The scientific problem of fusion analysis for geospatial big data is to obtain the structure, composition, spatial distribution and variation law of geological bodies from interdisciplinary data. The key technology is to carry out a geospatial three-dimensional reconstruction and visualization, and perform spatial analysis, prediction and decision. Spatial distribution and variability analysis of Ordos sandstone uranium mining is an important concern for uranium formation research. This study is based on big sample data of typical drilling and constructs a digital frequency spectrum model reflecting uranium oxidation-reduction space and lithofacies combination relationship, to describe uranium change of Zhiluo formation strata combination and spatial coupling relationship with the surrounding rock. Modeling process involves a series of key assessments which are structure analysis for big sample data of typical drilling, scale calculation, digital frequency spectrum and variation function principle, geological interpretation. Using stratum as minimum geological unit by spatial frequency spectrum calculation, this study can directly make frequency spectrum synthesis calculation with different levels such as stratum combination, sedimentary facies combination and basin, so as to build geological spatial digital feature model, which can be used as a basis for quantitative analogy and extrapolation prediction in the northern sandstone uranium basin.