Toward an internally consistent calibration for 11 AVHRR instruments in a 20-year time series of satellite data

The AVHRR (Advanced Very High Resolution Radiometer) data record comprises the longest existing daily satellite dataset. The CSIRO AVHRR time series includes data over Australasia from all of the NOAA polar orbiters since NOAA-6 in 1981 through the currently operational NOAA-17. Both inter- and intrasatellite factors affect the usability of these data for the generation of consistent time series of meteorological and geo/bio-physical parameters. Not only do the particular AVHRR instruments on the individual satellites differ in terms of spectral response and sensitivity, but the sensitivity varies with time, and the evolution of each satellite's orbit results in changes in target illumination and viewing geometry that must be taken into account. To improve our knowledge of the measurement variation of the different AVHRR instruments we have attempted to develop a fully operational method, based on multivariate alternate detection in comparison with pseudo-invariant features. A comparison of the operability and effectiveness of these methods is discussed and their relative impact on the overall consistency and long term calibration of the time series is described.