|
Titel |
Parameter estimation in a large scale Dutch Continental Shelf Model by Proper Orthogonal Decomposition |
VerfasserIn |
Muhammad Umer Altaf, Arnold W. Heemink, Martin Verlaan, Ibrahim Hoteit |
Konferenz |
EGU General Assembly 2011
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 13 (2011) |
Datensatznummer |
250057011
|
|
|
|
Zusammenfassung |
Accurate forecasting of the storm surges is very important in the Netherlands since large
areas of the land lie below sea level. Timely water level forecasts are necessary to support the
decision of the proper closure of the movable storm surge barriers. Dutch continental shelf
model (DCSM), is a shallow sea model of the entire European continental shelf, which is
used in the Netherlands to forecast the storm surges in the North Sea. The forecasts are
necessary to support the decision of the timely closure of the moveable storm surge barriers
to protect the land. The adjoint method has often been used for the calibration of the large
scale numerical flow models. A number of unknown parameters is introduced into the
numerical model. Using the given data these parameters are identified by minimizing a cost
function that measure the difference between model results and data (observations).
The drawback of the adjoint method is the programming effort required for the
implementation of the adjoint model code. In this study, we have implemented a
newly developed model calibration method MRVDA (model reduced variational
data assimilation) for the estimation of the depth values for the model DCSM with
approximately 106 operational grid points. The advantage of this method is that it shifts
the minimization into lower dimensional space and avoids the implementation of
the adjoint of the tangent linear approximation of the original nonlinear model. A
number of calibration experiments is performed. The effectiveness of the algorithm is
evaluated in terms of the accuracy of the final results as well as the computational costs
required to produce these results. In doing so, comparison is made with a simultaneous
perturbation stochastic approximation (SPSA) method. The main findings are: (1) A low
dimensional model of much smaller size can be constructed as compared to the
original model. (2) An overall improvement of more than 50% is obtained with
respect to the initial DCSM. (3) The POD calibration approach efficiently solves the
minimization problem without the burden of implementation of the adjoint code. |
|
|
|
|
|