|
Titel |
Rainfall frequency analysis using a hourly rainfall model calibrated on weather patterns: application on Reunion Island |
VerfasserIn |
Yoann Aubert, Patrick Arnaud, Jean-Alain Fine, Philippe Cantet |
Konferenz |
EGU General Assembly 2014
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 16 (2014) |
Datensatznummer |
250092856
|
Publikation (Nr.) |
EGU/EGU2014-7220.pdf |
|
|
|
Zusammenfassung |
The National Research Institute of Science and Technology for Environment
and Agriculture (Irstea) has developed an original method for regional
rainfall frequency analysis applied on the whole French territory: the
SHYREG1
method. It is based on a stochastic hourly rainfall generator. The parameters of the rainfall
generator were regionalized at the spatial resolution of 1 km2 thus allowing for the
implementation of the model for every 1 km2. Frequency distributions were then derived
from long simulated rainfall series for each pixel. Therefore statistical rainfall estimates of
various durations (from 1h to 72h) and return periods (from 2 to 1000 years) are made
available in a rainfall risk database (intensity-duration-frequency) for the entire French
territory.
This article presents the application of the SHYREG method in Reunion Island. Reunion
Island (with a 2500-km2 surface area) is located in the south-west Indian Ocean.
The climate is tropical and characterised by cyclonic rainfall. Tropical cyclones
generate heavy rains: during the last one (Bejinsa) in January 2014, rainfall
observed exceeded 1000 mm in Cilaos station. Likewise, world records of rainfall,
lasting between 5 days (4301 mm in Commerson) and 15 days (6433 mm in
Commerson), were observed in Reunion Island during the Hyacinthe Cyclone in January
19802.
In mainland France, the calibration of the hourly rainfall generator depends on two
seasons (winter from December to May and summer from June to November). However, in
order to account for different types of events during a same season, a specific calibration of
the hourly rainfall model was necessary. Four types of rainfall event were defined by
Météo-France: cyclones, storms, hard rain and rain. Météo-France rainfall data, evenly
located over the Island (52 hourly rain gauge stations and 98 daily rain gauge stations), were
used to calibrate the hourly rainfall generator. The SHYREG parameters were regionalized
based on 17 physiographic maps of the Island (relief, ocean distance, etc.) with a 1-km2
spatial resolution.
For return periods of up to 10 years, the SHYREG-estimated rainfall
frequency values are consistent with estimates from the GPD law according
to the Nash-Sutcliffe criteria. For extreme return periods, we validate
SHYREG-based rainfall frequency estimates according to criteria of reliability and
stability3
and compare with the GPD performance. The stability of the frequency analysis method is
defined by its capacity to produce similar results when calibrated with different data samples,
its reliability by its capacity to assign accurate probabilities of occurrence to observations.
Results from applying both criteria have shown that the SHYREG method is highly stable
and reliable compared to the GPD law. |
|
|
|
|
|