ACCELERATED HIGH-RESOLUTION RADIATIVE TRANSFER SIMULATION FOR CO2 CONCENTRATION ESTIMATION FROM NANOCARB MEASUREMENTS

Jordan Lontsi Tedongmo, Pablo Musé, Andres Almansa, Gabriele Facciolo

TL;DR: We propose a feedforward MLP surrogate for efficient radiative transfer modeling in the CO₂ weak band, trained with combined radiance and Jacobian losses, enabling precise and fast CO$_2$ concentration retrievals using the NanoCarb concept.

Abstract:

Studying climate change requires reducing uncertainties in CO$_2$ and CH$_4$ emission estimates to better distinguish anthropogenic from natural sources, which motivates spaceborne measurements with improved revisit and coverage. In this context, the Horizon Europe SCARBOn project assesses a low-cost satellite constellation featuring the NanoCarb imaging interferometer as its core sensor for monitoring CO$_2$ and CH$_4$ emissions in the atmosphere. However, estimating CO$_2$ and CH$_4$ concentrations from NanoCarb measurements poses significant challenges, as the full-physics retrieval algorithm commonly use rely on repeated high-resolution radiative transfer (RT) simulations, which are expensive with line-by-line RT models. As an alternative, we propose in this study to train a feedforward multilayer perceptron (MLP) surrogate to predict top-of-atmosphere radiances in the CO$_2$ weak band, using a combined MAE loss on radiances and RT Jacobians to preserve both spectral accuracy and sensitivities to geophysical parameters. Coupling the MLP
RT model with the NanoCarb instrumental response yields an efficient and precise forward model for NanoCarb measurements, which gives promising results for CO$_2$ concentration retrievals.

Submission Number: 38