|
Titel |
Effect of Parametrization in a Grid based mesoscale Hydrologic model on the Streamflow Prediction |
VerfasserIn |
R. Kumar, L. Samaniego, S. Attinger |
Konferenz |
EGU General Assembly 2009
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 11 (2009) |
Datensatznummer |
250029551
|
|
|
|
Zusammenfassung |
Distributed hydrologic models have the potential to simulate the spatial distribution of
hydrological processes and provide the estimate on streamflow at all points along the river
network within the catchment. While such models can explain the variability of spatially
distributed hydrological process, they have often complex structure and contains significant
number of unknown model parameters that need to be defined for every spatial element. To
reduce the complexity of these models, in terms of number of free parameters that need to be
estimated through the calibration process, several parametrization schemes have been
introduced in the recent past.
The purpose of this study is to investigate and compare the performance of two different
parameterization schemes, namely: Hydrological Response Units (HRU) and Multiscale
Parameter Regionalization (MPR) employed in the grid based mesoscale Hydrologic
Model (mHM), for the daily streamflow prediction. The HRU concept works on the basis
that groups the modeling cells, in which the dominant hydrological processes are represented,
into homogenous units based on the available catchment characteristics (elevation, slope,
landcover, soil textural information, geological characteristics, etc). The unique sets of
parameters are assigned to each HRU through the calibration process. In the case of the
MPR method the model parameters at coarser resolution (modeling cell) are linked
to their corresponding ones at a finer scale, in which the datasets are available,
through upscaling operators such as harmonic mean, average mean, amongst others.
Parameters at the finer scale are linked to catchment characteristics through different
nonlinear transfer functions. The global parameters (very few as compared to the total
number of free parameters) of these transfer functions are found via calibration
process.
The proposed study was carried in the upper catchment of the Neckar River (Germany)
covering an area of approximately 4000Â km2. The finer and coarser resolution
were fixed at (100 à 100) m and (4000 à 4000) m, respectively. The modeling
cells at the coarser resolution were grouped into 15 HRUs by k-means clustering
algorithm. The free parameters of both parametrization schemes were calibrated with the
simulated annealing algorithm using the discharge data of the catchment outlet.
Results obtained in this study indicated that the MPR method is more robust and
reliable than the HRU method. The Nash Sutcliffe Efficiency (NSE) of the MPR
method at the calibration and internal gauging stations were on average 5% and 10%
greater than that obtained with the HRU method but has 60% less free parameters to
calibrate. |
|
|
|
|
|