![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Model calibration in the spectral domain: opportunities for model development and confusion |
VerfasserIn |
Bettina Schaefli, Dmitri Kavetski |
Konferenz |
EGU General Assembly 2011
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 13 (2011) |
Datensatznummer |
250054349
|
|
|
|
Zusammenfassung |
Although spectral methods remain relative uncommon in catchment hydrology, there are
several application areas, in particular, karst hydrology, where model performance
is assessed in the spectral domain. Here, instead of tuning the model parameters
to reproduce the observed system behaviour directly in the time- and/or space-
domains, the model output and the reference samples are first transformed to the
spectral domain, followed by optimization of some performance metric based on
the transformed data. The switch from the original (time or space) domain to a
spectral domain is usually accomplished by applying a Fourier Transform and some
normalization to the modeled and observed series. While the application of this
transform is numerically straightforward (through the use of Fast Fourier Transform
algorithms), the interpretation of calibration results obtained in the spectral domain can be
counter-intuitive, and requires some digging into a mathematical field where engineers and
mathematicians speak different languages, creating room for confusion on both
sides.
In this contribution, we review several key assumptions and technical aspects underlying
spectral analysis methods, before discussing the opportunities it offers for the development
and calibration of catchment models. These include the definition of new types of modeling
errors, the detection of model structural errors, or model calibration in the absence of
concomitant input-output time series. |
|
|
|
|
|