Introduction

One of the main objectives of the study of the oceanic carbon cycle is to quantify the present and future role of the ocean in the absorption of anthropogenic carbon dioxide (CO₂). In situ data are typically used to quantify the present anthropogenic CO₂ concentrations in the ocean (Brewer 1978; Chen and Millero 1979; Chen 1993; Wallace 1995; Gruber et al. 1996; Gruber 1998; Peng et al. 1998; Sabine et al. 1999; Goyet et al. 1999). Global ocean models are mainly used in a prognostic mode to estimate the future penetration of anthropogenic CO₂ on the global scale (Sarmiento et al. 1992; Bhaskaran et al. 1995; Washington and Meehl 1996). Yet, accurate global initialization fields of the CO₂ properties in seawater, such as total CO₂ (TCO₂) and total alkalinity (TALK), do not exist.

In order to study the oceanic carbon cycle and to accurately describe and quantify the TCO₂ and TALK fields on the global scale, TCO₂ and TALK were measured with high accuracy throughout the water column of the major oceans. These measurements were mainly performed over the last two decades during intensive national and international field programs. Most of the data of these field programs are now freely available to the scientific community. However, these data need to be interpolated on a regular grid before they can easily be used in global ocean models.

The purpose of this work is therefore to best interpolate these data on a regular grid for use in ocean models. The interpolation is based on each measured profile from the base of the mixed layer to the bottom of the ocean. The data within the mixed layer are not considered here because they are subject to large spatial and monthly variations that are still difficult to accurately quantify. The variations of the CO₂ properties in the mixed layer are controlled by ocean circulation, evaporation/precipitation, dissolution of calcium carbonate, photosynthesis and oxidation of organic matter, and CO₂ flux across the ocean-atmosphere interface including penetration of anthropogenic CO₂. Many independent studies are currently designed to best quantify and parameterize each of these processes and the overall variations of the CO₂ properties in the mixed layer (Takahashi et al. 1997; Millero et al. 1998).

Below the mixed layer, TCO₂ and TALK are controlled by ocean mixing, formation/dissolution of calcium carbonate, and oxidation of organic matter (Brewer 1978). In other words, short-timescale processes do not significantly affect TCO₂ and TALK below the mixed layer. Thus it is possible to interpolate the data measured below the mixed layer at different times of year to acquire a reasonable understanding of the TCO₂ and TALK fields. In ocean areas where anthropogenic CO₂ is present (mainly in the upper 2000 m), it is also necessary to specify if and how data from different years are adjusted to a specific year before interpolation.

In practice, the distribution of anthropogenic CO₂ concentrations in the ocean is not accurately known. Estimates can differ significantly (Coatanoan et al. 2000) according to the various assumptions used. Until these differences are understood and considerably reduced, it will be very difficult to estimate pre-anthropogenic TCO₂ fields on the global scale. Consequently, in this paper the authors interpolate the measured TCO₂ and TALK data without adjustment for the variations in anthropogenic CO₂ concentration for a given year. Because most of the data were measured within the past twenty years, such small adjustment to the different data sets (except for the North Atlantic Ocean) would mainly be within the uncertainty of the interpolated field. The results provide an estimate of these fields for the mid-1990s, when most of the accurate measurements were performed.