In situ Measurements of Momentum Fluxes in Typhoons - AMS Journals

104

JOURNAL OF THE ATMOSPHERIC SCIENCES

VOLUME 72

In situ Measurements of Momentum Fluxes in Typhoons HENRY POTTER,* HANS C. GRABER, NEIL J. WILLIAMS, CLARENCE O. COLLINS III,1 RAFAEL J. RAMOS,# AND WILLIAM M. DRENNAN Rosenstiel School of Marine and Atmospheric Science, University of Miami, Miami, Florida (Manuscript received 7 February 2014, in final form 8 September 2014) ABSTRACT One of the scientific objectives of the U.S. Office of Naval Research–sponsored Impact of Typhoons on the Ocean in the Pacific (ITOP) campaign was improved understanding of air–sea fluxes at high wind speeds. Here the authors present the first-ever direct measurements of momentum fluxes recorded in typhoons near the surface. Data were collected from a moored buoy over 3 months during the 2010 Pacific typhoon season. During this period, three typhoons and a tropical storm were encountered. Maximum 30-min sustained wind speeds above 26 m s21 were recorded. Data are presented for 1245 h of direct flux measurements. The drag coefficient shows evidence of a rolloff at wind speeds greater than 22 m s21, which occurred during the passage of a single typhoon. This result is in agreement with other studies but occurs at a lower wind speed than previously measured. The authors conclude that this rolloff was caused by a reduction in the turbulent momentum flux at the frequency of the peak waves during strongly forced conditions.

1. Introduction The marine boundary layer is a dynamic region of Earth in which the ocean constantly interacts with the atmosphere. This interaction facilitates the exchange of momentum, mass, and heat between these fluids through turbulent processes. Tropical cyclones gain their energy from this interaction by extracting heat from the underlying ocean through enthalpy fluxes (Riehl 1950) and lose energy from wind stress on the surface of the water (Chen et al. 2007). Consequently, at high wind speeds, turbulent momentum and enthalpy fluxes are responsible for the genesis, conservation, and dissipation of tropical cyclones (Malkus and Riehl 1960; Emanuel 1986). The fluxes of momentum and enthalpy across the air–sea interface are typically represented in

* Current affiliation: Remote Sensing Division, Naval Research Laboratory, Washington, DC. 1 Current affiliation: Oceanography Division, Naval Research Laboratory, Stennis Space Center, Mississippi. # Current affiliation: Woods Hole Group–Houston, Stafford, Texas.

Corresponding author address: Henry Potter, Naval Research Laboratory, 4555 Overlook Ave. SW, Washington, DC 20375. E-mail: [email protected] DOI: 10.1175/JAS-D-14-0025.1 Ó 2015 American Meteorological Society

terms of nondimensional bulk transfer coefficients for drag CD and enthalpy CK. Ooyama (1969) and Emanuel (1986) both indicated the importance of CD and CK, and the latter theorized that the intensity of a tropical cyclone is proportional to the ratio of these bulk transfer coefficients, CK/CD. Comparing observations to results obtained from a simple axisymmetric model with idealized environmental conditions led Emanuel (1995) to hypothesize that the most likely range of CK/CD during a tropical cyclone is 1.2–1.5, with a lower bound of 0.75. Few campaigns have set out to directly measure fluxes in tropical cyclone conditions; thus, the data available to determine the value of these bulk transfer coefficients is limited. For wind speeds between approximately 5 and 20 m s21, there is general agreement as to the behavior of CK and CD. However, at high wind speeds, where measurements are made increasingly difficult because of harsh environmental conditions, data are scarce and have a large amount of scatter. The dedicated campaigns that have investigated enthalpy fluxes [e.g., Humidity Exchange Over the Sea (HEXOS; DeCosmo et al. 1996), GasEx (McGillis et al. 2004), and Surface Wave Dynamics Experiment (SWADE; Katsaros et al. 1993)] have general agreement that CK has no dependence on wind speeds between 5 and 20 m s21. This result was extended to 30 m s21 by the Coupled Boundary Layer Air–Sea Transfer (CBLAST) field

JANUARY 2015

105

POTTER ET AL.

program, during which the first direct measurements of enthalpy fluxes in the hurricane boundary layer were recorded using measurements from aircraft (Zhang et al. 2008a). Laboratory experiments reported by Haus et al. (2010) and Jeong et al. (2012) extend the findings to 40 m s21. Using a total energy budget approach, CBLAST data were also used to deduce CK for wind speeds between 52 and 72 m s21. These results suggest that it is probable that the magnitude of CK is not dependent on wind speed in major hurricane conditions (Bell et al. 2012). For wind speeds between 5 and 20 m s21, it has been shown by multiple campaigns that CD increases linearly with wind speed [e.g., Adverse Weather Experiment (AWE; Drennan and Shay 2006), HEXOS (Smith et al. 1992), SWADE (Donelan et al. 1997), Large and Pond (1981), Smith (1980)]. Outputs from the Coupled Ocean–Atmosphere Response Experiment (COARE) 3.5 algorithm (Edson et al. 2013) have a roughly linear wind speed dependence on CD under neutral conditions for wind speeds up to 25 m s21. The few measurements of momentum fluxes at or approaching tropical cyclone force winds indicate that CD may reach saturation or even decrease at higher wind speeds. Powell et al. (2003) used GPS dropsonde profiles of wind speeds extrapolated to the surface to determine CD. This study was the first to report a leveling off, or possible decrease, of CD at wind speeds above hurricane force (33 m s21). Further evidence was put forth by Donelan et al. (2004), who, from measurements made in a wind wave tank, also supported a saturation of CD at wind speeds above 33 m s21. As part of the CBLAST campaign, French et al. (2007) reported the first open-ocean eddy covariance measurements in hurricanes. This study determined the drag coefficient using a Rosemount 858Y probe and a Best Aircraft Turbulence gust probe mounted on a boom on an aircraft that obtained high-frequency measurements of pressure distribution and temperature. These measurements were used to determine three-dimensional wind velocities during a series of stepped descents within the boundary layer that were extrapolated to the surface. They found that, for wind speeds greater than 22 m s21 up to 30 m s21, CD begins to level off or decrease. However, because of the highly variable nature of the measurements and the limited number of individual flux estimates, they were not able to provide a definitive description of the behavior of CD at tropical storm–force winds. CBLAST data were also used by Bell et al. (2012), who deduced the momentum exchange for wind speeds between 52 and 72 m s21 using the conservation of azimuthally averaged angular momentum. They estimated that CD does not continue to increase beyond

