A suite of shipboard and satellite observations are analyzed and synthesized to investigate the three-dimensional structure of clouds and influences from sea surface temperature fronts over the western North Pacific. Sharp transitions are observed across the Kuroshio Extension (KE) front in the marine atmospheric boundary layer (MABL) and its clouds. The ocean’s influence appears to extend beyond the MABL, with higher cloud tops in altitude along the KE front than the surroundings.
In winter, intense turbulent heat release from the ocean takes place on the southern flank of the KE front, where the cloud top penetrates above the MABL and reaches the midtroposphere. In this band of high cloud tops, frequent lightning activity is observed. The results of this study suggest a sea level pressure mechanism for which the temperature gradient in the MABL induces strong surface wind convergence on the southern flank of the KE front, deepening the clouds there.
In early summer, sea fog frequently occurs on the northern flank of the subtropical KE and subarctic fronts under southerly warm advection that suppresses surface heat flux and stabilizes the surface atmosphere. Sea fog is infrequently observed over the KE front even under southerly conditions, as the warm ocean current weakens atmospheric stratification and promotes vertical mixing. The KE front produces a narrow band of surface wind convergence, helping support a broad band of upward motion at 700 hPa that is associated with the eastward extension of the baiu rainband from Japan in June–July.
The western North Pacific east of Japan displays rich structures in sea surface temperature (SST) at the confluence of several major ocean currents. The Kuroshio, a western boundary current of the North Pacific subtropical gyre, advects warm/saline water from the tropics, flowing northeastward along the continental slope of the East China Sea (ECS) and south coast of Japan. It separates from the Japanese coast around 35°N, forming an inertial current called the Kuroshio Extension (KE) that appears as a sharp SST front with meanders east of Honshu Island of Japan (Mizuno and White 1983). To the north, the Oyashio, a western boundary current of the North Pacific subarctic gyre, transports cold/freshwater from the subarctic region along the south coast of the Kuril Islands and Hokkaido of Japan. After leaving the east coast of Japan around 39°N, this cold current creates a robust northeastward-elongated SST front (Yuan and Talley 1996), which we refer to as “subarctic front.” Both KE and subarctic fronts are accompanied by strong northward-decreasing SST gradient steeper than 3°C per 100 km in the climatological mean (Fig. 1), forming the interfrontal zone (Yasuda et al. 1996) in the Kuroshio–Oyashio Extension (KOE) region.
The western North Pacific is also characterized by vigorous exchange of heat between the ocean and atmosphere (Qiu et al. 2004; Kubota et al. 2008; Bond and Cronin 2008). In winter the ocean loses a huge amount of heat by strong northwesterly winds with a cold/dry air mass, while in summer, surface latent and sensible heat fluxes are significantly suppressed under warm/humid advection by the prevailing southerlies. On the basin scale, wind-induced surface heat flux variations are a major mechanism for SST anomalies (Xie 2004, and references therein). Such an atmospheric forcing generally leads to a negative correlation between SST and wind speed, for example, as observed in association with the Pacific decadal oscillation (Mantua et al. 1997). Near the KE and subarctic fronts, on the other hand, heat transport by the geostrophic current and meridional migration of the fronts mainly control decadal SST anomalies (Qiu 2000; Xie et al. 2000; Schneider et al. 2002; Mochizuki and Kida 2006; Nonaka et al. 2006). Indeed, such ocean dynamical-induced SST anomalies play a primary role in determining mid- and late-winter turbulent heat flux over the KOE region on decadal time scales (Tanimoto et al. 2003). These studies suggest a two-way interaction between the ocean and atmosphere in the western North Pacific.
Recent high-resolution satellite measurements provide new insights into extratropical ocean–atmosphere interaction. For example, the Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) indicates that surface wind speed increases over the KE’s warm meanders while decreasing over detached cold eddies in spring (Nonaka and Xie 2003). This in-phase relationship between SST and wind is opposite to the conventional view that strong wind speed cools the ocean surface, indicative of an ocean-to-atmospheric influence. Similar covariability of SST and surface wind is observed in eastern Pacific tropical instability waves (TIWs), which cause the SST front to meander (Hayes et al. 1989; Xie et al. 1998; Liu et al. 2000; Chelton et al. 2001; Hashizume et al. 2001). Wallace et al. (1989) proposed a vertical mixing mechanism for such an in-phase SST–wind relationship: On the warmer flank of the front, the surface atmosphere is unstable and the intensified vertical mixing transfers large momentum from above to accelerate the surface wind. Direct atmospheric soundings and satellite observations provide some support for this vertical mixing mechanism over TIWs (Hashizume et al. 2002), the KE region (Tokinaga et al. 2006), the Gulf Stream (Sweet et al. 1981), the Brazil–Malvinas Confluence (Pezzi et al. 2005), and the Agulhas Return Current (O’Neill et al. 2005).
The sharp SST front also drives an anomalous surface wind through its effect on the sea level pressure (SLP) gradient (Lindzen and Nigam 1987). For example, numerical modeling results for the TIW SST front suggest that SLP anomalies develop downwind and drive anomalous winds in phase with the SST (Small et al. 2003), a mechanism consistent with limited Tropical Atmosphere Ocean (TAO) buoy observations (Cronin et al. 2003). Meanwhile, a recent momentum budget analysis with high-resolution atmospheric general circulation model (AGCM) experiments indicates that the effects of the SLP gradient and vertical mixing are comparable over the winter KE region (Koseki and Watanabe 2009, manuscript submitted to J. Climate). While such a SLP adjustment mechanism is supported mostly by numerical models or model-assimilated analyses, more in situ observational evidences should be pursued to test the mechanism.
