Abstract: To enhance the forecasting capability of the China Meteorological Administration Mesoscale (CMA-MESO) model in high-resolution domains and complex terrain, a vertical layering scheme (xu scheme) was developed. The scheme is designed for height-based terrain-following coordinates using continuous functions. It allows flexible adjustment of vertical layering and model top height and introduces a parameter b to control the bottom-layer thickness. The scheme is numerically stable and maintains continuous and smooth layer thickness, with denser layers near the surface. The xu_71 scheme (implemented 71 layers using xu scheme) was implemented in CMA-MESO V6.0 and showed characteristics similar to those of the meso_71 scheme (operational 71 layers scheme), but with better vertical profile continuity. Its performance was evaluated through batch forecast comparisons using key metrics, including the root-mean-square error on isobaric surfaces, 2-m temperature forecasts, and threat scores for accumulated precipitation. Results indicated that the xu_71 scheme slightly outperformed the meso_71 scheme across these key metrics. Additionally, the influence of b on 2-m temperature forecasts was analyzed using surface-layer parameterization, revealing that uncertainties inherent in stable similarity functions can increase diagnostic errors in 2-m temperature forecasts, especially when combined with thicker lowest model layer height (controlled by b) and complex terrains. Numerical analyses of dense layering during heavy rainfall forecasts showed that the xu_91 scheme (91 layers) performed better than coarser configurations. Specifically, the xu_91 scheme’s forecasts were closer to the actual observations for accumulated-precipitation areas exceeding 100 mm. Compared with the xu_51 scheme (51 layers), xu_91 predicted deeper updraft zones, more distinct vertical circulations, better vertical hydrometeors coordination, higher boundary layer heights, and stronger sensible heat fluxes. Overall, the designed vertical layering scheme is effective, easy to apply, and has the potential for enhanced application in physical parameterization studies.