about 30 m s21 wind speeds, but because of uncertainties in the estimate, could not rule out entirely the possibility of some continued increase. Wada and Kohno (2012) also reported a leveling of CD when applying the roughness length scheme of Taylor and Yelland (2001) to their numerical simulation of Typhoon Fanapi. Evidence is mounting that shows CD plateaus or decreases at high wind speeds. Even beyond the rolloff limits of Donelan et al. (2004) and Powell et al. (2003), CK/CD remains around 0.5, much lower than the 0.75 threshold suggested by Emanuel (1995). Zhang et al. (2008a), combining their results with French et al. (2007), found the mean value of CK/CD 5 0.63. Bell et al. (2012), using a budget analysis, found CK/CD likely to be less than 1.0 and perhaps as low as 0.4 for wind speeds to 72 m s21. The significant variability of this ratio highlights the uncertainty in our understanding of the drag coefficient at high wind speeds. Much of this uncertainty is due to the lack of direct flux measurements made at the air–sea interface during tropical cyclones. To ascertain the behavior of CD in tropical cyclones, there remains an imperative need for the direct measurement of momentum fluxes at the air–sea interface. One of the objectives of the Impact of Typhoons on the Ocean in the Pacific (ITOP) campaign was to make flux measurements in typhoon conditions to understand the behavior of bulk exchange coefficients at high wind speeds. This was achieved using moored surface buoys deployed for approximately 3 months during the 2010 Pacific typhoon season. The work presented in this paper focuses on momentum fluxes collected at the air–sea interface during the ITOP campaign. In section 2, we present the measurement theory. Section 3 is a discussion of the experiment, instruments, and data processing. Results are shown in section 4, and section 5 is a discussion. Conclusions are reserved for section 6.

2. Theory In stationary and homogeneous conditions, the momentum flux t is assumed to be constant within the surface flux layer. Using the eddy covariance method, the momentum flux is calculated above the viscous sublayer from high-frequency measurements of the wind velocity components: t 5 r[(2u0 w0 )i 1 (2y 0 w0 )j].

(1)

Here, r is the air density, i and j are unit vectors along and perpendicular to the mean wind direction; and u, y, and w correspond to horizontal-downwind, horizontalcrosswind, and vertical components of wind velocity, respectively. The overbar represents a time average

106


VOLUME 72

[O(30) min] and primes denote fluctuating components such that u0 5 y 0 5 w0 5 0. The momentum flux can also be represented in terms of the friction velocity: u* 5 jt/rj1/2 . The friction velocity was introduced by Monin and Obukhov (1954) as a turbulent scaling parameter. Monin–Obukhov similarity theory states that the mean gradient of wind speed is related to the friction velocity through a universal dimensionless gradient function uu

Because of the isolation and harsh environmental conditions encountered, measuring fluxes at sea is very challenging and expensive, especially at high wind speeds. As such, it is common for the momentum flux to be determined instead by using the nondimensional drag coefficient and the mean wind speed. The drag coefficient is typically expressed in terms of 10-m neutral equivalent conditions U10N and formulated as

›U u* 5 u (z) , ›z kz u

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 (2u0 w0 ) 1 (2y 0 w0 )

(2) CD10N 5

where U is the mean wind speed; z is height; k is the von Kármán constant (’0.4); and z 5 z/L, where L is the Obukhov length (Obukhov 1946). The variable L accounts for the effects of buoyancy on the wind profile and is given by u3 uy . L52 * kg(w0 u0y )

(3)

Here g 5 9.8 m s22 is acceleration due to gravity, uy is the mean virtual temperature, and w0 u0y is the virtual temperature flux. In magnitude, L represents the height near the surface at which shear and buoyancy turbulence are equal. The value of L is positive for stable conditions and negative for unstable conditions. Integrating Eq. (2), the velocity profile becomes u* z 2 cu (z) . Uz 2 U0 5 ln z0 k

(4)

Here Uz refers to mean wind speed at height z, U0 is mean wind speed at the surface (typically set equal to zero), and z0 is the surface roughness length for momentum. Physically, z0 is a measure of the upwind terrain roughness experienced by the surface wind and is the height where, theoretically, U 2 U0 goes to zero. Also cu (z) is the integrated form of uu (z): cu (z) 5

ðz

1 2 uu (z) dz . z z0

(5)

In neutral conditions where buoyant forcing becomes negligible, jz/Lj goes to zero, whereby the last term in Eq. (4) becomes zero, yielding the classic logarithmic mean wind profile. Although Monin and Obukhov (1954) predicted the existence of the gradient functions, they did not offer any suggestions as to their form, believing that they must be determined experimentally. For this study, we have applied the terms for stability correction provided by Donelan (1990).

2 U10N

5

u2* . 2 U10N

(6)

3. Experiment overview The ITOP campaign took place between August and December 2010. During this time, two pairs of fully instrumented buoys were deployed in the Philippine Sea at 19.638N, 127.258E and 21.238N, 126.968E, referred to as ‘‘South’’ and ‘‘North,’’ respectively. The buoys were approximately 180 km apart and 750 km east of Taiwan. Each pair consisted of an Extreme Air–Sea Interaction (EASI) buoy (Drennan et al. 2014) and an Air–Sea Interaction Spar (ASIS) buoy, (Graber et al. 2000). The buoys were deployed in about 5500 m of water with EASI anchored to the seabed by an approximately 3100-kg cast iron anchor attached to about 7000 m of line. This allowed each EASI to move relatively unhindered on the surface but remain restricted to a circular region of about 5-km radius. ASIS was attached to EASI by a floating 12-mm wire rope tether with a horizontal separation of about 60 m. The buoys were deployed from the Research Vessel (R/V) Roger Revelle in early August 2010. On 17 September [day of year (DOY) 260], during Typhoon Fanapi, the northernmost ASIS broke free from its tether and had to be recovered. Likewise, on 22 October (DOY 295), during Typhoon Megi, the southernmost ASIS broke free of its tether and also had to be recovered. Because of inclement environmental conditions, neither of the EASI buoys was able to be recovered in November 2010, as planned. They were subsequently retrieved by the R/V Roger Revelle during a return cruise in March 2011. By the time of recovery, the EASI buoys’ data acquisition systems had exhausted their power and memory. Because of the harsh conditions encountered during ITOP, multiple instruments failed or were damaged during the campaign. Following a rigorous quality-control procedure, the high-frequency wind speed measurements collected on EASI-South were deemed unreliable. As such, they are not considered in the following analysis, and all further discussion is only related to data collected on EASI-North.

