We propose an available method to generate land surface shortwave albedo at a 30 m spatial resolution in eight-day i nterval. The method uses at least one pair of fine-resolution remote sensing images（ e. g.

Landsat data） and coarse-resolution r emote sensing albedo products（MODIS albedo） to estimate the 30 m albedo on the date when the MODIS albedo is given without Landsat data. When MODIS albedo time-series is given

the time-series of a 30 m resolution albedo can be estimated. The method consists of three main steps:（1） generating fine-resolution shortwave albedo images based on Landsat ETM + data.（2） generating coarse-resolution shortwave albedo images based on MODIS albedo products.（3） using STARFM model to blend spatial i nformation from fine-resolution shortwave albedo images and temporal information from coarse resolution shortwave albedo i mages to estimate high spatio-temporal resolution albedo. Comparison with actual ETM + albedo shows that the predictions are of reasonable accuracy

and maintain a high level of spatial detail. For the problem of degrading estimation accuracy in the STARFM model for heterogeneous r egions

this study presents an improvement by introducing the idea of mixed pixel decomposition. In u sing this method the predicted albedo shows bigger correlation coefficients and smaller deviations with actual ETM + albedo for heterogeneous regions.