Machine Learning‐Based Denoising of Surface Solar Irradiance Simulated With Monte Carlo Ray Tracing

• Published in: Journal of geophysical research. Machine learning and computation Vol. 2; no. 3
• Main Authors: Reeze, M., Veerman, M. A., van Heerwaarden, C. C.
• Format: Journal Article
• Language: English
• Published: 01.09.2025
• DOI: 10.1029/2024JH000515
• ISSN: 2993-5210

Simulating radiative transfer in the atmosphere with Monte Carlo ray tracing provides realistic surface irradiance in cloud‐resolving models. However, Monte Carlo methods are computationally expensive because large sampling budgets are required to obtain sufficient convergence. Here, we explore the use of machine learning for denoising direct and diffuse surface solar irradiance fields. We use Monte Carlo ray tracing to compute pairs of noisy and well‐converged surface irradiance fields for an ensemble of cumulus cloud fields and solar angles, and train a denoising autoencoder to predict the well‐converged irradiance fields from the noisy input. We demonstrate that denoising diffuse irradiance from 1 sample per pixel (per spectral quadrature point) is an order of magnitude faster and twice as accurate as ray tracing with 128 samples per pixel, illustrating the advantage of denoising over larger sampling budgets. Denoising of direct irradiance is effective in sunlit areas, while errors persist on the edges of cloud shadows. For diffuse irradiance, providing additional atmospheric information such as liquid water paths and solar angles to train the denoising algorithm reduces errors by approximately a factor of two. Our results open up possibilities for coupled Monte Carlo ray tracing with computational costs approaching those of two‐stream‐based radiative transfer solvers, although future work is needed to improve generalization across resolutions and cloud types. As atmospheric models move toward higher resolutions and the demand for solar energy forecasts increases, there is an urgent need for accurate modeling of solar radiation. One of the most accurate techniques to simulate the transfer of radiation through the atmosphere is Monte Carlo ray tracing. However, this technique requires many computational resources, limiting its application in operational contexts. This problem has been extensively researched in the movie and gaming industry, where state‐of‐the‐art solutions use machine learning algorithms to approximate illumination in artificial scenes. Here, we explore the use of these algorithms in the context of atmospheric modeling, with a focus on incoming solar radiation under broken clouds. Using this approach, we are able to deliver close approximations of realistic surface irradiance at a fraction of the original computational cost. We train a denoising autoencoder for Monte Carlo ray tracing estimates of surface solar irradiance fields below cumulus clouds Denoising diffuse irradiance exceeds accuracy of higher sampling budgets that require over tenfold the computational cost Denoising direct irradiance is effective in sunlit areas but significant noise remains near cloud shadow edges