|
Titel |
A spatially consistent seamless predictions of continental-scale hydrologic fluxes and states |
VerfasserIn |
Rohini Kumar, Juliane Mai, Oldrich Rakovec, Matthias Zink, Matthias Cuntz, Stephan Thober, Sabine Attinger, Martin Schroen, David Schaefer, Luis Samaniego |
Konferenz |
EGU General Assembly 2016
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 18 (2016) |
Datensatznummer |
250122530
|
Publikation (Nr.) |
EGU/EGU2016-1582.pdf |
|
|
|
Zusammenfassung |
One of the major challenges in the contemporary hydrology is to establish a continental-scale
hydrologic model that can provide spatially consistent, seamless prediction of hydrologic
fluxes and states to better characterise extreme events like floods and droughts. This requires,
among other things, 1) a robust parameterization technique that allows the model to
seamlessly operate across a range of spatial resolutions and 2) an efficient parameter
estimation technique to derive a representative set of spatially consistent model parameters
that avoid inconsistencies in simulated hydrologic fields (e.g., soil moisture). In this study, we
demostrate the applicability of a mesoscale hydrologic model parameterized using a
multiscale regionalization technique to derive daily gridded fields of hydrologic fluxes/states
over the Pan-EU domain since 1950. A multi-basin parameter estimation (MBE) strategy
that utilizes observed streamflows from a set of hydrologically diverse basins is
introduced to infer a representative set of regional calibration parameters which is
applicable over the entire domain. We tested three sampling schemes to select a set of
calibration basins incremented sequentially from 2 to 20 basins, based on the 1) random
selection procedure, 2) gradient along the hydro-climatic regimes, and 3) diversity in
hydro-climatic and basin physiographical properties (e.g., terrain, soil, land cover
properties).
Results of the MBE approach are contrasted against the benchmark at-site calibration
strategy across 400 EU basins varying from approximately 100 to 500,000 km2.
At-site calibrated parameters performed best for site-specific streamflow predictions,
but their transferability to other sites resulted in poor performance. Moreover, the
at-site calibration strategy generated a patchy, spatially inconsistent distribution of
parameter fields that further induced large discontinuities in simulated hydrologic
fields of soil moisture among other sates/fluxes. These limitations were overcome
by the MBE strategy that provided a compromise solution with improved model
performance compared to at-site cross-validated estimates. The gridded fields of
hydrologic parameters, states and fluxes from MBE were spatial continuos and
much more meaningful compared to those of the at-site calibration strategy. The
selection of calibration basins that include diversity in both hydro-climatic and
basin physical properties provided consistently better results compared to other two
strategies. Overall, our study highlights the limitations of the at-site calibration
strategy and demostrate the potentials of the MBE strategy as a way forward for a
spatially consistent seamless predictions of continental hydrological fluxes and
states. |
|
|
|
|
|