Clouds may also respond to robust SST fronts (Small et al. 2008). During spring cold-air outbreaks behind synoptic storms, mesoscale cloud bands develop from the Gulf Stream north wall and trail downstream (Young and Sikora 2003). Liu et al. (2007) and Minobe et al. (2008) suggest that the cloud response to surface wind convergence can extend well into the free troposphere near sharp SST fronts maintained by the Agulhas Extension current and Gulf Stream, respectively. For the western North Pacific, a cloud band extends northeastward along the warmer flank of the Kuroshio Front in ECS during winter (Xie et al. 2002), supporting the findings by earlier studies during the Air Mass Transformation Experiment (AMTEX) 1974–75 (Ninomiya 1975; Ninomiya and Akiyama 1976; Nitta 1976). Moreover, Kobashi et al. (2008) detect a deep convective response to the subtropical SST front in spring. However, it remains unclear whether a similar cloud response exists over the KE region and how it varies seasonally.
For the summer season in the western North Pacific, previous studies have showed an interaction between SST and low cloudiness characterized by negative correlation: a SST decrease increases low cloudiness owing to strengthened atmospheric stratification, reducing solar radiation at the sea surface and amplifying the initial cooling (Klein and Hartmann 1993; Norris et al. 1998; Bond and Cronin 2008). This interaction is interpreted as a positive feedback between SST and low cloudiness. Atmospheric thermal advection and synoptic vertical motion also play an important role in determining the cloud type over the western North Pacific (Norris 1998a,b; Norris and Iacobellis 2005). Northerly induced cold advection up the SST gradient and synoptic descent behind an eastward-moving midlatitude cyclone tend to produce cumulus and stratocumulus by developing an inversion-capped marine atmospheric boundary layer (MABL). Southerly induced warm/moist advection over the northward-decreasing SST, on the other hand, increases the stratification of the surface layer, forming sea fog in summer. Unfortunately, coarse spatial resolution (>2° × 2° grid) of ship-based cloud and other atmospheric datasets used in these studies was too coarse to detect imprints of narrow SST fronts on the atmosphere.
Tanimoto et al. (2009) use high-resolution surface meteorological observations and atmospheric soundings during the Kuroshio Extension System Study (KESS) (Donohue et al. 2008) cruise in 2005 and show a sharp cross-frontal transition of low clouds. Their cloud-base height measurements by a laser ceilometer and atmospheric soundings successfully capture that warm southerly (cold northerly) advection across the KE front generates sea fog (stratocumulus with elevated cloud base) on the colder (warmer) flank of the front. Thus, the presence of an ocean front plays an important role in cloud regime transition as winds shift direction across the front.
The present study investigates SST frontal effects on clouds in the western North Pacific by synthesizing a number of different in situ and satellite observations in the region. Specifically we take advantage of high-resolution (one minute) surface meteorological observations on board research vessels. Some of these research cruises are equipped with laser ceilometers and conduct GPS soundings to probe the vertical structure of the atmosphere (Tokinaga et al. 2006; Tanimoto et al. 2009). To improve the spatiotemporal coverage, we use historical ship reports in the International Comprehensive Ocean–Atmosphere Data Set (ICOADS) (Worley et al. 2005). The midlatitude western North Pacific includes many busy shipping lanes, and the coverage and resolution of ICOADS resolve SST fronts reasonably well. For example, the data coverage for ICOADS monthly SST on a 1° grid for 1950–2006 exceeds 95% from the south coast of Japan to the KE and subarctic frontal regions (Fig. 2a). In such data-rich regions, surface air temperature (SAT) and SLP data in ICOADS, together with high-resolution satellite observations, will be analyzed in detail to study atmospheric adjustments to narrow SST fronts. Our focus is on the three-dimensional development of clouds of various types—ranging from sea fog to deep convection—and their seasonal variations. To extract the KE frontal effect on the atmosphere, we adopt a composite analysis referenced to the major SST front. We will show that cross-frontal advection in the MABL is important for cloud development and variations.
a. In situ observations
The R/V Mirai of the Japan Agency for Marine-Earth Science and Technology (JAMSTEC) provides continuous, along-track measurements of SST, SAT, surface wind, cloud-base height (by a ceilometer), and ocean current velocity [by an acoustic Doppler current profiler (ADCP)]. On the way to the western tropical Pacific and eastern tropical Indian Ocean, and back to Japan, R/V Mirai traverses the KE region (Fig. 1). From those cruises that crossed the KE front, we choose seven (six) ship tracks for winter (summer) analyses. The selected winter cruises are MR99–01, MR00–01, MR00–08, MR01–01, MR02–02, leg 3 of MR06–05, and MR07–01 (Fig. 1a). The summer cruises are MR00–04, MR01–02, MR02–04, legs 1 and 2 of MR04–03, and MR07–03 (Fig. 1b). We first define a local maximum in the meridional SST gradient south of 38°N as the axis of the KE front using the shipboard SST data from each cruise and then composite a cross-frontal section for each meteorological parameter based on the latitudinal distance from the KE front (Figs. 3 and 10). Thus, the 0° tick mark on the y axis of the cross section corresponds to the maximum gradient of SST. This composite method is also applied to cross-frontal sections of satellite observations (Figs. 5a, 7a,c,d, 9, and 14a). We use only cross sections that cut through the SST front with its southward gradient greater than 1.5°C per degree latitude between 143° and 155°E. To calculate surface latent and sensible heat fluxes, the Coupled Ocean–Atmosphere Response Experiment (COARE) 3.0 flux algorithm (Fairall et al. 2003) is used.
