 |
| Titel |
A bottom-up approach to derive the closure relation for modelling hydrological fluxes at the watershed scale |
| VerfasserIn |
Ekkamol Vannametee, Derek Karssenberg, Martin Hendriks, Marc Bierkens |
| Konferenz |
EGU General Assembly 2014
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 16 (2014) |
| Datensatznummer |
250096515
|
| Publikation (Nr.) |
 EGU/EGU2014-12021.pdf |
|
|
|
|
|
| Zusammenfassung |
| Physically-based hydrological modelling could be considered as an ideal approach for
predictions in ungauged basins because observable catchment characteristics can be used to
parameterize the model, avoiding model calibration using discharge data, which are not
available. Lumped physically-based modelling at the watershed scale is possible with the
Representative Elementary Watershed (REW) approach. A key to successful application of
this approach is to find a reliable way of developing closure relations to calculate fluxes from
different hydrological compartments in the REWs. Here, we present a bottom-up approach as
a generic framework to identify the closure relations for particular hydrological processes that
are scale-independent and can be directly parameterized using the local-scale observable
REW characteristics. The approach is illustrated using the Hortonian runoff as an
example.
This approach starts from developing a physically-based high-resolution model
describing the Hortonian runoff mechanism based on physically-based infiltration theory and
runoff generation processes at a local scale. This physically-based model is used to
generate a synthetic discharge data set of hypothetical rainfall events and HRUs (6•105
scenarios) as a surrogate for real-world observations. The Hortonian runoff closure
relation is developed as a lumped process-based model, consisting of the Green-Ampt
equation, a time-lagged linear reservoir model, and three scale-transfer parameters
representing the processes within REWs. These scale-transfer parameters are identified by
calibrating the closure relations against the synthetic discharge data set for each scenario
run, which are, in turn, empirically related to their corresponding observable REW
properties and rainstorm characteristics. This results in a parameter library, which
allows direct estimation of scaling parameter for arbitrary REWs based on their
local-scale observable properties and rainfall characteristics, potentially avoiding
calibration.
The Hortonain runoff closure relation is evaluated using field discharge observations from
16 km2 catchments in the French Alps. The catchments are disaggregated to 60 REWs.
Scaling parameters for each REW are derived from the parameter library. Discharge is
simulated from individual REWs, routed over the stream network, and summed at the
catchment outlets to obtain the catchment-scale responses. The results show that our closure
relation is capable of reproducing the observed hydrograph and discharge volume
without calibration, i.e. Nash-Sutcliffe index up to 0.8, 10% errors in discharge
volume. Our closure relation outperforms a simple lumped rainfall-runoff model
that does not have scaling components. A brute-force calibration for an optimal
local-scale REW observable (i.e. saturated hydraulic conductivity; Ks), using a
constant pre-factor for all REWs, however significantly improves the prediction.
The calibrated Ks values are comparable to the local-scale observations in the
study catchment, implying that calibration may be unnecessary if the local-scale
observable REW properties can be correctly estimated. The bottom-up approach
for derivation of closure relation, including the parameter estimation scheme, in
this study is robust and shows promising applicability for the REW-based models. |
|
|
|
|
|
|