A Verification of the Ozone Monitoring Instrument Calibration Using Antarctic Radiances

A technique for evaluating the radiometric calibration of satellite-born radiometers has been developed utilizing the stable reflectance of the Antarctic land mass. Previously, the scene radiances measured over the Antarctic and Greenland land masses were used to monitor time evolution of sensor radiometric sensitivity and to compare instruments with similar spectral responses. Through the use of radiative transfer modeling, we can now evaluate the albedo calibration (Earth radiance sensitivity divided by solar irradiance sensitivity) of any nadir-viewing, polar orbiting sensor measuring in the 300-800 nm range. Our evaluation of the Ozone Monitoring Instrument on the NASA Aura spacecraft confirms a 2% calibration uncertainty. The Ozone Monitoring Instrument The Ozone Monitoring Instrument (OMI) launched aboard the Earth Observing System (EOS) Aura satellite on 15 July 2004 is intended as the successor to the Total Ozone Mapping Spectrometer (TOMS) operated by NASA over the past 25 years. With improved horizontal resolution and an expanded wavelength range (270-500 nm), the OMI capabilities extend well beyond those of TOMS. Traditional TOMS products (column ozone, surface reflectance, aerosol index, UV surface flux) depend upon knowledge of absolute albedo calibration and the relative albedo calibration between wavelengths. Establishing an accurate OMI calibration and maintaining calibration consistency with the TOMS instruments are of equal importance. Furthermore, future data products will likely utilize OMI-measured radiances combined with radiances and products from other sensors, on Aura as well as on spacecraft in similar orbits. Given that no attempt was made to inter-calibrate OMI with other sensors prior to launch, a relative adjustment based upon in-flight data may be required to improve these products. The ice radiance technique Our approach to verifying a UV/VIS/NIR sensor’s albedo calibration is to compare measured and computed radiances for Earth scenes with known backscatter properties. We have chosen the surface and atmosphere of Antarctica for their highly predictable behavior over a broad range of wavelengths and viewing conditions (Jaross et al.). The spatial extent of the continent provides good sample statistics for sensors with very large footprints, a characteristic that provides for verification of older data sets. The stable nature of the Antarctic radiances arises from the high surface reflectivity (> 90%), the regenerative nature of the pure snow coverage, and the relatively low aerosol concentrations. The reflective properties of the Antarctic surface have been measured by numerous groups, most thoroughly and convincingly by Warren et al. They have published total hemispheric reflectance (Grenfell et al.) and a parameterization of reflectance anisotropy (Warren et al.) based on a variety of ground locations and viewing conditions. Their results compare well with model predictions for deep snow-covered surfaces. One important conclusion is that reflectance varies little between 300 nm and 650 nm, a range of primary interest for many Earth backscatter-measuring sensors. Reflectance anisotropy is an important characteristic to include in radiative transfer models of snow surfaces. The top-of-the-atmosphere (TOA) radiance difference between a Lambertian and a forward scattering snow surface with the same reflectance grows from 2% at 350 nm to 7% at 600 nm. We have derived a mean bidirectional reflectance distribution function (BRDF) for the Antarctic surface using the Warren et al. anisotropy. We employed combinations of reciprocity and interpolation to generate a model covering 2 steradians in reflected angles and solar incidence angles down to 0°. Though the continent never receives direct illumination at less than 45° incidence, the radiative transfer model must account for diffuse scattered light that can arrive at any incident angle. Jaross et al. estimated that BRDF knowledge represents the largest error component in the TOA radiance predictions, approximately 2% at moderate solar zenith angles (SolZA). In order to compare measured and calculated TOA radiances we created a look-up table of radiances using a Gauss-Seidel atmospheric modeling code (Herman et al.). This table has nodes covering all possible satellite viewing conditions, sensor wavelengths, surface pressures, and column ozone amounts. It does not include minor absorption species, aerosols, clouds, or a correction for rotational Raman scattering in the atmosphere. Monochromatic calculations were used to represent measurements with finite bandwidths, which are 0.4-0.7 nm in the case of OMI. A difference is computed between every valid Antarctic radiance measurement and a corresponding interpolated value from the table.