U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications

Near-Surface Shear Flow in the Tropical Pacific Cold Tongue Front

M.F. Cronin and W.S. Kessler

NOAA, Pacific Marine Environmental Laboratory, Seattle, Washington, 98115

Journal of Physical Oceanography, 39, 1200–1215
Published by the American Meteorological Society. Further electronic distribution is not allowed.

3. Methods and data

After the conclusion of the Eastern Pacific Investigation of Climate Processes (EPIC) experiment (Cronin et al. 2002), five current meters were available for an opportunistic study of near-surface shear. To quantify the near-surface shear in the poleward branch of the meridional overturning cell, the study needed to be off the equator and at a longitude that had strong equatorial upwelling and a well-developed cold tongue during boreal fall. A NOAA/Pacific Marine Environmental Laboratory (PMEL) Engineering Development Division test mooring scheduled for deployment near the 2°N, 140°W Tropical Atmosphere Ocean (TAO) mooring in May 2004 satisfied these requirements and was available for this study.

The test mooring carried a wind sensor, a sea surface temperature module at 1-m depth, and five Sontek current meters that measured horizontal currents in 1-m bins at 5 m, 10 m, 15 m, 20 m, and 25 m. Sontek current meters measure currents at a 1-Hz rate for 2 min every 20 min. Speed and direction resolution are 0.1 cm s-1 and 0.1°, with nominal accuracy of ±5 cm s-1, and ±5° for daily-averaged data, although current meters above 80 m appear to be more accurate than this (Freitag et al. 2003). Each Sontek current meter was mounted with a TAO temperature module that had a thermistor 3.3 m below the center of the current meter bin (i.e., at 8.3 m, 13.3, 18.3, 23.3, and 28.3 m). The TAO sea surface temperature module has an accuracy of 0.03°C, while subsurface modules have an accuracy of 0.09°C (Freitag et al. 1994).

Linear trends between pre- and postcalibrations were applied to all temperature data. Based upon the post-calibrations and intercomparisons with the other sensors, it appeared that the 25-m current meter and the 23.3-m thermistor had unacceptable calibration drifts. Thus, the 23.3-m thermistor data are not used in this analysis. Inspection of the 25-m current meter compass pitch and roll data showed that the quality deteriorated abruptly on 8 October 2004. Thus, for analysis of the mean shear and the influence of the diurnal cycle, we focus on the period in which all five current meters were functioning (24 May–7 October 2004). Shears during this period are computed by referencing the velocity at each level to the velocity at 25 m. This was deemed more robust than taking differences between each 5-m interval. Because the 25-m Sontek malfunctioned during the period of strong tropical instability waves (TIWs), currents were referenced to 20 m for the TIW shear analysis (from 1 November 2004 to 28 February 2005).

Wind stress was estimated using the Coupled Ocean–Atmosphere Response Experiment (COARE) version 3 bulk algorithm (Fairall et al. 2003). This algorithm requires not only wind speed and direction estimates relative to the ocean surface, but also measurements of the air and sea surface temperature and surface specific humidity. Although the test mooring carried a wind sensor, it did not carry an air temperature and relative humidity sensor. Thus, wind stress was calculated using hourly data from the 2°N, 140°W TAO mooring that was 6 n mi away from the test mooring.

To estimate the temperature gradient, we use version 4 TMI SST data obtained from the Asia–Pacific Data Research Center (APDRC) data delivery site (see online at http://apdrc.soest.hawaii.edu/data/data.php). The APDRC TMI daily fields are 3-day running means on a 0.25° × 0.25° spatial grid. Because the buoy temperature at 28.3 m nearly always differs from the 1-m SST by less than 0.2°C, the top 25 m can be considered well mixed for this purpose and we assume that the TMI SST gradients represent the horizontal gradients of temperature within the top 25 m. Geostrophic thermal wind is expected to respond slowly to changes in the horizontal temperature field and the Rossby radius of deformation at 2°N is roughly 210 km. Thus, the TMI data were filtered with a 5-day low-pass filter and the horizontal gradients were computed by fitting a straight line to the SST over eight grid points (i.e., 222 km).

To compute the geostrophic component of the near-surface shear (3b), we assume that the buoyancy gradient can be estimated from the temperature gradient. However, the temperature front on the northern edge of the cold tongue is collocated with a strong salinity front separating the freshwater beneath the intertropical convergence zone from the upwelled cool, salty equatorial water (Ando and McPhaden 1997; McPhaden et al. 2008). Covariations of daily-averaged SST and sea surface salinity (SSS) measurements at the nearby 2°N, 140°W TAO buoy over the period in which all five current meters functioned had a correlation of −0.74 and a mean slope (dS/dT) of −0.018 psu K−1. Thus, with a thermal expansion coefficient α = 3.0 × 10−4 K−1, and a haline contraction coefficient of 7.4 × 10−3 psu−1 (both of which are based upon the equation of state for mean surface conditions at 2°N, 140°W), the effective thermal expansion coefficient is estimated as αeff = αβ dS/dT ∼ 4.3 × 10−4 K−1 and the buoyancy gradient in (3b) is then estimated as b = eff T. During the TIW period (1 November 2004–28 February 2005), co-variations of coincident SST and SSS daily measurements were less correlated (−0.3) and the effective thermal expansion coefficient was estimated to be 3.4 × 10−4 K−1.

With the geostrophic shear estimated from the TMI SST, the observed mean total currents relative to 25 m are decomposed into geostrophic and residual ageostrophic components for the period during which all five current meters functioned (24 May–7 October 2004). The standard error for the mean shear is estimated as the standard deviation divided by the square root of the degrees of freedom. The integral time scale for the currents relative to 25 m, estimated as the integral of the temporal autocorrelation function, was found to be about 2 days. Thus, the mean shear computed over the 136-day period had approximately 68 degrees of freedom.

The resulting ageostrophic currents relative to 25 m are then compared to simulations by the classical Ekman model (4), the frontal Ekman model (5), and the generalized Ekman model (7). A novel result of this study is that the ageostrophic Ekman currents depend, not only upon the wind, but also upon the strength and orientation of the front relative to the wind. Observations during the passage of tropical instability waves (Willett et al. 2006) offer an opportunity to test this idea since the orientation of the front varies over the course of a wave, while the orientation of the wind is relatively steady.

To compute a composite TIW, an index was created using complex demodulation on the SST for a 30-day periodicity. Complex demodulation (Bloomfield 1976) is a type of bandpass that represents a time series as a near sinusoid with time-varying amplitude and phase that is equivalent to allowing slow frequency variation. It is thus appropriate for a nearly periodic signal whose variability wanders within the frequency band. The complex demodulation phase of buoy SST provided a good index of the TIW signal and was used as the basis for compositing the other quantities (currents, currents relative to 20 m and their geostrophic and agoeostrophic components, winds, and temperature). All variables were decomposed using complex demodulation around a central period of 30 days for the period when TIW were prominent in the record, 1 November 2004–28 February 2005. The near-sinusoidal representation of these were binned according to the concurrent SST phase to create a 30-day composite.

Shear is also expected to be sensitive to viscosity. Because stratification and mixing have their largest variations over the diurnal cycle, while wind and frontal forcing have much weaker or negligible variations at this time scale, the effects of viscosity on shear can essentially be isolated from other processes by considering the composite diurnal cycle. To compute diurnal composites of wind, temperature, and velocity, the time series over the 4.5-month period when all five current meters were available were binned into local time of day and then averaged.


Return to Previous section

Go to Next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage