Climatologies and Variability from the WOCE High-Resolution XBT Network in the Pacific

Bruce D. Cornuelle, Dean H. Roemmich, Janet Sprintall, Lynne Talley (Scripps Institution of Oceanography, UC San Diego), Michele Morris (NIWA), Rick Bailey (CSIRO)

The World Ocean Circulation Experiment (WOCE) repeated high-resolution expendable bathythermograph (HRXBT) network in the Pacific is used as the basis for climatological anaylses. The network presently consists of 7 sections, spanning the Pacific basin, which repeat the same Volunteer Observing Ship tracks approximately every 3 months (See Figure 1). A technician or scientist is accommodated on board the ship in order to sample around the clock using an automatic XBT launcher. Precise XBT locations are controlled by GPS navigation. Temperature sampling is to 800 m using Sippican Deep Blue XBT probes. Expendable conductivity and temperature profiler (XCTD) profiles are taken by the observer at key locations to supplement the historical T-S relations. Over 1300 XBT profiles are obtained from each repeat of the network. Sampling begins and ends near the shelf break in waters between 100 and 200 m depth. Probe spacing is as large as 50 km in mid-ocean quiet areas, increasing near boundaries, the Equator, islands, and other regions of enhanced variability to a maximum of about 10 profiles per degree. The sampling is designed to capture most of the variance associated with mesoscale features and boundary currents. The first HRXBT sampling began in 1986, with new sections started in 1991, 1993, and 1997.

The HRXBT sections with at least 6 years of data are used to estimate the seasonal march of temperature, heat content, and steric height. After removal of the seasonal march, the residuals are smoothed in time to separate the mesoscale variability from the interannual variability in the sections. The residuals show many examples of persistent small-scale structure, including horizontal variations in the annual cycle. Computing the covariances and correlations between points on the sections show significant non-local and small-scale features in both the correlations and the EOFs used to decompose the horizontal variability. These statistics can be used to compare to model simulations and as simple checks on surfacing forcing climatologies.

Figure 1: Tracks of all HRXBT sections to date.

Figure 2: Comparison of section-average temperature with section-average transport. In the p50/p34 transect, the temperature has been corrected for variability of the trackline using the meridional temperature gradient from Levitus. Black is net geostrophic transport northward (left hand scale). Red is average temperature (right hand scale). Note that the temperature scale is inverted.

Figure 3 is the matrix plot.


Time Mean

Time mean, using all section repeats, estimated as part of the seasonal to interannual time scales. The mean is estimated independently at every point in the section, without smoothing. The estimator avoids bias due to irregular distribution of cruises in time and fights crosstalk from the seasonal march. Contour interval is 1 degree C.


Root-mean-square (RMS) variability around the time mean at every point in the section. Contour interval is 0.25 degree C.

Annual Cycle Amplitude

Amplitude of annual cycle (12 month period) component. Contour interval is 0.25 degree C.

Annual Cycle Phase

Phase of the annual cycle in months relative to January 1. A phase of 0 means that the annual cycle is at a maximum at January 1. A phase of 3 means that the annual cycle is at a maximum at April 1. Contour interval is 1 month.

Average Temperature

Section average temperature, at the surface and averaged from the surface to 200, 400, 600, and 800 m. All time axes are the same.


The time series figures on the far right of the matrix of plots (Figure 3) show the extent of the time coverage, and an indication of the large-scale interannual variability on the section. The surface temperature has the largest fluctuations, and may be aliased by the sampling, but the deeper averages show a smooth progression through the years, with the longer sections showing smoother variations than the shorter sections, as expected. There is a sudden warming in the Tasman Box sections (PX31, PX06, PX34) which may be related to the aftermath of the 1997-1998 El Nino, but the series are not long enough to show multiple events.

Figure 2 shows a comparison of section-average temperature with section-average transport. There is a similarity between the curves that suggests a relationship. One explanation of this relationship is advection of the meridional temperature gradient by the transport.

There is considerable detail in the sections; here only a few of the most prominent points are discussed. For the mean temperature sections, the lines including western boundary currents (PX37, PX31, PX34, and PX50) show persistent recirculation features. In PX50 the recirculation feature is the East Cape Eddy, as documented by Roemmich and Sutton (1998). In PX06, a series of minor fronts, at about 31.5 S, 27 S, and 21 S indicates filaments of the currents associated with the Tasman Front (Morris, et al., 1995).

The RMS plots show a familiar pattern of strong surface variability, driven by the surface heat flux, and subsurface thermocline variability, much of which is presumably driven by the wind through Ekman pumping. There is also deep variability associated with Western boundary currents, and PX50 shows surprising deep variability in the center of the section which is not associated with annual forcing. This may be affected by the variability of the track, and does not show much annual cycle component. PX37 shows the lack of variability in the warm pool, and the strong variability in the layers just below it.

The annual cycle amplitudes show a remarkable uniformity in depth of penetration of the heat-forced cycle, which appears to be about 75 m for almost all regions. The deep variability in PX50 does therefore not project well on the annual band, even though deep mixed layers are found there and in parts of the other sections. The annual cycle is only one component of the seasonal march, and the late spring mixed layers may project more strongly on the higher harmonics of the annual cycle.

The comparison of the annual cycle to RMS for PX37 shows a minimum in the annual cycle response just below the surface layer which is not anticipated by the RMS plot. This can perhaps be rationalized as a nodal minimum due to interference between the annual cycle of heating at the surface and of wind forcing of the thermocline, but this minimum does not occur in the other tropical sections, such as PX06 or PX31.

These figures are meant as a starting point for analysis of the dynamics controlling the variability, and comparisons are underway with Pacific circulation models.


Roemmich, D. and P. Sutton, 1998. The mean and variability of ocean circulation past northern New Zealand; determining the representativeness of hydrographic climatologies. Journal of Geophysical Research, 103, 13041-13054.

Morris, M., D. Roemmich and B. Cornuelle, 1995. Observations of variability in the South Pacific subtropical gyre. Journal of Physical Oceanography, 26, 2359-2380.