JANUARY 2015

107

POTTER ET AL.

a. Instruments EASI is an adaptation of the Navy Oceanographic Meteorological Automatic Device (NOMAD) buoy. The NOMAD 6-m hull was originally designed in the 1940s for the U.S. Navy’s offshore data collection program but has since been used by the National Data Buoy Center (NDBC), among others. Figure 1 shows EASI deployed during ITOP. EASI was equipped to measure air–sea fluxes of momentum, heat, and mass, as well as mean meteorological and oceanographic parameters. A complete list of the instruments attached to EASI during ITOP is shown in Table 1, and an exhaustive description of EASI buoy and mooring design can be found at Drennan et al. (2014). The momentum flux package mounted on EASI consisted of two Gill R2A sonic anemometers and a K-Gill (e.g., Katsaros et al. 1993). Each sonic anemometer was installed on a separate mast, one fore and one aft, and was connected to one of the dual, redundant data acquisition systems. The fore mast sonic anemometer was positioned 5 m above mean sea level and recorded at 5 Hz; the aft mast anemometer was stationed at 5.45 m above sea mean level and recorded at 20 Hz. Both were positioned to minimize the effects of flow distortion. Because of the harsh conditions encountered, the fore mast R2A failed during the experiment, likely because of spray or rain penetration damaging the electronics. The K-Gill failed when it lost a blade. The rear-mast R2A operated successfully for over 4 months. Two full-motion packages, each consisting of three orthogonally mounted rate gyros (Systron Donner models QRS11 and SDG1000), a triaxis linear accelerometer (Columbia Research Laboratory model SA-307HPTX), and a compass (Precision Navigation TCM-2), were installed in EASI’s fore and aft hulls. Recording the motion of EASI served two purposes. First, it enabled EASI to operate as a single-point triplet surface follower

FIG. 1. EASI deployed in the Philippine Sea during the 2010 ITOP campaign. The bow is at the left.

wave buoy. All 6 degrees of freedom were recorded in local buoy reference frame, along with compass heading. The heave was tilt corrected following Anctil et al. (1994) and double integrated to produce sea surface elevation. Significant wave height Hs was calculated as 4 times the square root of the integral of surface elevation spectrum. The compass was used to transform pitch and roll to an Earth-fixed coordinate frame and integrated to produce north–south and east–west sea surface slopes and hence wave directional information. More information about the performance of EASI as a wave buoy can be found in Collins et al. (2014a,b). Second, the motion of the buoy was used to correct the wind velocity

TABLE 1. Meteorological and oceanographic instruments installed on EASI during the ITOP campaign. Additional thermistors were affixed to EASI’s mooring line. Instrument

Quantity

Parameter

Gill R2A sonic anemometer K-Gill anemometer RM Young anemometer LI-COR LI-7500 infrared gas analyzer Rotronic humidity–temperature probe MP101A-T7 Setra 278 barometer Compact Lightweight Aerosol Spectrometer Probe Eppley precision spectral pyranometer Eppley precision infrared radiometer Campbell 107-L thermistor WaDaR temperature logger TL-HA RBR TR1000 and TR1050

2 1 1 2 2 2 1 2 2 1 2 2

u, y, w wind speed, virtual temperature u, w wind speed u, y wind speed CO2, H2O gas (one open, one close path) Relative humidity, air temperature Atmospheric pressure Marine aerosol Solar radiation Infrared radiation Buoy hull temperature Air temperature Air (one) and ocean (one) temperature

108


VOLUME 72

spectra, the magnitude of which is a unique function of the wave energy in this frequency range. Following motion correction, the wind vector was rotated so that u points into the mean wind direction and y 5 w 5 0.

b. Data processing

FIG. 2. Wind spectra for (top) u, (middle) y, and (bottom) w before (black) and after (blue) motion correction. Environmental conditions are (left) developing seas (U10N 5 14.8 m s21, HS 5 3.6 m, U10N/Cp 5 1.0) and (right) well-developed seas (U10N 5 10.8 m s21, HS 5 6.3 m, and U10N/Cp 5 0.46). Black line shows the inertial subrange. The low-frequency peaks visible in the u spectra coincide with the peak wave frequency. Peaks centered at around 0.45 Hz are due to wave-induced resonance that was inherent in all precorrected power spectra. They have a magnitude that is a unique function of the wave energy at this frequency.

by subtracting the measured platform motion from the anemometer signal, also following Anctil et al. (1994). In Fig. 2, the u, y, and w velocity spectra Suu, Syy, and Sww, before and after motion correction, are plotted against frequency for two typical runs. Each plot represents the velocity spectra for a 30-min period, and the solid black lines correspond to the inertial subrange. The left column is taken from a time when U10N 5 14.8 m s21, HS 5 3.6 m, and inverse wave age U10N/Cp 5 1.0, where Cp is the phase speed of the peak of the wave spectrum. Plots in the right column represent a run when U10N 5 10.8 m s21, HS 5 6.3 m, and U10N/Cp 5 0.46. These periods, representing developing sea and swell-dominated conditions, respectively, were chosen to illustrate the motion correction during two distinct sea states. The lowfrequency peaks visible in the u spectra coincide with the peak wave frequency fp in their respective runs. The higher energy levels in the non-motion-corrected spectra approximately between 0.3 and 0.6 Hz are due to waveinduced resonance in the motion of EASI (Drennan et al. 2014). These were present in all precorrected power

Data were collected and stored in blocks, or runs, of 60 min on a custom PC-based data acquisition system. A roughly 5-min pause in acquisition occurred approximately every 3 h while summary meteorological data and buoy location were transmitted to shore via the Advanced Research and Global Observation Satellite (ARGOS). Subsequent analysis was carried out in 30-min blocks. Following Eq. (4), mean winds were raised from their 5.45-m measured speeds to 10-m neutral equivalent values. Drennan et al. (2014) investigated the effect of EASI’s motion on wind speed due to changes in anemometer height caused by platform tilt. They found that, despite the high instantaneous pitch and roll angles, reduction in mean anemometer height was at most 1%, which had a negligible effect on measured wind speed. Another potential error discussed by Drennan et al. (2014) relates to the small-tilt-angle assumption used by motion correction algorithms such as Anctil et al. (1994). At the highest tilt angle measured during ITOP, 288, the instantaneous error in the rotation matrix was under 6%, and typically much less. Sea surface temperature (SST), measured at a 1-m depth with an RBR Inc. temperature and depth series 1000 logger (TDR1000), was only recorded until the originally scheduled recovery date. Because SST is required for converting to U10N, the conversion could only be made for the first 87 days of deployment. As such, only data from DOYs 218–305 are considered here. This amounts to 4097 runs of 30 min each. The time series of each run was analyzed for spikes in u, y, and w. Isolated spikes in the data, with values greater than four standard deviations from their respective means, were interpolated through. Any runs for which more than 0.5% of values fulfilled this criterion were discarded, accounting for the removal of 17 runs. Overall, spikes proved rare, with the highest wind speed runs averaging about 20 isolated spikes. Because of our interest in high-wind-speed fluxes, runs for which U10N is less than 5 m s21 were also removed from analysis, accounting for the removal of 1025 runs. Data quality of individual runs was assured by inspecting both the linear cumulative summations of the covariance and the cumulative integrals of cospectrum (ogives) for downwind and crosswind stress (French et al. 2007). A total of 572 runs were removed in this manner, accounting for 19% of the runs not already eliminated. Of those removed, 13% had U10N . 10 m s21 and none had U10N . 15 m s21. Many of the

JANUARY 2015

POTTER ET AL.