To examine the vertical structure of MABL, we analyze a total of 561 radiosonde soundings over the western North Pacific (30°–40°N, 141°–155°E) from a series of shipboard observational campaigns from 2003 to 2007, which is a unique dataset in the present study (Figs. 1b and 1d). The soundings were obtained on board R/Vs Kofu-maru and Ryofu-maru of the Japan Meteorological Agency (JMA), R/Vs Roger Revelle and Melville of the Scripps Institution of Oceanography, R/V Kaiyo-maru of the Japan Fishery Agency (JFA), and R/Vs Mirai and Hakuho-maru of the Ocean Research Institute (ORI) of the University of Tokyo/JAMSTEC. All variables in the soundings are linearly interpolated to vertical levels at 25-m intervals.
We also use the ICOADS release 2.4, which contains surface meteorological measurements and weather information from ships (merchant, navy, and research), moored and drifting buoys, coastal stations, and other marine platforms. Because commonly used gridded ICOADS products show some unrealistic features (Minobe and Maeda 2005), we first remove such suspicious data from the International Maritime Meteorological Archive (IMMA) format of ICOADS by the same subjective quality control as Minobe and Maeda (2005). Next, we remove data that depart from the climatological monthly mean by more than 2.5 standard deviations at each 1° × 1° grid box and month and construct a monthly dataset on a 1° × 1° grid. Finally, we apply a weighted average using values at the grid point and eight surrounding points. Despite this moderate spatial smoothing, the resultant dataset yields KE and subarctic SST fronts much stronger than the commonly used gridded ICOADS (Fig. 2b). We construct the monthly mean climatology of SST, SAT, surface wind, and SLP by averaging data from 1950 to 2006. In addition to the meteorological parameters, weather information based on visual observations has been archived in the IMMA format in accordance with the World Meteorological Organization (WMO) code 1860 and 4677 for the period from 1982 to 2006. We use the weather codes to detect summer sea fog.
b. Satellite observations
We analyze a suite of satellite observations of SST, sea surface wind, total cloud liquid water, cloud-top pressure, and lightning flash frequency as well as air temperature and humidity soundings by several different sensors on different platforms. The Advanced Microwave Scanning Radiometer for Earth Observing System [EOS (AMSR-E)] aboard the NASA Aqua spacecraft, launched in May 2002, measures a number of important geophysical parameters including SST and total cloud liquid water (Wentz and Meissner 2000). We use the monthly product for SST and total cloud liquid water available from Remote Sensing Systems (RSS) for September 2002 to April 2007 on a 0.25° × 0.25° grid. For SST data prior to September 2002, the global, high-resolution SST product that uses the Advanced Very High Resolution Radiometer (AVHRR) Pathfinder SST (Vazquez 2005) is available from the NASA/Jet Propulsion Laboratory Physical Oceanography Distributed Active Archive Center (JPL PO DAAC). We use a daily mean product on a 0.25° × 0.25° grid from January 1999 to December 2004 for comparison with the cloud-top pressure from the International Satellite Cloud Climatology Project (ISCCP) described in this section.
The Aqua spacecraft also carries the Atmospheric Infrared Sounder (AIRS) and its two companion microwave instruments, the Advanced Microwave Sounding Unit (AMSU) and the Humidity Sounder for Brazil (HSB), measuring atmospheric temperature and humidity under both clear and cloudy conditions (Aumann et al. 2003). The AIRS/AMSU/HSB observations provide a three-dimensional view of the troposphere over the global ocean in detail never possible before. Although the HSB observation stopped in February 2003, the AIRS/AMSU level-3 monthly mean products (version 5) without HSB are available from September 2002 to April 2007 at the NASA Goddard Space Flight Center Data Active Archive Center. These products have 24 (12) vertical levels for air temperature (humidity) on a 1° horizontal grid. In the AIRS/AMSU level-3 products, moreover, the microwave (MW)-only standard product retrieved by the MW retrieval stage of the AIRS algorithm is also available with the same spatial and temporal resolutions. To avoid sampling bias due to the presence of cloud, we use the MW-only standard product for air temperature. In analyzing the satellite sounding data, we apply a high-pass filter of 7° running mean in the meridional direction to remove the large-scale background field and extract structures on shorter spatial scales that are likely induced by SST fronts. (This high-pass spatial filter is applied for both the satellite and ICOADS observations in Figs. 7 and 8.)
The microwave scatterometer, SeaWinds, on the NASA Quick Scatterometer (QuikSCAT) satellite measures daily surface wind velocity over the global ocean. It has revealed rich wind structures on shorter spatial scales around the world (Chelton et al. 2004; Sampe and Xie 2007). We use the monthly wind velocity product available at RSS from August 1999 to April 2007 on a 0.25° × 0.25° grid.
The cloud-top pressure data of the ISCCP pixel level cloud product (DX) is used for the period from January 1999 to December 2004 (Rossow and Schiffer 1999). The periods from January 1999 to April 2003 and from May 2003 to December 2004 were based on the observations by the Geostationary Meteorological Satellite-5 (GMS-5) and the Geostationary Operational Environmental Satellite-9 (GOES-9), respectively. This dataset is available originally at about 30-km horizontal resolution every 3 h, but we interpolate the data onto a 1° × 1° grid. Frequency of cloud occurrence is calculated at each 50-hPa layer using 3-hourly data.
