The WRF model often struggles to accurately replicate specific characteristics of the atmospheric boundary layer, particularly under highly stable conditions. In this study, we reconstructed an OBS-nudging module using meteorological data with high spatiotemporal resolution, then coupled it in the WRF model (WRF-OBS) to improve stable boundary layer (SBL) simulation over the North China Plain (NCP). The results showed that WRF-OBS improved the simulation of SBL characteristics and reduced the deviation from observations significantly. The correlations (R2) between WRF-OBS simulations and observations of 2 m temperature, relative humidity, and 10 m wind speed at 460 stations across the NCP were 0.72, 0.56, and 0.75, respectively, which were much higher than the values for results from the unassimilated WRF model (WRF-BS). The simulated vertical profiles of temperature, relative humidity, and wind were generally consistent with observations at Pingyuan station. The meteorological factors which caused heavy air pollution was also investigated based on WRF-OBS simulation. The SBL characteristics obtained from WRF-OBS showed that light wind persisted over the NCP region during the period of heavy pollution, and Pingyuan was affected by warm and humid air. Vertically, the persistent temperature inversion at Pingyuan station was one of the main drivers of the heavy pollution. The WRF-OBS simulation captured the characteristics of the two temperature inversion layers very well. The two inversion layers covered the NCP, with a horizontal scale of approximately 200 km, and created very stable conditions, preventing the vertical diffusion of pollutants and maintaining high PM2.5 concentrations.