1. Introduction
In situ observations and numerical model studies have clearly shown wind changes near the surface and throughout the depth of the marine atmospheric boundary layer (MABL) across well-defined SST frontal zones with cross-frontal length scales of O(50–1000 km), which loosely defines the oceanic mesoscale (see Small et al. 2008 for a detailed review). On these scales, near-surface winds are stronger over the warmer sides of the SST fronts compared to the cooler sides. Features along SST fronts that have been shown to generate these wind perturbations are meanders in ocean currents, ocean eddies, and instability waves. While in situ observations and numerical simulations have been of enormous benefit in characterizing these interactions, in situ observations are generally limited to localized case studies over relatively short time scales (≲1 week), while numerical models only offer approximations of the atmospheric response to SST fronts that are sensitive to specifications of the SST boundary conditions, model spatial resolution, and subgrid-scale mixing parameterization (Warner et al. 1990; Song et al. 2009).
Many of these limitations can be overcome through investigation of the wind–SST coupling using satellite observations. Satellite surface wind stress and 10-m equivalent neutral wind (ENW; as defined in section 2c) measurements from the Quick Scatterometer (QuikSCAT) and SST and ENW measurements from the Advanced Microwave Scanning Radiometer for Earth Observing System (AMSR-E) on board the Aqua satellite have the advantage of long accumulated geophysical data records (over 10 complete years’ worth of surface vector winds from QuikSCAT and currently 9 years’ worth of SST and ENW data from AMSR-E), frequent sampling (1 ~ 3 times daily in midlatitudes and 0 ~ 2 times daily in the tropics), and broad spatial coverage over most of the ice- and precipitation-free global oceans. Many studies have documented the effects of SST on the QuikSCAT surface wind stress on the oceanic mesoscale (e.g., Xie et al. 1998; Chelton et al. 2001; O’Neill et al. 2003; White and Annis 2003; Chelton et al. 2004; Vecchi et al. 2004; Tokinaga et al. 2005; O’Neill et al. 2005; Song et al. 2006; Chelton et al. 2007; Haack et al. 2008; O’Neill et al. 2011, manuscript submitted to J. Climate, hereafter OCE). However, because of large variations in near-surface atmospheric stability and surface ocean currents near SST frontal zones, SST-induced changes in the actual surface wind speed cannot technically be inferred from QuikSCAT surface wind stress and ENW observations alone since scatterometers fundamentally estimate surface wind stress rather than the actual surface wind speed over the ocean (e.g., Ross et al. 1985; Liu and Tang 1996). One objective of this study is to quantify the responses of the satellite ENW and surface wind stress magnitude to SST-induced changes in surface wind speed and surface layer stability near strong, well-defined SST frontal zones using surface meteorological and SST observations from moored buoys near the Gulf Stream and the eastern equatorial Pacific regions. Long data records, ancillary meteorological observations, and fixed positioning of the moored buoys allow a quantitative assessment of the influence of surface wind speed and surface layer stability on the coupling of scatterometer-estimated winds to SST fronts. Using these collocated moored buoy and QuikSCAT estimates of ENW and surface wind stress, it is shown that the QuikSCAT ENW and wind stress responses to mesoscale SST perturbations are consistent with an SST-induced response of the actual surface wind speed rather than to cross-frontal variations of surface layer stability.
The mesoscale wind–SST coupling described here has dynamically important repercussions for the ocean–atmosphere system (for a recent review, see Chelton and Xie 2010). Minobe et al. (2008) has shown a strong correspondence between the surface convergence associated with the SST-induced wind perturbations and the upward vertical motion over the Gulf Stream, which significantly affects clouds (Minobe et al. 2010), precipitation (Minobe et al. 2008; Kuwano-Yoshida et al. 2010; Minobe et al. 2010), and lightning flash rate (Minobe et al. 2010). Similar interactions between the boundary layer and free atmosphere have also been shown to occur over the Kuroshio extension (Tokinaga et al. 2009) and the Agulhas Return Current (Liu et al. 2007). Additionally, Sampe and Xie (2007) have shown from scatterometer ENW observations that high-wind events are much more frequent over warm SST perturbations over midlatitudes. Furthermore, satellite observations of the surface wind response to mesoscale SST perturbations have been used as an important diagnostic tool to evaluate mesoscale ocean–atmosphere interactions in numerical weather prediction models, including climate, operational, mesoscale, reanalysis, and fully two-way coupled ocean–atmosphere models, and to evaluate improvements in the model formulations (e.g., Small et al. 2005; Chelton 2005; Song et al. 2006; Maloney and Chelton 2006; Seo et al. 2007; Haack et al. 2008; Song et al. 2009; O’Neill et al. 2010b; Minobe et al. 2010).
Spatial variability of the SST-induced wind stress can generate wind stress curl perturbations related to the local crosswind SST gradient (Chelton et al. 2001; O’Neill et al. 2003; Chelton et al. 2004; O’Neill et al. 2005). Since the wind stress curl drives Ekman upwelling in the ocean over midlatitudes, there are potentially significant feedbacks onto the ocean from the wind stress perturbations coupled to the mesoscale SST field. SST-induced features in the wind stress field are expected to create significant perturbations in the upper-ocean vertical mixing and Ekman upwelling. Evidence is beginning to emerge from ocean modeling studies suggesting significant effects of the SST-induced wind stress field on local and large-scale ocean circulation. Specifically, SST-induced feedbacks can modify the pathway and transport of gyre circulations (Milliff et al. 1996; Hogg et al. 2009), the growth rate of baroclinic instabilities associated with oceanic frontal zones (Spall 2007), and the upwelling and dynamics of eastern boundary current systems (Jin et al. 2009). These feedbacks occur primarily through modulation of Ekman upwelling, wind-driven basin-scale volume transports, surface heat and momentum fluxes, and modification of baroclinic instabilities in ocean currents. Because of the dynamical implications of the mesoscale wind–SST coupling, an evaluation of the satellite-derived mesoscale wind–SST coupling is needed since these observations have played a large role in motivating the dynamical studies.
The influence of SST-induced wind stress perturbations on the ocean does not depend on whether the surface wind stress response to SST is derived from wind speed, atmospheric surface layer stability, or surface ocean currents. In contrast, however, the SST influence on the free troposphere does depend on these factors, since vertical motion induced by surface mass convergence or divergence depends on the SST influence on the actual surface winds relative to a geographically fixed coordinate system. Linking a free-tropospheric response to mesoscale SST fronts via vertical motion (e.g., Liu et al. 2007) implies that the surface mass convergence, and hence the actual surface wind, responds to SST. In this analysis, it is established from moored buoy observations that the mesoscale SST influence on the scatterometer-derived wind is attributable mainly to the response of the actual surface wind to mesoscale SST fronts. This conclusion clarifies the interpretation of the wind–SST coupling derived from scatterometer observations, and supports the continued use of scatterometer winds for study of the deep tropospheric response to SST-induced surface wind perturbations.
