typhoon (Lin et al., 2009). Thus, the upper ocean could not have provided enough heat for the maintenance of the typhoon's intensity. On the other hand, the typhoon moved very slowly in this region, which would have required an even higher UOHC to produce a sufficient heat flux (Lin et al., 2009). In fact, the observed local enthalpy flux (0.39–0.7 × 10⁸ J m⁻²) was even smaller because the vertical mixing decreased the UOHC. What then is the energy source for the maintenance of the typhoon's intensity for such a long time? We conclude that the main energy source for the typhoon was the release of latent heat from the precipitation (2.6–4.4 × 10⁸ J m⁻²; Fig. 2b) from 28 July to 2 August. The two-day average heat flux power resulting from the latent heat release during the passage of the typhoon was estimated to be 1.5–2.6 × 10³ W m⁻².

Such a conclusion can also be drawn from the differences between region B and C (Fig. 1a). It is notable that the surface enthalpy flux (90–120 W m⁻²) was smaller in region B than in region C (150–180 W m⁻²; Fig. 2b), though the typhoon was more intense in region B than in region C.

The responses of the SST and the SSHA can be seen from satellite-derived observations (Fig. 3). Before the passage of typhoon Namtheun, the sea surface was very warm with SSTs >26°C in the study area (Fig. 3a). After the passage of the typhoon, the cyclonic eddies deepened and the SSHA decreased (Figs 3b to 3d). Also after Namtheun's departure, significant sea surface cooling appeared along the track, with SSTs dropping to 22–24°C (Figs 3b to 3d). There were two cooling centres, one located in region B where the typhoon moved slowly and another at the cyclonic eddy in region C. The changes in sea surface pressure and sea surface winds with time near float 2900141 are shown in Fig. 4a. The wind speed increased before the arrival of the typhoon, which is similar to the findings of a previous study (Shi & Wang, 2007). The SSHA decreased immediately after the arrival of the typhoon and remained that way for the rest of the study period (until 12 August; Fig. 4b). The time series of SSTs near the floats are also depicted in Fig. 4c, where the SST measured by TMI was consistent with the temperature observed at 5 db by the Argo floats. The greatest amount of sea surface cooling (3–4°C) observed by TMI occurred a few days after the passage of the typhoon. Although the Argo profiles missed this cooling on 30 and 31 July, the vertical profiles of the floats captured snapshots of the oceanic conditions before, during and after the passage of the typhoon. Thus, these profiles show the oceanic responses.

Fig. 3 The sea surface temperature (shaded, °C) and the sea surface height anomalies (contour, cm) for the periods (a) 07/24–07/26, (b) 07/27–07/28, (c) 08/01–08/03 and (d) 08/04–08/06.

Fig. 4 The time series of (a) the sea surface pressure (SSP, solid line) and the sea surface wind (SSW, dashed line) of the typhoon, and the sea surface wind (SSW, squares) at Argo float 2900141, (b) the sea surface height anomalies at Argo float 2900141, and (c) the sea surface temperature measured by satellite and the near-surface temperature measured by Argo floats.

c Upper Ocean Responses to Namtheun

The responses in the upper ocean (within 200 db) can be separated into two sub-processes. The first is the fast spin-up of the upper ocean stirring by typhoon winds, which induces or enhances a cyclonic eddy. During this stage, one-dimensional vertical mixing and entrainment processes exist where the surface water mixes locally with the deeper cold water. Thus, the total vertical integral of latent heat is changed very little during these processes. In addition, stratification breaks down in this sub-process. The second upper ocean response is a slow spin-down of surface water after the passage of the typhoon. During this stage, sustentative surface waters from the ambient converge to the cyclone eddy. Thus, the changes in the latent heat and salinity cannot be balanced by local fluxes.

