Long‐Term Trends and Solar Responses of the Mesopause Temperatures Observed by SABER During the 2002–2019 Period

The global distribution and variations of the monthly mesopause temperature are presented during 2002–2019 covering the latitudes of 83°S to 83°N based on Sounding of the Atmosphere using Broadband Emission Radiometry (SABER) observations. To investigate the long‐term trend and solar response of the mesopause temperature, a three‐component harmonic fit is first applied to remove the seasonal variation from the monthly temperature data series. Then a multiple linear regression model is performed to residual temperatures versus constant, linear trend, solar activity, and geomagnetic activity terms. In this study, the mesopause temperature shows a cooling trend through all latitudes ranging from ~0 to −0.14 K/year with a mean of −0.075 ± 0.043 K/year. The cooling trends in the Southern Hemisphere are stronger than those in the Northern Hemisphere. For high latitudes (60–80°), significant negative trends can be observed during nonsummertime, while no significant trends are found for summertime. The mesopause temperature shows apparent positive responses to solar activity through all latitudes ranging from 3.03 to 4.80 K per 100 solar flux units (sfu) with a mean of 3.99 ± 0.49 K per 100 sfu, which is more significant and stable in the Northern Hemisphere. There is a pronounced drop for mesopause height at polar latitudes, which reflects the shrinking effect at lower altitudes mainly caused by greenhouse gas cooling. We show that the length of the time interval analyzed strongly influences the results. Our results, obtained from 18‐year SABER observations, are expected to be a robust measure of the mesopause temperature variability. Key Points The mesopause temperature during 2002–2019 shows a cooling trend through all latitudes ranging from ~0 to −0.14 K/year The mesopause temperature shows a positive response to solar activity through all latitudes ranging from 3.03 to 4.80 K per 100 sfu A sufficiently long time interval is necessary for such analyses