One challenge when evaluating the wind–SST coupling from discrete in situ observations is to define the ambient large-scale state of the wind and SST fields unrelated directly to the mesoscale wind–SST coupling. In this analysis, pairs of moored buoys near SST frontal zones are used to test the hypothesis that cross-frontal variations in wind speed are related to those of SST; while confirming this hypothesis, it is shown that this method also yields estimates of the mesoscale wind–SST coupling that are consistent with those derived from the spatially filtered satellite ENW and SST fields. This simple methodology is furthermore used to quantitatively assess the ENW response to SST from satellite observations collocated to the buoy locations.
The buoy meteorological and SST observations used here are described next in section 2a. In section 2b, the methodology for evaluating the surface wind response to SST from 17 pairs of moored buoys located near the Gulf Stream and the eastern equatorial Pacific is described. The wind–SST coupling is investigated on periods longer than 10 days, which is consistent with the time averaging applied to satellite and model wind and SST fields in previous studies of mesoscale wind–SST coupling. With these buoy observations, the influence of surface-layer stability on the responses of the ENW and surface wind stress magnitude are investigated quantitatively in sections 2c and 2d, respectively. The QuikSCAT and AMSR-E satellite wind and SST observations are described in detail in section 3. After collocating the gridded satellite observations in space and time with those from the buoy pairs, it is then shown that the satellite-derived coupling agrees well with that obtained from the buoys over most locations. The satellite-derived wind–SST coupling is shown to be consistent with a response of the actual surface wind speed to SST.
2. Wind–SST coupling from moored buoys
a. Description of moored buoy observations
Multiyear records of in situ meteorological and SST data central to this analysis were obtained from moored buoys collected and distributed by the National Data Buoy Service (NDBC), the Canadian Department of Fisheries and Oceans (CDFO), and the National Oceanic and Atmospheric Administration/Pacific Marine Environmental Laboratory (NOAA/PMEL) Tropical Atmosphere Ocean (TAO) array. As shown in Fig. 1, these sites consist of a set of 3 NDBC and 6 CDFO buoys straddling the northern Gulf Stream in the North Atlantic and 15 TAO buoys along the northern and southern sides of the equatorial Pacific cold tongue between 2°S–2°N and 95°–155°W. Descriptions of the moored buoys used here are summarized in Table 1, including the location, observational period, measurement frequency, instrument measurement heights, and SST measurement depth. As further summarized in Table 1, the observations used here were collected during the period 1 August 1999–31 July 2009 at most sites. A small number of the CDFO and NDBC buoys were relocated during the course of their deployment, so only buoy data records collected within a 50-km radius of their position on 31 July 2009 were used. For the NDBC buoys, the historical standard meteorological data were analyzed; these data contain wind, air and dewpoint temperature, and barometric pressure observations averaged over an 8-min interval once per hour. For the TAO buoys, the high-resolution observations at 10-min intervals were used, which represent 2-min averages during each 10-min interval. For the Canadian buoys, observations are reported as 10-min averages at hourly intervals. Listed in Table 1 are the numbers of independent observations where valid wind speed, air temperature, and SST were measured simultaneously and used here.
Maps of the (top) north Atlantic and (bottom) eastern equatorial Pacific with the locations of moored buoys used in this study marked. The solid black lines connect the buoy pairs used here, which are listed in Table 1. The gray contours are the AMSR-E satellite SST averaged over the period 1 Jun 2002–31 May 2010 with a contour interval of 1°C. The solid black SST contours represent (top) the 18°C isotherm and (bottom) the 25°C isotherm. The gray vectors show the vector-averaged QuikSCAT surface wind over the period 1 Jun 2002–31 May 2009. Abbreviations used in the key include the Canadian Department of Fisheries and Oceans (CDFO), National Data Buoy Center (NDBC), and Tropical Atmosphere Ocean (TAO).
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
Description of the moored buoys used in this study, including location coordinates, time periods of data used, numbers of observations, measurement frequencies, anemometer heights, air temperature Ta and humidity qa measurement heights, and SST Ts measurement depths. The numbers of observations reported here are the total numbers of observations where wind speed, SST, and air temperature were available simultaneously. The measurement frequency is the frequency and duration of recorded observations; for instance, the NDBC buoys average and record observations for an 8-min period once per hour while the TAO buoys average and record observations for a 2-min period once every 10 min.
Buoy pairs were chosen near sharp, well-defined SST frontal zones, as shown by the mean SST field for the period 1 June 2002–31 May 2009 from the AMSR-E satellite in Fig. 1. The lines connecting individual buoys represent the buoy pairs used in this analysis; overall, there are 7 buoy pairs over the Gulf Stream and 10 buoy pairs over the equatorial Pacific. For each buoy pair, the buoys were required to be relatively close (separation distances δL are listed in Table 2, which range from 155.6 to 343.0 km), to have several years of overlapping wind, SST, and air temperature records, and to have identical wind and air temperature measurement heights (listed in Table 1). The buoy pairs listed in Table 2 satisfy these requirements. The TAO buoys along 2°N are just far enough south to be minimally affected by the intertropical convergence zone.
Buoy-pair separation distances (δL), numbers of concurrent observations for each buoy pair (N) in which valid wind speed and SST observations were available, cross-correlation coefficients between 
For this analysis, all buoy observations were carefully quality controlled by visual inspection and by removing outliers that exceeded three standard deviations from a detrended 1-month running mean, where the standard deviations were computed from a detrended running 1-month mean. Measurements at times when the differences in wind speed, SST, or air temperature between the buoys in each pair exceeded three standard deviations from the mean difference were also removed. These steps were especially important for the CDFO buoys, whose measurements were not quality controlled as extensively as those from NDBC or TAO. Observations from the CDFO buoys are thus noisier than the NDBC and TAO buoys. Finally, the time series of wind speed 
b. SST-induced response of the actual wind speed
To quantitatively assess the influence of SST fronts on surface wind speed, differences in wind speed 
(a) The actual wind speed difference between buoy pairs 
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
As in Fig. 2, but for the north equatorial Pacific cold tongue between the equator and 2°N.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
As in Fig. 2, but for the south equatorial Pacific cold tongue between 2°S and the equator.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
These binned scatterplots also show that for all buoy pairs, 
As summarized in Table 2, 
Finally, the sensitivity of the correlation between 
Correlation coefficients between the actual wind speed differences 
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
c. SST-induced response of the ENW
1) ENW definition and significance
Conversion of the buoy wind to ENW is necessary because scatterometers do not measure the actual near-surface wind speed, but rather the backscattered microwave energy from wind-generated capillary-gravity waves, which are in near equilibrium with the local surface wind stress. Scatterometers thus provide estimates of surface wind stress over the ocean. The surface wind speed and stress are not related uniquely to each other because the surface wind stress also depends on near-surface stability, surface ocean currents, surface waves, and surface air density in addition to wind speed. Computing the actual surface wind speed from scatterometer-estimated surface wind stress thus requires coincident surface measurements of stability (for which air–sea temperature and specific humidity differences are needed), ocean currents, waves, and surface air density, all of which are unfortunately impractical to make on the time and space scales of the scatterometer stress measurements. For this reason, scatterometer backscatter measurements are calibrated to the vector ENW.
