Global atmospheric temperature and moisture profiles are two of the most important products that are currently produced by the AIRS/AMSU/HSB instruments aboard the Aqua satellite and that will be produced by the CrIS/ATMS instruments aboard the NPP satellite to be launched in 2009. The general objective of the proposed work is to: 1) improve existing facility algorithm performance (product accuracy, yield, and latency) by incorporating advanced signal processing and data fusion techniques recently developed by the proposal team, 2) develop and validate precipitation (surface rain rate) algorithms using current and future Aqua/NPP microwave sounding data, and 3) validate the algorithm performance of future NPP sounders using comprehensive data records generated by the Aqua sounders and the NPOESS Aircraft Sounding Testbed-Microwave instrument.
A new class of statistical estimation techniques has recently been developed [Staelin/Cho 2006, Blackwell 2005] that retrieve both cloud-cleared infrared sounding radiances and atmospheric profiles from combined infrared and microwave radiances observed in cloudy conditions, yielding accuracies that are commensurate with current model-based, physical algorithms at a substantially reduced computational complexity and can be readily integrated into existing algorithms. Based on previous experiments with AIRS/AMSU, it is expected that the new algorithms will provide superior accuracy, yield, robustness, and speed for both AIRS/AMSU/HSB and CrIS/ATMS EDRs/CDRs. The planned research will further explore the combination of heritage, physical techniques with the new statistical techniques.
In parallel with the atmospheric sounding work described above, we plan to develop a class of statistical precipitation algorithms that are fast, accurate, and flexible enough to accommodate a variety of spectral bands and resolutions, spatial sampling and geometry configurations, and temporal coincidences. Such algorithms will improve satellite precipitation retrieval capabilities for present and future NASA/NOAA satellites. Perhaps the primary disadvantage to the existing model-based, physical precipitation algorithms is the inherent difficulty in dealing with different data types-this is precisely the area that we hope to improve with this study. The precipitation work of [Chen 2003] and the IR/MW data fusion work of [Blackwell 2005] is the cornerstone that will be built upon. The statistical algorithms developed here will complement the model-based, physical algorithms that are in use and under development.
Finally, the rich array of data products generated by AIRS/AMSU/HSB provides a unique opportunity to validate the algorithm performance of NPP/NPOESS. Specifically, pre-launch simulations of the NPP/NPOESS EDR algorithms can be carried out and validated using proxy data generated with AIRS/AMSU and NAST-I/M. Post-launch algorithm cross-validation exercises will also be defined as part of this work.
