 |
| Titel |
Inferring the theoretical limitations of model structures to simulate runoff observations with a runoff-rainfall (sic!) model |
| VerfasserIn |
M. Herrnegger, H.-P. Nachtnebel |
| Konferenz |
EGU General Assembly 2012
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 14 (2012) |
| Datensatznummer |
250065820
|
|
|
|
|
|
| Zusammenfassung |
| This contribution presents a method to analyse the theoretical limitations of model structures
to map observed runoff. This procedure may precede further meaningful hypothesis testing.
Model structures, which do not comply with the precondition to potentially simulate observed
runoff exactly, can be ruled out beforehand.
For the presented method a runoff-rainfall model is used, as precipitation is modelled
from runoff. A conceptual, HBV-type rainfall-runoff model is embedded in an iteration
algorithm, in which for every time step a catchment precipitation value is determined, which
results in a simulated runoff value that corresponds to the observed value. In this contribution
we do not focus on the modelled precipitation, but use the runoff-rainfall model to diagnose
model deficits. In the framework of the iteration procedure it is possible to test
the hypothesis, whether a model is generally capable of simulating observations
exactly.
Hydrological models are abstract and simplified representations of real world systems,
with the consequence, that most model parameters cannot be inferred through direct
observations in the field, but can only be estimated through a conventional calibration
procedure. With these given model parameters, it is assumed, that potentially a perfect
agreement between simulated and observed runoff is possible, ignoring possible intrinsic
mathematical constraints of the model, to identically simulate observations. At this
point, the failure to meet observations could be associated with inadequate model
parameters.
With the runoff-rainfall model it is possible to localise the range of model parameters
(ΦInvers) which theoretically enable a perfect fit of simulated to observed runoff. Comparing
ΦInvers with model parameters derived in a conventional manner (e.g. with GLUE-type
analysis) (ΦForward) enables the inference, whether the model structure is meaningful or not.
Differing parameters between ΦInvers and ΦForward and their linked physical
processes can be identified, enabling the diagnosis of model structural deficits. |
|
|
|
|
|
|