A Generic Software Framework for Data Assimilation and Model Calibration

A Generic Software Framework for Data Assimilation and Model Calibration

Van Velzen, N.

Heemink, A.W. (promotor)

Electrical Engineering, Mathematics and Computer Science

Delft Institute of Applied Mathematics

2010-06-07

The accuracy of dynamic simulation models can be increased by using observations in conjunction with a data assimilation or model calibration algorithm. However, implementing such algorithms usually increases the complexity of the model software significantly. By using concepts from object oriented programming, this thesis proposes a software framework for data assimilation and model calibration. The framework is strictly separated from the dynamic model implementation such that the models and the methods can de developed independent from each other. Three main studies are performed in order to prove the applicability of the suggested approach. In the first study it is shown that an equivalent of an existing data assimilation application of the operational shallow water model WAQUA/TRIWAQ can be realized using the suggested approach. In a second study it is shown, using the LOTOS-EUROS air quality model, that the framework adds a significant flexibility by allowing a model to be combined with a number of off-the-shelf data assimilation methods without the need of any additional programming. Finally it is shown how the proposed framework allows computations to be automatically parallelized and how models that are parallel themselves can be coupled to the framework.

data assimilation
problem solving environment
parallel computing
LOTOS-EUROS
WAQUA/TRIWAQ

http://resolver.tudelft.nl/uuid:373b19c3-2f4a-4ec2-bb32-8b37bd5b6a9a

2010-06-07

9789085705178

Institutional Repository

doctoral thesis

(c) 2010 Van Velzen, N.