The daily data files contain 9 columns. These files contain the average of flask replicates collected at a given station and time. Columns 1 and 2 are the sampling date and time (UTC). Column 3 is the relevant concentration (e.g. CO2 in ppm, O2/N2 and APO in per meg units). Columns 4 and 5 are the standard deviation of the retained replicates and the number of retained replicates. Column 6 and 7 are the dates in several redundant formats. Column 8 is a flag for identifying uncertainty in the reported sampling time. All sampling date/times are nominally corrected to UTC. But in some cases the relevant correction is uncertain due to incomplete information on the original sample log sheets. In this case, Column 8 gives the (assumed) correction applied (in hours). The flag is otherwise set to "UTC", indicating that the time is not uncertain. Column 9 is a 3-letter station identifier.
The monthly data files contain 6 columns. These files are based on daily data which is combined and averaged to yield a monthly mean centered on the middle of the month. Column 1 is the date in fractional years. Column 2 is the relevant concentration (e.g. CO2 in ppm, O2/N2 and APO in per meg units). Column 3 gives the data after a seasonal adjustment to remove the quasi-regular seasonal cycle. The adjustment involves subtracting from the data a 4-harmonic fit to the seasonal cycle. Column 4 is a smoothed version of the data generated from a stiff cubic spline function plus the 4-harmonic function. Column 5 is the same smoothed version with the seasonal cycle removed. Column 6 indicates the number of sampling dates that are incorporated into the monthly average. Missing values are set to zero.