Simple stochastic modelling of the eruption history of a basaltic volcano: Nyamuragira, Zaire

Abstract

Three simple models of the behaviour of a series of basaltic eruptions have been tested against the eruptive history of Nyamuragira. The data set contains the repose periods and the volumes of lava emitted in 22 eruptions since 1901. Model 1 is fully stochastic and eruptions of any volume with random repose intervals are possible. Models 2 and 3 are constrained by deterministic limits on the maximum capacity of the magma reservoir and on the lowest drainage level of the reservoir respectively. The method of testing these models involves (1) seeking change points in the time series to determine regimes of uniform magma supply rate, and (2) applying linear regression to these regimes, which for models 2 and 3 are the determinsstic limits to those models. Two change points in the time series for Nyamuragira, in 1958 and 1980, were determined using a Kolmogorov-Smirnov technique. The latter change involved an increase in the magma supply rate by a factor of 2.5, from 0.55 to 1.37 m3s-1. Model 2 provides the best fit to the behavior of Nyamuragira with the ratio of variation explained by the model to total variation. R2, being greater than 0.9 for all three regimes. This fit can be interpreted to mean that there is a determinstic limit to the elastic strength of the magma reservoir 4–8 km below the summit of the volcano.