The upper ocean heat balance in the western equatorial Pacific warm pool during September-December 1992

Meghan F. Cronin and Michael J. McPhaden

Pacific Marine Environmental Laboratory, NOAA, Seattle, Washington

Journal of Geophysical Research, 102(C4), 8533-8553 (1997)
This paper is not subject to U.S. copyright. Published in 1997 by the American Geophysical Union.

2. Data

2.1. Buoy Array and Instrumentation

Figure 1 shows the TOGA-COARE enhanced monitoring array of TAO buoys. The data in this analysis derive primarily from a PROTEUS (PROfile TElemetry of Upper ocean currentS) buoy which was deployed at 0°0.1'S, 156°2.3'E on September 15, 1992. On December 20, 1992, the buoy broke away from its anchor, but within several days the buoy and subsurface mooring instrumentation were successfully recovered. By comparing data from the mechanical current meters and the downward-looking acoustic Doppler current profiler (ADCP) attached to the buoy's bridle, it was determined that the mooring line had become entangled with itself on deployment such that instruments below 50 m were shifted up 18 m. A schematic of the 0°, 156°E PROTEUS buoy with the depth-shifted instrumentation is shown in Figure 2. Table 1 lists the depths and duration of the subsurface temperature, salinity, and current velocity instrumentation. As shown in Figure 2 and Table 2, surface data from the buoy include vector winds, air temperature, relative humidity, rain rate, and shortwave radiation. SST and sea surface salinity were measured at 1 m depth. All surface and subsurface PROTEUS data were averaged hourly and subsampled once per hour centered on the half hour. Collocated temperature and salinity were used to estimate potential density ( + 1000) according to the UNESCO 1983 polynomial [Fofonoff and Millard, 1983]. Figures 3-4 show the surface and subsurface time series at 0°, 156°E during the 3-month deployment.

Figure 1. The enhanced monitoring array. The heat balance is estimated at the central 0°, 156°E PROTEUS mooring. The four nearest ATLAS buoys at 0°, 154°E; 0°, 157.5°E; 2°N, 156°E; and 2°S, 156°E are used to estimate horizontal temperature gradients. The 0°, 156°E PROTEUS mooring is on the northern edge of the COARE intensive flux array (dashed region).

Figure 2. Diagram of the PROTEUS buoy which was deployed at 0°, 156°E from September 15 to December 20, 1992. The vector averaging current meters (VACMs) are located 1 m above the listed SEACAT depths.

Figure 3. Hourly surface data: (a) eastward (dark line) and northward (light line) winds, (b) shortwave radiation, (c) relative humidity, (d) rain, (e) 1-m sea surface salinity (SSS), and (f) air temperature (light line) and 1-m sea surface temperature (SST) (dark line).

Figure 4. Daily averaged subsurface (a) temperature, (b) salinity, (c) potential density, (d) ADCP zonal flow, and (e) meridional flow. The surface layer is defined as the depth of the 21.8 kg m density surface (ATLAS moorings use the 28.5°C isotherm as the surface layer depth). The subsurface temperature contour interval (CI) is 1°C for values less than 28°C and 0.5°C for warmer values. The salinity CI is 0.2 psu. The density CI is 0.1 kg m for values less than 21.4 kg m and 0.4 kg m for denser values. The velocity CI is 25 cm s. Westward and southward directed currents are contoured with dashes.

If your browser cannot view the following table correctly, click this link for a GIF image of Table 1

Table 1. Subsurface PROTEUS Instrumentation Used in This Analysis

As shown in Table 1, subsurface temperatures on the 0°, 156°E PROTEUS mooring were measured with mini-temperature recorders (MTR) designed and manufactured at the NOAA Pacific Marine Environmental Laboratory, and with Seabird Electronics, Inc. SEACATS. In order to estimate the horizontal temperature gradients, daily averaged subsurface temperature data were used from nearby ATLAS moorings at 0°, 154°E; 0°, 157.5°E; 2°N, 156°E; and 2°S, 156°E, shown in Figure 1. ATLAS moorings have 11 thermistors located at depths of 1, 25, 50, 75, 100, 125, 150, 200, 250, 300, and 500 m. On the basis of laboratory predeployment and postdeployment calibrations [McCarty and McPhaden, 1993; Freitag et al., 1994], SST accuracy is 0.01°C for the PROTEUS mooring and 0.03°C for ATLAS moorings; subsurface temperature accuracy is 0.01°C for the PROTEUS mooring and 0.09°C for ATLAS moorings.

SEACATs were also used to measure subsurface salinity on the 0°, 156°E PROTEUS mooring. Precalibrations and post-calibrations showed drifts in some SEACAT conductivity sensors equivalent to a salinity drift of up to -0.03 psu in 4 months. By using a linear weighted average of the precalibration and postcalibration coefficients, nearly all density inversions within the mixed layer were eliminated, and an in-situ comparison with 90 CTDs taken within 5 km of the mooring by the R/V Hakuho-Maru during November 10-25, 1992, was significantly improved.

