We developed a fully-coupled meso-scale model WRF-Chem-SMOKE by incorporating a selection of smoke emission models and improving the representations of aerosol-cloud interactions in the microphysics scheme. We find that the difference in smoke emissions between different datasets, even in one fire cluster, could lead to significant discrepancies in modeled AODs. The integrated smoke emission dataset improves the prediction of modeled AODs. We find that the modeled cloud properties and precipitation are extremely sensitive to the smoke loadings. Higher smoke loadings suppress precipitation initially, because of smoke-induced reduction of the collision-coalescence and riming processes, but ultimately cause an invigoration of precipitation.