Reconciling Observation and Model Trends in North Atlantic Surface CO2, 1992-2014, by multiple linear regression.

Monthly files of surface ocean fCO2 for the North Atlantic were generated for 1992 to 2014. They were generated by Multiple Linear Regression of interpolated observation-based products of surface ocean fCO2, split into seven latitude bands of 60, 30, 20, 10, 5, 2 and 1 degrees width. The multiple linear regression method was also applied to 19 CMIP5 model fields to calculate the interpolation error, in an approach known as Observation Sampling Experiments. Additionally, global files of monthly surface ocean fCO2 were generated using 5 ensembles of a coupled ocean-biogeochemical model with different surface flux forcing for each ensemble: ERA-Interim (observation-based forced experiment), CanESM2, GFDL-ESM2M, HadGEM2-ES and IPSL-CM5A-LR (CMIP5-forced simulations experiments).