FY 1997

The stepwise linear regression method for calculating anthropogenic CO₂ invasion into the north Pacific Ocean

Slansky, C.M., R.A. Feely, and R. Wanninkhof

In Biogeochemical Processes in the North Pacific, Proceedings of the International Marine Science Symposium on Biogeochemical Processes in the North Pacific, S. Tsunogai (ed.), Japan Marine Science Foundation, Mutsu, Japan, 12–14 November 1996, 70–79 (1997)

In order to quantify temporal and spatial variation of dissolved inorganic carbon (DIC) inventories resulting from invasion of anthropogenic carbon dioxide (CO), NOAA's Ocean-Atmosphere Carbon Exchange Study is designed to acquire high-quality data on temperature, salinity, oxygen, nutrients, and the carbon system in the major ocean basins. Direct determination of carbon inventories requires that DIC variations due to water mass motions and biological metabolism be taken into account. We have used the 1991 Climate and Global Change (C&GC '91) transect data of salinity, theta, apparent oxygen utilization (AOU), silicate, and phosphate along 152°W to construct a forward linear stepwise regression statistical model to predict DIC. The model gives a standard error of 3.8 µmol DIC kg, or about 2 times the precision of the analytical method. The uncertainty generally decreases with increasing pressure. The regression model was applied to eight stations from the 1973 GEOSECS dataset. The results show observed-predicted differences in the upper 800 db that are consistent with an anthropogenic CO increase of approximately 1.3 ± 0.7 µmol kg yr in the mixed layer of the northeastern Pacific subarctic and subtropical waters. This number agrees with estimates by other methods. This symposium proceedings paper is a shortened version of a paper submitted for publication elsewhere (Slansky et al., submitted for publication).