The first sub-process began at the time of the passage of the typhoon on 28 July (Figs 5a and 5b). The warm surface water was entrained below the seasonal thermocline in the vicinity of float 2900141, with the surface temperature decreasing by 1°C and the average temperature within the upper 30 db increasing by 1.5°C. Thus, the stratification was unstable between 6 and 20 db (Fig. 5c). It was found that the cooling in the mixed layer (1°C in the upper 10 db) was caused primarily by the entrainment of deep colder water at the surface. This surface heat loss compensated for the heat gain in the subsurface, so the average temperature in the upper 200 db changed very little (0.001°C from 23 July to 28 July; Table 1). This is similar to the results of a previous study (e.g., about 85% of the surface cooling being caused by entrainment; Price, 1981). Later, vertical mixing occurred in the layer with the strongest stratification (Fig. 5d) as a result of instability, which agrees well with previous observations in which a strong current shear induced mixing events through shear instabilities at the top of the thermocline (Shay et al., 2000). The temperature and salinity were homogeneous within the waters above 40 db because of wind-induced vertical mixing. As a consequence, the seasonal thermocline was enhanced and shoaled slightly (approximately 5 m; Fig. 5c) because of the superposed warm water (Fig. 5d).

Fig. 5 The Argo (a) temperature profiles (the SSTs detected by TMI and Argo float), (b) salinity profiles, (c) density profiles, and (d) the Brunt-Väisälä frequency profiles in the upper 100 db from 23 July to 7 August.

This local process continued for a few days, and SSTs cooled by up to approximately 4°C in the 2–3 days after the passage of the typhoon (Sun et al., 2010; Yang, Fu, et al., 2010). However, this cooling can only be seen from SSTs measured by TMI, because there were no measurements from the Argo float during that time period (Fig. 4c). The mixed layer cooling at float 2900139 was also a result of vertical mixing. Because of this, the total heat content changed very little (an increase of 0.2 × 10⁸ J m⁻²; Table 2) within the entire upper ocean (200 db), although there was a heat loss within the mixed layer (upper 40 db) of 4.21 × 10⁸ J m⁻² (Table 2).

Table 2. The 5-day (e.g., ‘07/23–07/28’ represents ‘from 23 July to 28 July’) anomalies of MLH/ULH (mixed layer latent heat/upper ocean latent heat), MLF/ULF (mixed layer freshwater flux/upper ocean freshwater flux), EF/LR (enthalpy flux by OAFlux /latent heat release by precipitation) and P (precipitation according to GPCP) for Argo floats 2900139 and 2900141.

Time interval (float)	MLH/ULH(10^8 J/m^2)	MLF/ULF (mm)	EF/LR(10^8 J/m^2)	P (mm)
07/28–08/02 (139)	−1.04/0.2	207/240	0.45/3.0	122
07/23–07/28 (141)	−0.23/0.08	−102/−92.6^a	0.38/0.59	23.9
07/28–08/02 (141)	−3.62/−4.21	146.4/112.8	0.37/2.24	91.4
08/02–08/07 (141)	6.02/6.65	25.9/43.7	0.30/0.32	13.2
08/07–08/12 (141)	0.30/0.18	52.6/69.9	0.38/0.005	0.2
07/28–08/12 (141)	2.70/2.62	225/226	1.05/2.56	105
^aThe mixed layer and the upper ocean heat fluxes due to estimated evaporation from 23 July to 28 July derived from salinity anomalies and precipitation are −3.09 × 10^8 J m^−2 and −2.85 × 10^8 J m^−2, respectively.

In the first sub-process, strong winds blew the surface salt water away from the centre of the typhoon. Consequently, the subsurface salt water accumulated and converged in front of the typhoon track. The salinity in the mixed layer increased in the vicinity of float 2900141 before the passage of the typhoon on 28 July (from 34.796 to 34.836; Table 1). This can be seen from the heat and salt budgets. The local heat loss of 0.34 × 10⁸ J m⁻² (with an average flux of 80 W m⁻²; Fig. 2) from evaporation can only explain about 10% of the increased salinity in the mixed layer. Thus, the salinity increase should result more from the subsurface higher salinity mode water convergence than from local evaporation.