109

c. Environmental conditions during ITOP

FIG. 3. Data from two (top) accepted and (bottom) discarded runs. (left) Cumulative summations for the downwind covariance plotted against percentage of total run flux. It is normalized by the total covariance. (right) The downwind ogive for the same runs.

discarded runs were measured at times when the EASI buoy heading changed significantly with respect to the mean wind. Figure 3 shows a comparison between accepted (top row) and discarded (bottom row) runs for downwind portions of cumulative summations and ogives. For the accepted runs, the cumulative summations increase with a near-constant slope for the entire run reflecting stationary conditions, thereby satisfying this assumption of eddy correlation theory. Here negative values reflect transfer of momentum downward to the ocean. The normalized ogives for these runs show good agreement and reveal that the majority of the energy in the momentum fluxes occurs between 0.01 and 1.0 Hz, as expected by the universal curves (e.g., Miyake et al. 1970). The plateau at the highfrequency end of the spectrum at around 2 Hz shows that measurements made at 20 Hz are sufficient to include the energy from all turbulent scales present. Likewise, the plateau at around 0.015 Hz shows largest-scale eddies having periods of about 11 min, whereby a 30-min sampling interval is adequate to capture all the turbulent length scales present. The cumulative summations of the rejected runs appear to be nonstationary, as much of the energy is contained in abrupt bursts over short periods. There is also significant variability in the direction of turbulent transfer among the rejected runs. The inconsistencies of these runs are reflected in the ogives, which do not mimic the accepted runs’ smooth slopes and concentration of energy in the middle frequency range. Zhang et al. (2008b) showed that atmospheric rolls in hurricanes can manifest themselves through cospectral irregularities such as these. Atmospheric rolls may be at play here, but further investigation is required before this claim can be substantiated.

According to the Joint Typhoon Warning Center (JTWC; Angrove and Falvey 2010), 19 storms of various intensities formed in the western North Pacific during the 2010 season. Of these, 14 occurred during the ITOP experimental period and four tracked favorably for EASI to record boundary layer characteristics under high wind speeds. These storms included the three most intense typhoons of the season—Typhoon Fanapi, Super Typhoon Megi, and Typhoon Chaba—as well as the second most intense tropical storm of the season, Tropical Storm Dianmu. JTWC estimated maximum sustained (1-min average) wind speeds for these storms ranged from 28 m s21 for Dianmu to 82 m s21 for Megi. Dianmu formed as a tropical depression during deployment operations approximately 200 km south of EASI and propagated at about 3.5 m s21 along a clockwise path roughly equidistant from the buoy over the following 2 days before departing to the north. Dianmu passed closest to EASI on DOY 220, at which time maximum wind speeds were estimated by JTWC to be 13 m s21. Roughly 1 month following Dianmu, Fanapi approached EASI from the southeast, propagating at about 9.5 m s21. On DOY 258, EASI had its closest encounter with a tropical storm of the season when Fanapi’s radius of maximum winds (RMW) passed 12 km to EASI’s east. Fanapi had slowed to an estimated 2 m s21 propagation speed and was supporting JTWCestimated 23 m s21 winds as it passed. It then turned abruptly and departed toward the northeast. The strongest storm of the season, Super Typhoon Megi, propagated east to west at 6 m s21 and passed closest to EASI on DOY 290. At that time, EASI was 284 km from Megi’s very narrow 20-km RMW, which had JTWCestimated wind speeds of 82 m s21. Megi was closely followed by Chaba, which approached EASI from the southeast on a very similar trajectory to Fanapi. While still to EASI’s east, Chaba turned north on DOY 300, its RMW 48 km from EASI at its closest pass. At that time Chaba had JTWC-estimated maximum wind speeds of 49 m s21, which increased to 59 m s21 over the following 24 h as it dawdled northward, moving at an average pace of 3 m s21. Environmental conditions for each of these storms according to the JWTC are summarized in Table 2, and best-track estimates can be seen in Fig. 4.

4. Results Figure 5 shows U10N for the entire experiment (30-min mean). The four major storms encountered are easily identifiable from their wind speed peaks; their names are shown at the top of the figure. For all three typhoons,

110


VOLUME 72

TABLE 2. JTWC-estimated maximum wind speeds, minimum sea level pressure, radius of maximum winds, and speed of propagation for the four storms encountered during ITOP. Also shown is the DOY and distance between EASI and RMW at each storm’s closest pass. Storm name

Max wind speed (1-min avg; m s21)

Min sea level pressure (hPa)

Radius of max winds (km)

Speed of propagation (km h21)

DOY (min distance [km]) from EASI to RMW

Dianmu Fanapi Megi Chaba

28 54 82 59

982 944 903 937

28–93 28–56 19–56 28–111

6–63 5–34 5–28 6–53

220 (201) 258 (12) 290 (284) 300 (49)

30-min-average wind speeds above 15 m s21 were recorded at EASI. Maximum wind speeds for the entire experiment were recorded during Chaba on DOY 299. A 30-min-mean U10N of 26.44 m s21, 1-min sustained maximum U10N of 31.5 m s21, and maximum instantaneous wind speed reaching 40.64 m s21 were recorded during that time. Figure 6 shows CD10N as a function of U10N for the entirety of the experiment (save the rejected runs) together with a histogram of data distribution. The drag coefficient is seen to increase with increasing wind speed up to about 22 m s21, at which point CD10N rolls off and decreases slightly. This negative slope was found to be significantly different from zero (.99% confidence).

However, its onset and magnitude were also found to be sensitive to both the bin size and run averaging period of the data. For comparison, processing the data in runs of 15 rather than 30 min extended the maximum U10N to 27.22 m s21. When fit into 2 m s21 bins, as before, the results (not shown) indicate a rolloff at slightly lower wind speed, followed by plateauing of the CD10N, indicating no wind speed dependence greater than 24 m s21. Variability was also found when altering bin sizes. In Fig. 7 the drag coefficient is plotted alongside results from Powell et al. (2003), French et al. (2007), and Donelan et al. (2004) up to wind speed of 42 m s21. The Smith (1980) bulk relation, extended to 40 m s21, well beyond the wind speed range of Smith’s initial data,

FIG. 4. Best-track estimate according to the JTWC (Angrove and Falvey 2010) for the storms of interest during ITOP. EASI-North is represented by a large black dot about 750 km east of Taiwan. The port of Kaohsiung is indicated by a red star. Text color represents individual storms: Dianmu (purple), Fanapi (green), Megi (pink), and Chaba (brown). Daily storm locations at 0000 UTC are represented by black dots on storm tracks. Accompanying numbers refer to DOY and distance from EASI to storm’s RMW (in parentheses).

JANUARY 2015

POTTER ET AL.

111

FIG. 5. The 30–min-average 10-m neutral wind speeds recorded during ITOP. Peaks in wind speed occur during storms, the names of which are provided at the top.

is also included for reference. Both Powell et al. and Donelan et al. determined that the drag coefficient’s dependence on wind speed goes through a transition at about 33 m s21. Powell et al. (2003) observed a drag coefficient decrease beyond this transition (measurements were made in wind speeds above 50 m s21 but not shown here), whereas Donelan et al. (2004) determined that it plateaus, no longer exhibiting wind speed

dependence. The French et al. (2007) results, presented here in 2.5 m s21 bins, support the leveling off or decreasing drag coefficient claims of Powell et al. (2003) and Donelan et al. (2004), although at much lower wind speed. CBLAST and ITOP values of CD10N both terminate their ascents at approximately equal wind speeds. In Fig. 8 each storm track is shown with their respective U10N, HS, and CD10N. Estimated CD10N from the

FIG. 6. (top) CD10N as a function of U10N for the 2490 runs analyzed. Green dots represent average values for bins that are 2 m s21 wide, and red crosses show one standard deviation on either side of U10N and CD10N. Bins are joined by solid black line. (bottom) Histogram of data distribution in 1 m s21 bins.

