 |
| Titel |
Multi-variate spatial explicit constraining of a large scale hydrological model |
| VerfasserIn |
Oldrich Rakovec, Rohini Kumar, Luis Samaniego |
| Konferenz |
EGU General Assembly 2016
|
| Medientyp |
Artikel
|
| Sprache |
en
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 18 (2016) |
| Datensatznummer |
250123691
|
| Publikation (Nr.) |
 EGU/EGU2016-2987.pdf |
|
|
|
|
|
| Zusammenfassung |
| Increased availability and quality of near real-time data should target at better understanding
of predictive skills of distributed hydrological models. Nevertheless, predictions of regional
scale water fluxes and states remains of great challenge to the scientific community. Large
scale hydrological models are used for prediction of soil moisture, evapotranspiration and
other related water states and fluxes. They are usually properly constrained against river
discharge, which is an integral variable. Rakovec et al (2016) recently demonstrated that
constraining model parameters against river discharge is necessary, but not a sufficient
condition. Therefore, we further aim at scrutinizing appropriate incorporation of readily
available information into a hydrological model that may help to improve the realism of
hydrological processes. It is important to analyze how complementary datasets
besides observed streamflow and related signature measures can improve model skill
of internal model variables during parameter estimation. Among those products
suitable for further scrutiny are for example the GRACE satellite observations. Recent
developments of using this dataset in a multivariate fashion to complement traditionally
used streamflow data within the distributed model mHM (www.ufz.de/mhm) are
presented. Study domain consists of 80 European basins, which cover a wide range
of distinct physiographic and hydrologic regimes. First-order data quality check
ensures that heavily human influenced basins are eliminated. For river discharge
simulations we show that model performance of discharge remains unchanged when
complemented by information from the GRACE product (both, daily and monthly time
steps). Moreover, the GRACE complementary data lead to consistent and statistically
significant improvements in evapotranspiration estimates, which are evaluated using
an independent gridded FLUXNET product. We also show that the choice of the
objective function used to estimate model parameters leads to considerable changes in
the partitioning of precipitation into runoff components, while maintaining total
runoff estimates unaltered. Objective functions that take into account the spatial
patters of GRACE estimates perform better than those constrained only against
discharge. Improvements in parameter estimation based on multiple data sources will
enhance the community efforts towards spatially consistent large scale seamless
predictions.
Reference: Rakovec, O., Kumar, R., Mai, J., Cuntz, M., Thober, S., Zink, M., Attinger, S.,
Schäfer, D., Schrön, M., Samaniego, L. (2016): Multiscale and multivariate evaluation of
water fluxes and states over European river basins, J. Hydrometeorol., 17, 287–307, doi:
10.1175/JHM-D-15-0054.1. |
|
|
|
|
|
|