During a dome collapse, the source of ground vibration can move down a valley for several minutes. At Soufrière Hills Volcano I worked with colleagues to follow pyroclastic flows using the changing strength of their seismic signals at different stations.
At the Montserrat Volcano Observatory (2000–2003), I developed a real-time system that estimated energy magnitude and used amplitude source location (ASL) to track pyroclastic flows. I am rebuilding the system with Python and ObsPy. Maps of the estimated source positions through time help visualize dome-collapse trajectories and investigate what information might be useful for early warning.
Following a flow through the network
We compare amplitudes across stations over a short time window and ask which source position best explains that pattern. Repeating the comparison gives a sequence of estimated positions. Station coverage, local amplification, and attenuation all affect how confidently we can interpret the path.
Art Jolly, Glenn Thompson, and G. E. Norton described this application at Soufrière Hills in 2002. See my publication profile.
A related experiment in the Belham Valley
For a rain-triggered flow on Montserrat, stations close to the channel let us compare arrival times along the valley. The Belham Valley field study shows those recordings and how timing was used to estimate flow speed.