The Marine Virtual Laboratory (version 2.1): enabling efficient ocean model configuration
Abstract. The technical steps involved in configuring a regional ocean model are analogous for all community models. All require the generation of a model grid, preparation and interpolation of topography, initial conditions, and forcing fields. Each task in configuring a regional ocean model is straightforward – but the process of downloading and reformatting data can be time-consuming. For an experienced modeller, the configuration of a new model domain can take as little as a few hours – but for an inexperienced modeller, it can take much longer. In pursuit of technical efficiency, the Australian ocean modelling community has developed the Web-based MARine Virtual Laboratory (WebMARVL). WebMARVL allows a user to quickly and easily configure an ocean general circulation or wave model through a simple interface, reducing the time to configure a regional model to a few minutes. Through WebMARVL, a user is prompted to define the basic options needed for a model configuration, including the model, run duration, spatial extent, and input data. Once all aspects of the configuration are selected, a series of data extraction, reprocessing, and repackaging services are run, and a “take-away bundle” is prepared for download. Building on the capabilities developed under Australia's Integrated Marine Observing System, WebMARVL also extracts all of the available observations for the chosen time–space domain. The user is able to download the take-away bundle and use it to run the model of his or her choice. Models supported by WebMARVL include three community ocean general circulation models and two community wave models. The model configuration from the take-away bundle is intended to be a starting point for scientific research. The user may subsequently refine the details of the model set-up to improve the model performance for the given application. In this study, WebMARVL is described along with a series of results from test cases comparing WebMARVL-configured models to observations and manually configured models. It is shown that the automatically configured model configurations produce a good starting point for scientific research.