Synthetic Study of a Surrogate-Model-Based CO2/CH4 Retrieval Algorithm from Infrared Reflectance Measurements

Abstract

Simulating the solar radiation spectrum reflected by the Earth atmosphere for trace gas retrievals, such as CO2 and CH4, typically involves solving the radiative transfer equation (RTE), a computationally intensive process due to the multiple scattering term. To accelerate computations, we utilize a fast Python-based radiative transfer model FALCAS that approximates multiple scattering by introducing a virtual layer with isotropic scattering. This allows us to mimic the results of models with full multiple scattering treatment, while significantly reducing the computation time. We develop an algorithm for retrieving CO2 and CH4 concentrations using FALCAS and evaluate its accuracy against the discrete ordinate model PYDOME. The CO2 and CH4 retrievals are performed using FALCAS from synthetic spectra generated by PYDOME in the 1500–1750 nm spectral range, considering various aerosol and geometric conditions. The error in retrieved concentrations is about 2.0% for an aerosol optical thickness of about 0.9. This result remains robust under noise and convolution effects as well as across varying geometries, with errors under 5% at a signal-to-noise ratio of 50.