Abstract: The Ordos Basin is rich in oil and gas resources. However, due to strong absorption of seismic waves by the thick unconsolidated loess formation, seismic data suffer from severe high-frequency attenuation, low dominant frequency, and poor wavelet consistency, which severely hinder the detailed prediction of deep thin reservoirs. To address this issue, this study proposes a viscoelastic medium Q-processing workflow that integrates full-stratum compensation from the near-surface to mid-deep zones. For near-surface compensation, the peak frequency shift method is applied to double-hole uphole survey data to obtain the Q-V relationship, and a high-precision 3D Q-field model is constructed using tomographic velocity fields. Furthermore, a stepwise compensation method in the shot and receiver domains is employed to accurately compensate for spatially variable near-surface absorption, effectively enhancing the resolution and wavelet consistency of the raw data. On this basis, a mid-deep Q-field model is built through VSP data constraints and Q tomographic inversion, and Q migration is then applied to achieve amplitude and phase compensation along the actual propagation paths of seismic waves. Field data application demonstrates that the frequency band is broadened by 15 Hz, deep weak signals are effectively restored, thin coal-seam reflections become clearer, and well-seismic consistency is improved by an average of 0.05. This effectively resolves the challenge of absorption and attenuation compensation in seismic data from the loess tableland area, providing reliable technical support for the exploration and development of thin tight gas and coalbed methane reservoirs in the Ordos Basin.