Since significant air–sea temperature and humidity differences are often present near strong SST fronts, and since surface ocean currents are often most significant near ocean frontal zones, it has been suggested that ENW perturbations near SST fronts observed from scatterometers are more strongly affected by cross-frontal variations of surface layer stability and surface ocean currents rather than the actual surface wind speed (Liu et al. 2007; Liu and Xie 2008). In contrast to these suggestions, however, the buoy wind speed analysis above showed that SST fronts do significantly modify the actual surface wind speed.
To quantify the significance of surface layer stability in determining the ENW response to SST, the responses of the ENW and actual wind speed to SST from the buoy pairs is compared. As succinctly stated on page 53 of Garratt (1992), “The effects of buoyancy can be interpreted as a deviation of the wind speed from the neutral value.” Motivated by the relative lack of long-term in situ observations of the ENW response to SST frontal zones, some characteristics of the buoy-derived ENW near the Gulf Stream and equatorial Pacific cold tongue are first described here. The analysis from the previous subsection is then repeated for 
The ENW was computed according to Eq. (1), where u* and z0 were estimated using the Coupled Ocean–Atmosphere Research Experiment (COARE) bulk flux algorithm version 3.0 (Fairall et al. 2003) using as inputs buoy measurements of wind speed, SST, air temperature, air pressure, and specific humidity. No temporal smoothing was applied to the input variables to the COARE algorithm. Kara et al. (2008) has shown that the ENW computed from the COARE algorithm is very close to that computed from the Liu–Katsaros–Businger (Liu et al. 1979) and Bourassa–Vincent–Wood (Bourassa et al. 1999) flux algorithms. When buoy measurements of air pressure or specific humidity were unavailable, constant values of 1013 hPa for pressure and 75% for relative humidity were used; while these assumptions have no qualitative impact on the results presented here, they allow significantly more observations to be available for this analysis. Additionally, humidity measurements were not made routinely on the CDFO buoys used here and air pressure measurements were not reported on the TAO buoys used here. Liu (1990) and Geernaert and Larsen (1993) have found that the computation of the ENW was sensitive to the surface water vapor flux for very warm SSTs and low relative humidities. These conditions were rarely encountered at the CDFO buoys (which did not measure humidity and are thus most influenced by this assumption), however, since the SST rarely exceeded 15°C and the relative humidity at nearby NDBC buoys was relatively high (the mean was 76% and the standard deviation was 9%). Additionally, the ENW means and variances for the CDFO buoys were relatively insensitive to a range of specified relative humidities between 60% and 90%.
2) Observations of V10n − V10
Over the sharp SST frontal zones associated with the Gulf Stream, V10n − V10, and thus stability, are affected significantly by the wind direction relative to the front; this directional sensitivity is shown by the two-dimensional histograms in Fig. 6 for each of the Gulf Stream buoys, which shows V10n − V10 along the y axis and buoy wind direction along the x axis. To preserve variations over all time scales, these histograms were computed from the unsmoothed ENW and wind direction. The dynamic ranges of the V10n − V10 distributions are somewhat smaller using the 10-day running averages of V10n − V10 compared to the unsmoothed V10n − V10 shown here, although the means and medians are similar (not shown).
Two-dimensional histograms of V10n − V10 (y axis) and wind direction (x axis) computed from each of the 24 buoys near the Gulf Stream and eastern equatorial Pacific as indicated above each plot. Unsmoothed time series of V10n − V10 and wind direction were used in the histogram computation to preserve variations over all time scales.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
Consistent with the mean QuikSCAT wind field shown in Fig. 1, the winds are predominantly westerly in the northwest Atlantic. Winds with a northerly component (negative wind direction) are most prevalent and typically correspond to offshore surface flow from cool to warm SST across the northern edge of the Gulf Stream. These histograms show that V10n − V10 is mostly positive for northerly winds, consistent with unstable conditions generated by flow from cool to warm SST. Conversely, winds with a southerly component (positive wind direction) correspond to V10n − V10 that is mostly negative, consistent with stabilization of the surface layer as air blows from warm to cool SST. The absolute differences in V10n − V10 between northerly and southerly winds are not symmetric, as is evident from the smaller |V10n − V10| in northerly flow than for southerly flow. For instance, in southerly flow, V10n is as much as 1 m s−1 smaller than V10, while for northerly flow, V10n is less than 0.5 m s−1 greater than V10. These histograms show that southerly winds are much less common than northerly winds and thus all of the V10n − V10 distributions over the Gulf Stream are centered over slightly positive values of V10n − V10. Over the equatorial Pacific, the wind has a mostly easterly component, and V10n is only about 0.25 m s−1 greater than V10 for all buoys (Fig. 6). This represents slightly unstable conditions as surface air blows westward from the equatorial cold tongue to warmer waters (Fig. 1).
Seasonal variations of V10n − V10 are quite significant over the Gulf Stream while nearly nonexistent over the eastern equatorial Pacific, as shown by monthly medians and interquartile ranges of V10n − V10 over the Gulf Stream (Fig. 7a) and eastern equatorial Pacific (Fig. 7b) buoys. In these plots, the black points represent the medians and the colored bars represent the interquartile ranges; like the histograms in Fig. 6, unsmoothed winds were used. Over the Gulf Stream, the largest positive values of V10n − V10 occur during December (with median values of ~0.25 m s−1), and the largest negative values occur during June (with median values between about 0 and −0.75 m s−1). Note that buoy 44004 has a much smaller seasonal cycle of V10n − V10 compared to the others. Additionally, little month-to-month variability exists at any of the Gulf Stream buoys for the 6 months between September and February. Unstable conditions during winter over the Gulf Stream are also likely exacerbated by cold continental air advecting off the Eastern Seaboard. The intermonthly variability of V10n − V10, as represented by the interquartile range, shows counterintuitively that V10 − V10, and thus stability, is most variable during the early summer rather than winter. The reason for this result is explained below. Over the equatorial Pacific, there is very little intermonthly variability of V10n − V10, which is mostly less than 0.25 m s−1 for all months.
Monthly statistics of unsmoothed V10n − V10: median (black points) and interquartile range (colored bars) for the (a) Gulf Stream and (b) eastern equatorial Pacific. Each buoy is color coded according to the legend on the right for each month. These statistics were computed for the time periods shown in Table 1. (c) Behavior of V10n − V10 (y axis) as a function of V10n (x axis) for various air–sea temperature differences Ts − Ta shown by the key on the right side of the panel. These curves were computed from the COARE flux algorithm using an RH of 75% and an air temperature (Ta) of 18°C.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
It is clear from Figs. 6 and 7 that the seasonal variability of V10n − V10, and hence surface-layer stability, over the Gulf Stream is associated mainly with the seasonal evolution of the large-scale wind direction relative to the Gulf Stream frontal zone. This point can also be seen from maps of the mean ENW vectors from QuikSCAT and SST from the AMSR-E averaged separately for June 2002–09 and December 2002–08 (Figs. 8a,b, respectively). Also shown by the gray contours in these maps are mean sea level pressure fields from the National Centers for Environmental Prediction (NCEP) FNL 4 times daily analyses. In December, winds blow mainly offshore from the northwest from cooler to warmer SST, generating an unstable surface layer. Conversely, in June, the winds blow either along the Gulf Stream or with a weak cross-isotherm component from warmer to cooler SST, generating a stable surface layer. Additionally, the winds are much weaker during June compared to December. The seasonal changes in the large-scale wind direction are consistent with the seasonal progression of the mean large-scale sea level pressure field, which ultimately has a large influence on the seasonal variations of V10n − V10.
