Recalibration and merging of SSU observations for stratospheric temperature trend studies

Abstract

AbstractLong‐term observations from the Stratospheric Sounding Unit (SSU) during 1979–2006 onboard NOAA historical polar orbiting satellites were recalibrated for climate change investigation. A two‐point linear calibration equation, with cold space and an internal blackbody warm target as end‐point references, was used to transfer SSU raw counts data into radiances. The warm target temperature was represented by measurements from the space side thermistor on the blackbody, and the cold space radiance was assumed to be zero. Space view corrections due to an electrical interference were applied. Intersatellite calibration was conducted simultaneously by applying calibration offsets determined from residual intersatellite biases. The recalibration reached an accuracy of 0.1–0.2 K for global means and thus is expected to improve the consistency in stratospheric temperature time series in climate reanalyses. The recalibrated SSU radiances were further adjusted to develop Version 2 of the NOAA stratospheric temperature time series. The effects being adjusted included those from changes in instrument cell pressure and atmospheric carbon dioxide concentration, viewing angle differences, and semidiurnal tides due to orbital drift. Intersatellite biases were carefully removed to ensure smooth transitions between satellite pairs. Differences from Version 1 included improved radiance calibration, improved adjusting schemes for diurnal drift and intersatellite biases, removal of time‐varying cell pressure adjustment for NOAA‐9 channel 1, and excluding NOAA‐7 channel 2 in the time series. In addition to the final merged data set, intermediate synthetic time series corresponding to different adjustments were also created to quantify their impact on the final trend as well as its reliability and uncertainty. Excellent matching between satellite pairs, especially the 7 year overlaps between NOAA‐11 and NOAA‐14 during 1997–2004, in intermediate as well as the final time series provided strong evidence on the validity of adjustments and thus confidence on the resulting trends. The Version 2 global mean trends for 1979–2006 were −0.69 ± 0.18, −0.77 ± 0.15, and −0.85 ± 0.15 K/decade for SSU channels 1, 2, and 3, representing temperatures of middle stratosphere, upper stratosphere, and stratosphere‐mesosphere, respectively. Among these, cooling of channel 2 was stronger and channel 3 weaker than those in UK Met Office (UKMO) data by about 1 K during the entire SSU period from 1979 to 2006. Finally, the average of the channel 1 and channel 3 anomalies in Version 2 was close to channel 2 anomalies to within 0.2 K for the entire 1979–2006 period with identical trends. This feature was found consistent with chemistry‐climate model simulations.