The aerosol single-scattering albedo (SSA) refer to the characterization of aerosol absorption, which determines the positive and negative aerosol radiative forcing and accurate assessment of aerosol radiative forcing in climate change is of great significance. Based on AERONET (AERosol RObotic NETwork) and OMI (Ozone Monitoring Instrument) data from October 2004 to December 2016, the trend of aerosol SSA and comparison agreement with AERONET and OMI in the north of China (Beijing, Xianghe, Xinglong, and Lanzhou) were analyzed. The SSA annual mean of four sites were: Beijing 0.89±0.04 (AERONET) and 0.90±0.04 (OMI); Xianghe 0.89±0.04 (AERONET) and 0.91±0.04 (OMI); Xinglong 0.92±0.04 (AERONET) and 0.91±0.04 (OMI); Lanzhou 0.91±0.04 (AERONET) and 0.90±0.04 (OMI). All four sites showed the same seasonal variations, with high values in summer and low values in winter. Due to the limited data, the monthly average can not simply use to analyze the annual trend, we need to screen the data and remove the seasonal variation. The result showed that ground-based network and satellite had different time scales in both Beijing and Xianghe, while the OMI in Xinglong and AERONET in Lanzhou were meet the data requirements for trend analysis. The SSA of all four sites had the significant increases, indicating that the absorption of aerosol in the northern China have weaken and scattering have been enhanced in recent years. In particular, the AERONET and OMI retrievals of SSA were found an upward trend in four seasons at Beijing, but the absorption in autumn and winter was enhanced at Xianghe. In addition, due to the slight difference between AERONET and OMI, the data of AERONET and OMI were compared to accuracy of the results. The data of Xianghe revealed much bigger gap between the two inversions, which 30% of data agree within the absolute difference of ±0.03, and 55% agree within the difference of ±0.05; and 46% of Beijing data within ±0.03 and 68% within ±0.05, Xinglong 50% within ±0.03 and 78% within ±0.05; the data of Lanzhou was much better consistent, which 51% (86%) of data made agreement within the difference of ±0.03 (±0.05). In total, the data at sites which were more affected by human activities were consistently poor.