Abstract: The development of thin sand bodies (thickness: 4~6 m) in H₃ reservoir, the Huoshaoshan Oilfield, Junggar Basin, faces vertical and horizontal prediction challenges. Given this, this paper proposed a technology that combines spectrum bluing for frequency broadening with dominant frequency band fusion of multi-source inversion data. Based on the blue spectrum of logging-derived reflection coefficient, the proposed technology expanded the pre-stack seismic high-frequency band to 120 Hz through well-seismic modeling. By doing so, it can enhance reservoir responses above 60 Hz and suppress low-frequency noises. The frequency-specific processing was then conducted for the three types of data volumes. Specifically, the low-frequency model (0~8 Hz) retained the structural trend, while the intermediate-frequency pre-stack waveform indicator inversion V_P/V_S volume (8~60 Hz) depicts the lithological boundary. The high-frequency pre-stack geostatistical inversion V_P/V_S volume (60~120 Hz) captures details of thin sand bodies. The adaptive weighting allocation achieved full-band synergistic integration, overcoming the identification limitations (8‒10 m seismic tuning thickness) of traditional inversion methods. The application of this technology shows that the fused data volumes can clearly identify thin sand bodies (4 m), with the lateral inter-well connectivity error reduced by 25% compared to conventional inversion methods. This technology established a multi-scale collaborative model of "trend controlling in low-frequency bands, boundary controlling in intermediate-frequency bands, and detail capturing in high-frequency bands", breaking through the bottleneck of multi-solutions from single-source inversion. The information fusion of full-frequency bands provides a replicable technical path for the detailed assessment of continental clastic reservoirs with thin interbeds. It also offers methodological inspirations for the development of similar complex reservoirs.