As an indicator of atmospheric convective activity, we use the monthly climatology of lightning activity observed by the Optical Transient Detector (OTD) onboard the MicroLab-1 satellite from April 1995 to March 2000 and the Lightning Imaging Sensor (LIS) onboard the TRMM satellite from January 1998 to December 2005. Christian et al. (2003) have reported that lightning is predominant in the North Atlantic and western North Pacific Oceans year round where atmospheric convection is produced as cold air blows over the warm ocean surface. The LIS/OTD merged monthly climatology is available at the NASA Global Hydrology Resource Center on a 0.5° × 0.5° grid.
The Japanese Ocean Flux datasets with Use of Remote Sensing Observations version 2 (J-OFURO2) is a new product capable of representing ocean mesoscale variability. It uses SST and surface wind from spaceborne microwave radiometers and scatterometers, surface air specific humidity estimated from the Special Sensor Microwave Imager (SSM/I), and surface air temperature from the National Centers for Environmental Prediction (NCEP) reanalysis. The COARE3.0 flux algorithm is used. Both latent and sensible heat fluxes of this product show good agreement with in situ observations from the Kuroshio Extension Observatory (KEO) buoy (Kubota and Tomita 2007). This dataset is available from January 2002 and December 2005 on a 0.25° × 0.25° grid, and we analyze the monthly mean product.
c. Reanalysis product
We use the monthly fields of pressure vertical velocity and horizontal divergence of the Japanese 25-yr Reanalysis (JRA-25) (Onogi et al. 2007) produced jointly by the JMA and Central Research Institute of Electric Power Industry (CRIEPI). These fields are available on a 1.25° × 1.25° grid at 23 vertical levels from 1979 to 2004. The product is updated for the subsequent period using the same climatic assimilation system, called the JMA Climate Data Assimilation System (JCDAS). Note that the JRA-25 started to assimilate high-resolution surface winds observed by the European Remote Sensing (ERS) and QuikSCAT satellites from 1995. For comparison with satellite observations, we analyze the JRA-25 for 7 years from 2000 to 2006 when the ERS and QuikSCAT scatterometer data are assimilated.
3. Winter cloud response
a. Low clouds
Figure 3 exhibits cross-frontal sections composited from R/V Mirai observations from January to March. In winter the prevailing winds are northwesterly (Fig. 3b). The ship-mounted ADCP measurements capture a strong eastward current on the warmer flank of the sharp SST front with maximum speed exceeding 1.2 m s−1 near the sea surface (Fig. 3e). This strong eastward current forms a SST front with the SST changing 7°C in less than 100 km (black line in Fig. 3c). By contrast, the meridional gradient of SAT is much more relaxed (red line in Fig. 3c). As a result, an atmospheric stability parameter (SAT − SST) displays a pronounced minimum of less than −8°C in a narrow width of 100 km downwind/south of the KE front, with a background value of about −5°C on the upwind side. Upward surface latent and sensible heat fluxes reach local maxima of 390 and 170 W m−2, respectively, just south of the SST front (Fig. 3d). The minimum in (SAT − SST) and maximum in turbulent heat flux are collocated with the eastward current maximum, indicative of the strong KE influence on the atmosphere.
Further evidence for the KE influence is found in the cloud-base height observed by laser ceilometer (Fig. 3a). The cloud-base height increases from 0.4 km at y = 0.5° to 1.2 km at y = −1° across the KE front, suggestive of vertical deepening of the MABL southward across the KE front under northerly cold advection. This sharp transition appears to be associated with a rapid increase in upward surface heat fluxes. The largest frequency of cloud occurrence appears at 1 km altitude about 1° south of the ocean current maximum, suggesting that the SST influence tends to emerge on the downwind side of the SST front due to the northerly advection. A close examination indicates that the southward cloud-base elevation was associated with one single atmospheric synoptic disturbance. The MR07–01 cruise cut through an atmospheric cold front at 37°N on 1700 UTC 18 February 2007 and the KE front at 36.2°N at 2100 UTC on the same day. Surface heat flux rapidly increased by 200 W m−2 at the former and 400 W m−2 at the latter front, with clouds observed frequently south of each front.
Composite atmospheric soundings also support the rapid transition of the MABL across the KE front. We employ SAT − SST as a criterion for the following composite because it represents the KE heating (Fig. 3d) and is available in most ship observations. Figure 4 shows vertical profiles of virtual potential temperature (θυ), relative humidity, and zonal and meridional winds, each of which has been composited separately for the highly unstable (SAT − SST < −6°C; 40 soundings) and the stable to weakly unstable (SAT − SST > −2°C; 39 soundings) categories. The θv profile for the unstable composite (Fig. 4a) features a well-defined surface mixed layer of 0.6–0.8 km high, capped by a high relative humidity (RH > 70%) layer between 0.8 and 1.4 km high (Fig. 4b). Moreover, this layer of high RH is in good agreement with cloud-base observations on the warmer flank of the KE front (Fig. 3a). Above the frictional surface layer as thin as 0.15 km, the vertical shear of zonal wind is weak throughout the mixed layer (Fig. 4c). Although the zonal wind is weaker above the 0.5 km height in the unstable composite, it becomes stronger by 1 m s−1 at the sea surface than in the stable composite, possibly as a result of enhanced vertical mixing. While infrequent in winter, a strongly stratified MABL is sometimes observed under warm advection by southerly winds (Fig. 4d). The stable composite (Fig. 4a) shows no surface mixed layer with stratification developed right above the surface. Because of the suppressed mixing, the vertical shear of the zonal wind above the frictional surface layer is substantially stronger than in the unstable composite (Fig. 4c). The high relative humidity (RH > 70%) layer is lower in altitude by 0.2–0.3 km than in the unstable composite (Fig. 4b). These composite profiles corroborate that the increased instability and vertical mixing in the MABL cause the cloud base to rise as northerly winds blow across the KE front.
