Temporal evolution and extreme value analysis of precipitations in Burkina Faso

Modeling of damaging extreme events is of great interest for the risk managers who want to implement some protections. It is based on the so-called generalized extreme values model. In the environmental sciences, we need to model the extremes of two or more processes like the daily rainfall recorded in various points of a given region. Each individual process could be modeled using univariate technique, but this strategy could not take into account for dependence of extreme values. For instance, consider two sites A and B fairly close to one another. If we observe an extreme value of the daily rainfall at site A, this information impacts the probability to observe an extreme value at site B even if we never recorded such value at B. The basic modeling tool is formed by so-called max-stable processes which take into account the possible spatial dependence through so-called extremal coefficient. We illustrate with annual maximum daily rainfall values recorded in 41 stations in Burkina Faso.