112


VOLUME 72

FIG. 7. CD10N as a function of U10N from this study (black), French et al. (2007; green, 2.5 m s21 bins), Smith (1980; red solid, measured; red dashed, extended beyond original measurements), Donelan et al. (2004; blue), and Powell et al. (2003; purple). Error bars denote 95% confidence intervals of the means.

Smith (1980) bulk relation is also plotted for comparison. Peak U10N was recorded on DOY 299.6, roughly 9 h prior to when Chaba passed closest to EASI. The period extending approximately 4 h on either side of this wind speed maximum is when the high-wind-speed-CD10N rolloffs occurred. This is one of several episodes when the drag coefficient decreased significantly from Smith’s bulk values (e.g., see DOY 260.2, 290.5, and 301.3). Holthuijsen et al. (2012) showed that storm sector wave state can influence the drag coefficient and may be at play here. The effect of storm sector wave state on CD10N during ITOP will be discussed extensively elsewhere. When Chaba’s highest U10N were recorded, JTWC estimated maximum wind speeds Vmax were 46 m s21. Over the following 6 h, peak storm winds intensified and Vmax reached 49 m s21. Despite this, U10N at EASI decreased over this period, with a rapid drop from 25.5 to 20 m s21 during a single hour around DOY 299.8. As seen in Fig. 8, this abrupt U10N decrease coincided with a sudden change in Chaba’s propagation direction as it encountered steering flows from the continent. During this time, the storm, which was approaching from the southeast, swung around and headed north. As a consequence of this direction change, EASI’s location relative to the storm track moved from the right side to the left side, repositioning EASI from Chaba’s strong to its weak side. Using JWTC estimates, it was found that on DOY 299.8, when Chaba changed direction, its propagation

speed VP was approximately 2.1 m s21. On the strong side, the propagation direction and wind direction are aligned, whereby the storm’s relative wind speed is increased by VP. On the weak side, the opposite is true; the propagation speed counters the wind speed, thereby reducing the speed by VP. As such, Chaba’s VP and direction change contributed to an approximately 4.2 m s21 difference in U10N and effectively accounts for the sudden reduction observed. In Fig. 9, boundary layer characteristics recorded on EASI during Chaba are shown: U10N and HS (top); u* (second row); recorded and Smith (1980) bulk CD10N (third row); wind and wave direction at peak frequency (fourth row); and inverse wave age U10N/CP, where CP is wave speed at peak frequency (bottom). The figure is divided into six panels by vertical black lines. The solid line indicates the time when Chaba passed closest to EASI. Column I is a period of 18 h, during which Chaba approached EASI. Columns II and III are periods of 6 h each, during which the typhoon was still approaching. Of these, column III represents the period when maximum sustained winds speeds were reached and the drag coefficient rolloff is observed (see Fig. 6). Column IV is a 4-h period immediately prior to Chaba’s closest approach to EASI. During this time, a rapid reduction in wind speed occurred, as discussed above. Columns V and VI are both periods of 6 h when Chaba began to move away from EASI; U10N is approximately constant during period V and decreases during VI.

JANUARY 2015

POTTER ET AL.

113

FIG. 8. (top) Storm paths, (middle) U10N and HS, and (bottom) CD10N for the four named storms of interest in this study: (from left to right) Dianmu, Fanapi, Megi, and Chaba. Storm paths show JTWC-estimated wind speeds (Angrove and Falvey 2010). (top) Axes are degrees of longitude (y axis) and latitude (x axis). Dots on storm track show positions every 6 h and numbers refer to DOY. Large black dot shows location of EASI. (middle) Recorded U10N (black) and HS (blue). (bottom) CD10N plots show recorded values in black and theoretical values from the Smith (1980) bulk relation in red. Solid vertical lines denote the time when each storm was closest to EASI.

At the start of period I, the peak waves were aligned with the wind: both approached from the northnortheast and within about 108 of each other. During this period, the inverse wave age shows that conditions were nearing full development, although a swell peak is also visible in the wave spectra (not shown). Over 12 h from DOY 298.75 to DOY 299.25 (still period I), this swell peak became the dominant frequency, but the wind sea peak was still evident in the wave spectra. This swell approached from the southeast directly ahead of Chaba, traveling about 908 from the wind direction with cp ; 17 m s21. During this time, CD10N is in general agreement with the Smith (1980) bulk value. On DOY 299.25, with U10N at 17.5 m s21 and HS above 6 m, the wind waves became dominant. Over the following roughly 12 h (periods II and III), the waves continued to be strongly forced, and inverse wave age reached 1.5

while U10N and HS continued to increase. During these periods, CD10N reached a plateau, causing it to diverge from the Smith (1980) bulk values, which continued to increase following the wind speed. This was especially true during period III, when the maximum U10N was recorded and the CD10N rolloff was observed. During the 4 h immediately prior to EASI’s closest encounter with Chaba (period IV in Fig. 9), both U10N and u* decreased abruptly. However, CD10N remained consistent with those recorded during period III and lingered below the Smith (1980) bulk values, which dropped because of decreased wind speed. This period also exhibited a rapid reduction in inverse wave age, with the winds beginning to turn as Chaba moved to within 49 km of EASI. When Chaba began to move away from EASI (period V in Fig. 9), U10N stayed at around 20 m s21, while u* increased (discussed below).

114


VOLUME 72

FIG. 9. Boundary layer characteristics recorded during typhoon Chaba: (from top to bottom) U10N and HS; u*; recorded and Smith (1980)-estimated CD10N; wind direction and wave direction at the peak frequency; and inverse wave age U10N/CP, with level of full development indicated by the horizontal dashed line. Plots are divided into six columns by vertical lines; the solid line in each panel shows the time when Chaba passed closest to EASI. Each column is referred to in the discussion section.

This caused CD10N to increase and align with the Smith (1980) bulk values. During this period, the wind continued to turn and began approaching from the northnorthwest while the waves also began to turn, lagging behind the wind. During this period, the angle between the wind and waves increased to about 458. As Chaba continued to move away from EASI (period VI), wind and waves continued to turn, U10N and u* both continued to drop, and CD10N remained approximately equal to the Smith (1980) bulk values. During the 46 h examined in Fig. 9, turbulent stress was supported largely by the downwind component of the flux. During periods I, V, and VI, when the wind was not aligned with the peak waves, the crosswind flux contributed a mean of 22% to the total. During periods II, III, and IV, when the wind was acutely aligned with the peak waves, the crosswind flux contributed an average of just 10% to the total. These values were found to be significantly different (99% confidence level). The

mean angle between the wind and stress directions during these 46 h were 12.18 for periods I, V, and VI and 5.68 for periods II, III, and IV, and they were significantly different at greater than the 95% confidence level. While there does appear to be a clear turning of the wind stress due to the swell (cf. Grachev et al. 2003; Zhang et al. 2009), the effect is small, perhaps because of the lack of frequency separation between the wind sea and swell systems. This will be the subject of a future publication. In Fig. 10, the u–w cospectra for selected periods are plotted in the dimensionless scaling of Miyake et al. (1970). The cospectra and frequency f were normalized using mean wind speed U, measurement height z 5 5.45 m, and the friction velocity squared u2* and then averaged in equally spaced logarithmic frequency bins. The blue lines are mean values for each period, black lines are plus or minus one standard error, and green lines are the curves of Miyake et al. (1970). Also shown

JANUARY 2015

POTTER ET AL.

