RESULTS

NCAR's CSM has been used to simulate early Paleogene climate with realistic boundary conditions including 560 ppm pCO₂. At ~3500 m depth in the Pacific, temperatures are ~8°C, which is much warmer than modern values but in the cooler end of the range of values predicted from benthic foraminiferal calcite (closer to middle Eocene values). The meridional SST gradient within 30° of the equator is very similar to the modern value. The meridional SST gradient is much smaller than the modern one in the extratropical regions, which is consistent with the phenomenon of high-latitude amplification noted in many previous climate-change studies. Although the mean state of the simulation shows a strong Walker cell and associated warm pool (cold-tongue structure), the tropics do not collapse onto a steady La Niña or ENSO mode, and the ITCZ in the eastern Pacific is in approximately the same location as it is today.

To set up the context of this simulation, the annual average global upwelling distribution and Pacific Ocean current vectors at ~100 m depth are shown in Figure F1. There is clearly a large body of upwelled water in the eastern Pacific, and westward spanning tongues of upwelled water associated with the equator, the ITCZ in the Northern Hemisphere, and a probably spurious southern branch associated with a southern ITCZ. This latter feature is further discussed below. The current structure clearly reveals vigorous NECC and EUC currents as well as the larger-scale gyre structures throughout the Pacific (Fig. F1).

The net freshwater flux distribution is close to modern distributions; the locations of net precipitation (i.e., the ITCZ and warm pool and South Pacific Convergence Zone [SPCZ]) are at near-modern locations and magnitudes (top of Fig. F2). The seasonal cycle of net freshwater flux at 115°W (bottom of Fig. F2) is similar to that produced by the coupled models for modern day, as well as being close to observed values with the exception of a brief period (April-May) when the ITCZ is placed south of the equator.

In Figure F3, quantities are averaged through the upper 30 m of the model, which corresponds roughly to the Ekman layer and the mixed-layer depth. In Figure F3A, annual average temperature, salinity, and currents in this layer are shown. There is a large east-west gradient in Ekman layer temperature (7°C) along the equator, and the warm pool is displaced southward in the west. A tongue of warm (30°C) water is displaced north of the equator in the east. Salinity gradients are predominately zonal (east-west), and low salinity values are largely associated with the precipitation in the warm pool. The currents in this layer are very similar to those observed today and are quite vigorous (~30 cm/s).

Because of current and upwelling variability over the seasonal cycle, there is an impressive degree of seasonal variation in temperature, salinity, and flow streamlines, despite the fact that the tropics are sometimes considered homogenous (e.g., Barron, 1995). I will mostly discuss and contrast the seasonal extremes which in the tropics are approximately April-May-June (AMJ) and October-November-December (OND). In AMJ, the western Pacific warm pool is at its extreme in terms of temperature and southward displacement and there is substantial cross-equatorial transport in both the east and west Pacific in the Ekman layer (Fig. F3B). During OND, flow is much more zonal and vigorous (Fig. F3), and upwelling of cold water in the eastern Pacific and a Northern Hemisphere ITCZ is best developed (Fig. F3C).

In Figure F3D, annual average upwelling, streamlines of the flow, and an estimate of seawater ¹⁸O (hereafter _w) are plotted. The _w distribution in Figure F3D is calculated from the empirical relation between seawater salinity and _w proposed by Fairbanks et al. (1997) for the equatorial region. Fortunately, this modern correlation is (1) excellent (R² = 0.92), (2) was developed for this specific region, and (3) is accurate for the kind of variations in temperature and precipitation that occur in this region over ENSO cycles. In keeping with the model-produced salinity distribution, the calculated _w shows little meridional variation; therefore, there should be little need to correct for _w variations along the tropical transect when foraminiferal oxygen isotope ratios derived from Leg 199 are eventually processed (provided suitable foraminifers are found).

The streamlines in Figure F3D highlight the overall current structure (de-emphasizing the current strength) and show the correspondence between gyres and expected upwelling (both are a result of the Ekman transport divergence). The mean annual depth of the 24°C isotherm, which corresponds to the mid- to upper thermocline in this simulation, shows the same east-west tilt characteristic of the present and a similar seasonal cycle of this quantity as produced in a modern-day simulation along the equator (Fig. F3E, F3F, F3G). The relationship between the thermocline distribution and the upwelling distribution is clear, and thermocline depth can therefore be considered an indicator in the simulation of upwelling changes as well (Fig. F3E, F3F, F3G).

Since this simulation is meant to be a straw man, I have made a prediction of the ¹⁸O ratio that should be expected from two different depth levels, mixed-layer dwellers (top of Fig. F3H), and above-thermocline dwellers (bottom of Fig. F3H) based on these model results (hereafter _c). To do this, I used the calculated _w and model-predicted temperatures and inverted the Bemis et al. (1998) temperature (_c - _w) to produce an estimated relation for _c (in a manner similar to Schmidt, 1999). I chose the Bemis et al. (1998) equation as the estimator for two reasons. Most importantly, this relation produces the same results as the commonly used Erez and Luz (1983) relationship within this temperature range, so my results are directly comparable to previous proxy data studies. Secondly, because the relationship is linear and has the same slope as their equation, it is simple to immediately assess the possible effects of different light levels (and possibly alkalinity) and depth habits by subtracting 0.333 from the values here. It is clear that the predicted values show substantial structure and variability, which remain to be tested against data. Importantly, these results indicate that the vast majority of the signal recorded by foraminiferal _c in this region is due to temperature variations and not salinity (_w) variations. Additionally, depth habit will be a crucial factor to constrain, as it generates variations in predicted _c of over 1.0. More details on that issue follow in the discussion of the meridional transects.

As shown in the annual averages in Figure F4A, at a mean depth of ~100 m, upwelling in the east is associated with cool temperatures and the downward-dipping thermocline in the west (Fig. F3G) is associated with warm temperatures. At this level, the EUC is still quite vigorous (20 cm/s). Over the seasonal cycle, there are large changes in the magnitude of the EUC and NECC/NSCC complex (Fig. F4B, F4C).

In the region below the upper thermocline (190-550 m), where the seasonal cycle is less evident (not shown), the tracer properties are relatively homogenous (12°C; 34.6 ppt), but the current shows interesting structure (Fig. F5). The presence of a deep NECC dominates the deep tropics and is responsible for transport water from the western Pacific north and eastward at 3 cm/s. This pattern is pronounced in deeper waters (900-2000 m) (Fig. F6), where variations in temperature and salinity are due to the deep and vigorous (1 cm/s) NECC.

In keeping with the straw man approach, specific predictions are presented in Figure F7 to be tested against the results of Leg 199. This figure presents a transect along 115°W from 15°S to 20°N of the seasonal extremes of east-west (zonal) velocity and temperature and my estimate of _c. Upwelling clearly plays an important role within 4° of the equator. Similarly, the importance of zonal advection of plankton and water mass properties from the warm pool or from the Atlantic must not be neglected. As this figure shows, the currents are extremely vigorous and variable. Again, it should be noted that the structure of the results south of ~5°S is probably spurious given the fact this model (and all CGCMs) produce a spurious Southern Hemisphere ITCZ. This makes it likely that the currents and water mass properties in that region are not correct in the simulations of the past. Not clearly visible on Figure F7 is that fact that benthic conditions at 115°W may reflect input of a "young" Atlantic water mass entering into tropical current systems via the Panamanian Seaway (Huber and Sloan, 2001; Huber et al., submitted).