Maps of the AMSR-E SST (colored), QuikSCAT vector-averaged ENW (vectors), and the NCEP sea level pressure (gray contours) averaged for (a) June 2002–09 and (b) December 2002–08 over the northwest Atlantic. The locations of the buoys are also shown. The contour interval for the sea level pressure fields is 1 hPa.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
To aid in the interpretation of the histograms in Fig. 6 and the seasonal cycles in Figs. 7a,b, the behavior of V10n − V10 computed from the COARE algorithm is illustrated as a function of V10n for a range of air–sea temperature differences (Ts − Ta) between −4° and 8°C, a fixed air temperature of 18°C, and a relative humidity of 75% (Fig. 7c). For the range of Ts − Ta considered here, the absolute magnitude of V10n − V10 decreases steadily as the wind speed increases. For unstable stratification (Ts − Ta > 0), the differences are generally less than 0.5 m s−1 for the range of wind speeds considered here. The differences are progressively more significant for light winds and stable stratification, which accounts for the larger V10n − V10 when southerly winds blow across the north wall of the Gulf Stream from warm to cool SST as shown in Fig. 6.
Figure 7c also shows that while stability determines the sign of V10n − V10, its magnitude depends on both V10n and stability, especially for wind speeds below ~5 m s−1. As the wind speed increases for a given Ts − Ta, V10n − V10 decreases while becoming less sensitive to further increases in wind speed. Additionally, the magnitude of V10n − V10 is not symmetric about the air–sea temperature difference, with a much larger amplitude response for stable Ts − Ta compared to an equally unstable Ts − Ta (e.g., cf. V10n − V10 of about 0.25 and −0.4 m s−1 for Ts − Ta = +2 and −2°C, respectively, when V10n = 8 m s−1). During winter, while Ts − Ta is positive and relatively large over the Gulf Stream, so too is the wind speed, which compensates the temperature-induced surface instability and reduces its effect on V10n − V10. In contrast, during summer, weak winds and stable stratification combine to amplify the negative V10n − V10, which results in the larger |V10n − V10| encountered during the Gulf Stream summer shown in Fig. 7a. Two-dimensional histograms of Ts − Ta and V10n shown for June (Fig. 9a) and December (Fig. 9b) for all nine Gulf Stream buoys bear these scenarios: weak winds and marginally stable Ts − Ta during June, and strong winds and mainly unstable Ts − Ta during December. In these histograms, Ts − Ta is along the y axis and V10n is along the x axis, and the unsmoothed buoy wind and temperatures were used.
Two-dimensional histograms of Ts − Ta (y axis) and V10n (x axis) combined from all nine Gulf Stream buoys for (a) June and (b) December. These histograms were computed from the unsmoothed buoy wind and temperature observations.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
Finally, the larger interquartile ranges of V10n − V10 during the summer months over the Gulf Stream can be attributed to the increased sensitivity of V10n − V10 to wind speed and stability at low wind speeds and stable stratification, as shown in Fig. 7c. The increased variability of V10n − V10 during summer can be seen more clearly from two-dimensional histograms of V10n − V10 as a function of month for each of the Gulf Stream buoys, which are shown in Fig. 10. For all nine buoys, the distributions of V10n − V10 all have negative skewness (i.e., having longer tails for V10n − V10 < 0), which is most apparent during summer; this skewness is consistent with the increased sensitivity of V10n −V10 to small changes in wind speed and stability in the low wind speed and stable conditions encountered during summer. Note that the histograms in Fig. 9 show qualitatively that the variabilities of the air–sea temperature difference and wind speed are much larger during winter; however, for the reasons noted above, the ENW is much less sensitive to stability in the unstable stratification and strong winds typical of winter.
Two-dimensional histograms of V10n − V10 (y axis) and month (x axis) for each of the nine Gulf Stream buoys during the entire period considered here. These histograms were computed from the unsmoothed buoy winds.
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
3) Results
To quantify the SST-induced responses of the ENW V10n, binned scatterplots of 
Bar charts of the coupling coefficients from the buoy-pair differences for the (a) actual wind speed and ENW (
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
Since the magnitudes of the responses of the actual wind speed and ENW to SST differ by only 10%–30%, the SST-induced response of the ENW is thus attributable mainly to the response of the actual wind speed rather than to surface-layer stability. This result is consistent with earlier conclusions based on numerical model simulations (Wai and Stage 1989; Small et al. 2003; O’Neill et al. 2010b) and sensitivity analyses (O’Neill et al. 2005; Small et al. 2008).
In addition to this important result, the linear response of the buoy 
d. SST-induced response of the surface wind stress magnitude
In this section, the surface wind stress magnitude response to SST is related to the ENW response to SST. The surface wind stress magnitude was also computed from the buoy observations using the COARE flux algorithm and then smoothed with a 10-day running average. Binned scatterplots of 
It was shown earlier that the SST-induced response of δV10n was caused mainly by the response of the actual wind speed 
e. Seasonal variability of the wind–SST coupling over the North Atlantic
Histograms of 
Histograms derived from the seven buoy pairs over the Gulf Stream for the 6-month periods November–April (blue curves) and May–October (red curves): (a) δTs, (b) δV10n (solid) and 
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
Seasonal variations of the responses of the actual wind speed 
Since 
3. Comparison of satellite and buoy wind–SST coupling
The mesoscale wind–SST coupling observed by QuikSCAT and AMSR-E is now assessed using the buoy observations. Most previous studies investigating this coupling using satellites have relied on spatial high-pass filtering to isolate the mesoscale SST influence on surface winds. It is difficult to objectively assess these spatially high-pass-filtered satellite wind and SST fields using buoys since it is not possible to spatially filter the buoy data in the same manner. This shortcoming is overcome here using the simple methodology developed in section 2, whereby the wind and SST differences between pairs of buoys are compared to those from the spatially unfiltered satellite observations collocated in space and time.
a. Satellite data description
1) QuikSCAT scatterometer wind data
In nonprecipitating, ice-free oceans, the QuikSCAT instrument infers surface vector winds from Ku-band microwave radar measurements of backscattered power from the wind-roughened ocean surface. Over the ocean, surface roughness has been found to be related to the surface wind stress rather than the near-surface wind speed (e.g., Weissman et al. 1994; Bourassa 2006). Since there are few direct observations of surface stress over the ocean, and since a fundamental physical relationship between microwave radar backscatter and surface wind stress does not yet exist, scatterometer backscatter measurements are calibrated empirically to buoy measurements of V10n (e.g., Freilich and Dunbar 1999).