In the second sub-process, the sustaining surface fresh waters from the ambient converge on the cyclonic eddy without the support of the strong cyclonic winds. For example, the upper ocean water near float 2900139 became fresh (34.777; Table 1). Local precipitation of 122 ± 6 mm cannot account for the net freshwater increase of 240 ± 6 mm within the upper ocean (Table 2). In the region of float 2900141, the upper ocean seawater freshened by 0.04 from 28 July to 12 August (Table 1). Local precipitation of 91.4 ± 6 mm was notably smaller than the freshwater flux (net increase of 225 ± 6 mm accounting for 0.04) derived from the salt budget (Table 2). It is estimated from Table 1 that the net increase in fresh water within the entire mixed layer was about 146 ± 6 mm from 28 July to 2 August and 78.5 ± 6 mm from 2 August to 12 August (Table 2), which is larger than the observed precipitation (approximately 105 ± 6 mm from 28 July to 12 August) according to GPCP (Table 2).

At a later stage, the accumulated fresh surface water gradually converges from the mixed layer and spreads to the entire upper ocean. On 2 August, the 146 ± 6 mm freshwater flux was confined mainly within the upper 40 m. The freshwater reduction in the surface layer was compensated for by the freshwater increase in the subsurface layer, for the fresh water entrained to within 200 m depth. This can be seen in Table 2. From 28 July to 12 August, the total freshwater flux (approximately 226 ± 6 mm) and latent heat flux (2.62 × 10⁸ J m⁻²) in the upper ocean within 200 db agreed well with the fluxes entering the mixed layer in the upper 90 db (approximately 225 ± 6 mm and 2.70 × 10⁸ J m⁻²). The total latent heat loss of about 4 × 10⁸ J m⁻² (Table 2) from 28 July to 2 August, with a maximum cooling of about 2°–3°C at approximately 10 db (Fig. 5a), was caused more by the convergence of cold surface water than by the local enthalpy flux (0.1–0.2 × 10⁸ J m⁻² in Fig. 2a). Such convergence also brings warm water into the upper ocean. The latent heat increased by approximately 6.3 × 10⁸ J m⁻² with an average heat flux of 1.4 × 10³ W m⁻² from 2 August to 7 August (Table 2), which was also much larger than the net shortwave heat input. Because the vertical mixing process made the water above 80 db more homogeneous, the stratification was re-established at 80 db (dotted curve in Figs 5c and 5d). Although at the location of the Argo float there was a warm water sink, the depth of the sinking was less than 200 db. Thus, this result implies that such fresh water could not be transported below the permanent thermocline. The latent heat budget also suggests this.

Finally, the trajectory of float 2900141 also implies that it was located at the centre of the convergence. The float was always brought back to its original position by the converging water, so it moved only about 10 km in 20 days. The convergence of the sea water might be a result of a decrease in the sea surface height (Table 1). It seems that precipitation at the sea surface was entrained into the mixed layer by local convergence. However, this convergence occurred in a shallow layer above the mode water (200 db).

d Deep Ocean Responses to Namtheun

During the passage of the typhoon, not only did the entrainment and mixing take place above the seasonal thermocline but the water below the mode water also upwelled (Fig. 6a). As the winds blew the surface water away, the mode water was lifted up. Within the mode water, strong entrainment moved the warm and fresh water from the upper ocean to the cold and salty mode water, and the mixing of the waters broke down the internal surface (between the upper and lower layers) of the mode water (Figs 6a and 6b). Thus, the weakest stratification (Brunt-Väsälä frequency of 10⁻⁶) above 250 db became notably stronger (dashed curve in Fig. 6c), which can also be seen in the T-S profiles of the Argo float (Fig. 6b), where the T-S structure above 250 db significantly changed on 28 July (dashed curve in Fig. 6b). As a consequence, the upper mode water above 200 db became lighter (squares in Fig. 6c). However, this strong entrainment lasted only a few days; after 7 August the mode water returned to 200 db. The density variation implies that there was less upwelling above the mode water, thus the mixed layer cooling (from 28 July to 2 August) cannot be a result of the upwelling beneath the permanent thermocline.