115

FIG. 10. Mean u–w cospectra plotted in the scaling of Miyake et al. (1970). U is mean wind speed, f is frequency (Hz), and z is measurement height (5.45 m). Plots II, III, IV, and V refer to periods displayed in Fig. 9 and show the evolution of the spectra recorded during Chaba. Blue lines are mean values, black lines are 61 standard error, and green lines are the Miyake et al. (1970) universal curve. The red lines are 1D wave frequency spectra nondimensionalized using z and U as above. The wave spectra are normalized to a maximum of 0.3. Note that panels I and VI (from Fig. 9) are not shown.

are the 1D wave frequency spectra plotted in dimensionless frequency, with their magnitude normalized to a maximum of 0.3. Figure 10 is divided according to periods II, III, IV, and V as previously defined, representing the immediate approach and departure of Chaba in relation to EASI. These periods are explored in order to understand the reason behind the drag coefficient rolloff recorded during period III. While generally adhering to the universal curve, the cospectra at II, III, and IV have dips at the peak wave frequency. This is most prevalent during period III, when the drag coefficient rolloff was observed. During period V, as Chaba moved away from EASI, this dip is gone. These cospectral dips occurred when HS was between 8 and 11 m and peak wave direction was essentially aligned with the wind. As U10N and HS grew between periods II and III, the dip became more pronounced and CD10N increasingly deviated from the Smith (1980) bulk curve. During period IV, the winds decreased by 8 m s21, the

inverse wave age dropped by 30%, HS reduced from over 10 m to less than 9 m, the waves were becoming misaligned from the wind, and the spectral dip was curbed significantly. During period V, the angle between the wind and waves continued to open, the cospectral dip was gone, and CD10N fell on top of the Smith (1980) values. A cospectral analysis of the y–w flux (not shown) found that for periods II, III, and IV the crosswind component of the flux did not resemble the universal curve of Miyake et al. (1970). During these times, the crosswind energy was concentrated in minor peaks aligned with the wave frequency. During period V, the crosswind flux was spread across all measured frequencies. Save the wave frequency dips, the u–w cospectra in Fig. 10 resemble the universal curve of Miyake et al. (1970). However, they have the appearance of being blueshifted to higher frequencies. This shift increases in prevalence from period II to period III and then decreases during period IV. After the storm has passed, the

116


blueshift has effectively gone (period V). Zhang (2010) used CBLAST data recorded in the boundary layer between outer rainbands at wind speeds above 20 m s21 to examine cospectra of momentum fluxes. He also found that cospectra fell into blueshifted universal curves, though to a greater extent than seen here. As seen in Fig. 10, the appearance of a blueshift is supported in the inertial subrange more so than a shift in peak frequency. A similar comparison to the universal curves of Kaimal et al. (1972; not shown) revealed greater agreement in the inertial subrange and provided no indication of a blueshift. No relationship was found between shift extent and typhoon location, or U10N; as such, it is not evident that the blueshift is a consequence of typhoon dynamics or intensity. It is likely that the apparent shift is an effect of the u* nondimensionalization term. Because of the presence of the wave-coherent spectral dip, u* is reduced, causing the nondimensionalized spectrum to shift toward higher energy. This idea is supported by the variation in the spectral shift in Fig. 10 from period III to period IV, which is seen to diminish with the reduction of the cospectral dip.

5. Discussion The presence of these wave-coherent dips reduced the magnitude of the friction velocity and therefore the drag coefficient. It is clear that this feature is related to the surface wave field. One possibility is that the EASI measurements are made within the wave boundary layer (WBL). The WBL is the layer above the surface where wave-coherent components of the wind field are significant. EASI measurements, made 5.45 m above the water level, were indeed well below the crests of the largest waves observed during typhoons. However, modeled results of Janssen (1989) and Makin and Mastenbroek (1996) indicate that the expected height of the WBL is well under 1 m at the friction velocities measured during the typhoons in this study. While the results of Janssen (1989) and Makin and Mastenbroek (1996) are strictly true for pure wind sea conditions, typhoon measurements here were during mixed wind sea-swell conditions. However, Drennan et al. (2005) demonstrated that WBL effects are expected to be negligible as long as the overall wave energy is dominated by wind sea (i.e., contribution of the total wave energy from wind sea is more than 50%), as was the case here. While evidence suggests that the EASI wind speed measurements were made above the WBL during the typhoons, this could only be confirmed with near-surface wind profile measurements, which were not recorded on EASI. Dropsondes, which were deployed from aircraft during the ITOP experiment, have been successfully

VOLUME 72

used to provide boundary layer wind profiles in previous campaigns (e.g., Drennan et al. 2007; Zhang et al. 2011). However, dropsondes do not provide the reliable nearsurface profile measurements that are required here. The possibility that the wave-coherent dips are spurious, somehow related to the motion of the platform on the extreme waves, can be discounted. The wave field was very similar in both period and height during periods II– IV (when the dip was present) and period V (when it was not). Hence, motion-related errors seem unlikely. Finally, we consider the idea that the wave-coherent dips are due to flow separation. Donelan et al. (2004) suggested that flow over breaking waves separates from the surface and reattaches near the crest of the preceding wave. They went on to propose that the separated flow does not ‘‘see’’ the wave troughs, thereby limiting the aerodynamic roughness of the surface and hence the drag. Here we would expect a higher degree of flow separation when the peak waves are aligned with the wind (as during II–IV when the dip was prominent) than when they are not (as during V). This is consistent with the observations. Although Donelan’s work was completed in a tank where waves are long crested, the same mechanism may be at play here where the waves are short crested [see, e.g., Fig. 9 of Drennan (2005)].