For this analysis, we use the complete QuikSCAT data record, which spans the 10+ year period 19 July 1999–19 November 2009. Processed QuikSCAT swath data were provided by the Remote Sensing Systems (version 3) and were gridded here onto a 0.25° spatial grid by fitting in-swath wind measurements to a quadratic surface using locally weighted regression [“loess” smoothing; Cleveland and Devlin (1988)] with a half span of 80 km. Based on the filtering properties of the loess smoother (Schlax et al. 2001; Chelton and Schlax 2003), the resulting gridded fields have a spatial resolution of approximately 50 km. QuikSCAT ENWs are collocated spatially to the buoy locations through bilinear interpolation of the four closest satellite grid cells within ±30 min of the buoy observation time. Rain-contaminated observations were not used here, as determined by the standard QuikSCAT multidimensional histogram rain flag (MUDH; Huddleston and Stiles 2001) and by contemporaneous radiometer rain estimates from a combination of the Special Sensor Microwave Imager (SSM/I) F13, F14, and F15 satellites. Since QuikSCAT is in a sun-synchronous polar orbit, it crosses the equator at nearly the same local solar time each day: ascending orbits (traversing south to north) cross the equator at about 0600 LST and descending orbits at about 1800 LST, and these times vary relatively little over the duration of the QuikSCAT mission.
The accuracy of QuikSCAT ENW measurements has been quantified previously in terms of vector component errors in the alongwind and crosswind directions (Freilich and Dunbar 1999); for QuikSCAT, these accuracies have been determined to be 0.75 m s−1 in the alongwind component and 1.5 m s−1 in the crosswind component, with a total ENW accuracy of about 1.7 m s−1 (Chelton and Freilich 2005). The wind direction accuracy at low ENWs increases rapidly with increasing ENW, and is about 14° for ENWs greater than 6 m s−1. Additionally, there is no evidence of systematic SST-dependent errors in the ENW from SeaWinds on QuikSCAT (e.g., Ebuchi et al. 2002) or from a nearly identical SeaWinds scatterometer on the Advanced Earth Observing Satellite 2 (ADEOS-II; Ebuchi 2006).
One objective of this analysis is to compare satellite measurements of the coupling between surface winds and SST with moored buoy measurements. It is thus of interest to assess the quality of the QuikSCAT ENW near SST frontal zones, as wind measurements from moored buoys are the primary means for determining the accuracy of satellite surface winds. Comparison statistics of collocated QuikSCAT and buoy V10n measurements are shown in Table 4. For all buoys, the correlation coefficients between the collocated QuikSCAT and buoy V10n measurements range from 0.83 to 0.97, with the overall correlation of 0.96 over the North Atlantic and 0.91 over the equatorial Pacific. The RMS differences are between 0.73 and 1.42 m s−1, with overall RMS differences of 1.22 m s−1 over the North Atlantic and 0.88 m s−1 over the equatorial Pacific. These correlations and RMS differences are consistent with previous evaluations of the QuikSCAT winds using moored buoys (e.g., Ebuchi et al. 2002; Pickett et al. 2003; Chelton and Freilich 2005; Portabella and Stoffelen 2009). QuikSCAT winds near land have been found to be more uncertain (Pickett et al. 2003), which may explain the higher RMS differences for the Gulf Stream buoys. The biases in V10n (defined as satellite minus buoy) are between −0.13 and +0.47 m s−1 over the North Atlantic, while over the equatorial Pacific, they are about −0.2 m s−1 for the buoys along the equator and 2°S and between −0.30 and −0.63 m s−1 along 2°N. Overall biases are 0.13 m s−1 over the North Atlantic and −0.38 m s−1 over the equatorial Pacific. The consistent negative biases in V10n over the eastern equatorial Pacific are very similar to other satellite wind comparisons from the European Remote Sensing Satellite-1 and -2 (ERS-1 and -2) scatterometers (Quilfen et al. 2001) and from passive microwave imagers (e.g., Plate 1 in Mears et al. 2001). Additionally, as shown in the next subsection, there are very similar biases in the AMSR-E wind measurements in this region.
Satellite–buoy comparison statistics of the ENW V10n from QuikSCAT and AMSR-E, and of the SST Ts from AMSR-E at each individual buoy location, including the number of concurrent observations N used in the satellite–buoy comparisons, correlation coefficient, RMS difference, and bias. The biases were computed as satellite minus buoy. Units for the V10n and Ts bias and RMS statistics are m s−1 and °C, respectively.
As summarized in Ross et al. (1985), there are essentially three different factors that can contribute to differences in wind measurements from satellites and in situ platforms: “These effects are 1) actual errors in the measurements of the winds by the satellite, 2) actual errors in the measurement of the winds by the anemometers, and 3) real differences between the two measurements that result from differing averaging times, the finite areas sampled by the radar, lack of homogeneity within the footprint, and the lack of space-time coincidence of the two measurements.” Besides these sources of uncertainty, there are at least two others that can contribute to wind measurement differences between moored buoys and satellites. The first effect is the relative motion between the air–sea interface caused by surface ocean currents (e.g., Kelly et al. 2001, 2005; Park et al. 2006; Kara et al. 2007; Liu et al. 2007). Over the tropical Pacific, Kelly et al. (2005) showed that differences in scatterometer and moored buoy winds were well correlated with estimates and direct observations of near-surface ocean currents, particularly between 2°S and 2°N where surface currents are typically strong. This can account for the consistent biases of V10n in the equatorial Pacific from the various satellite datasets as described above. The second effect is the state of the surface wave field, which decouples the surface wind from the surface wind stress (e.g., Geernaert 1990). Sea state effects are usually not taken into account when converting the actual buoy wind speed to ENW, but are naturally part of the surface roughness that scatterometers measure. Buoy wind measurements have been observed to be affected by surface wave distortion of the near-surface wind profile by a factor of about 40% for wind speeds exceeding 10 m s−1 (e.g., Large et al. 1995). Since most of the buoys used here do not routinely measure ocean surface currents or the wave state, these effects are usually unavoidable sources of uncertainty in wind comparisons between buoys and satellites.
2) AMSR-E wind and SST data
The satellite SST fields used in this study were measured over most of the global oceans in nonprecipitating and ice-free conditions by AMSR-E. ENW measurements made by AMSR-E are also utilized and compared with those from QuikSCAT. The AMSR-E SST and wind fields used here were also processed by Remote Sensing Systems (version 5). One of the main capabilities of AMSR-E is its ability to measure winds and SST through clouds and atmospheric aerosols, which are essentially transparent to the particular microwave frequencies used by AMSR-E. The spatial resolutions of individual AMSR-E SST and wind measurements are about 58 and 37 km, respectively, and were gridded onto the same 0.25° spatial grid as the QuikSCAT wind measurements. AMSR-E is in a sun-synchronous polar orbit, with equatorial crossing times of ascending orbits at about 1330 LST and descending orbits at about 0130 LST, and to date these times have varied little over the course of the AMSR-E mission. The AMSR-E data record used here spans the 7-yr period 1 June 2002–19 November 2009. For comparisons with the buoys, the AMSR-E SST and wind fields were bilinearly interpolated to each buoy location using gridded measurements from the four closest grid points within ±30 min. While QuikSCAT infers vector ENW from measurements by an active microwave radar, AMSR-E infers the scalar ENW using a passive multichannel microwave radiometer based on changes of surface emissivity induced by surface wind stress.
