You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/paper/paper.md
+9-7Lines changed: 9 additions & 7 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -47,10 +47,10 @@ Marine science relies on fieldwork for data collection, yet sea-going opportunit
47
47
48
48
`VirtualShip` goes beyond simply extracting grid-cell values from model output. Instead, it uses programmable behaviours and sophisticated interpolation techniques (with `Parcels` underpinnings) to access data in exact locations and timings, as if they were being collected by real-world instruments. `VirtualShip` shares some functionality with existing tools, such as `OceanSpy` [@Almansi2019] and `VirtualFleet` [@Maze2023], but extends capabilities to mesh many different instrument deployments into a unified expedition simulation framework. Moreover, `VirtualShip` exploits readily available, streamable data via the Copernicus Marine Data Store, removing the need for users to download and manage large datasets locally and/or arrange for access to remote servers. `VirtualShip` can also integrate coordinate files exported from the [Marine Facilities Planning](https://www.marinefacilitiesplanning.com/cruiselocationplanning#) (MFP) tool, giving users the option to define expedition waypoints via an intuitive web-based mapping interface.
49
49
50
-
# Functionality
51
-
52
50
`VirtualShip` simulates the deployment of virtual instruments commonly used in oceanographic fieldwork, with emphasis on realism in how users plan and execute expeditions. For example, users must consider ship speed and instrument deployment/recovery times to ensure their expedition is feasible within given time constraints. Possible instrument selections include surface `Drifter`[@Lumpkin2017], `CTD` (Conductivity-Temperature-Depth; @Johnson2007), `Argo float`[@Jayne2017], `XBT` (Expendable Bathythermograph; @Goni2019), underway `ADCP` (Acoustic Doppler Current Profiler; @Kostaschuk2005), and underway `temperature/salinity`[@Gordon2014] probes. More detail on each instrument is available in the [documentation](https://virtualship.readthedocs.io/en/latest/user-guide/assignments/Research_proposal_intro.html#Measurement-Options).
53
51
52
+
# Software design
53
+
54
54
The software can simulate complex multidisciplinary expeditions. One example is a virtual expedition across the Agulhas Current and the South Eastern Atlantic that deploys a suite of instruments to sample physical and biogeochemical properties (\autoref{fig:fig1}). Key circulation features appear early in the expedition track, with enhanced ADCP speeds marking the strong Agulhas Current (\autoref{fig:fig1}b) and drifters that turn back toward the Indian Ocean indicating the Agulhas Retroflection (\autoref{fig:fig1}c). The CTD profiles capture the vertical structure of temperature and oxygen along the route, including the warmer surface waters of the Agulhas region (\autoref{fig:fig1}d, early waypoints) and the Oxygen Minimum Zone in the South Eastern Atlantic (\autoref{fig:fig1}e, final waypoints).
55
55
56
56
The software is designed to be highly intuitive to the user. It is wrapped into three high-level command line interface commands using [Click](https://click.palletsprojects.com/en/stable/):
@@ -61,22 +61,24 @@ The software is designed to be highly intuitive to the user. It is wrapped into
61
61
62
62
A full example workflow is outlined in the [Quickstart Guide](https://virtualship.readthedocs.io/en/latest/user-guide/quickstart.html) documentation.
63
63
64
-
65
-
66
-
# Implementation
67
-
68
64
Under the hood, `VirtualShip` is modular and extensible. The workflows are designed around `Instrument` base classes and instrument-specific subclasses and methods. This means the platform can be easily extended to add new instrument types. Instrument behaviours are coded as `Parcels` kernels, which allows for extensive customisability. For example, a `Drifter` advects passively with ocean currents, a `CTD` performs vertical profiling in the water column and an `ArgoFloat` cycles between ascent, descent and drift phases, all whilst sampling physical and/or biogeochemical fields at their respective locations and times.
69
65
70
66
Moreover, the data ingestion system relies on Analysis-Ready and Cloud-Optimized data (ARCO; @Stern2022, @Abernathey2021) streamed directly from the Copernicus Marine Data Store, via the [`copernicusmarine`](https://github.com/mercator-ocean/copernicus-marine-toolbox) Python toolbox. This means users can simulate expeditions anywhere in the global ocean without downloading large datasets by default. Leveraging the suite of [physics and biogeochemical products](https://virtualship.readthedocs.io/en/latest/user-guide/documentation/copernicus_products.html) available on the Copernicus plaform, expeditions are possible from 1993 to present and forecasted two weeks into the future. There is also an [option](https://virtualship.readthedocs.io/en/latest/user-guide/documentation/pre_download_data.html) for the user to specify local `NetCDF` files for data ingestion, if preferred.
71
67
72
-
# Applications and future outlook
68
+
69
+
70
+
# Research impact statement
73
71
74
72
`VirtualShip` has already been extensvely applied in Master's teaching settings at Utrecht University as part of the [VirtualShip Classroom](https://www.uu.nl/en/research/sustainability/sustainable-ocean/education/virtual-ship) initiative. Educational assignments and tutorials have been developed alongside to integrate the tool into coursework, including projects where students design their own research question(s) and execute their fieldwork and analysis using `VirtualShip`. Its application has been shown to be successful, with students reporting increased self-efficacy and knowledge in executing oceanographic fieldwork [@Daniels2025].
75
73
76
74
The package opens space for many other research applications. It can support real-life expedition planning by letting users test sampling routes before going to sea. It also provides tooling to explore real-time adaptive strategies in which sampling plans shift as forecasts or observations update. The same workflow can also be used to investigate sampling efficiency, for example, examining how waypoint number or spacing shapes the ability to capture features of interest. Moreover, the software is well-suited for developing Observation System Simulation Experiments (OSSEs; e.g. @Errico2013) to test and optimise observational strategies in a cost- and time-efficient manner. This framework further enables instrument design experiments that are relevant to autonomous observing systems. There is potential for users to prototype and test control strategies for gliders, REMUS vehicles, and Saildrones, as well as explore concepts for new instruments at early stages of development. Future tutorials could demonstrate how to define custom instruments within the VirtualShip framework.
77
75
78
76
Both the customisability of the `VirtualShip` platform and the exciting potential for new ARCO-based data hosting services in domains beyond oceanography (e.g., [atmospheric science](https://climate.copernicus.eu/work-progress-our-data-stores-turn-arco)) means there is potential to extend VirtualShip (or "VirtualShip-like" tools) to other domains in the future. Furthermore, as the `Parcels` underpinnings themselves continue to evolve, with a future (at time of writing) [v4.0 release](https://docs.oceanparcels.org/en/v4-dev/v4/) focusing on alignment with [Pangeo](https://pangeo.io/) standards and `Xarray` data structures [@Hoyer2017], `VirtualShip` will also benefit from these improvements, further enhancing its capabilities, extensibility and compatability with modern cloud-based data pipelines.
79
77
78
+
# AI usage disclosure
79
+
80
+
Generative AI technologies (Gemini v2.0/2.5/3.0, ChatGPT v4o/5.0/5.1/5.2 and GitHub Copilot) were used for code generation, refactoring and test scaffolding. AI-assisted autocompletion tools (via GitHub Copilot) were used in the writing of this manuscript. Authors carefully reviewed and edited all AI-assisted content and made the core desigin decisions.
81
+
80
82
# Acknowledgements
81
83
82
84
The VirtualShip project is funded through the Utrecht University-NIOZ (Royal Netherlands Institute for Sea Research) collaboration.
0 commit comments