b. Middle clouds
Mirai observations clearly show the effect of the KE front on the MABL. This subsection examines whether this influence extends into the free atmosphere above. Figure 5a displays a cross-frontal section of frequency of cloud occurrence as a function of cloud-top pressure (CTP) and latitude, based on ISCCP, averaged from 143° to 155°E for December–February. In the lower atmosphere below 850 hPa, clouds tend to be frequently observed south of y = −2° with a maximum occurrence at 950 hPa, while there is a minimum over the warm KE. Low clouds like cumulus/stratocumulus with clear-sky openings are often observed over the western North Pacific in winter (Norris 1998b). In the midtroposphere above 700 hPa, the cloud occurrence has a single peak between y = −2° and 0°. The horizontal pattern of cloud occurrence in the 500–700-hPa layer exhibits a maximum greater than 30% centered at 35°N from 141° to 150°E (Fig. 5b). Thus, clouds tend to be more vertically developed and reach the midtroposphere on the warmer flank of the KE front. In Fig. 5b, middle clouds are also frequently observed along the northwestern coast of Honshu and Hokkaido Islands, Japan, due to orographic lift on the upwind side of mountains. On the subarctic front to the north, middle clouds are infrequent in this layer possibly because small moisture content limits the vertical development of clouds.
Lightning observations by the LIS/OTD corroborate the middle-cloud distribution (Fig. 5c). In winter, the North Pacific storm track covers a broad band from the KOE region to the west coast of North America (Nakamura et al. 2004). In this broad storm track, lightning activity is enhanced, preferentially occurring in eastward-traveling atmospheric fronts (Christian et al. 2003). Notably, the lightning flash rate displays a local maximum over the KE region between 143° and 155°E, slightly to the north of the maximum middle-cloud occurrence. The collocated increase in lightning activity supports the enhanced vertical development of clouds along the winter KE front.
Total cloud liquid water measured by AMSR-E displays a zonal band that stretches from the south coast of Japan eastward (Fig. 6a). Embedded in this basin-scale cloud band is a local maximum that is nearly collocated with the band of maximum middle-cloud occurrence. Underneath it, upward surface heat fluxes are enhanced with a maximum exceeding 450 W m−2 on the warmer flank of the KE front (Fig. 6b). This maximum in surface heat flux shows good agreement with the Mirai observation (Fig. 3d). QuikSCAT wind velocity also displays a band of convergence on the warmer flank of the KE front (Fig. 6c), presumably helping deepen the clouds. The collocation of vertically developed clouds with surface convergence and enhanced surface heat flux is analogous to what is observed over the Gulf Stream (Minobe et al. 2008).
c. Temperature and pressure adjustments
AIRS/AMSU soundings offer hints at how the atmosphere adjusts to the KE front. Figures 7a and 7b exhibit a vertical cross section of air temperature and its horizontal map averaged between 1000–600 hPa, respectively. A meridional high-pass filter has been applied to extract frontal structures. Both air temperature and humidity anomalies are positive from the surface to midtroposphere with a surface maximum on the warmer flank of the KE front (Fig. 7a). The positive maximum in lower-atmospheric temperature (Fig. 7b) is collocated with the local maxima in surface heat flux, cloud liquid water, and surface wind convergence (Fig. 6).
This atmospheric temperature structure leads us to a hypothesis of SLP adjustment: SST-induced surface heat flux and surface wind convergence cause atmospheric convection to reach the midtroposphere, lowering SLP on the warmer flank of the KE front. Indeed, high-pass filtered SLP and surface wind anomalies derived from ICOADS (Fig. 8) are consistent with the SLP adjustment: a band of negative SLP anomalies forms on the warmer flank of the KE front, toward which surface wind anomalies converge (Fig. 8a). These SLP anomalies are further consistent with the horizontal pattern of AIRS/AMSU air temperature, especially over the warm Kuroshio south of Japan and its extension (Fig. 7b). Surface wind anomalies are largely in geostrophic balance with SLP (Fig. 8a), decelerating the westerlies at the SST front and accelerating them on the flanks (Figs. 7d and 8c). Between 33° and 37°N (30° and 33°N), the northerlies (southerlies) in the high-pass filtered wind field may be a response to SLP gradient, or alternatively due to enhanced (suppressed) vertical mixing as atmospheric stability, SAT − SST, decreases (increases) in this region.
