An improved correction framework based on an optimized Modified Quantile-Quantile Adjustment (MQQA) method was proposed, with a novel two-step bridging strategy to ensure continuity into the pre-MODIS era (1995~1999). Taking Deep Blue AOD as the reference baseline to harmonize historical AVHRR and MODIS datasets, a 30-year (1995~2024) bias-corrected daily AOD dataset over China was reconstructed based on the MQQA-based framework. Validation against AERONET shows the corrected AVHRR AOD regression slope increased from 0.81 to 0.91, while the harmonized MODIS record achieved a 10-20% reduction in RMSE, with 74-79% data falling within the Expected Error envelope.Relative to the VIIRS baseline, the harmonized dataset reduced AOD RMSE by ~40% and bias standard deviation by 60-70%, effectively mitigating spatial artifacts and spurious trend shifts across sensor generations.These improvements collectively ensure the construction of a temporally consistent, high-continuity AOD record, thereby providing a robust dataset for long-term air quality assessment, climate studies, and environmental applications over China.