Résumé:
The results obtained are due to the series of observation of the floods, the more the series
considered is long the more the results will be representative, as well as the flow chosen
influences directly on the tendency of the quantiles for periods of return. Therefore, frequency
probability studies are only possible when a significant corpus of data is available, and they
are all the more reliable as this corpus is large. The problem in using these data, however,
comes from the fact that the readings are taken at a fixed time every day, every 24 hours, and
that this division differs from the division corresponding to the maximum flow.
In general, the use of the Flow. Model. Return Period (QMP) approach seems to us to be well
adapted. It is able to take into account the maximum flow, an essential notion when we talk
about floods, and it considers different statistical laws. Let us also note that the most adapted
law to this region is the Exponential law (Maximum likelihood), followed by the Log-normal
law (Maximum likelihood).