Note that the AIRS/AMSU air temperature maximum at 1000 hPa (T1000) is displaced about 100 km south of the SST maximum (Fig. 7c), presumably by the northerly advection. A similar downwind displacement of SAT is reported for tropical instability waves in the eastern Pacific (Small et al. 2003). Cross-frontal profiles of SST, SAT, SLP, and surface wind anomalies in ICOADS corroborate satellite observations, both in support of the SLP adjustment.
d. Seasonal cycle
Figure 9 shows the seasonal cycles of surface heat flux, cloud liquid water, and surface wind convergence, confirming the collocation of their maxima on the warmer flank of the KE front during winter. As the surface wind shifts to gain an increasingly strong northerly component during September–November, upward surface heat fluxes intensify, peaking from December to the following January, with the maximum exceeding 450 W m−2 (Fig. 9a). The KE front sustains this strong surface heating. From February on, the upward surface heat flux gradually decreases as northerly winds begin to wane. During November to the following March, when the meridional SST gradient is steepest, cloud liquid water content (Fig. 9b) and surface wind convergence (Fig. 9c) are both enhanced on the warmer flank of the KE front. The climatological seasonal cycle suggests that the KE heating causes surface wind convergence and promotes cloud formation.
In summer, structures in surface flux and cloud water become blurred across the KE front because of meridional migration of the baiu front. Nevertheless, strong surface wind convergence remains along the KE front, possibly affecting clouds. The next section examines the association between the SST fronts and cloud/fog formation over the western North Pacific in summer.
4. Summer cloud response
Figure 10 shows cross-frontal sections composited from R/V Mirai observations for May to July. In this season, most of the R/V Mirai ship tracks cross the sharp KE front west of 148°E (Fig. 1c). Compared with winter, surface winds are considerably weak with speeds less than 4 m s−1. The direction also changes to southeasterly south of the KE front (Fig. 10b). The surface southeasterlies gradually weaken northward and change to weak northeasterlies,1 resulting in surface wind convergence over the KE region. QuikSCAT satellite corroborates this surface wind convergence trapped near the KE front during summer (Fig. 9c). In common with winter, a strong eastward ocean current is still found south of the SST front (Fig. 10e) with maximum speed exceeding 1 m s−1 near the sea surface. Across the front the SST decreases northward, by 6°C, from 23°C at y = −0.25° to 17°C at y = 0.25°. The SAT profile is smoother than SST, with its value decreasing northward from 21° to 18°C across the same distance (Fig. 10c). As a result, SAT − SST is −2° ∼ −3°C on the warmer flank of the KE front, while it is slightly positive or nearly zero on the colder flank. Note that SAT − SST increases by 5° ∼ 6°C from the winter average on both sides of the KE front, partly because of the switch from the winter, cold northerly, to summer, warm southerly, advection.
The weakened surface wind and increased air − sea temperature difference suppress upward surface heat fluxes over the KE (Fig. 10d). The latent heat flux is 130 W m−2 on the warmer flank of the KE front, less than a third of the value in winter. A meridional gradient of 100 W m−2 in less than 50 km across the front is maintained. Sensible heat flux barely exceeds 30 W m−2 on the warmer flank, while it is nearly zero on the colder flank.
a. Low clouds
Cloud-base occurrence also displays a sharp transition across the KE front (Fig. 10a). The large frequency of cloud-base occurrence is found in the layer between 0.4 and 0.9 km south of the KE front, about 0.5 km lower than in winter. A bimodal distribution of cloud-base height is seen right over the KE front with peaks at 0.1–0.3 km and 0.6–0.8 km, respectively. The former corresponds to the mixed-layer top and the latter to the main inversion. This bimodal structure is consistent with the 2005 summer KESS observations by atmospheric soundings and a ceilometer (Tanimoto et al. 2009). North of the KE front, by contrast, the cloud base tends to be frequently observed near the sea surface, suggesting fog occurrence. Another peak of the cloud-base frequency is around the 1.5 km height, which is mainly due to one single MR02–04 cruise during 25–26 June 2002. The cause of this peak is not identified, but a temperature inversion formed by atmospheric subsidence may partly contribute (Meitín and Stuart 1977). JRA-25 shows that the R/V Mirai was sailing in a high pressure system accompanied by atmospheric subsidence (not shown).
We have composited summertime atmospheric soundings based on the SAT–SST stability parameter. A total of 67 (66) soundings are used for the neutral to unstable (stable) composite with SAT − SST below −1°C (above 1.5°C). Figure 11 shows the composite profile for the neutral to unstable (stable) category in solid (dashed) lines. While the mixed layer is not as well developed as in winter because of increased surface stability, marked differences between composites appear near the sea surface. In the neutral to unstable composite, the vertical gradient of θυ is weak from the surface to 0.4 km (Fig. 11a), capped by a high relative humidity layer (RH > 90%) between 0.3 and 0.7 km height (Fig. 11b). The base of this layer is in good agreement with ceilometer cloud-base observations on the warmer flank of the KE front (Fig. 10a), indicating low cloud formation. Near the surface, weak northeasterlies advect cold air from the subarctic region, promote vertical mixing, and result in weakened vertical shear in both zonal and meridional wind (Figs. 11c,d). In the stable composite, on the other hand, the vertical gradient of θv is strong in the surface layer (Fig. 11a), with relative humidity above 90% from the sea surface to 0.6 km height (Fig. 11b). Southwesterlies advect warm/moist air from the subtropics, suppress vertical mixing, and intensify the vertical shear (Figs. 11c,d). High relative humidity near the surface is indicative of fog formation. During the KESS 2005 and 2006 summer cruises, we visually confirmed fog (small cumulus/stratocumulus) formation over the stable sea surface under the southerlies (the neutral to unstable sea surface under the northerlies).
