LAUSR

Mount Cameroon is currently the only active volcano in Central Africa. It supports an active farming population due to the fertility of its soil. Its two most recent eruptions were in 1999 and 2000. They were associated with serious costly material damages that impoverished part of the local population. Some people were evacuated from their villages to neighboring localities. The population that lives on these flanks is therefore exposed to volcanic risk. Risk assessment starts with an evaluation of the hazard, which is here an estimation of the probability of an eruption of a given magnitude within a given time interval. The present study exploits the time series of quiescent time periods of Mount Cameroon volcano to predict the likelihood of its future eruptions. Only eruptions with VEI > 0 are considered in this study. The time series of repose intervals of Mount Cameroon is proved to be stationary based on the Kolmogorov–Smirnov (K–S) test on the cumulative number of eruptions and Dickey–Fuller unit root tests, and the successive repose time intervals are proved to be independent using the serial correlation scatterplot and Spearman correlation coefficient. Renewal processes (exponential, Weibull, log-logistic) are each best fitted and tested for goodness of fit. The parameters of the processes were calculated using the maximum likelihood functions. From the K–S and the chi-square tests, all three applied theoretical processes are found to be a good fit to the data. The exponential process is the best characteristic of the series as supported by the Akaike information criterion (AIC) best-to-fit test. The three models offer different predictions of the occurrences of future eruptions of Mount Cameroon. For elapsed time lower than or equal to 30 years, the probabilities estimated from the log-logistic and the Weibull distributions are higher than those calculated from the exponential. For higher elapsed time, the probabilities from the exponential distribution are intermediary between the probabilities from the log-logistic (lower) and the Weibull (higher). These two latter distributions predict different hazard rates than the exponential.