The AMSR-E and buoy SSTs are highly correlated, with cross-correlation coefficients for all buoys between 0.94 and 0.99 (Table 4). The RMS SST differences are largest for the buoys near the Gulf Stream, which are between 0.94° and 1.67°C and have an overall RMS difference of 1.36°C. These RMS uncertainties are consistent with those previously reported in this region (Hosoda et al. 2006), although they are larger than the previously reported 0.4°C by Chelton and Wentz (2005), whose results were based on global buoy–satellite comparisons. The RMS differences over the equatorial Pacific are much smaller, between 0.28° and 0.57°C with an overall RMS difference of 0.40°C, and decrease westward where the SST fronts are more diffuse on average (see Fig. 1). Over the Gulf Stream, the AMSR-E SST is biased warm relative to the buoy SST, between 0.04° and 1.15°C with an overall bias of 0.42°C for all buoys. Hosoda et al. (2006) has previously shown similar warm biases in multiyear comparisons of the AMSR-E SST measurements relative to infrared SST measurements just north of the Gulf Stream. In contrast, there is a consistent but slight cool bias in the AMSR-E SST relative to the buoy SST of −0.17°C over the equatorial Pacific.
The larger differences between the AMSR-E and buoy SST measurements in the Gulf Stream region are likely due at least in part to the relatively coarse 58-km spatial resolution of the AMSR-E 6.9-GHz footprint. Since the SST can change by as much as ~8°C across the width of the Gulf Stream, which can be as narrow as 50 km, AMSR-E will blur these sharp frontal features over a broader distance compared to the moored buoys, thus resulting in a mismatch with the buoy point measurements. In contrast, the SST gradients are usually much smaller in the equatorial Pacific, with typical SST changes of ~3°–4°C associated with TIWs over similar distances. Additionally, there are some differences between the buoy bulk SST, which is measured between 0.6- and 1-m depth, and the subskin temperature measured by AMSR-E, which represents the water temperature within the O(1 mm) attenuation depth of microwave radiation. The differences will be most apparent during the daytime in low wind speed conditions (e.g., Gentemann et al. 2004). A comparison of the wind–SST coupling between satellites and buoys that was codified for day and night and for various wind speed regimes did not show any consistent differences in the wind–SST coupling (not shown).
Biases and RMS differences of the AMSR-E V10n measurements relative to the buoy V10n are similar to those from QuikSCAT, as shown in Table 4. The largest RMS differences are in the Gulf Stream region, and nearly all of these are somewhat larger than the QuikSCAT RMS differences. Over the equatorial Pacific, the RMS differences of the AMSR-E and QuikSCAT V10n relative to the buoys are of very similar magnitude, all differing by less than 0.15 m s−1. Additionally, the biases of the QuikSCAT and AMSR-E V10n relative to the buoys are correlated. For instance, the same large negative biases in the QuikSCAT ENW for the buoys along 2°N also occur in the AMSR-E ENW. It is not presently known what causes these consistent biases in the satellite wind measurements compared to the buoy. It may be an indication of a combination of surface ocean currents, spatial variations in the surface wave field, or biases in the buoy measurements.
b. Results
Since the QuikSCAT and AMSR-E observation times differ by several hours, the AMSR-E SST was linearly interpolated to the QuikSCAT observation times. After interpolation, the satellite and buoy δV10n and δTs time series were smoothed with a 10-day running average and are denoted as before with an overbar. The coupling coefficients 
Bar chart showing the coupling coefficients 
Citation: Journal of Climate 25, 5; 10.1175/JCLI-D-11-00121.1
Comparisons of 
The 
To assess the robustness of the 
It is noteworthy that the 
To summarize the evaluation of the satellite-derived 
Mean, RMS, and normalized mean differences of 
4. Summary and conclusions
The response of the surface wind to mesoscale SST fronts associated with the Gulf Stream and equatorial Pacific cold tongue were investigated using buoy and satellite observations. This study was motivated by satellite observations and modeling studies showing large spatial variations of the surface wind stress near SST fronts, with enhanced stress over warmer SSTs and reduced stress over cooler SSTs. Two main questions were addressed in this analysis. The first was the relative roles that surface wind speed and surface layer stability play in determining the responses of the surface wind stress and 10-m equivalent neutral wind (ENW, as defined in section 2c) to SST. The second was the accuracy of the ENW–SST coupling derived from satellites. Both of these questions were addressed using buoy observations of surface wind and SST and collocated satellite observations of ENW and SST, from which the following major conclusions were reached:
In sections 2b and 2c, analysis of 17 buoy pairs near the Gulf Stream and eastern equatorial Pacific showed that the actual surface wind speed, ENW, and surface wind stress magnitude differences between two buoys in each pair satisfy linear relationships with the SST difference on time scales longer than 10 days. The slopes of these linear relationships from the buoys (denoted as
Overall, the buoy observations show that the response of the actual surface wind speed to SST on time scales longer than 10 days is not significantly different from the ENW response to SST, given that
The QuikSCAT ENW and AMSR-E SST fields accurately depict the ENW response to SST compared with that inferred from the buoy observations, as shown in section 3b and summarized by the bar chart of
These results are significant for several reasons. First, they clarify that the responses of the ENW and surface wind stress to mesoscale SST variability primarily involve a dynamical response of the actual surface wind rather than surface layer stability. The spatially high-pass-filtered satellite ENW perturbations associated with SST investigated previously can thus be attributable mainly to a response of the actual surface wind speed. This result supports, for instance, the interpretation that the SST-induced response of the scatterometer-derived ENW convergence is related closely to surface mass convergence, which appears to force vertical motion in the deep troposphere over SST frontal zones (Liu et al. 2007; Minobe et al. 2008; Tokinaga et al. 2009; Minobe et al. 2010); if the ENW response to SST were primarily a consequence of stability effects alone and not the actual surface winds, scatterometer ENW convergence would not be related to the actual surface mass convergence and thus could not directly explain the vertical motion over SST fronts through mass conservation. Second, the magnitude of the ENW response to SST derived from satellites is in fairly close agreement with that derived from the buoys, although biased slightly low, particularly over the south side of the equatorial Pacific cold tongue. Finally, the linear wind response to mesoscale SST variability is a feature common to both the satellite and buoy datasets.
Acknowledgments
I would like to thank Mark Bourassa, Dudley Chelton, Steve Esbensen, Kathie Kelly, and Walt McCall for several in-depth discussions throughout the course of this analysis. AMSR-E data are sponsored by the NASA Earth Science MEaSUREs DISCOVER Project and the AMSR-E Science Team. QuikSCAT data are sponsored by the NASA Ocean Vector Winds Science Team. Both the QuikSCAT and AMSR-E data used here were produced by Remote Sensing Systems and are available online (www.remss.com). The TAO buoy observations were provided by the TAO project office of NOAA/PMEL. Part of this research was performed while the author held a National Research Council Research Associateship Award at the Naval Research Laboratory in Monterey, California. This work was also partly supported by NASA Grant NNX11AF31G for funding of NASA’s Ocean Vector Winds Science Team activities.
REFERENCES
Bourassa, M. A., 2006: Satellite-based observations of surface turbulent stress during severe weather. Atmosphere–Ocean Interactions, Vol. 2, W. Perrie, Ed., Wessex Institute of Technology, 35–52.
Bourassa, M. A., D. G. Vincent, and W. L. Wood, 1999: A flux parameterization including the effects of capillary waves and sea state. J. Atmos. Sci., 56, 1123–1139.
