HR: 0800h
AN: H51B-0351 [Abstracts]
TI: Ensemble Approach to Establish the Effect of Data Uncertainty on Parameter Uncertainty within
Multi-objective Parameter Estimation Frameworks
AU: Pande, S
EM: saketpande@cc.usu.edu
AF: Utah Water Research Laboratory, Utah State University, 8200 Old Main Hill, Logan, UT 84322-8200
United States
AU: * Bastidas, L A
EM: Luis.Bastidas@usu.edu
AF: Utah Water Research Laboratory, Utah State University, 8200 Old Main Hill, Logan, UT 84322-8200
United States
AU: McKee, M
EM: mmckee@cc.usu.edu
AF: Utah Water Research Laboratory, Utah State University, 8200 Old Main Hill, Logan, UT 84322-8200
United States
AB:
We use an ensemble approach, within the strong constraint data assimilation framework i.e. perfect model assumption, to study
the effect of data uncertainty (input and output) on the multi-objective parameter estimation problem. To achieve this we
used the MOSCEM algorithm and applied it to two conceptual rainfall-runoff models: the Sacramento and a simplified version of
it the SixPar. The MOSCEM algorithm, because is based on the Markov chain search provides and idea of the underlying
parameter uncertainty and its distribution. We coupled the optimization algorithm with an ensemble of perturbed input and
output sequences to incorporate the additional uncertainty due to data error represented by an additive form of an
uncorrelated heteroscedastic measurement error while the mean, variance, and other statistical properties of the original
observational series are maintained. The use of the simplified version of the Sacramento model allows establishing the effect
of structural model complexity in the uncertainty assessment. The procedure was tested using 40 years of data from the Leaf
River in Mississippi using a 100 ensemble runs. Thus we have identified a collection of 50,000 `optimal', in the
multi-objective sense, parameter sets and correspondingly 50,000 objective function points. That allows us to establish
several regions of attraction, the uncertainty distribution of the parameter sets, and a collection of parameter sets
associated with a prescribed level of uncertainty for the observational data.
DE: 0550 Model verification and validation
DE: 1805 Computational hydrology
DE: 1846 Model calibration (3333)
DE: 1873 Uncertainty assessment (3275)
SC: Hydrology [H]
MN: Fall Meeting 2005