Examining methods used in extracting long-term thermospheric density trends

Numerous density trend studies have shown that thermospheric density is decreasing, and the declining trends were attributed to the increase of anthropogenic gas emissions in the lower atmosphere. Since solar variations are of much larger magnitude and on a time scale that is, relatively speaking, comparable to the time range of available data (~40 years), extracting the trends is not trivial. A model is constructed to investigate if the methods were effective to remove or minimize solar influences in their trend analysis. The phase and amplitude of the solar cycle and the time range of available data are the three parameters considered in our model. It is found that the amplitude of density variation due to solar influence affects the trend results the most and that the Difference and the Ratio method can deduce a reliable trend when there is small uncertainty or variation in the three parameters. Our error propagation analysis shows that the Ratio method would yield smaller propagated errors than the Difference method. We applied the two methods to deduce the trend in the mass densities derived from satellite drag measurements and found that they yielded the same trend. However, it is noted that the Difference method yielded a larger trend than the Ratio method when the first 100 data points were excluded. This indicates that data selection for trend analysis also affects the trend result.