Different parameters can be extracted from the spectral analysis of ambient vibrations for characterizing rocky prone-to-fall columns. 2D numerical modeling was applied to identify three seismic parameters (resonance frequency and two spectral ratio amplitudes) able to characterize the column decoupling. These numerical results were compared to data acquired at two rocky sites exhibiting cliff-like geometry, showing the influence of additional rear fractures. The resonance frequency, which appeared to be the more robust parameter, turned out to vary with temperature. The similarity between the two time series was used to separate the reversible thermal effects from those resulting from damaging.