A sampling strategy to determine the annual mean air-sea CO 2 fluxes in the high-latitude North Pacific and North Atlantic was developed using a combination of signal-to-noise ratios and 2D Fourier transforms in conjunction with a coupled climate carbon model. To account for unresolved mesoscale variability in our simulations, we used summer and winter cruises to determine the magnitude of this variability and estimate its impact on the air-sea fluxes. From our analysis, we propose a regular sampling strategy of every 6° in latitude and 10° in longitude every 3 months for the North Pacific and North Atlantic. Applying this sampling strategy to the simulation returned an annual mean air-sea flux value within ±15% of the total simulated value, when applied now and into the future. Our study highlights several key points for measuring annual mean basin-scale CO 2 fluxes: (1) the combination of temporal and spatial sampling dramatically reduces the noise in the observations and provides a good representation of the fluxes with far fewer measurements than required for resolving the spatial or temporal signals independently; (2) that sampling at higher than recommended frequencies in time and space provides little improvement in the estimated annual mean flux; (3) the uncertainty in the decadal annual mean uptake is limited by interannual variability and not by sampling error or unresolved mesoscale variability; and (4) our high-latitude sampling strategies remain valid until at least the end of this century.