AbstractPersistent cold air pools (CAPs) trap pollutants in valleys for extended periods, leading to reduced visibility and increased air pollution within these valleys. The structure of the persistent cold air pool that occurred in the Lanzhou Valley in December 2016 was simulated using different Planetary Boundary Layer (PBL) scenarios of the Weather Research and Forecasting (WRF) model, and the simulation of the persistent cold air pool was further optimized in these PBL scenarios. The simulation results indicated that weather‐scale dry subsidence and nighttime ground radiation cooling were significant factors contributing to the accumulation of persistent CAPs and pollutants in the Lanzhou Valley. In contrast, convective lifting from the ground led to the dissipation of persistent CAPs and a reduction in pollution within the valley. During persistent CAPs, the PM2.5 concentration and valley heat deficit (Q) were 66.7% and 62% higher, respectively, than during non‐CAP. In the original MYNN scheme, the average PBL height, double turbulent kinetic energy (QKE), and turbulence length scale during persistent CAPs decreased by 30.79%, 50.5%, and 34.4%, respectively, compared to non‐CAP. Compared with the original MYNN scheme, the optimized MYNN scheme shows a significant improvement in the turbulence simulation during the sustained CAPs, resulting in a more stable atmosphere. The PBL height during the sustained CAPs is reduced by 28 m, the diurnal turbulence length scale is reduced by 31.62%, the stability parameter is reduced by 39%, the diurnal mean QKE is reduced by 27.45%, and the QKE impact height is reduced by 100–400 m.