Fig. 6 (a) Temperature (T) and the salinity (S) profiles below 100 db with the arrows indicating the upwelling direction, and (b) the T-S profiles within the mode water on 28 July, 7 August and 12 August. (c) The density variation on 28 July (dσ, squares) and the Brunt-Väisälä frequency (N ², dashed curve) profiles, and (d) the upwelling height profiles of Argo floats on 28 July, 7 August and 12 August.

Meanwhile, the ocean below the mode water experienced significant upwelling, as is shown by the upward shift of the T-S profiles (marked by the upward arrows) after the passage of the typhoon (Fig. 6a). The temperature and salinity decreased by about 0.5°C to 1°C and 0.02–0.06, respectively, within the permanent thermocline. The rapid adjustment at the top of the permanent thermocline suggests a barotropic mechanism for the deep ocean to respond to the perturbations in the upper ocean, probably via gravity pressure (Chang, 1985). Consequently, the upwelling of the deep seawater made the in situ water (below 200 db) more dense after the passage of the typhoon (squares in Fig. 6c), which is quite different from the result produced by vertical mixing. The upwelling had three main features based on the upwelling height profiles (Fig. 6d). First, the upwelling occurred very quickly during the passage of the typhoon on 28 July, and the height of upwelling was inversely proportional to depth (i.e., the shallower the depth, the stronger the upwelling; solid curve in Fig. 6d). Second, the upwelling also occurred in the deep sea. There was a notable upwelling even at 800 db, which has not previously been reported. For example, the maximum upwelling depth caused by Typhoon Lingling (2001) was observed to be only 300 db (Shang et al., 2008), and the maximum upwelling depth was less than 200 db in the vicinity of Bermuda (Black & Dickey, 2008). Third, the upwelling of the thermocline had an obvious establishment process.

In Fig. 7a, the upwelling signal was located at 300 db on 28 July, then propagated downward to 700 db on 7 August. The downward-propagating upwelling signal had a speed of about 40 m d⁻¹ (approximately 2 m h⁻¹), which is similar to that of the downward inertial oscillation in previous studies (Morozov & Velarde, 2008; Shay & Elsberry, 1987). Although the low temporal resolution of the observations (approximately 5 days) cannot resolve the inertial oscillation (approximately 1 day), the downward-propagating upwelling signal should be associated with it. After 10 days, the inertial oscillation vanished, but the displacement of the thermocline (upwelling) still existed, and maintained a cyclonic eddy (see the SSHA decrease in Fig. 4b or Fig. 3 of Yang, Sun, et al. (2010)). The upwelling (Figs 6a and 6d) associated with low SSHAs (Fig. 4b) and low temperatures (Fig. 4c) lasted for more than 20 days (Fig. 7a), which was longer than the duration of the inertial oscillations observed in past studies (about 7 days in Shay & Elsberry (1987) and 10 days in Morozov & Velarde (2008)).

Fig. 7 (a) The upwelling height derived from density profiles (solid curve) and by Shay's formula (dashed) on 28 July, 7 August and 12 August. (b) The scheme of the ocean's responses includes typhoon passage processes and post-typhoon processes. The solid arrows indicate the convergent upwelling flow, and the dashed arrows indicate the downward propagating inertial signals.

It is interesting to compare the upwelling height predicted by empirical formulas to observations. Because Price's formula, Eq. (3), depends only on atmospheric conditions, it is useful when oceanic measurements are scarce. However, the absence of oceanic conditions makes the formula only a zero-order approximation. Moreover, Price's formula is approximately equivalent to the product of the potential Ekman pumping velocity and translation time and overestimates the real upwelling strength (see Eq. (5) in Sun et al. (2010)). In comparison, Shay's formula, Eq. (4), which does account for oceanic conditions, seems to be a better first-order approximation and can resolve the spatial variation of upwelling. The upwelling profiles obtained from Argo data and from Shay's formula (η = 61 mm calculated from SSHAs in Table 1, g′ = 0.02 m s⁻², thus h = 30 m) are depicted in Fig. 7a. It is clear that Shay's formula agrees quite well with the Argo data within the entire permanent thermocline layer after the upwelling signal has propagated across the entire permanent thermocline. However, the formula's agreement with observations is not as good at the top of the permanent thermocline. In fact, the deepening of the SSH is accompanied by both upwelling of the thermocline and downwelling of upper water (including mode water and seasonal thermocline).

