Water vapor is a key component of the Earth's hydrologic cycle that plays an important role in the intensity of severe weather and hurricanes. Numerical weather prediction models are expected to show improved forecast skill when water vapor remote sensing observations are assimilated from operational weather satellites. Accurate water vapor vertical profiles from satellites would also provide a valuable climate record for evaluation of NWP reanalysis products and for validating climate models.

The joint U.S./European polar weather satellite program will operate the JPSS CrIMSS sensor suite in the PM orbit while the METOP IASI/AMSU sensors fly in the AM. This paper describes the methodology developed for validation of the water vapor vertical profiles from the CrIMSS, a key JPSS EDR. The approach uses ground-truth measurements from the Department of Energy Atmospheric Radiation Measurement sites (Southern Great Plains, North Slope of Alaska, and Tropical Western Pacific) where special launches of Vaisala RS92 radiosondes will be conducted at the overpasses of the NPP satellite. Coincident with these radiosonde launches at the Southern Great plains site, a validation profile of water vapor mixing ratio profile will be obtained from the ground-based ARM Raman Lidar. These Raman profiles obtain their absolute calibration from a ground-based Micro-wave Radiometer (MWR) operating nearby. In a similar manner, the precipitable water vapor associated with the vertical integral of the water vapor EDR profile will be compared against both the ARM MWR and measurements from the SuomiNet network of ground-based GPS receivers. Matchups of these data will be analyzed to make an independent assessment of the accuracy of the operational CrIMSS EDRs.

Preliminary assessment of this methodology using NASA AIRS L2 and NOAA IASI as a proxy for the CrIMSS EDRs has been performed. Results show a dry bias in AIRS L2 version 5 for summer, night-time observations in the Southern Great Plains. This is confirmed using time-interpolated radiosonde profiles. The result is consistent with Bedka et al. 2010 that showed a similar dry bias in PWV from MWR data.