 |
| Titel |
Flood estimation and comprehensive hydrological assessment on basis of process-oriented continuous simulation |
| VerfasserIn |
Daniel Viviroli, Rolf Weingartner |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250045789
|
|
|
|
|
|
| Zusammenfassung |
| This contribution provides an overview of a comprehensive effort in process-oriented flood
estimation that makes extensive use of the advantages of continuous simulation. On
the basis of the hydrological modelling system PREVAH (Viviroli et al., 2009),
continuous hydrographs of 23 years duration each were simulated for 450 catchments in
Switzerland, laying the bases for a complementary national flood estimation procedure
for ungauged catchments. This extensive set of catchments covers approx. 90%
of the hydrologically relevant meso-scale catchments in the Swiss River Rhine
basin and represents a wide range of pluvial, nival and glacial regime types. To
ensure applicability for flood estimation, all simulations were performed at hourly
time-step.
To achieve modelling and flood estimation for ungauged sites, a robust regionalisation
scheme was developed on the basis of model calibrations for 140 sites with long-term hourly
gauge records. The main challenge met both in calibration and regionalisation was arriving at
simulations that show good skill in representing annual hourly peak flow and are still
hydrologically sound. This was achieved by employing a combination of selected
standard skill scores and peak-flow specific skill scores. Jack-knife validation of the
regionalisation scheme for 61 catchments with hourly long-term gauge records
shows a median Nash Sutcliffe efficiency of 0.71 for overall model performance
and a median error for estimation of a 100 year flood (HQ100) of 9%. There is
thus only a slight tendency for underestimation of large floods, and in almost half
of the catchments tested, the HQ100 estimated from continuous simulation lies
within the 90% confidence interval of the HQ100 extrapolated from the observed
record.
Thanks to the extensive simulation data available, it is possible to assess in great detail the
hydrological simulations as a whole as well as the peak flow estimations derived from them.
In particular, model and flood estimation results can be compared to simulation data from
catchments located in the same river basin as well as to peak flow and water balance
estimates derived from alternative modelling approaches. This range of auxiliary information
is unique yet for flood estimation and greatly enhances transparency and verifiability of the
flood estimations achieved as compared to present-day statistical and empirical flood
estimation procedures. It is furthermore possible to precisely identify limitations of this
implementation of the process-oriented flood estimation approach. The most important
restriction is a minimum drainage area of about 25Â km2, below which both process
representation in PREVAH as well as accuracy of precipitation data are insufficient. All of the
data mentioned are readily available to the end user in the form of pre-processed tables and
graphs.
Efforts to explore the added value resulting from using the process-based framework
instead of simple statistical and empirical approaches are under way and show
encouraging results. The hydrographs simulated have already been proven suitable for
estimation of extreme direct runoff volumes and exceedance of stage thresholds.
Further work in progress concerns implementation of anticipated changes in climate
and land use, allowing for the identification of catchments sensitive to the changes
mentioned.
Reference:
Viviroli, D., Zappa, M., Gurtz, J., and Weingartner, R. 2009. An introduction to the
hydrological modelling system PREVAH and its pre- and post-processing tools.
Environmental Modelling & Software, 24, 1209–1222. doi:10.1016/j.envsoft.2009.04.001 |
|
|
|
|
|
|