Detection of Extremes with AIRS and CrIS

Climate change is expected to be detected first as changes in extreme values rather than in mean values. The availability of data of from two instruments in the same orbit, AIRS data for the past eleven years and AIRS and CrIS data from the past year, provides an opportunity to evaluate this using examples of climate relevance: Desertification, seen as changes in hot extremes, severe storm, seen as a change in extremely cold clouds and the warming of the polar zone. We use AIRS to establish trends for the 1%tile, the mean and 99%tile brightness temperatures measured with the 900 cm(exp -1) channel from AIRS for the past 11 years. This channel is in the clearest part of the 11 micron atmospheric window. Substantial trends are seen for land and ocean, which in the case of the 1%tile (cold) extremes are related to the current shift of deep convection from ocean to land. Changes are also seen in the 99%tile for day tropical land, but their interpretation is at present unclear. We also see dramatic changes for the mean and 99%tile of the North Polar area. The trends are an order of magnitude larger than the instrument trend of about 3 mK/year. We use the statistical distribution from the past year derived from AIRS to evaluate the accuracy of continuing the trends established with AIRS with CrIS data. We minimize the concern about differences in the spectral response functions by limiting the analysis to the channel at 900 cm(exp -1).While the two instruments agree within 100 mK for the global day/night land/ocean mean values, there are significant differences when evaluating the1% and 99%tiles. We see a consistent warm bias in the CrIS data relative to AIRS for the 1%tile (extremely cold, cloudy) data in the tropical zone, particularly for tropical land, but the bias is not day/night land/ocean consistent. At this point the difference appears to be due to differences in the radiometric response of AIRS and CrIS to differences in the day/night land/ocean cloud types. Unless the effect can be mitigated by a future reprocessing the CrIS data, it will significantly complicate the concatenation of the AIRS and CrIS data records for the continuation of trends in extreme values.