A Theoretical Analysis for Improving Aerosol-Induced CO2 Retrieval Uncertainties Over Land Based on TanSat Nadir Observations Under Clear Sky Conditions

Aerosols significantly affect carbon dioxide (CO2) retrieval accuracy and precision by modifying the light path. Hyperspectral measurements in the near infrared and shortwave infrared (NIR/SWIR) bands from the generation of new greenhouse gas satellites (e.g., the Chinese Global Carbon Dioxide Monitoring Scientific Experimental Satellite, TanSat) contain aerosol information for correction of scattering effects in the retrieval. Herein, a new approach is proposed for optimizing the aerosol model used in the TanSat CO2 retrieval algorithm to reduce CO2 uncertainties associated with aerosols. The weighting functions of hyperspectral observations with respect to elements in the state vector are simulated by a forward radiative transfer model. Using the optimal estimation method (OEM), the information content and each component of the CO2 column-averaged dry-air mole fraction (XCO2) retrieval errors from the TanSat simulations are calculated for typical aerosols which are described by Aerosol Robotic Network (AERONET) inversion products at selected sites based on the a priori and measurement assumptions. The results indicate that the size distribution parameters (r(eff), v(eff)), real refractive index coefficient of fine mode (a(r)(f)) and fine mode fraction (fmf) dominate the interference errors, with each causing 0.2-0.8 ppm of XCO2 errors. Given that only 4-7 degrees of freedom for signal (DFS) of aerosols can be obtained simultaneously and CO2 information decreases as more aerosol parameters are retrieved, four to seven aerosol parameters are suggested as the most appropriate for inclusion in CO2 retrieval. Focusing on only aerosol-induced XCO2 errors, forward model parameter errors, rather than interference errors, are dominant. A comparison of these errors across different aerosol parameter combination groups reveals that fewer aerosol-induced XCO2 errors are found when retrieving seven aerosol parameters. Therefore, the model selected as the optimal aerosol model includes aerosol optical depth (AOD), peak height of aerosol profile (H-p), width of aerosol profile (H-w), effective variance of fine mode aerosol (v(eff)(f)), effective radius of coarse mode aerosol (r(eff)(c)), coefficient a of the real part of the refractive index for the fine mode and coarse mode (a(r)(f) and a(r)(c)), with the lowest error of less than 1.7 ppm for all aerosol and surface types. For marine aerosols, only five parameters (AOD, H-p, H-w, r(eff)(c) and a(r)(c)) are recommended for the low aerosol information. This optimal aerosol model therefore offers a theoretical foundation for improving CO2 retrieval precision from real TanSat observations in the future.