||Numerical air quality model (AQM) is the essential part of air quality management and planning since it is the only viable tool to evaluate the atmospheric responses to different control strategies through model perturbations. Errors from uncertainties in model inputs or internal module parameterizations may propagate through the model structure and undermine the feasibility of simulation results. In this study, the impact of uncertainties from emission, the representation of hydroxyl radical (OH) chemical mechanism as well as visibility regression formula to air quality modeling were investigated through the state-of-art MM5-SMOKE-CMAQ modeling system. The uncertainties in emission inventory (EI) estimates in Hong Kong were quantified for the first time using novel statistical methods featured in Bootstrap technique and Monte Carlo simulation. Motor vehicle, marine, and biogenic emission sources were identified as the major contributors to the overall EI uncertainty. Improvement of model performances by propagating the EI with uncertain bounds was limited and only appeared in few stations mainly due to the complex nature and nonlinearity of the modeling system. Two new chemical mechanisms, namely excited NO2 (NO2E) chemistry and nitrous acid (HONO) chemistry, with the potential to enhance OH formation and further ozone and particulate matters (PM) yields, were incorporated into standard CMAQ model. The impact of NO2E chemistry on ozone enhancement was moderate and only pronounced with the combination of abundant emission precursors and favorable meteorology conditions. Standard CMAQ model seriously underestimated HONO mixing ratio over PRD region by only considering gaseous phase formation pathway. Direct HONO emission, heterogeneous reactions, as well as surface photolysis reactions are needed to add into AQM as the additional important sources for HONO formation. Localized empirical regression formula for visibility simulation was implemented in this study to improve model performance and to understand the spatial-temporal variation of light extinctions over PRD region. Ammonia sulfate and black carbon dominated the light extinction budget over PRD region. Contributions from different emission sources to visibility degradation were apportioned using emission zero-out sensitivity studies. Model results suggested that beside local emission reduction, controlling the super regional (outside PRD region) sulfur source is also important for visibility improvement over PRD region.