An RD Instruments, downward-looking 153.6-kHz ADCP measured current velocity from 14 m to approximately 250 m in 8-m bins. At 10 m depth, current velocities were measured by a EG&G model 610 vector averaging current meter (VACM). For purposes of the analysis it was assumed that velocities shallower than 10 m had no shear. The 82- and 182-m VACMs were used only to determine the depth correction as discussed earlier. The errors in current speed and direction were assumed to be 3 cm s and 2°, respectively [Lien et al., 1994; P. Plimpton, personal communication, 1996].

Vector winds at 0°, 156°E were measured with a RM Young wind sensor and vane located 4 m above the ocean surface atop the buoy tower. As listed in Table 2, the sensor's resolution is 0.2 m s. Air temperature and relative humidity were measured 3 m above the surface by a Rotronic Instrument Corporation model MP-100 temperature-humidity probe with shielding to reduce the effects of radiative heating. The sensors, however, were unaspirated, so errors may be larger when winds are light on sunny days. No postcalibrations were obtained for the thermistor and relative humidity sensor because the thermistor failed during postcalibration procedure. However, in-situ comparisons with the R/Vs Le Noroit, Hakuho-Maru, Natsushima, and Moana Wave when these ships were within 3-15 km of the buoy suggest that the buoy's relative humidity measurements were 3% too low at deployment and had a linear drift of 0.9% month. Freitag et al. [1994] show that individual humidity sensors of this generation could have offsets and drifts of similar magnitude. Thus the relative humidity data have been adjusted to agree with the ship measurements. This correction to the relative humidity results in a more consistent heat balance, described in section 4.

If your browser cannot view the following table correctly, click this link for a GIF image of Table 2

Table 2. Standard Deviations of the 5-Day Triangular Filtered Surface Measurements, Typical Measurement Errors, and Resulting Errors in Bulk Fluxes

Instrument	Variable	5-Day Filtered Std Dev	Typical Error	Corresponding rms Error, W m
				Q	Q	Qrf	Q
RM Young	winds	1.41 m s	0.20 m s	4.7	0.5	0.1	5.1
Rotronics	air temperature	0.38°C	0.16°C	0.6	1.3	0.2	1.8
Rotronics	relative humidity	0.03	0.02	9.2	0.1	0.6	9.2
SEACAT	SST	0.30°C	0.014°C	0.4	0.1	0.0	0.5
ORG	rain rate	0.37 mm h	25%	0.0	0.0	2.9	2.9
Eppley	Qsw	40 W m	2-3.5%	0.2-0.3	0.1	0.0	6.4-11.3
The 2% error in the shortwave radiation, Qsw, is the standard error assuming the radiometer mast is vertical. This error increases to a maximum of 3.5% if the mast is tilted 15° from the vertical. To estimate the upper and lower limits of the resulting flux errors, the bulk heat flux algorithm was run with the observed variables with plus and minus typical errors. One half the rms difference between the upper and lower limit is listed as the corresponding rms error in the heat flux due to a typical error in the bulk parameter.

Hourly rain rates were measured by a Science Technology, Inc. model 105 optical rain gauge (ORG). The instrument computes a rain rate proportional to scintillations caused by rain drops falling through a beam of near-infrared light [Wang and Crosby, 1993]. A description of the ORG mounting, sampling, and signal-processing characteristics for application on TAO buoys is given by McPhaden and Milburn [1992]. The accuracy of moored ORG measurements has been difficult to determine for lack of sufficient calibration data in natural rain rates prior to the COARE field phase (the particular instrument used in this study was nonfunctional on return and could not be postcalibrated). Most potential error sources (e.g., vibration, drop-size dependence, and splash from the instrument housing) would lead to relative uncertainties of O(10%) [e.g., McPhaden, 1993a; Thiele et al., 1994; F. Bradley, personal communication, 1996]. Probably the greatest uncertainty is simply in the factory calibrations which, if used without further correction, can lead to differences between like instruments of 15-30% [Thiele et al., 1994]. For our purposes we assume an error of 25%.

Shortwave radiation was measured by an Epply radiometer with a calibration accuracy of 2%. Although the radiometer has a full record up to when the buoy broke away from its anchor, the radiometer and its mast were missing when the buoy was recovered. Photos taken from the R/V Hakuho-Maru on November 20, 1992, show that the radiometer mast had been bent and was tilted at about 15° from the vertical. The buoy had apparently been vandalized. Although it is not clear when the vandalism occurred, the fact that the November 13 noontime radiation exceeded 1000 W m suggests that the buoy had been vandalized sometime after November 13. The tilt could produce an error as large as 3%; however, on overcast days on which the radiation is diffuse, the error produced by the tilt should be negligible. We have therefore made no attempt to correct for this error. Instead, we assume that the shortwave radiation error is 2% prior to November 20 and thereafter is 3.5%.

