 |
| Titel |
Fix or Flex? Hypothesis testing with a flexible approach for conceptual model development |
| VerfasserIn |
Fabrizio Fenicia, Dmitri Kavetski, Hubert H. G. Savenije, Martyn P. Clark, Laurent Pfister |
| Konferenz |
EGU General Assembly 2011
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 13 (2011) |
| Datensatznummer |
250054667
|
|
|
|
|
|
| Zusammenfassung |
| Whether hydrological modelling is best pursued using a fixed model structure calibrated over
a wide range of catchments, or, instead, pursued using flexible model structures adapted to
the specific catchments of interest, remains a controversial question in the hydrological
community. On one hand, the generality of physical principles may appear to favour a
single fixed model, while on the other hand, environmental diversity has often led
to perceptions of “uniqueness of place”. In our opinion, these questions are best
explored by formulating competing hypotheses of catchment behaviour in different
conditions, and carefully comparing them. This is the motivation for SUPERFLEX,
a flexible approach for conceptual model development, based on generic model
components, where different model structures can be systematically generated and
compared. Importantly, SUPERFLEX is implemented within a robust numerical
framework, which helps meaningful hypothesis-testing by protecting against spurious
numerical artefacts, and making model calibration and parameter inference more
efficient.
In this study, the SUPERFLEX framework was used to generate a set of alternative
model structures for 4 different European and New Zealand catchments with distinct
climatology and physical characteristics. The hypothesized model structures were
broadly reflective of fieldwork insights available at these locations. Calibration,
including parameter optimization and uncertainty assessment, suggested that the
performance of the competing models differs significantly across the catchments.
Experimental knowledge is then shown to be useful in interpreting the differences in model
behaviour, and in lending additional evidence to the inferred mechanistic behaviour of
the 4 catchments. The estimated system complexity also differed substantially,
with some catchments displaying a much simpler behaviour than others. Overall,
this case study illustrates the development and application of SUPERFLEX for
hypothesis-testing, as a tool for improving our understanding and interpretation of both
differences and similarities of catchment dynamics in diverse environmental conditions. |
|
|
|
|
|
|