Using the COAsT Python Package to Develop a Standardised Validation Workflow for Ocean Physics Models

Validation is one of the most important stages of a model's development. By comparing outputs to observations, we can estimate how well the model isable to simulate reality, which is the ultimate aim of many models. During development, validation may be iterated upon to improve the modelsimulation and compare it to similar existing models or perhaps previous versions of the same configuration. As models become more complex, datastorage requirements increase and analyses improve, scientific communities must be able to develop standardised validation workflows for efficientand accurate analyses with an ultimate goal of a complete, automated validation. We describe how the Coastal Ocean Assessment Toolbox (COAsT) Python package has been used to develop a standardised and partially automated validation system. This is discussedalongside five principles which are fundamental for our system: system scaleability, independence from data source, reproducible workflows, expandablecode base and objective scoring. We also describe the current version of our own validation workflow and discuss how it adheres to the aboveprinciples. COAsT provides a set of standardised oceanographic data objects ideal for representing both modelled and observed data. We use the packageto compare two model configurations of the Northwest European Shelf to observations from tide gauge and profiles.