<![CDATA[Geotechnical construction failures and sinkhole risks in critical infrastructure areas are often exacerbated by the unpredictable nature of karst subsurface structures. The identification of subsurface cavities in limestone poses significant challenges due to material heterogeneity and the complexity of wave propagation. This study aims to evaluate the response of seismic tomography inversion models to real-world conditions, particularly in detecting the influence of cavities and secondary structures on P-wave velocity ( ) characteristics. Data acquisition was conducted in the Kalipalung Formation using five seismic refraction lines with a constant inversion scheme of 20 iterations. The results demonstrate that model accuracy is highly sensitive to surface geometry, where the Root Relative Mean Square Error (RRMSE) increased drastically from 7.97% on flat topography to 39.79% on steep slopes due to wavefield scattering phenomena. Physically, the inversion model successfully identified lithological zoning with a range of 300–3900 m/s. However, the presence of secondary structures, such as massive stalactites and calcite recrystallization, causes cavity anomalies (300–600 m/s) to appear fragmented (spotty) and discontinuous. This phenomenon confirms the presence of volume averaging and blind zone effects that mask the acoustic impedance contrast between air and the host rock. Thus, while seismic tomography is effective in mapping subsurface material heterogeneity, the internal complexity of karst systems and extreme topography remain the primary limiting factors in fully delineating cavity geometry. This evaluation highlights the necessity of topographic static corrections and high-density sensor arrays to reduce interpretation uncertainty in extreme karst zones]]>
Copyrights © 2026