b. Sea fog
We now use ICOADS to examine sea fog in summer over the broad western North Pacific. Figure 12 shows composite maps for frequency of fog occurrence in June based on the surface meridional wind at each 1° × 1° grid point. Near-surface stability is affected not only by SST fronts but also by horizontal thermal advection. Under surface southerly winds, SAT − SST tends to be positive over most of the basin due to the warm advection (Fig. 12a). From southeast of Hokkaido to a region around 42°N, 160°–170°E and over the northern Sea of Japan, SAT–SST rises above 1°C and the frequency of fog occurrence exceeds 40% (Fig. 12b). To the southeast, a zonal band of secondary maximum in fog occurrence (>10%) extends from 32°N, 145°E eastward, corresponding to a positive SAT − SST region on the colder flank of the subtropical SST front. Sandwiched in between are the Kuroshio and its extension where, despite southerly warm advection, SAT − SST remains weakly negative and the frequency of fog occurrence reaches a meridional minimum (<10%). The strong heating by SST and enhanced vertical mixing maintain this band of infrequent fog occurrence.
Surface northerlies cause cold advection and destabilize the lower atmosphere. SAT − SST drastically changes to negative, especially south of the subarctic SST front and along the vicinity of the Kuroshio in the ECS (Fig. 12c). Over these regions, the fog occurrence drops to below 10% while the fog band along the subtropical SST front disappears (Fig. 12d). Near the Kuril Islands northeast of Hokkaido, SAT − SST remains positive and fog occurrence is greater than 40% despite the northerly cold flow. There strong tidal mixing maintains very cold SST (Nakamura et al. 2000a,b) and keeps SAT − SST positive (Tokinaga and Xie 2009). The Yellow and East China Seas are another region where warm advection and ocean tidal mixing are important for sea fog formation (Zhang et al. 2009).
Figure 13 compares histograms of the meridional wind under all weather and foggy conditions for the subtropics and subarctic. In both regions, a southerly wind prevails during 70% of the time in summer. Overall, sea fog tends to occur under surface southerly winds, consistent with the composite analysis (Fig. 12). However, the peak under foggy conditions is clearly different between the subtropics and subarctic. The subtropics features a single peak for meridional winds from +3 to +5 m s−1 (∼20%) under foggy conditions (Fig. 13a), corroborating previous findings about the fog dependency on warm advection (Norris 1998a,b; Tanimoto et al. 2009). In the subarctic, on the other hand, the histogram under foggy conditions shows double peaks for meridional winds ranging from +3 to +5 m s−1 and from −1 to +1 m s−1 (Fig. 13b). While the former peak falls in the same range as in the subtropics, the latter major peak represents calm conditions with little horizontal advection. The subarctic histogram suggests that other factors contribute to fog occurrence besides southerly warm advection. The SST north of the subarctic front is too cold to destabilize the surface atmosphere even under the northerly cold advection (Fig. 12c). In the Sea of Okhotsk the mean SST in June is about 5°C, which is 17°C lower than in the KE region.
c. Mid–high clouds
Figure 14a exhibits a cross-frontal section of cloud-top occurrence from June to July, the baiu period. In the lower atmosphere below 700 hPa, the CTP frequency increases downward with its maximum near the sea surface south of y = −3°, probably associated with low clouds like cumulus/stratocumulus. Besides boundary layer clouds, cloud top is frequently observed in a layer between 500 and 350 hPa with a meridional maximum directly over the KE front. This maximum of high cloud-top occurrence appears in the horizontal map as a slightly northeastward tilted band extending from Kyushu to the KE region (Fig. 14b), characteristic of the baiu front where subtropical warm/moist and subarctic cold/dry air masses meet. Cloud-top occurrence in the 500–350-hPa layer is especially high along the southeastern coast of Japan, suggestive of land heating and/or orographic effects.
Over the ocean, the band of frequent cloud-top occurrence in the 500–350-hPa layer is roughly collocated with the KE front between 144° and 162°E, suggestive of a connection between the sea surface and free troposphere. QuikSCAT observations reveal a narrow band of surface wind convergence confined by and meandering with the KE front from the eastern coast of Japan to 155°E (Fig. 15a). Surface wind convergence broadens in the meridional direction farther eastward. The narrow band of surface wind convergence is roughly collocated with upward vertical velocities at 700 hPa and the baiu cloud band east of Japan. The JRA-25 captures this narrow band of surface convergence2 over the KE front around 35°N as well as the divergence both to the north and south (Fig. 15b). It further shows that this surface convergence is connected with a deep upward velocity structure, which broadens above the MABL to cover a broad meridional extent between 27° and 40°N. The collocation of the KE-induced surface wind convergence and upward motion in the baiu cloud band suggests that the SST front may play a significant role in extending baiu precipitation east of Japan, say, by moisture convergence, helping the growth of baroclinic disturbances.
We have analyzed a suite of in situ and satellite observations to study the climatological cloud response to SST fronts in the western North Pacific. Our synthesis of these diverse measurements reveals the three-dimensional structure of the ocean and atmosphere in detail not possible before. High-resolution ship observations capture sharp cross-frontal transitions in the MABL during both winter and summer. From satellite observations, we have detected the influence of the KE front on relatively deep clouds that reach the mid to upper troposphere.
In winter, the KE releases huge amounts of latent and sensible heat with flux greater than 450 W m−2 under strong surface northwesterlies. The intense turbulent heat flux warms the MABL on the southern flank of the KE front, lowering SLP and causing surface wind convergence. The resultant upward motion transports moisture, causing clouds to grow higher. Satellite-observed cloud-top height displays a meridional maximum along the KE front with strong vertical development of clouds corroborated by enhanced lightning activity.
