Authors: Ivan Dimov, Istvan Farago, Agnes Havasi, Zahari Zlatev
Addresses: Institute for Parallel Processing, Bulgarian Academy of Sciences, Acad. G. Bonchev Str. Bl. 25-A, Sofia 1113, Bulgaria. ' Eotvos Lorand University, Pazmany P. s. 1/c, Budapest 1117, Hungary. ' Eotvos Lorand University, Pazmany P. s. 1/c, Budapest 1117, Hungary. ' National Environmental Research Institute, Frederiksborgvej 399, P.O. Box 358, Roskilde DK-4000, Denmark
Abstract: Splitting techniques are commonly used when large-scale models, which appear in different fields of science and engineering, are treated numerically. Four types of splitting procedures are defined and discussed. The problem of the choice of a splitting procedure is investigated. Several numerical tests, by which the influence of the splitting errors on the accuracy of the results is studied, are given. It is shown that the splitting errors decrease linearly when (1) the splitting procedure is of first order and (2) the splitting errors are dominant. Three examples for splitting procedures used in all large-scale air pollution models are presented. Numerical results obtained by a particular air pollution model, Unified Danish Eulerian Model (UNI-DEM), are given and analysed.
Keywords: air pollution models; splitting techniques; error analysis; splitting order; partial differential equations; ordinary differential equations; environmental modelling; environmental pollution; Denmark.
International Journal of Environment and Pollution, 2008 Vol.32 No.2, pp.174 - 199
Published online: 13 Feb 2008 *Full-text access for editors Access for subscribers Purchase this article Comment on this article