|
The purpose of this paper is to formulate a suitable
hydrological simulationprocedure for flood analysis. The
estimation of model parameters with noiseeffect and the
influence of noise effect on the hydrological simulation
areinvestigated.The autoregressive model with exogenous input
(ARX) based on the concept ofthe manifold cell model is adopted
to simulate the rainfall-runoff process.The least-square
estimation of model parameters with colored noise effect andthe
shuffled complex evolution (SCE) algorithm are applied to
improve theparameter identification. The real-time least-square
method with colorednoise effect is adequately employed to the
runoff model. The model parametersare proved to be time-varying
to update and correct the system condition atany time. By this
way, the hourly discharge can timely be predicted in theproject
area. In case of lacking sufficient records , two alternative
methodsare used for initial value decision of verification in
rainfall-runoff process,one is equaling calibration from many
events and the other is calibrating fromfew data.To verify the
applicability of the proposed procedure, the upland watersheds
ofFeitsui and Shihmen Reservoirs in northern Taiwan are chosen
as the projectareas. Results show that the simulated streamflows
by considering colorednoise effect are in better agreement with
the recorded ones. The analyticalapproach adopted in this study
could be utilized as a reference for hydrologicalplanning and
reservoir operation of project areas in Taiwan.
|