Turbulent diffusion and volcanic gas dispersion in the Atmospheric Surface Layer: Insights from La Solfatara, Campi Flegrei, Italy

Numerical models are widely used to simulate volcanic gas dispersion and estimate local emission sources. However, significant uncertainties arise from the approximations inherent in their physical formulations. Recent advances in high-performance computing (HPC) have enabled high-resolution simulations with minimal numerical diffusion, revealing previously unnoticed limitations in the Monin-Obukhov Similarity Theory used within atmospheric gas dispersion models. One key issue is the determination of the minimum vertical turbulence diffusion coefficient ( Kz _min ) in the Atmospheric Surface Layer (ASL), which plays a crucial role in reducing biases in advection-diffusion models caused by inadequate turbulence representation. In this study, we refine the Eulerian passive gas transport model DISGAS (v. 2.5.1) using measured data on fumarolic and diffuse CO₂ fluxes and air concentrations, along with local wind measurements collected during an ad hoc field campaign from 4–10 May 2023. To account for uncertainties in gas flow rates and turbulent velocity fluctuations, we conducted a statistically robust set of simulations by varying CO₂ fluxes and Kz _min values. Model outputs were compared with in situ CO₂ concentration measurements at fixed monitoring stations. Results indicate that during stable atmospheric conditions, setting Kz _min within the range of 1.5–2 m² s ^− 1 significantly improves agreement with observations and reduces systematic biases in source estimation. These findings refine model parameterization to better represent turbulence under stable atmospheric conditions at La Solfatara crater during the May 2023 survey. Moreover, the proposed methodology can be adopted for automated data assimilation workflows aimed at constraining unknown fumarolic gas source fluxes in other volcanic settings.