2.2. Air-Sea Heat Fluxes at 0°, 156°E

The surface heat flux is estimated as

(1)

where is the albedo, Q_sw is the incoming shortwave radiation, Q_lw is the net longwave radiation, Q is the latent heat flux, Q is the sensible heat flux of the cool air over the warm water, and Q_rf is the sensible heat flux due to rain.

As described earlier, the shortwave radiation, Q_sw, was measured by an Epply radiometer. A constant albedo value of 0.055 is used based on measurements taken aboard the R/V Franklin [P. Coppin and F. Bradley, personal communication, 1994]. Net longwave radiation, Q_lw, is estimated using the Clark et al. [1974] bulk formula [Fung et al., 1984]:

where = 0.97 is the emissivity, = 5.67 × 10 W m K is the Stephan Boltzmann constant, T is the near-surface air temperature (in kelvins), e is the near-surface vapor pressure (in millibars), T_s is the 1-m-depth sea surface temperature (in kelvins), and C is the cloud cover index which ranges from 0 (clear sky) to 1 (cloud-covered sky). C is estimated by inverting Reed's [1977] expression for the daily averaged insolation Q_sw:

where Q_cs is the clear-sky radiation and n is the noontime solar altitude. These daily estimates of C are then interpolated to hourly values to estimate hourly net longwave radiation. Intercomparison with the shipboard measurements of incoming longwave radiation suggests that the root mean square (rms) error in Q_lw is 6 W m.

Latent and sensible heat loss rates are estimated using the COARE version 2.5 bulk formulae [Fairall et al., 1996a], which are based on Liu et al. [1979]. The algorithm includes a cool skin parameterization and a simplified Price et al. [1986] mixed layer model to extrapolate bulk SST measurements to the interface z = 0 [Fairall et al., 1996b]. Additionally, the wind speeds are measured relative to the sea surface velocity (assumed here to be identical to the velocity at 10 m depth), and a gustiness factor which depends upon the convective scaling velocity is included for low wind conditions. The COARE bulk algorithm also estimates the sensible heat flux due to rain Q_rf by assuming that the rain falls at the wet-bulb temperature. Time series of the heat fluxes are shown in Figure 5. Figure 6 shows the time series of the net surface heat flux, Q.

Figure 5. Hourly air-sea heat fluxes. (a) Shortwave radiation, Q_sw, (b) latent heat flux, Q, (c) sensible heat flux, Q, (d) net longwave radiation, Q_lw, and (e) sensible heat flux due to rain, Q_rf .

Figure 6. Hourly and 5-day triangular filtered net surface heat flux. A positive flux value represents a heat flux into the ocean.

To estimate rms errors in the turbulent heat fluxes due to typical errors in the bulk variables (wind speed, air temperature, relative humidity, SST, rain, and shortwave radiation), the flux algorithm was run sequentially with observed values ±1 standard error (i.e., twelve runs for the six bulk variables). The errors listed in Table 2 were then computed as one-half the rms difference between the flux time series estimated with values +1 standard error and -1 standard error of the corresponding input variable. Errors due to inadequate parameterization of the turbulent fluxes are not included, so our error estimates for bulk fluxes should be considered as lower bounds. Latent heat flux is sensitive primarily to errors in relative humidity and to a lesser extent to errors in wind speed (Table 2), and the resulting latent heat flux error is similar in magnitude to the rms error in the shortwave radiation (Table 3). Errors in shortwave radiation, air temperature, and SST have a relatively small effect on the turbulent heat fluxes. The record length mean net surface heat flux is not significantly different than zero. However, since the standard deviation of the 5-day filtered net surface heat flux is more than 4 times larger than the rms error, the data can resolve the variability in the surface heat fluxes on wind burst timescales which are highlighted by the 5-day triangular filter (cutoff frequency = (3 - day), half amplitude frequency = (6.77 - day)).

If your browser cannot view the following table correctly, click this link for a GIF image of Table 3

Table 3. Record Length Means, Standard Deviations of the 5-Day Triangular Filtered Fluxes, rms Error, and Cross Correlations Between 5-Day Filtered Fluxes and 5-Day Filtered Wind Speed
Heat Flux	Mean, W m	5-Day Filtered Std Dev, W m	rms Error, W m	5-Day Filtered Correlation with Wind Speed
(1-)Qsw	189	38	6.6-11.5	-0.57
Qlw	-48	5	6	0.70
Q	-118	34	10.3	-0.93
Q	-11	5	1.4	-0.69
Qrf	-2	2	3.0	-0.32
Q	10	68	14.2-16.9	-0.80

Return to previous section or go to next section

PMEL Outstanding Papers

PMEL Publications Search

PMEL Homepage