6. Conclusion Direct measurements of momentum fluxes taken from a floating platform during typhoons were presented. Measurements were made over 87 days in the Philippine Sea during the 2010 Pacific typhoon season. Over 1200 h of momentum fluxes were calculated using the eddy covariance method for wind speeds from 5 to 26.44 m s21. For wind speeds up to 20 m s21, the drag coefficient was found to agree well with previous studies. At wind speeds above about 22 m s21, there is evidence that the drag coefficient decreases or plateaus. This provides support to the previous studies of French et al. (2007), Donelan et al. (2004), and Powell et al. (2003), who also found a limit to the surface drag. We found that the drag coefficient rolloff is due to a reduction in the turbulent momentum flux at the frequency of the peak waves during strongly forced conditions when the wind and waves were closely aligned. This feature is consistent with some form of flow separation at the wave frequency. The plateau (or decrease) in drag coefficient at high wind speeds presented in this study is of critical importance for the understanding and modeling of tropical cyclones and other intense storms. Modifying tropical cyclone models to represent a reduced CD in wind speeds above about 22 m s21 will lead to higher values of CK/ CD and more intense storms. Consequently, surface

JANUARY 2015

POTTER ET AL.

layer parameterization used to predict tropical cyclone tracks, ocean mixing, and waves may be invalid, which will have crucial implications for risk assessment of landfalling storms. It has been shown that tropical storm–induced surface waves can have marked impact on surface stress. These results exemplify the need to couple waves into tropical storm prediction models, such as the work of Chen et al. (2013), who have shown that fully coupled wind–wave parameterization improves model-simulated surface wind speed and inflow angle, which are important to storm evolution and structure. Alongside modeling efforts, our work also highlights an inherent need for more concurrent measurements of momentum fluxes and waves during tropical cyclones. Our results substantiate previous claims of a limit in surface drag but stem from a single event during Typhoon Chaba. Additional measurements in tropical cyclones are needed in wind speeds above 20 m s21 to better understand the relationship between the drag coefficient and the wave field in typhoons. Acknowledgments. ITOP was funded by ONR under Grant N0014-09-1-0392. We thank this agency for their support. We also acknowledge the contributions of Mike Rebozo at RSMAS, Joe Gabriele and Cary Smith of Environment Canada, and the WHOI mooring group led by John Kemp. We are also grateful for support and assistance provided by the captains and crew of the R/V Roger Revelle. We acknowledge additional support from NSF (Grant OCE-0526442) for the development of the EASI buoy and ONR (Grant DURIP N00014-090818) for funding construction of the second EASI buoy. We also thank the three anonymous reviewers and the editor. REFERENCES Anctil, F., M. A. Donelan, W. M. Drennan, and H. C. Graber, 1994: Eddy-correlation measurements of air-sea fluxes from a discus buoy. J. Atmos. Oceanic Technol., 11, 1144–1150, doi:10.1175/ 1520-0426(1994)011,1144:ECMOAS.2.0.CO;2. Angrove, M. D., and R. J. Falvey, 2010: Annual tropical cyclone report 2010. U.S. Naval Maritime Forecast Center/Joint Typhoon Warning Center, Pearl Harbor, HI, 109 pp. [Available online at www.dtic.mil/dtic/tr/fulltext/u2/a561986.pdf.] Bell, M. M., M. T. Montgomery, and K. A. Emanuel, 2012: Air–sea enthalpy and momentum exchange at major hurricane wind speeds observed during CBLAST. J. Atmos. Sci., 69, 3197– 3222, doi:10.1175/JAS-D-11-0276.1. Chen, S. S., J. F. Price, W. Zhao, M. A. Donelan, and E. J. Walsh, 2007: The CBLAST-Hurricane program and the nextgeneration fully coupled atmosphere–wave–ocean models for hurricane research and prediction. Bull. Amer. Meteor. Soc., 88, 311–317, doi:10.1175/BAMS-88-3-311. ——, W. Zhao, M. A. Donelan, and H. L. Tolman, 2013: Directional wind–wave coupling in fully coupled atmosphere–wave–ocean

117

models: Results from CBLAST-Hurricane. J. Atmos. Sci., 70, 3198–3215, doi:10.1175/JAS-D-12-0157.1. Collins, C. O., III, B. Lund, R. Ramos, W. M. Drennan, and H. C. Graber, 2014a: Wave measurement intercomparison and platform evaluation during the ITOP (2010) experiment. J. Atmos. Oceanic Technol., 31, 2309–2329, doi:10.1175/ JTECH-D-13-00149.1. ——, ——, R. T. Waseda, and H. C. Graber, 2014b: On recording sea surface elevation with accelerometer buoys: lessons from ITOP (2010). Ocean Dyn., 64, 895–904, doi:10.1007/ s10236-014-0732-7. DeCosmo, J., K. B. Katsaros, S. D. Smith, R. J. Anderson, W. A. Oost, K. Bumke, and H. Chadwick, 1996: Air-sea exchange of water vapor and sensible heat: The Humidity Exchange Over the Sea (HEXOS) results. J. Geophys. Res., 101, 12 001– 12 012, doi:10.1029/95JC03796. Donelan, M. A., 1990: Air-sea interaction. Ocean Engineering Science, B. Le Méhauté and D. M. Hanes, Eds., The Sea— Ideas and Observations on Progress in the Study of the Seas, Vol. 9A, John Wiley and Sons, 239–292. ——, W. M. Drennan, and K. B. Katsaros, 1997: The air–sea momentum flux in conditions of wind sea and swell. J. Phys. Oceanogr., 27, 2087–2099, doi:10.1175/1520-0485(1997)027,2087: TASMFI.2.0.CO;2. ——, B. K. Haus, N. Reul, W. J. Plant, M. Stiassnie, H. C. Graber, O. B. Brown, and E. S. Saltzman, 2004: On the limiting aerodynamic roughness of the ocean in very strong winds. Geophys. Res. Lett., 31, L18306, doi:10.1029/2004GL019460. Drennan, W. M., 2005: On parameterisations of air-sea fluxes. Atmosphere-Ocean Interactions, Vol. 2, W. Perrie, Ed., Advances in Fluid Mechanics, Vol. 33, WIT Press, 1–33. ——, and L. K. Shay, 2006: On the variability of the fluxes of momentum and sensible heat. Bound.-Layer Meteor., 119, 81– 107, doi:10.1007/s10546-005-9010-z. ——, P. K. Taylor, and M. J. Yelland, 2005: Parameterizing the sea surface roughness. Phys. Oceanogr., 35, 835–848, doi:10.1175/ JPO2704.1. ——, J. A. Zhang, J. R. French, C. McCormick, and P. G. Black, 2007: Turbulent fluxes in the hurricane boundary layer. Part II: Latent heat flux. J. Atmos. Sci., 64, 1103–1115, doi:10.1175/ JAS3889.1. ——, H. C. Graber, C. O. Collins III, A. Herrera, H. Potter, R. J. Ramos, and N. J. Williams, 2014: EASI: An air–sea interaction buoy for high winds. J. Atmos. Oceanic Technol., 31, 1397– 1409, doi:10.1175/JTECH-D-13-00201.1. Edson, J. B., and Coauthors, 2013: On the exchange of momentum over the open ocean. J. Phys. Oceanogr., 43, 1589–1610, doi:10.1175/JPO-D-12-0173.1. Emanuel, K. A., 1986: An air–sea interaction theory for tropical cyclones. Part I: Steady-state maintenance. J. Atmos. Sci., 43, 585– 605, doi:10.1175/1520-0469(1986)043,0585:AASITF.2.0.CO;2. ——, 1995: Sensitivity of tropical cyclones to surface exchange coefficients and a revised steady-state model incorporating eye dynamics. J. Atmos. Sci., 52, 3969–3976, doi:10.1175/ 1520-0469(1995)052,3969:SOTCTS.2.0.CO;2. French, J. R., W. M. Drennan, J. A. Zhang, and P. G. Black, 2007: Turbulent fluxes in the hurricane boundary layer. Part I: Momentum flux. J. Atmos. Sci., 64, 1089–1102, doi:10.1175/JAS3887.1. Graber, H. C., E. A. Terray, M. A. Donelan, W. M. Drennan, J. C. Van Leer, and D. B. Peters, 2000: ASIS—A new air-sea interaction spar buoy: Design and performance at sea. J. Atmos. Oceanic Technol., 17, 708–720, doi:10.1175/ 1520-0426(2000)017,0708:AANASI.2.0.CO;2.