The above results show the ocean's responses to the typhoon. These responses can be illustrated by a conceptual model (Fig. 7b). During the passage of the typhoon precipitation, evaporation and heat flux occur, which change the total salt and total latent heat of the entire water column. In the upper ocean, strong winds stir the water and weaken the seasonal thermocline through vertical mixing. The responses within the thermocline are barotropic. Upwelling is well established at the top of the permanent thermocline within several hours. However, the response in the permanent thermocline is a downward baroclinic process. The perturbation propagates from the upper ocean (as a downward inertial oscillation signal) with a velocity of 2 m s⁻¹ and lasts about 10 days. The upwelling is not vertically uniform. It should be pointed out that the upwelling still existed (the low SSHA in Fig. 4b) after the downward inertial oscillation vanished and, thus, may maintain a long-term eddy (Sun et al., 2010; Yang, Sun, et al., 2010) more effectively than it maintains inertial waves. The eddy may have a longer and more important impact on the ocean environment (e.g., long-term cooling, chlorophyll-a concentration enhancement and production of a strong advection current (Sun et al., 2010; Yang, Sun, et al., 2010)).

4 Discussion

In this investigation, Q_s and S_e resulting from entrainment were assumed to vanish at certain depths (i.e., 90 db and 200 db). The choices of such depths are based on vertical profiles (Figs 5 and 6). It can be seen from Fig. 5d that notable vertical mixing and/or horizontal convection occurred above 90 db, so that the stratification changed significantly within this depth. In addition, the entrainment should be very small beneath 90 db. The choice of 200 db as the upper ocean is based on the same reasoning and on the T-S diagrams (Fig. 1d, Fig. 6a and Fig. 6b), where the upper mode water (separating the mixed layer from the thermocline) changed very little during the passage of the typhoon. This assumption that the entrainment depth was 200 db could be acceptable in this study because the water properties seldom changed at this depth during the entire typhoon-ocean interaction process (Fig. 6b) and because the net fluxes between the upper ocean and the deeper ocean as a result of entrainment can be ignored.

The intensity of Typhoon Namtheun remained unchanged from 28 July to 30 July (Fig. 1a), and the maintenance of the typhoon's intensity in this case is very useful for understanding the energy balance in the air-sea interactions. Our calculations estimated that the typhoon obtained a vast amount of energy (3–5 × 10⁸ J m⁻²) during these two days; this thermal energy must be transported with the typhoon to maintain the typhoon's intensity. It is obvious that this thermal energy is eventually translated into the typhoon's mechanical energy because typhoons are known to be a thermal engine, and this mechanical energy in turn is transported to the ocean via wind stress. Given the amount of energy obtained by the typhoon and subsequently released to the ocean, and also given the size of the region affected by the typhoon (400 km by 500 km, Fig. 2a), there is about 10²⁰ J of total mechanical energy transferred to the ocean by this single typhoon over the two days. This input energy was used to supply the surface waves, currents and vertical mixing in the upper ocean, as well as the inertial oscillations (Liu et al., 2008).

In this case, the latent heat release resulting from the precipitation was the main energy source for the typhoon, which compensated for the energy loss due to friction and resistance from 28 July to 30 July. Because the local air-sea heat flux was the main focus of previous studies (Lin et al., 2008, 2009; Shay et al., 2000), it is worth pointing out that precipitation, accompanied by local air-sea heat flux, plays an essential role in the energy budget for the maintenance of the typhoon intensity, especially in regions of low UOHC.

