Calibrating and validating the ISIS-Fish model of the Bay of Biscay demersal mixed fishery

The Bay of Biscay is a highly productive ecosystem submitted to an intense anthropic pressure and supporting an economically important demersal fishery. Although several modelling works concerning this fishery have been published in the past, the recent implementation of a pluri-annual management plan, as well as important concerns on mixed fishery issues, encourage the development of new operational models capable of making projections at seasonal and fleet scales.   
In this study, we present a new ISIS-Fish model of the Bay of Biscay which is built on recently published description of the main species' biology  (SEANOE https://doi.org/10.17882/102675) and of fishing activity in the Bay. We present the calibration of the species' accessibility at age profiles and of the targetting behaviour of the different métiers involved in the fishery. We then provide a detailed assessment of the model's prediction skills, in which the simulated values for several variables during two distinct periods of time (the calibration period and a distinct "hindcast" period)  were compared to reference data thanks to four different metrics. 
The data project includes: 
- a README file that provides details on the calibration method, and a step-by-step guide to reproduce all the analysis (calibration and skill assessment) presented here
- raw and processed reference data to be compared to simulated data in order to make diagnostics and corrections on the model
- all the successive versions of the ISIS-Fish model with associated files useful for their setting (mainly parameter files and scripts)
- simulation objects (so that the user is able to re-run all the simulation on his own computer) as well as processed simulated data associated to each model, to be compared to reference data in order to make he diagnostics and corrections
- files containing the diagnostics (reference vs. simulation comparison at different scales) of the different versions of the model, along with the R code used to produce these diagnostics
Then, the materials provided here allow to 1) get the calibrated model and use it, 2) have a clear sight of the model's strength and weaknesses, and 3) to reproduce the model building based on raw data and a previously available ISIS-Fish model of the Bay of Biscay.