118


Grachev, A. A., C. W. Fairall, J. E. Hare, J. B. Edson, and S. D. Miller, 2003: Wind stress vector over ocean waves. J. Phys. Oceanogr., 33, 2408–2429, doi:10.1175/1520-0485(2003)033,2408: WSVOOW.2.0.CO;2. Haus, B. K., D. Jeong, M. A. Donelan, J. A. Zhang, and I. Savelyev, 2010: Relative rates of sea-air heat transfer and frictional drag in very high winds. Geophys. Res. Lett., 37, L07802, doi:10.1029/2009GL042206. Holthuijsen, L. H., M. D. Powell, and J. D. Pietrzak, 2012: Wind and waves in extreme hurricanes. J. Geophys. Res., 117, C09003, doi:10.1029/2012JC007983. Janssen, P. A. E. M., 1989: Wave-induced stress and the drag of air flow over sea waves. J. Phys. Oceanogr., 19, 745–754, doi:10.1175/1520-0485(1989)019,0745:WISATD.2.0.CO;2. Jeong, D., B. K. Haus, and M. A. Donelan, 2012: Enthalpy transfer across the air–water interface in high winds including spray. J. Atmos. Sci., 69, 2733–2748, doi:10.1175/JAS-D-11-0260.1. Kaimal, J. C., J. Wyngaard, Y. Izumi, and O. R. Coté, 1972: Spectral characteristics of surface-layer turbulence. Quart. J. Roy. Meteor. Soc., 98, 563–589, doi:10.1002/qj.49709841707. Katsaros, K. B., M. A. Donelan, and W. M. Drennan, 1993: Flux measurements from a SWATH ship in SWADE. J. Mar. Syst., 4, 117–132, doi:10.1016/0924-7963(93)90003-5. Large, W. G., and S. Pond, 1981: Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr., 11, 324–336, doi:10.1175/1520-0485(1981)011,0324: OOMFMI.2.0.CO;2. Makin, V. K., and C. Mastenbroek, 1996: Impact of waves on airsea exchange of sensible heat and momentum. Bound.-Layer Meteor., 79, 279–300, doi:10.1007/BF00119442. Malkus, J. S., and H. Riehl, 1960: On the dynamics and energy transformations in steady-state hurricanes. Tellus, 12A, 1–20, doi:10.1111/j.2153-3490.1960.tb01279.x. McGillis, W. R., and Coauthors, 2004: Air-sea CO2 exchange in the equatorial Pacific. J. Geophys. Res., 109, C08S02, doi:10.1029/ 2003JC002256. Miyake, M., R. Stewart, and R. W. Burling, 1970: Spectra and cospectra of turbulence over water. Quart. J. Roy. Meteor. Soc., 96, 138–143, doi:10.1002/qj.49709640714. Monin, A. S., and A. M. Obukhov, 1954: Basic laws of turbulent mixing in the surface layer of the atmosphere. Tr. Geofiz. Inst., Akad. Nauk SSSR, 151, 163–187. Obukhov, A. M., 1946: ‘Turbulentnost’ v temperaturnoj–neodnorodnoj atmosfere (Turbulence in an atmosphere with a non-uniform

VOLUME 72

temperature). Tr. Akad. Nauk. SSSR Inst. Teorel. Geofiz., 1, 95–115. Ooyama, K., 1969: Numerical simulation of the life cycle of tropical cyclones. J. Atmos. Sci., 26, 3–40, doi:10.1175/ 1520-0469(1969)026,0003:NSOTLC.2.0.CO;2. Powell, M. D., P. J. Vickery, and T. A. Reinhold, 2003: Reduced drag coefficient for high wind speeds in tropical cyclones. Nature, 422, 279–283, doi:10.1038/nature01481. Riehl, H., 1950: A model of hurricane formation. J. Appl. Phys., 21, 917–925, doi:10.1063/1.1699784. Smith, S. D., 1980: Wind stress and heat flux over the ocean in gale force winds. J. Phys. Oceanogr., 10, 709–726, doi:10.1175/ 1520-0485(1980)010,0709:WSAHFO.2.0.CO;2. ——, and Coauthors, 1992: Sea surface wind stress and drag coefficients: The HEXOS results. Bound.-Layer Meteor., 60, 109–142, doi:10.1007/BF00122064. Taylor, P. K., and M. J. Yelland, 2001: The dependence of sea surface roughness on the height and steepness of the waves. J. Phys. Oceanogr., 31, 572–590, doi:10.1175/1520-0485(2001)031,0572: TDOSSR.2.0.CO;2. Wada, A., and N. Kohno, 2012: Impact of surface roughness length on typhoon simulations. 30th Conf. on Hurricane and Tropical Meteorology, Ponte Vedra Beach, FL, Amer. Meteor. Soc., P2.22. [Available online at https://ams.confex.com/ams/30Hurricane/ webprogram/Paper204931.html.] Zhang, F. W., W. M. Drennan, B. K. Haus, and H. C. Graber, 2009: On wind-wave-current interactions during the Shoaling Waves Experiment. J. Geophys. Res., 114, C01018, doi:10.1029/ 2008JC004998. Zhang, J. A., 2010: Spectral characteristics of turbulence in the hurricane boundary layer over the ocean between the outer rain bands. Quart. J. Roy. Meteor. Soc., 136, 918–926, doi:10.1002/qj.610. ——, P. G. Black, J. R. French, and W. M. Drennan, 2008a: First direct measurements of enthalpy flux in the hurricane boundary layer: The CBLAST results. Geophys. Res. Lett., 35, L14813, doi:10.1029/2008GL034374. ——, K. B. Katsaros, P. G. Black, S. Lehner, J. R. French, and W. M. Drennan, 2008b: Effects of roll vortices on turbulent fluxes in the hurricane boundary layer. Bound.-Layer Meteor., 128, 173–189, doi:10.1007/s10546-008-9281-2. ——, R. F. Rogers, D. S. Nolan, and F. D. Marks, 2011: On the characteristic height scales of the hurricane boundary layer. Mon. Wea. Rev., 139, 2523–2535, doi:10.1175/MWR-D-10-05017.1.