Method for in‐flight satellite calibration in the ultraviolet using radiative transfer calculations, with application to Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY)

Abstract

[1] We present a reflectance validation method, using a polarized radiative transfer model as a reference. The method involves calculating the complete radiative transfer in the ultraviolet for a carefully selected, natural, cloud-free Earth target. An extensive sensitivity study shows that in the ultraviolet, the ozone profile (mainly below 330 nm) and the surface albedo (mainly above 330 nm) are the important parameters that affect the reflectance and are therefore the major sources of error in the modeled reflectances. The accuracy with which the reflectance in the UV range of 240-400 nm can be calculated, because of uncertainty in input parameters, is 2-6%, depending on wavelength. The method is applied to Scanning Imaging Absorption Spectrometer for Atmospheric Chartography (SCIAMACHY), using a cloud-free scene over the Sahara desert. The reflectance measured by SCIAMACHY turns out to show a scan-angle-independent calibration error of 10-25% for this particular Sahara target. The result is confirmed for ocean and vegetation targets.Citation: Tilstra, L. G., G. van Soest, and P. , Method for in-flight satellite calibration in the ultraviolet using radiative transfer calculations, with application to Scanning Imaging Absorption Spectrometer for Atmospheric Chartography

Large-scale validation of SCIAMACHY reflectance in the ultraviolet

“…It gives rise to a positive error of about 0.02 in d R , smaller for shorter wavelengths. The effect can be compensated in principle by adjusting the ground albedo, but our study of the albedo sensitivity of the TOA radiance (Tilstra et al, 2005) demonstrates that this is not straightforward, as for most common surface types the sensitivity is wavelength dependent.…”

“…We only apply this method in the UV because this spectral range can be modelled well with relatively coarse guesses for the atmospheric state and ground albedo. The sensitivity of the simulation to input parameters, as well as an end-to-end error estimate of the use of RTM simulations as a validation method, are the subjects of another paper (Tilstra et al, 2005). For a full discussion we refer to that work.…”

“…Its main conclusion is that the standard deviation of the error in the simulation due to input parameters is approximately 3% over the wavelength range 250-400 nm, and up to 8% around 305 nm where the sensitivity to the ozone profile and column is largest. The sensitivity study of Tilstra et al (2005) does not include clouds. Our cloud filtering procedure adds an offset between 0% and 2% to their error estimate for λ>310 nm, which is largest for the longer wavelengths in the study.…”

“…The RTM we use is LidortA (van Oss and Spurr, 2002). This is a fast code based on the discrete ordinates algorithm (Chandrasekhar, 1950).…”

“…The estimated parameters are the input for the RTM, which calculates the spectrum expected from the satellite instrument. We choose to employ this technique in this paper, and apply it to the UV part of the spectrum, for two reasons: first, it complements the validation by comparison with other satellite instruments, and second, the sensitivity of the UV spectrum to the ground albedo and clouds, which are the main uncertainties in the input parameters, is small, especially for λ<350 nm (Tilstra et al, 2005). The major error source around 300 nm is ozone, which can be estimated accurately.…”