Report No FR0459
SURVEY AND TESTING OF MARINE MODELS
Improvement of the way in which marine and estuarine water quality modelling work is carried out by or on behalf of members of the Foundation for Water Research (FWR).
Taking account of the recommendations of FWR/National Rivers Authority (NRA) Report FR 0374, "A framework for marine and estuarine model specification in the UK", improved knowledge of the suitability of different modelling packages. The report will help members of FWR to choose packages on the basis of their appropriateness, resolution, accuracy and user-friendliness.
To test marine and estuarine modelling software against criteria established in FR 0374.
Numerical water quality models of marine and estuarine waters are extensively used by the Water Service Companies and by the NRA for strategic design purposes and also to assist in discussions relating to the granting of consents to discharge. Modelling to various standards of intricacy and accuracy is used in different parts of the country; there is a need to establish the performance and usability of available software packages relative to the guidelines and sample specification set out in FR 0374, by applying them to the same physical configurations with the same scenarios of loading.
Some of the model manufacturers have already produced modelling systems which largely satisfy the guide lines set out in FR 0374-. This was established both by market survey (Questionnaire) and by testing certain packages against a Test Specification at a single site: Bideford Bay, Devon.
Nevertheless, for the particular type of model tested (2-D depth-averaged), there remain opportunities for improvement, and these are outlined in the text. We also conclude that it is not possible to rank the models that we tested or which were submitted for tests. The reader must instead take account of the various advantages and disadvantages of the models and weight these according to individual circumstances.
No such test of practical 2-D modelling packages on data from a single site has ever been carried out before in the UK. The task proved more complicated and more time-consuming than had been originally envisaged. Key issues arose as to what exactly was being tested: the model itself, the appropriateness of the survey data, the boundary conditions or the ability of the modelling house to surmount all of the problems associated with a typical practical modelling investigation.
We conclude that all of these aspects are important and therefore recommend that test programmes for model packages should extend over a period sufficient to allow careful definition of the problems to be solved and location and preparation of an adequate data set.
We recommend the preservation of the Questionnaire, with appropriate updates, as a useful document for re-assessing model packages in future.
We recommend the preservation of the Bideford Bay Test Specification for the time being, so that modelling organisations which wish to test their models, but at their own pace, may do so and report along the lines requested in this project. It is not clear, however, what the forum for this reporting should be, nor how it would be funded.
We recommend that provision be made, in the funding of future modelling contracts, for the data set: bathymetry, boundary conditions and survey data to be archived in as permanent and accessible manner as possible. The British Oceanographic Data Centre (BODC) might provide suitable archiving facilities. In this way a modelling exercise could be revisited at a later date, either for new design studies or for bench-mark testing, without excessively high start-up costs.
VI RESUME OF CONTENTS
This report is presented in two volumes. Volume 1 consists of the body of the report, while Volume 2 is made up of Appendices, including the Questionnaire, the Responses, the Test Specification and Reports on the Performance of the Models Tested. It is possible to use Volume 1 without access to Volume 2, because certain key figure s and extracts from tables are included. Obviously the reader will obtain a more complete picture if Volume 2 is available for reference.
After an introductory section which describes the purpose and content of the report, Volume 1 continues with a section describing the Questionnaire and the Responses to it. There follows a section which reviews the task of testing the models: defining the problem and choosing or finding a suitable test site. Section 4 describes the data set available for the testing, the Test Specification itself and the problems encountered with the data set when the modelling houses set to work on it. The results from the modelling houses are presented in two sections: the first, Section 5, is concerned with the numerical accuracy of the modelling packages and the second, Section 6, with the ease-of-use of the packages. Conclusions are presented in Section 7 - and they include lessons learnt about running a project of this sort as well as those about the models themselves.
Copies of the report are available from FWR, price £35.00, less 20% to FWR Members.