Global observations of the carbon budget, 2, CO2 column from differential absorption of reflected sunlight in the 1.61 μm band of CO2

Abstract

This paper investigates the impact of uncertainty in atmospheric composition and state upon the feasibility of measuring the CO2 column from spectral analysis of sunlight reflected to space in the 1.61 μm absorption band of CO2. In principle, measurements of clear sky radiance at two frequencies, one where CO2 absorbs strongly and the other weakly, allow the difference between the optical thicknesses of the atmosphere at the two frequencies to be determined precisely. That difference, denoted by L, is a linear functional of the CO2 density profile, which depends strongly on the CO2 amount and only weakly upon its vertical distribution, thus suggesting that the CO2 column may be estimated from L. A simple model for the radiance reflected to space is used to estimate the magnitude of the error in the CO2 column inferred from L when the atmosphere contains thin cloud and aerosol. It emerges that measurements in two channels in the 1.61 μm CO2 absorption band are too sensitive to cloud and aerosol to allow the CO2 column to be inferred with precision better than a few percent in the presence of thin cloud and aerosol. However, simultaneous measurements of optical thickness in the nearby 1.27 μm absorption band of O2 are tightly correlated with those for CO2, even in the presence of aerosol and thin cirrus, and therefore may allow the CO2 column to be determined relative to the O2 column, provided that the latter is known independently from surface pressure. The correlation between O2 and CO2 optical thicknesses depends upon the mean scattering height, but this quantity may be estimated with sufficient accuracy from radiances measured in the O2 band. A prototype algorithm is developed to estimate the CO2 column from data in two CO2 channels and three O2 channels. The algorithm is used to estimate the probable bias and standard error of measurements of CO2 column from space under conditions where the optical thicknesses of aerosol and cirrus may be as large as 0.2 and 0.1, respectively, and where the temperature profile is known to within ±1 K. The simulations suggest that the error in the estimated CO2 column caused by these sources is approximately 0.5%. This conclusion is interpreted cautiously because the analysis assumes inter alia that the spectroscopic properties of both CO2 and O2 are known accurately, that the surface reflectance and the scattering properties of aerosol and cirrus vary predictably between 1.27 μm and 1.61 μm, and that difficult technical issues associated with high spectral resolution measurements can be resolved. Nevertheless, the importance of global measurements of CO2 is such that the method warrants further investigation.