Trends in CO₂, CH₄, N₂O, SF₆

PDF Version

PDF Version

The table summarizes annual increases in atmospheric SF₆ based on globally averaged marine surface data.

PDF Version

The annual increase in atmospheric SF₆ in a given year is the increase in its abundance (mole fraction) from January 1 in that year to January 1 of the next year, after the seasonal cycle has been removed (as shown by the black lines in the figure above). It represents the sum of all SF₆ added to the atmosphere by human activities during the year (losses are negligible). Our first preliminary estimate for the annual increase of a particular year is reported in April of the following year, using available data from the previous year. It is important to recognize that the initial, April estimate of the annual increase is likely to change significantly as more data are added to the analysis. That estimate will be updated in subsequent months as more samples are measured for SF₆ and included in the analysis. By autumn of the following year the annual increase will typically converge toward a “final” value.

Estimates of the globally-averaged SF₆ abundance (monthly- and annually-averaged means), and the annual increase, are updated every month as new samples are returned to Boulder, measured for SF₆, and added to the analysis. Adding new, more recent data improves the accuracy of the initial estimate by increasing the spatial density of data and eliminating “end effects” of the curve fitting procedures used. We’ve investigated the impacts of adding new data to the parameters reported here, and a summary of the results follows:

Initial estimates of the SF₆ annual increase made for the previous year are biased compared to those that follow using additional data. The average bias in the initial estimate is +0.02±0.01 ppt yr^-1 (1 standard deviation shown). Over the next few months, the average bias slowly decreases until it is negligible by July or August. In any given year though, bias in the initial estimate of the annual increase can be much larger than the average, with bias up to ±0.04 ppt yr^-1; that is, it can be positive or negative. In other words, until late in a year, bias in the annual increase can be much larger than the uncertainty reported based on the bootstrap method described below.

Behavior of initial annually-averaged means and monthly-averaged means are similar (see links to files below). For monthly mean SF₆, the initial value is typically too high, by up to 0.05 ppt.

The estimated uncertainty in the global annual SF₆ increase varies by year. It is estimated using two terms: The first is a "bootstrap" (resampling) method that varies the sites in our network. Each bootstrap realization of the network is constructed by randomly picking sites, with restitution, from the existing marine boundary layer sites in the NOAA/GML cooperative global air sampling network (Dlugokencky et al., 1994). Each member of the ensemble of networks has the same number of sites as the real network, but some sites are missing, while others are repeated. An additional condition is that at least one site is present from high southern latitudes, one from the tropics, and one from high northern latitudes, because we have always maintained broad latitude coverage in the real network. Temporal data gaps at individual sites are present in the bootstrap networks. The second term is a Monte Carlo method that randomly modifies the data to account for measurement uncertainty. The modifications are based on assessment of random uncertainty in the measurements, and it varies over time. In both cases, 100 globally-averaged time series are created. We calculate the mean and standard deviation for each year's annual increase from the ensemble members, and one standard deviation from the two terms (network and analytical) is taken in quadrature to give the reported uncertainty at each time step. As mentioned earlier, bias in our first estimates of annual increase, monthly mean, and annual mean can be significantly greater than the stated uncertainty.