In early summer, turbulent heat flux is much suppressed over the western North Pacific as the southerly warm advection stabilizes the surface atmosphere. A weak, albeit significant, meridional gradient in surface heat flux remains across the KE front. A sharp transition across the KE front is observed from low clouds with elevated base on the southern flank to fog on the northern flank. In ICOADS sea fog tends to be frequently observed on the colder flanks of the western North Pacific SST fronts under southerly warm advection, while its frequency of occurrence decreases under northerly cold advection. Sea fog occurrence is low on the warmer flank of KE front even with warm advection, possibly because the warm ocean current destabilizes the surface atmospheric stratification and the vertical mixing prevents fog from forming. North of the subarctic front, on the other hand, the climatological frequency of fog occurrence is above 40% with the maximum exceeding 50% along the Kuril Islands where strong tidal mixing maintains a very cold ocean surface. The low (high) fog occurrence over the Kuroshio Extension (the Kurils) illustrates the importance of subsurface ocean processes for fog formation.
The baiu rainband is an important summer climate phenomenon for Japan, a feature poorly simulated by atmospheric models. Our results suggest a possible role of the KE front in baiu formation east of Japan by producing a narrow band of surface wind convergence. Atmospheric reanalysis, assimilating scatterometer wind data, shows that the surface convergence maintains a band of upward motion extending through the entire troposphere. Further studies are necessary to firmly establish this oceanic influence on the baiu rainband.
Minobe et al. (2008) have reported similar enhanced vertical development of upward motion and precipitation on the warmer flank of the Gulf Stream front. They find that the surface wind convergence pattern induced by the Gulf Stream front is consistent with the SLP adjustment mechanism. Liu et al. (2007) also detect a deep atmospheric response to oceanic mesoscale eddies produced by the Agulhas Current meander. Both studies use annual-mean data. While both the Gulf Stream and Agulhas Current maintain robust SST fronts year round, the SST gradient of the KE front has strong seasonality in magnitude. The region where the southward SST gradient is steeper than 2°C per 100 km extends from the east coast of Japan to 151°E in winter but is confined to the west of 146°E in summer (Fig. 1). The surface wind also changes from strong northwesterly in winter to weak southerly. A seasonal perspective, as performed here, is needed to study the SST influence on the atmosphere over the western North Pacific.
While our analysis focuses on cloud variations, the SST front may affect extratropical storm tracks through its effect on air temperature and baroclinicity (Inatsu et al. 2002; Tanimoto et al. 2003; Nakamura et al. 2004, 2008). Indeed, our analysis of AIRS/AMSU soundings shows such a SST influence on the lower-atmospheric temperature and hence baroclinicity. Adachi and Kimura (2007) demonstrate that surface cyclogenesis is frequently observed over the Kuroshio Extension in winter, a result consistent with lower SLP (Fig. 8a), strong vertical development of clouds (Fig. 5b), and enhanced lightning activity (Fig. 5c) detected in the present study. Numerical simulations with regional atmospheric models show that the SST front of the Kuroshio and its extension affects the growth of an extratropical cyclone (Xie et al. 2002; Taguchi et al. 2009, manuscript submitted to J. Climate). In the Southern Hemisphere, sharp SST fronts maintain strong meridional gradients of surface heating (Tokinaga et al. 2005; Nonaka et al. 2009, manuscript submitted to J. Climate) favoring atmospheric cyclogenesis (Sinclair 1994). Nakamura and Shimpo (2004) present observational evidence for SST frontal effects on atmospheric storm tracks, with some support from AGCM experiments (Inatsu and Hoskins 2004). The interaction of SST fronts and atmospheric storm tracks is a subject of our ongoing research.
We wish to thank the captains and crews of the R/V Kaiyo-maru of JFA, R/Vs Kofu-maru and Ryofu-maru of JMA, R/Vs Roger Revelle and Melville of SIO, and R/Vs Mirai and Hakuho-maru of ORI/JAMSTEC for atmospheric sounding and sea surface meteorological data; K. Kai, H. Okajima, M. Nonaka, B. Taguchi, F. Kobashi, and H. Nakamura for their cooperation in the cruise surveys, without which this study would not have been possible. Careful comments from M. F. Cronin and an anonymous reviewer helped improve an early version of the manuscript. The R/V Mirai data were obtained from the JAMSTEC Data Site for Research Cruises; the AMSR-E and QuikSCAT data from RSS; the AIRS/AMSU and AVHRR SST data from NASA GSFC and JPL PO DAAC, respectively. We also thank M. Kubota for providing the J-OFURO2 surface flux dataset. This work was supported by NASA, NSF, JAMSTEC, the Sumitomo Foundation (043426), and Grant-in-Aid for Scientific Research (18740307, 17340137, and 15340153) defrayed by the Ministry of Education, Culture, Sports, Science and Technology of Japan.
Corresponding author address: Hiroki Tokinaga, International Pacific Research Center, University of Hawaii at Manoa, Honolulu, HI 96822. Email: email@example.com
On the colder flank of the KE, the weak southeasterlies in the QuikSCAT climatology (Fig. 1c) differ from the weak northeasterlies in the ship composite because of insufficient sampling in the latter.
The narrow band of surface convergence and upward motion is not present in JRA-25 prior to 1995, the year when scatterometer wind observations began to be assimilated.
* International Pacific Research Center Publication Number 599 and School of Ocean and Earth Science and Technology Publication Number 7666.