Atmospheric CO2 measurements and error analysis on seasonal air-sea CO2 fluxes in the Bay of Biscay

Atmospheric molar fraction of CO2 (xCO(2)(atm)) measurements obtained on board of ships of opportunity are used to parameterize the seasonal cycle of atmospheric xCO(2) (xCO(2)(atm)) in three regions of the eastern North Atlantic (Galician and French offshore and Bay of Biscay). Three selection criteria are established to eliminate spurious values and identify xCO(2)(atm) data representative of atmospheric background values. The filtered data set is fitted to seasonal curve, consisting of an annual trend plus a seasonal cycle. Although the fitted curves are consistent with the seasonal evolution of xCO(2)(atm) data series from land meteorological stations, only ship-board measurements can report the presence of winter xCO(2)(atm) minimum on Bay of Biscay. Weekly air-sea CO2 flux differences (mmol C(.)m(-2) day(-1)) produced by the several options of xCO(2)(atm) usually used (ship-board measurements, data from land meteorological stations and annually averaged values) were calculated in Bay of Biscay throughout 2003. Flux error using fitted seasonal curve relative to on board measurements was minimal, whereas land stations and annual means yielded random (-0.2 +/- 0.3 unnol C.m(-2) -day(-1)) and systematic (-0.1 +/- 0.4 mmol C(.)m(-2) day(-1)), respectively. The effect of different available sources of sea level pressure, wind speed and transfer velocity were also evaluated. Wind speed and transfer velocity parameters are found as the most critical choice in the estimate of CO2 fluxes reaching a flux uncertainty of 7 mmol C(.)m(-2.)day(-1) during springtime. The atmospheric pressure shows a notable relative effect during surnmettime although its influence is quantitatively slight on annual scale (0.3 +/- 0.2 nurrol C(.)m(-2.)day(-1)). All results confirms the role of the Bay of Biscay as CO2 sink for the 2003 with an annual mean CO2 flux around - 5 +/- 5 mmol, C m(-2) day(-1). (c) 2006 Elsevier B.V. All rights reserved.