Refining sea-bed process models for aquaculture

Kenny Black, Trevor Carpenter, Andrew Berkeley, Kevin S Black, Carl L. Amos

Research output: Book/ReportOther report

13 Downloads (Pure)

Abstract

DEPOMOD was funded by a NERC-MAFF Link programme and was a collaboration between SAMS (NERC DML then), Marine Harvest and SEPA from 1997 to 1999. Its purpose is to predict the benthic impacts on sediment dwelling animals from marine cage farms given farm (configuration, feeding rate) and environmental information (bathymetry, water currents). The purpose of AutoDEPOMOD (2005) was to iteratively run the DEPOMOD from an initial biomass, automatically assessing the result then re-running the model as many times as required, until a compliant solution was obtained. AutoDEPOMOD was adapted to determine medicine residues in sediments. Rather than duplicating functions available in 3rd party software, AutoDEPOMOD incorporated processes which called on these functions in an automated way.
Windows and the other 3rd party applications utilised have been updated many times causing functional problems. These have been worked around by running AutoDEPOMOD on virtual machines (e.g. XP in a VirtualBox or VMware VM), but these older systems and applications are not being supported and may soon not even be available to future users. The coding language used in AutoDEPOMOD is obsolete and the software design was determined by the computing constraints of the 1990s. Moreover, while giving good results at the majority of Scottish sites, it has been increasingly obvious that it underestimates the impacts at dispersive sites. This is important as the salmon farming industry is increasingly looking for larger sites in more dispersive environments.
The goals of the present project were:
1. To completely rewrite the model code in a modern language (Java) in a modular form that will facilitate future maintenance and development. Hereafter, this model will be referred to as NewDEPOMOD to distinguish it from its successors.
2. To remove dependencies on third party software (e.g. excel, MS Excel) as far as possible
3. To carry out field and laboratory studies to improve our understanding of resuspension processes around fish farms and to use these to refine the modelling of resuspension processes and hence improve the predictive ability of the model.
Recoding the model was done in 2 phases. The first aimed to reproduce the AutoDEPOMOD model coded in Borland Delphi and in Java code retaining the existing structure. The second aimed to refactoring1 the code in order for it to be much better able to be adapted in future developments. In addition the refactored model was designed to take advantage of modern computing engineering (e.g. multiple processors) and to include insights from fieldwork on resuspension process (annexes 1 and 2). All third party dependencies have been removed (except Java, which is likely to be a long-lived platform). This has demanded the development of a new User Interface to complement the new model engine.
Three research cruises were implemented at fish farm sites chosen by the project management team. The first cruise obtained detailed multi-beam bathymetry at 8 salmon farms on the west Coast of Scotland. The second 2 cruises engaged Partrac Ltd who deployed state of the art benthic flumes which were used to determine erosion/resuspension properties of the seabed around these same farms.
Original languageEnglish
Place of PublicationOban
PublisherScottish Association for Marine Science
Number of pages200
VolumeSAM/004/12
Publication statusPublished - 30 Aug 2016

Fingerprint Dive into the research topics of 'Refining sea-bed process models for aquaculture'. Together they form a unique fingerprint.

Cite this