A method for calibration of historical satellite sensors in the solar region using statistics of deep-convective cloud observations from MODIS, VIRS/TRMM and SEAWIFS

The long-term satellite data records available from the optical channels of the Advanced Very High Resolution Radiometer (AVHRR) and other sensors are a valuable source of information for research on climate variability and operational monitoring. This has been successfully demonstrated in many studies. Their application for detecting climate trends and use in climate change related studies is more difficult, requiring a high level of accuracy and consistency of sensor calibration to derive trends. The absence of onboard calibration and in-flight degradation of sensors with time constitute two major challenges. Despite numerous efforts, there are still substantial discrepancies between various sources of AVHRR calibration that may reach up to 20 % or more.

In the absence of onboard calibration, the vicarious calibration techniques are the only alternative. Vicarious calibration using observations of surface targets is vulnerable to uncertainties related to variability of atmospheric state, cloud contamination and surface characterization. Approaches based on sensor intercalibration require matching observational geometry, space-time collocation and spectral correction for differences in band spectral response. This may be too resource intensive and is limited to periods where suitable conditions exist for such comparisons. Further, the intercalibration approach still requires knowledge of absolute calibration for at least one sensor participating in the comparison.

A reasonable alternative is to use high altitude optically thick deep convective clouds. The impact of atmosphere in this case is much smaller, as only ozone and stratospheric aerosol contribute to major uncertainties of spectral measurements at this altitude. To reduce the impact of 3-D effects and uncertainty related to angular distributions, we propose to use only nadir view observations. To demonstrate the feasibility of this approach, we analyzed 5 years of MODIS, VIRS/TRMM and SEAWIFS observations of high altitude (T<210K) optically think clouds in the tropical zone over the Atlantic and Pacific oceans. These three sensors are known to be well calibrated. It is found that the distribution of reflectances within small solar zenith angle (SZA) bins is quite stable over the years (within +/- 0.03) and consistent among all three sensors. This allows us to build an empirical model of nadir reflectance against SZA for cold thick deep convective clouds. Assuming the invariant properties of these clouds over time, this model then can be applied as the reference calibration standard to derive calibration of historical AVHRR and other sensors in the solar region.