This report is the 4th deliverable in WP4 and studies the performance of two models commonly used in this WP, Ecopath with Ecosim (EwE) and Gadget. An Atlantis model was constructed for Icelandic waters and used as an operating model, i.e. simulated data from Atlantis was fed into EwE and Gadget. Thus, Atlantis was considered as the “true” ecosystem and EwE and Gadget studied to see if they were able to replicate the truth. Such a comparison is not possible e.g. using survey data as such data are not the true population but measurements of one. A methodology was developed to automatically extract data from Atlantis and import to the other models. Also, balancing and fitting routines were written for EwE to make the modelling process more automatic and less subjective. This study was considered data rich as the best possible knowledge of the ecosystem was known, e.g. the true biomass was known. The EwE model had 42 functional groups and time-series fitting was done for the 25 vertebrate groups. A single-species Gadget model was constructed for cod and haddock. The EwE model was able to simulate the magnitude and the trends for both biomass and catches for most of the vertebrate groups. It did however overestimate the biomass of the juvenile stanza groups. The Gadget model did replicate the trends in biomass very accurately but did overestimate the biomass for the both groups. A further study is needed to investigate why the Gadget model overestimated the biomass. Also, further study is needed to test the performance of the model when the information of the ecosystem is limited but that will be addressed in the 7th deliverable in this WP.
