Generating Long-Term Temperature and Humidity Profiles from HIRS Observations

Temperature and humidity profiles are derived based on NOAA polar orbiting satellites' High-resolution Infrared Sounder (HIRS) measurements from 1979 to present. To achieve homogeneity of the time series, HIRS longwave channel data are inter-calibrated to a base satellite. Temperature and humidity are subsequently derived using a backpropagation neural network technique. Increasing carbon dioxide (CO2) concentrations must be accounted for in the profile retrievals. Radiative transfer model simulations show that the effect of increasing CO2 on HIRS temperature sounding channel observations during the last several decades can be more than 1.5 K for some channels. A two-tiered approach is used to remove cloud-contaminated HIRS observations. Cloudy pixels are first identified and removed by using a neighboring difference method in both spatial and temporal dimensions. The remaining pixels are further screened by co-located Advanced Very High Resolution Radiometer cloud products. Flags are assigned to HIRS pixels to indicate the likelihood of cloud contamination. Radiosonde data and profiles derived from Global Positioning System Radio Occultation and from radiosonde observations are incorporated to reduce retrieval biases. The developed algorithm derives profiles at standard pressure levels from the surface to lower stratosphere for temperature and from the surface to upper troposphere for humidity. The algorithm, data flow, and analysis results will be shown.