![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Temporal resolution of precipitation as a control of runoff model performance |
VerfasserIn |
Anna Sikorska, Jan Seibert |
Konferenz |
EGU General Assembly 2016
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 18 (2016) |
Datensatznummer |
250132771
|
Publikation (Nr.) |
EGU/EGU2016-13307.pdf |
|
|
|
Zusammenfassung |
In most cases the choice of the temporal resolution of the data used for model calibration is
driven by the data availability. Yet, a carefully chosen resolution could greatly improve the
identifiability of catchment processes and enhance the model predictability, while saving
costs and time for too detailed measurement protocols or simulations. In this work we
formulate this issue as an inverse problem and investigate how data of different
temporal resolutions, ranging from one hour to two days, affects the predictability of a
precipitation-runoff model (HBV) for both high and low flow conditions. We employ a
Bayesian inference with a formal likelihood function, which enables a quantitative
comparison. Our study is based on the sample of five heterogeneous Swiss meso-scale
catchments with two different datasets of precipitation fields: radar-based (RB) and ground
station network (GSN).
Our results showed that a high data resolution was preferable in small and medium size
catchments, with catchments smaller than 60 km2 having a stronger tendency towards an
hourly resolution. As opposed, a high data resolution was not so important in catchments
larger than 200 km2 and daily data seemed to be detailed enough. A similar effect was
observed for predictive uncertainty bands, which were significantly reduced with finer data in
small and medium size catchments, whereas they remained similar in larger catchments
despite increasing the temporal resolution. Moreover, RB resulted in a better model
performance for both high and low flows in small and medium size catchments, whereas a
similarly good model performance was obtained with GSN in the larger catchments. |
|
|
|
|
|