The precise characterization of the internal structures of small fracture-cavity clusters in carbonate rocks in the Tahe Oilfield is crucial for achieving efficient and large-scale development in this region. However, due to limitations of seismic resolution, predicting small-scale fracture-cavity units remains challenging. Furthermore, the complexity of the seismic wavefield in such reservoirs leads to ambiguous correlation between seismic anomalies and geological features, which impedes the accurate placement of hydrocarbon exploration targets. To address these issues, a study was conducted to evaluate the seismic identification capability of the internal structures of small-scale cavity clusters using forward modeling. By quantitatively comparing the performance of seismic data at different frequencies with full-frequency-band data in identifying the internal structures of these small cavity clusters, the resolution limits of various frequency data for delineating such structures were defined under specific conditions. The results demonstrate that prestack frequency-division (high-frequency) data can better identify closely-spaced cavities that are indistinguishable on conventional full-frequency-band data. Consequently, this approach allows for a more refined characterization of the scale and spatial distribution of fracture-cavity units, providing a reliable basis for optimizing well trajectory design.