UCL Earth Sciences


Forecasting Volcanic Eruptions

Conventional analyses of the precursory trends for developing deterministic forecasts

Volcanic eruptions are commonly preceded from days to months by increasing rates of ground deformation and of local fracturing of the crust. Both rates of change have shown repeatable variations with time, suggesting a potential for yielding reliable forecasts of eruptions. Conventional analyses of the precursory trends have developed either deterministic forecasts from empirical models, or probabilistic forecasts by applying statistical models to the empirical trends. In both cases, the reliability of forecasts has been limited by the empirical nature of the primary analyses. Since 1997, research at UCL has focused on identifying the physical controls on the empirical trends to identify the limits within which deterministic analyses are valid, to increase the confidence in deterministic forecasts when applicable, and to improve understanding of catastrophic rock failure at volcanoes and elsewhere. The results were used successfully to forecast the 2011 eruption of El Hierro (Canary Islands). 

Forecasting Volcanic Eruptions
A natural approach to forecasting eruptions is to examine the variation of common precursors with time and, by extrapolating these trends into the future, to identify when an eruption is likely to occur. However, the two most common precursors – increasing rates of ground deformation and of local fracturing of the crust – are not independent factors: both are expressions of how a rock responds to a change in applied stress or a change in rock resistance. To understand precursory behaviour, therefore, it is essential first to understand how fracturing and deformation vary with each other and, second, to identify how changes with time in applied stress or rock resistance can produce the observed time series among precursors.

The analyses of precursory trends has been developed in three key stages: to identify the nature of evolutionary trends with time; to use limiting trends to identify the macroscopic and microscopic controls on combined rock deformation and fracturing; and to identify common pairs of precursory times series for contemporaneous ground deformation and local seismicity. The time-dependent behaviour of fracturing and strain in experiments resembles that recorded in the field and strongly suggests that analogous processes operate over the corresponding range of length scales. During all three stages, therefore, data from the field have been compared with published results from analogue laboratory experiments on brittle rock failure. In addition to the cited publications, the research provided the basis for the EU Project Volcalert, co-ordinated by Kilburn (2001-03).

Related Links:

  • Aon Benfield UCL Hazard Centre
  • Bell A, Kilburn CRJ 2011. Precursors to dyke-fed eruptions at basaltic volcanoes: insights from the spatiotemporal patterns of volcano-tectonic seismicity at Kilauea volcano, Hawaii. Bull Volcanol, doi: 10.1007/s00445-011-0519.
  • Smith R, Kilburn CRJ 2010. Forecasting eruptions after long repose intervals from accelerating rates of rock fracture: The June 1991 eruption of Mount Pinatubo, Philippines. J Volcanol Geotherm Res,  191: 129–136.
  • Kilburn, CRJ 2003.  Multiscale fracturing as a key to forecasting volcanic eruptions. J Volcanol Geotherm Res, 125: 271-289.
  • Kilburn, CRJ, 2012. Precursory deformation and fracture before brittle rock failure and potential application to volcanic unrest. J Geophys Res, doi: 10.1029/2011JB008703.
  • Kilburn CRJ, Voight B, 1998. Slow rock fracture as eruption precursor at Soufriere Hills volcano, Montserrat, Geophys Res Letters, 25, 3665-3668.
  • McGuire WJ, Kilburn CRJ, 1997. Forecasting volcanic events: Some contemporary isues, Geol Rundschau, 86, 439-445.