Accurate modeling of rockfalls and pyroclastic flows is still an open issue, partly due to a lack of measurements related to their dynamics. Using seismic data from the Soufrière Hills Volcano, Montserrat, and granular flow modeling, we show that the power laws relating the seismic energy Es to the seismic duration ts and relating the loss of potential energy ΔEp to the flow duration tf are very similar, like the power laws observed at Piton de la Fournaise, Reunion Island. Observations showing that tf≃ts suggest a constant ratio Es/ΔEp≃10−5. This similarity in these two power laws can be obtained only when the granular flow model uses a friction coefficient that decreases with the volume transported. Furthermore, with this volume‐dependent friction coefficient, the simulated force applied by the flow to the ground correlates well with the seismic energy, highlighting the signature of this friction weakening effect in seismic data.

[1] This paper investigates the variation of the first resonance frequency of the Chamousset limestone column (21,000 m 3 , Vercors, French Alps) before its collapse in November 2007. The site was instrumented with seismometers and extensometers during a 4-month period with some gaps in the measurements. Experimental results and numerical modeling showed that the resonance frequency of a prone-to-fall column can be derived from the spectra of continuous seismic noise records. At the Chamousset site, the evolution of the resonance frequency appeared to be strongly controlled by the temperature. When temperatures were positive, slight resonance frequency variations correlated well with thermal fluctuations. Irreversible damage can occur during freezethaw cycles and to a lesser extent during strong wind. It coincided with a significant drop in resonance frequency, which was interpreted as the result of rock bridge breakage. This hypothesis is supported by fresh rupture observations after the collapse, seismic event records, and numerical modeling. This study suggests that seismic noise recording could be used for assessing the potential failure of unstable columns in rigid rocks.