Tropical sensitivity to Vertical Grid Spacing in the Korea Integrated Model

Abstract

This study investigates the sensitivity of simulated temperature to vertical grid spacing in the troposphere using medium-range forecasts with the Korea Integrated Model (KIM). To examine the impacts of finer vertical resolution, model performance was compared between configurations employing 91 and 137 vertical layers. Increasing the vertical resolution produces persistent cooling throughout the troposphere, with particularly strong signals in the lower and upper troposphere. This cooling cannot be explained solely by the ability to resolve an initial condition or the dynamical adjustments. Instead, physically generated cooling becomes apparent in the early stage of the simulation, indicating that physical parameterizations play the dominant role in generating the tropical cooling signal. A key physical structural change associated with enhanced vertical resolution is an increase in boundary layer height over oceanic regions, together with enhanced cloud-water and cloud-ice mixing ratios. Finer vertical grid spacing more effectively resolves the nonlinear potential temperature gradient near the boundary layer top, producing a colder and moister environment above the boundary layer and strengthening the low-level cloud-water and longwave radiative cooling. In the upper troposphere, increased ice mixing ratio is found to originate primarily from processes represented within the cumulus parameterization framework. However, the increase occurs regardless of whether convective activity strengthens or weakens, suggesting that the response is not controlled solely by enhanced convection intensity. Instead, the results indicate an important contribution from numerical discretization effects within the mass-flux-type cumulus parameterization under refined vertical resolution. These discretization effects promote the retention of more subgrid-scale condensates within convective plumes, leading to enhanced hydrometeors and radiative cooling throughout the troposphere. The findings indicate that higher vertical resolution does not necessarily guarantee improved forecast performance, emphasizing the need for careful evaluation and retuning of physical parameterizations when model resolution is modified.