Chelton, D. B., 2005: The impact of SST specification on ECMWF surface wind stress fields in the eastern tropical Pacific. J. Climate, 18, 530–550.
Chelton, D. B., and M. G. Schlax, 2003: The accuracies of smoothed sea surface height fields constructed from tandem altimeter datasets. J. Atmos. Oceanic Technol., 20, 1276–1302.
Chelton, D. B., and M. H. Freilich, 2005: Scatterometer-based assessment of 10-m wind analyses from the operational ECMWF and NCEP numerical weather prediction models. Mon. Wea. Rev., 133, 409–429.
Chelton, D. B., and F. J. Wentz, 2005: Global microwave satellite observations of sea surface temperature for numerical weather prediction and climate research. Bull. Amer. Meteor. Soc., 86, 1097–1115.
Chelton, D. B., and S.-P. Xie, 2010: Coupled ocean–atmosphere interaction at oceanic mesoscales. Oceanogr. Mag., 4, 52–69.
Chelton, D. B., and Coauthors, 2001: Observations of coupling between surface wind stress and sea surface temperature in the eastern tropical Pacific. J. Climate, 14, 1479–1498.
Chelton, D. B., M. G. Schlax, M. H. Freilich, and R. F. Milliff, 2004: Satellite measurements reveal persistent small-scale features in ocean winds. Science, 303, 978–983.
Chelton, D. B., M. G. Schlax, and R. M. Samelson, 2007: Summertime coupling between sea surface temperature and wind stress in the California Current System. J. Phys. Oceanogr., 37, 495–517.
Cleveland, W. S., and S. J. Devlin, 1988: Locally weighted regression: An approach to regression analysis by local fitting. J. Amer. Stat. Assoc., 83, 596–610.
Ebuchi, N., 2006: Evaluation of marine surface winds observed by SeaWinds and AMSR on ADEOS-II. J. Oceanogr., 62, 293–301.
Ebuchi, N., H. C. Graber, and M. J. Caruso, 2002: Evaluation of wind vectors observed by QuikSCAT/SeaWinds using ocean buoy data. J. Oceanic Atmos. Technol., 19, 2049–2062.
Fairall, C. W., E. F. Bradley, J. E. Hare, A. A. Grachev, and J. B. Edson, 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate, 16, 571–591.
Freilich, M. H., and R. S. Dunbar, 1999: The accuracy of the NSCAT 1 vector winds: Comparisons with National Data Buoy Center buoys. J. Geophys. Res., 104, 11 231–11 246.
Garratt, J. R., 1992: The Atmospheric Boundary Layer. 1st ed. Cambridge University Press, 316 pp.
Geernaert, G., 1990: Bulk parameterizations for the wind stress and heat flux. Current Theory, Vol. 1, Surface Waves and Fluxes: Theory and Remote Sensing, G. Geernaert and W. Plant, Eds., Kluwer Academic, 91–172.
Geernaert, G., and S. Larsen, 1993: On the role of humidity in estimating marine surface layer stratification and scatterometer cross section. J. Geophys. Res., 98, 927–932.
Gentemann, C. L., F. J. Wentz, C. A. Mears, and D. K. Smith, 2004: In situ validation of Tropical Rainfall Measuring Mission microwave sea surface temperatures. J. Geophys. Res., 109, C04021, doi:10.1029/2003JC002092.
Haack, T., D. Chelton, J. Pullen, J. D. Doyle, and M. Schlax, 2008: Summertime influence of SST on surface wind stress off the U.S. west coast from the U.S. Navy COAMPS model. J. Phys. Oceanogr., 38, 2414–2437.
Hayes, S. P., M. J. McPhaden, and J. M. Wallace, 1989: The influence of sea-surface temperature on surface wind in the eastern equatorial Pacific. J. Climate, 2, 1500–1506.
Hogg, A. C. M., W. K. Dewar, P. Berloff, S. Kravtsov, and D. K. Hutchinson, 2009: The effects of mesoscale ocean–atmosphere coupling on the large-scale ocean circulation. J. Climate, 22, 4066–4082.
Hosoda, K., H. Murakami, F. Sakaida, and H. Kawamura, 2006: Difference characteristics of sea surface temperature observed by GLI and AMSR aboard ADEOS-II. J. Oceanogr., 62, 339–350.
Huddleston, J. N., and B. W. Stiles, 2001: Multidimensional Histogram (MUDH) rain flag product description, version 2.1. Jet Propulsion Laboratory (JPL) Tech. Rep., 8 pp. [Available online at http://podaac.jpl.nasa.gov/quikscat/qscat_doc.html.]
Jin, X., C. Dong, J. Kurian, J. C. McWilliams, D. B. Chelton, and Z. Li, 2009: Wind–SST interaction in coastal upwelling: Oceanic simulation with empirical coupling. J. Phys. Oceanogr., 39, 2957–2970.
Kara, A. B., E. J. Metzger, and M. A. Bourassa, 2007: Ocean current and wave effects on wind stress drag coefficient over the global ocean. Geophys. Res. Lett., 34, L01604, doi:10.1029/2006GL027849.
Kara, A. B., A. J. Wallcraft, and M. A. Bourassa, 2008: Air-sea stability effects on the 10 m winds over the global ocean: Evaluations of air–sea flux algorithms. J. Geophys. Res., 113, C04009, doi:10.1029/2007JC004324.
Kelly, K. A., S. Dickinson, M. J. McPhaden, and G. C. Johnson, 2001: Ocean currents evident in satellite wind data. Geophys. Res. Lett., 28, 2469–2472.
Kelly, K. A., S. Dickinson, and G. C. Johnson, 2005: Comparisons of scatterometer and TAO winds reveal time-varying surface currents for the tropical Pacific Ocean. J. Atmos. Oceanic Technol., 22, 735–745.
Kuwano-Yoshida, A., S. Minobe, and S.-P. Xie, 2010: Precipitation response to the Gulf Stream in an atmospheric GCM. J. Climate, 23, 3676–3698.
Large, W. G., J. C. McWilliams, and S. C. Doney, 1994: Oceanic vertical mixing: A review and a model with a nonlocal boundary layer parameterization. Rev. Geophys., 32, 363–403.
Large, W. G., J. Morzel, and G. B. Crawford, 1995: Accounting for surface wave distortion of the marine wind profile in low-level Ocean Storms wind measurements. J. Phys. Oceanogr., 25, 2959–2971.
Liu, W. T., 1990: Remote sensing of surface turbulence heat flux. Remote Sensing, Vol. 2, Surface Waves and Fluxes, G. L. Geernaert and W. J. Plant, Eds., Kluwer Academic, 293–309.
Liu, W. T., and W. Tang, 1996: Equivalent neutral wind. JPL Publ. 96-17, 16 pp.
Liu, W. T., and X. Xie, 2008: Ocean–atmosphere momentum coupling in the Kuroshio extension observed from space. J. Oceanogr., 64, 631–637.
Liu, W. T., K. B. Katsaros, and J. A. Businger, 1979: Bulk parameterization of air–sea exchanges of heat and water vapor including the molecular constraints at the interface. J. Atmos. Sci., 36, 1722–1735.
Liu, W. T., X. Xie, and P. Niiler, 2007: Ocean–atmosphere interaction over Agulhas extension meanders. J. Climate, 20, 5784–5797.
