Homogeneity testing of multivariate hydrological records, using multivariate copula L-moments

Recently, there have been an increasing number of studies dealing with change detection in multivariate series. However, a major drawback with most of the currently used methods is the lack of flexibility. Indeed, these methods are only able to detect changes in the strength of dependence assuming invariant shape of the dependence structure. However, under a changing climate, the shape of dependence might change as well. Furthermore, in the multivariate setting, heterogeneities can occur in the margins and/or in the dependence structure. The most existing approaches for multivariate change detection deal with the whole distribution. In this paper, we propose a novel statistical test for multivariate heterogeneity detection, based on copula and multivariate L-moments. A simulation study is conducted to evaluate the performance of the proposed test and to compare it with those of existing tests. Results indicate that the proposed test has an interesting power especially when the dependence strength remains invariant, with power ranging between 45 and 93% whereas for the existing tests the power is lower than 14% in this case. An application to a real data set is also provided. Results show the ability of the proposed test to discriminate homogeneous and inhomogeneous series.