Outline of our approach
Preparation of data
Basically, for any national database, the first operation to perform in order to use COST methods will be to export the data in a single pre-defined format. The COST data format will be specified in collaboration with the FishFrame developers at an early stage of both projects. Consequently, those countries uploading their data in the FishFrame database will be able to run the COST methods either outside FishFrame within R environment, or within FishFrame environment (to be implemented as part of the FishFrame project). Any national or international database having access to raw validated data will have the possibility to create a linkage to R and the COST methods and, if need be, a user-friendly interface. The main advantage of an integrated approach database-COST is that the users can always get access to the most recent developments of the COST methods.
An important point to keep in mind is the need for COST to work on the raw validated data. However COST methods will not modify the raw validated data stored in national or international databases. The management of the data (addition/modification/deletion) is the sole responsibility of the ad hoc database managers. The exception that confirms the rule is that COST methods will read the data stored in FishFrame as they will both work with the same format, but although management methods will be developed for the COST objects, no modification of the dataset will be possible from COST.
Genericity of COST
COST methods will not take into account all the local/regional sampling specificities, nor will they access all the different data formats stored in European databases. In order to ensure the genericity of the COST methods, and consequently their high quality , a fixed number of options will be available to describe the sampling designs. The options taken into account are those summarised in the WKSCMFD report (ICES, 2005) which were meant to summarise the current situations all over Europe. The alternatives are shown in the table below, where in any given situation, the user should describe one sampling design by choosing one option per line.