Quantifying the Sensitivity of NCEP's GDAS/GFS to CrIS Detector Differences

Design of Infrared (IR) sounding instruments has advanced to use large arrays of detectors to make simultaneous observations. Different bias, noise and correlation properties between detectors can result in the need to treat observations from each detector differently. Having to treat each detector as an independent instrument is not desirable for Numerical Weather Prediction (NWP) centers. Current radiometric uncertainty and noise-equivalent changes in radiance (NEdN) requirements for IR sounder instruments do not fully address or constrain the noise characteristics between detectors. Detector properties, such as quadratic nonlinearity and detector noise, will contribute to inter-detector bias. No systematic studies have been conducted to determine the degree of match needed between detectors in an array used for IR sounding instruments that support NWP. The goal of this work is to understand what level these inter-detector biases begin to affect NWP analysis and forecast systems and help define the inter-detector design requirements. Instrument providers will need this information to assure that all instrument field-of-views (FOVs) are matched well enough to support NWP radiance assimilation.

A single detector is perturbed with several different detector quadratic nonlinearity parameter characterizations to determine their effect on analyses. Clear sky CrIS spectra radiances are simulated from the NASA GEOS-5 analyses assuming an aqua planet. Instrument noise and/or systematic bias are then added to these radiances. We will focus on one CrIS surface channel. A control run will be generated. It assimilates conventional data, microwave satellite radiances and simulated CrIS observations with constant instrument noise on all FOVs. The FOV 5 detector (center of a 3x3 array of detectors) NEdN will be used. Experiments will be conducted with the same configuration as the control but will assimilate perturbed CrIS observations at FOV 7. The magnitude of the perturbations will be multiples of the FOV 5 NEdN. Statistics such as FOV selected, O-B bias and RMS, will be determined with respect to each detector and presented.