All of these physical responses in the ocean can have a great impact on phytoplankton blooms (D'Asaro, 2003; Sun et al., 2010; Yang, Fu, et al., 2010), especially when the translational speed of the typhoon is relatively slow (Sun et al., 2010). The biophysical responses to this typhoon were investigated by Yang, Sun, et al. (2010) who concluded that typhoon-induced entrainment, vertical mixing and upwelling resulted in a significant enhancement of chlorophyll-a concentration.

5 Conclusions

In this study, precipitation data, accompanied by heat flux data, provided comprehensive insights into the energy budget of a typhoon. Typhoon Namtheun occurred in July 2004, received heat from the sea, produced heavy precipitation over the sea and cooled the sea surface. During this time, the latent heat release (2.6–4.4 × 10⁸ J m⁻²) from the precipitation was more important than the local air-sea enthalpy flux (0.39–0.7 × 10⁸ J m⁻²) in supplying the typhoon's energy budget. The largest sea surface cooling (3–4°C), located in region B (the typhoon moved slowly) and in region C (the location of a cyclonic eddy), occurred about 2–3 days after the passage of the typhoon.

In the upper ocean (within 200 db), the oceanic responses can be separated into two sub-processes, the fast spin-up accompanied by one-dimensional vertical mixing and the slow spin-down accompanied by convergence of the surface water. From the Argo profile on 28 July during the passage of the typhoon it can be seen that the typhoon-induced surface mixing broke down the seasonal thermocline (approximately 20 db) immediately, and the sea surface cooling (0.5–1°C) at that time was primarily caused by vertical mixing, thus the total latent heat in the upper ocean (within 200 db) changed very little (Table 2). In addition, the convergence of sea surface fresh water from precipitation also made the post-typhoon water fresher than before. However, this convergence occurs in the layer above the mode water.

Meanwhile, the water beneath the permanent thermocline upwelled in a vertically non-uniform manner by about 70 m on 28 July. Then, the upwelling signals within the permanent thermocline propagated downward as an inertial oscillation, until the upwelling heights were uniform 10 days later. It shows that Shay's formula (h = gη/g′) is a good approximation for upwelling within the entire permanent thermocline. The rapid response at the top of the permanent thermocline suggests a barotropic mechanism, probably via gravity pressure. On the other hand, the propagation of the upwelling signal within the permanent thermocline clearly shows a baroclinic propagation mechanism. The mechanism was more effective in maintaining a cyclonic eddy than in maintaining an inertial wave, thus the upwelling and low SSHA (>20 days) lasted much longer than the inertial oscillation (approximately 10 days). This change in vertical structure and long-term upwelling could have some impacts on the ocean environment and even on the short-term climate.

It should be noted that the present temporal resolution of the Argo profiles are not sufficient for typhoon investigations. For example, the lack of sufficient vertical profiles leaves the details of the largest cooling process, from 30 July to 31 July, unknown.

Nevertheless, because of the high vertical resolution of the Argo profiles, they provide some new insights into the response of the ocean to tropical cyclones. For example, in this study they illuminated the heat and salt balances in the upper ocean and captured the vertical structure of upwelling heights in the deep ocean. Further studies are necessary to determine how much energy is needed to maintain a typhoon of a specific category and how much mechanical energy induced by typhoons is injected into the deep ocean below the thermocline.

Acknowledgements

We thank the anonymous reviewers for their useful comments and suggestions. This work is supported by the National Basic Research Program of China (Nos. 2007CB816004, 2012CB417402 and 2013CB430303), the Knowledge Innovation Program of the Chinese Academy of Sciences (No. KZCX2-YW-QN514), and the Open Fund of the State Key Laboratory of Satellite Ocean Environment Dynamics (No. SOED1209). We also thank the China Argo Real-time Data Center for float profiles, STI for providing typhoon track data, Aviso for SSHA data, NASA/Goddard Space Flight Center's Laboratory for the 1DD GPCP data, Remote Sensing Systems for TMI/AMSR-E SST data, and the WHOI for supplying the air-sea heat flux data.