Temperature Trends and Anomalies in Modern Satellite Data: Infrared Sounding and GPS Radio Occultation.

Trends in monthly average, zonal average temperature in the stratosphere as retrieved from highly accurate modern satellite data are intercompared and compared with climate reanalyses from 2003 through 2014. The data sets used are those of Atmospheric Infrared Sounder and Global Positioning System (GPS) radio occultation, and the reanalyses are those of MERRA and ERA‐Interim. Trends produced by all data sets agree to within 0.02 K/year in the lower stratosphere and 0.05 K/year in the middle stratosphere. A number of retrieval errors are found that introduce incorrect trends and seasonal anomalies. Adding microwave data to the infrared retrieval changes trends by approximately 0.01 K/year, thus improving agreement with the other data sets. The signature of the quasi‐biennial oscillation in temperature and the annual cycle of temperature over Antarctica as retrieved from infrared data contain null‐space errors of more than 3 K due to erroneous priors used in retrieval. Nonuniformity in GPS radio occultation gives rise to errors because changes in received GPS signal strength alter the upper boundary initialization in radio occultation retrieval. Finally, an incorrect specification of atmospheric water vapor introduces an erroneous seasonal cycle of temperature as retrieved from GPS radio occultation data in the upper troposphere. All of these time‐dependent retrieval errors can be corrected with future research and improvements to spectral infrared and GPS radio occultation retrieval systems. Plain Language Summary: Satellite instruments that have been observing the atmosphere since 2003 are highly calibrated and particularly rich in information. Two of them have been measuring temperature trends in the atmosphere between 7 and 30 km altitude, and their results differ by an uncomfortable amount. Because the measurement techniques have nothing in common, it is possible to trace the causes of the differences to fairly well‐known problems in their formulations that translate basic measurements to air temperature. The results of this paper will direct those responsible for retrieving temperature from modern satellite data to fix these problems and thereby bring multiple, independent observations of temperature trends into better agreement with each other. Key Points: Temperature trends in the lower and middle stratosphere as retrieved from modern satellite data differ by 0.02 and 0.05 K/yearNull‐space error affects temperature anomalies in infrared data, and receiver performance affects trends in GPS RO dataTime‐dependent biases inherent to infrared and GPS RO retrieval systems can be removed by modification of the retrieval systems [ABSTRACT FROM AUTHOR]