Maloney, E. D., and D. B. Chelton, 2006: An assessment of the sea surface temperature influence on surface wind stress in numerical weather prediction and climate models. J. Climate, 19, 2743–2762.
Mears, C. A., D. K. Smith, and F. J. Wentz, 2001: Comparison of Special Sensor Microwave Imager and buoy-measured wind speeds from 1987 to 1997. J. Geophys. Res., 106, 11 719–11 729.
Milliff, R. F., W. G. Large, W. R. Holland, and J. C. McWilliams, 1996: The general circulation responses of high-resolution North Atlantic ocean models to synthetic scatterometer winds. J. Phys. Oceanogr., 26, 1747–1768.
Minobe, S., A. Kuwano-Yoshida, N. Komori, S.-P. Xie, and R. J. Small, 2008: Influence of the Gulf Stream on the troposphere. Nature, 452, 206–209.
Minobe, S., M. Miyashita, A. Kuwano-Yoshida, H. Tokinaga, and S.-P. Xie, 2010: Atmospheric response to the Gulf Stream: Seasonal variations. J. Climate, 23, 3699–3719.
O’Neill, L. W., D. B. Chelton, and S. K. Esbensen, 2003: Observations of SST-induced perturbations on the wind stress field over the Southern Ocean on seasonal timescales. J. Climate, 16, 2340–2354.
O’Neill, L. W., D. B. Chelton, S. K. Esbensen, and F. J. Wentz, 2005: High-resolution satellite measurements of the atmospheric boundary layer response to SST perturbations over the Agulhas Return Current. J. Climate, 18, 2706–2723.
O’Neill, L. W., D. B. Chelton, and S. K. Esbensen, 2010a: The effects of SST-induced surface wind speed and direction gradients on midlatitude surface vorticity and divergence. J. Climate, 23, 255–281.
O’Neill, L. W., S. K. Esbensen, N. Thum, R. M. Samelson, and D. B. Chelton, 2010b: Dynamical analysis of the boundary layer and surface wind responses to mesoscale SST perturbations. J. Climate, 23, 559–581.
Park, K.-A., P. C. Cornillon, and D. L. Codiga, 2006: Modification of surface winds near ocean fronts: Effects of Gulf Stream rings on scatterometer (QuikSCAT, NSCAT) wind observations. J. Geophys. Res., 111, C03021, doi:10.1029/2005JC003016.
Pickett, M. H., W. Tang, L. K. Rosenfeld, and C. H. Wash, 2003: QuikSCAT satellite comparisons with nearshore buoy wind data off the U.S. west coast. J. Atmos. Oceanic Technol., 20, 1869–1879.
Portabella, M., and A. Stoffelen, 2009: On scatterometer ocean stress. J. Atmos. Oceanic Technol., 26, 368–382.
Quilfen, Y., B. Chapron, and D. Vandemark, 2001: The ERS scatterometer wind measurement accuracy: Evidence of seasonal and regional biases. J. Atmos. Oceanic Technol., 18, 1684–1697.
Risien, C. M., and D. B. Chelton, 2008: A global climatology of surface wind and wind stress fields from eight years of QuikSCAT scatterometer data. J. Phys. Oceanogr., 38, 2379–2413.
Ross, D. B., V. J. Cardone, J. Overland, R. D. McPherson, W. J. Pierson Jr., and T. Yu, 1985: Oceanic surface winds. Advances in Geophysics, Vol. 27, Academic Press, 101–140.
Sampe, T., and S.-P. Xie, 2007: Mapping high sea winds from space. Bull. Amer. Meteor. Soc., 88, 1965–1978.
Schlax, M. G., D. B. Chelton, and M. H. Freilich, 2001: Sampling errors in wind fields constructed from single and tandem scatterometer datasets. J. Atmos. Oceanic Technol., 18, 1014–1036.
Seo, H., A. J. Miller, and J. O. Roads, 2007: The Scripps Coupled Ocean–Atmosphere Regional (SCOAR) model, with applications in the eastern Pacific sector. J. Climate, 20, 381–402.
Small, R. J., S.-P. Xie, and Y. Wang, 2003: Numerical simulation of atmospheric response to Pacific tropical instability waves. J. Climate, 16, 3722–3740.
Small, R. J., S.-P. Xie, Y. Wang, S. K. Esbensen, and D. Vickers, 2005: Numerical simulation of boundary layer structure and cross-equatorial flow in the eastern Pacific. J. Atmos. Sci., 62, 1812–1830.
Small, R. J., and Coauthors, 2008: Air–sea interaction over ocean fronts and eddies. Dyn. Atmos. Oceans, 45, 274–319, doi:10.1016/j.dynatmoce.2008.01.001.
Song, Q., P. Cornillon, and T. Hara, 2006: Surface wind response to oceanic fronts. J. Geophys. Res., 111, C12006, doi:10.1029/2006JC003680.
Song, Q., D. B. Chelton, S. K. Esbensen, N. Thum, and L. W. O’Neill, 2009: Coupling between sea surface temperature and low-level winds in mesoscale numerical models. J. Climate, 22, 146–164.
Spall, M. A., 2007: Effect of sea surface temperature–wind stress coupling on baroclinic instability in the ocean. J. Phys. Oceanogr., 37, 1092–1097.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. 1st ed. Kluwer Academic, 666 pp.
Tokinaga, H., Y. Tanimoto, and S.-P. Xie, 2005: SST-induced surface wind variations over the Brazil–Malvinas confluence: Satellite and in situ observations. J. Climate, 18, 3470–3482.
Tokinaga, H., Y. Tanimoto, S.-P. Xie, T. Sampe, H. Tomita, and H. Ichikawa, 2009: Ocean frontal effects on the vertical development of clouds over the northwest Pacific. J. Climate, 22, 4241–4260.
Vecchi, G. A., S.-P. Xie, and A. S. Fischer, 2004: Ocean–atmosphere covariability in the western Arabian Sea. J. Climate, 17, 1213–1224.
Wai, M., and S. A. Stage, 1989: Dynamical analysis of marine atmospheric boundary layer structure near the Gulf Stream oceanic front. Quart. J. Roy. Meteor. Soc., 115, 29–44.
Warner, T. T., M. N. Lakhtakia, J. D. Doyle, and R. A. Pearson, 1990: Marine atmospheric boundary layer circulations forced by Gulf Stream sea surface temperature gradients. Mon. Wea. Rev., 118, 309–323.
Weissman, D. E., K. L. Davidson, R. A. Brown, C. A. Friehe, and F. Li, 1994: The relationship between microwave radar cross-section and both wind speed and stress: Model function studies using Frontal Air–Sea Interaction Experiment data. J. Geophys. Res., 99, 10 087–10 108.
White, W. B., and J. L. Annis, 2003: Coupling of extratropical mesoscale eddies in the ocean to westerly winds in the atmospheric boundary layer. J. Phys. Oceanogr., 33, 1095–1107.
Xie, S.-P., M. Ishiwatari, H. Hashizume, and K. Takeuchi, 1998: Coupled ocean–atmosphere waves on the equatorial front. Geophys. Res. Lett., 25, 3863–3866.
