In order to study the impact of land use change on the surface albedo accurately and quantitatively, a long-term global spectral surface albedo dataset (near-infrared, visible, shortwave) is obtained by using land use types dataset from 850 A.D. to 2100 A.D. provided by land-use harmonization (LUH2) project, and then detailed analysis and comparative verification are carried out. The results show that, the spatial distribution characteristics of spectral surface albedo obtained in this work are consistent with those of MODIS. The visible albedo in non-snow and non-ice covered area varies from-0.0081 to 0.0029 with high precision, while the largest difference is found in snow and ice covered area inthe middle and high latitudes of the southern and northern hemisphere. It is also shown that, since 1860, the urban built-up area has increased year by year in all typical regions (including Eastern China, Southeastern Europe, Mid-east Unite States and Southern Brazil), while the natural vegetations have decreased in Southern Brazil. From 1860 to 1980, natural vegetations were mainly transformed into urban and croplands in each region. In 1980～2015, croplands continued to increase in Southern Brazil, while the other regions showed a trend of gradual restoration of natural vegetations. In addition, the spectral albedo of different land cover types varies greatly. For example, the visible albedo in spring is 0.5069 on the surface of ice/snow, while itis 0.0444 in mixed forest and 0.0870 in urban. Under the influence of different latitudes and underlying surface properties, even the spectral albedo of the same land use type also has obvious spatial differences. For example, the maximum visible albedo of grassland in spring is 0.4915, and the minimum is 1.127×10-4. It is believed that the long-term global spectral albedo dataset can lay a data foundation for the quantitative research on the climate change driven by land use change.