Selection of the most adequate frequency model to estimate the flood extreme values in the North West of Algeria

摘要

This work is intended to determine the most appropriate model to estimate the extreme values of flood in the North West of Algeria. The study is conducted in two steps. The first one consists on the determination of the type of laws which may represents better the data series using the log-log plot. The second step extends on the establishment of a ranking of the laws of the selected type. This has become possible with the application of different choice criteria: the Bayesian information criterion, the Akaïk information criterion corrected and the Schwarz criterion. In this step, we have first established a ranking based on the unique information gets from the observations of each site independently and secondly, to explore the information from the adjacent sites through a Bayesian approach. The results obtained put the Gamma law in the first position.