To investigate potential temporal lags, LSCE first performed a basic lag-correlation analysis between the North Atlantic Oscillation (NAO), the major driver of climate variability in the Atlantic, and the air-sea CO2 flux. That analysis revealed the complex spatiotemporal NAO structure that cannot be resolved with traditional methods, such as EOF analysis.
Thus LSCE used a method known as Multi-channel Singular Spectrum Analysis (MSSA) to analyse interannual-to-decadal variability of climate and biogeochemical variables. MSSA accounts, simultaneously, for temporal as well as spatial variability among a suite of variables. Delays in the air-sea CO2 flux relative to climate forcing were identified, and had maximum correlation when the flux lags the forcing by 1 and 3 years. The 1-year lag may be due to the 1-year equilibration time required for perturbations in atmospheric CO2 to mix throughout the mixed layer [Broecker and Peng, 1974]. Lags of up to 3 years have been documented previously as being linked to interior ocean dynamics [Hakkinen, 1999; Gulev et al., 2003].
One cause may be export production linked with circulation-driven control factors. Evidence already exists that lateral advection partially controls air-sea CO2 fluxes in both the subtropical and sub-polar gyres based on a Lagrangian model of mixed layer DIC content [Follows and Williams, 2004]. Lateral advection from higher latitude mode waters retains a signature of thermo-cline nutrient anomalies that is delivered to the subtropics [Palter et al., 2005]. For sub-polar air-sea CO2 fluxes, lags may derive from advection within spatially heterogeneous tracer fields that stem in part from competition between entrainment and net carbon export production. More quantitative, process-oriented studies are needed to better resolve the mechanisms responsible for the lags in the air-sea CO2 flux relative to the climate forcing.