On this page
If you searched for an online version of Ocean Data View, the first thing worth knowing is that one exists and is made by the same people: webODV, developed at the Alfred Wegener Institute alongside the desktop program. So “browser or desktop” is not a choice between the real tool and a substitute.
The more useful question is which stage of work you are at. Screening whether a region has the structure you expect is a different job from building a section figure for a paper, and browser tools and desktop tools are not equally good at both. This guide compares them by the work rather than by the feature list.
For a broader survey of what to plot in the first place, Ocean Data Visualization: Methods, Examples, and Tools covers the visualization types themselves.
What Desktop Ocean Data View Is For
Ocean Data View is a software package by Reiner Schlitzer at AWI, free of charge for non-commercial, non-military research and teaching, and running on Windows, macOS, and Linux. It has been widely used for exploring and publishing profile, time-series, and trajectory data for decades, and it is worth being clear about what it does well:
- Building and curating collections from many cruises, formats, and sources, then treating them as one dataset
- Colour sections along arbitrary cruise tracks, isosurface maps, and property-property plots
- Gridding, including two fast weighted-averaging algorithms and the DIVA software for more careful interpolation
- Derived variables, geostrophic velocity sections, difference fields, and animations
- Producing figures at publication quality, with full control over the result
A browser tool does not replace all of that. If your output is a section figure for a manuscript, or your input is a heterogeneous collection you need to curate once and reuse, ODV is the tool for the job.
webODV: The Same Lineage, in a Browser
webODV brings ODV’s analysis into the browser as two services.
The Data Extractor handles subsetting: you step through station, variable, and format selection and receive the result as text, an ODV collection, or NetCDF. Extraction runs in the background and notifies you when it is ready, which suits requests large enough that you would not want to hold a browser tab open.
The Data Explorer reproduces much of the desktop look and feel — maps, surface plots, section plots, scatter plots, filtering — against datasets already hosted on the server. Views can be saved to an account and reopened from any browser.
Several instances exist for different data collections, including an Argo webODV and the general webODV Explore service. The important structural point is that the data lives on the server: you are analysing hosted collections, not files you brought with you.
Argo-Specific Browser Routes
For Argo specifically, several other browser tools cover parts of the same ground:
- The Euro-Argo Data Selection tool is a search-and-download front end to the GDAC, organised around region, date, float, and parameter.
- The Argo Fleet Monitoring dashboard answers questions about a specific float — where it is, what it carries, what state its data is in.
- Argovis provides a map-based browser over Argo and related datasets, with an API behind it.
- OceanGraph focuses on reading and comparing preprocessed Argo profiles interactively, with derived indicators such as mixed layer depth available on the map.
These overlap, but not evenly. Some are primarily search tools that end in a download; others are primarily reading tools. Knowing which one you have opened saves a lot of time.
What a Browser Actually Costs You
The trade is not “less powerful”. It is more specific than that.
You work with what the server hosts. A browser tool analyses collections someone else has prepared. If your question needs a dataset that is not there — your own CTD casts, a mooring record, an unpublished cruise — that is a desktop or scripted workflow.
How freely you can define derived quantities varies by tool. Desktop ODV ships a large set of derived variables and lets you define your own through expressions and macros, then apply them across a collection. webODV lets you add and edit derived variables too, but Argo-specific browser tools generally expose a fixed set. If your analysis depends on a quantity nobody has implemented for you, you will be writing it yourself.
Export is narrower than the display. A tool that only saves images gives you a figure to paste into a slide, but not the numbers behind it. One that writes CSV, NetCDF, or an ODV collection hands the result straight to ODV or a script — which is why webODV’s Data Extractor offers text, ODV collection, and NetCDF. The export formats on offer decide whether the browser step feeds the next one or ends there.
Reproducibility has to be arranged. A desktop or scripted workflow leaves artefacts — a collection file, a script, a container — that reconstruct the result. A sequence of clicks does not, unless the tool saves views or you record what you did. This is the cost that is easiest to ignore and most expensive later; How to Make a Research Analysis Reproducible with Docker covers what recovering it involves.
Session limits are real. Browser tools cap how much data a session will load. That cap is usually generous for exploration and restrictive for bulk work, and hitting it is a reliable signal that you have moved past the stage the tool is built for.
Matching the Tool to the Task
| Task | Where it belongs |
|---|---|
| Check whether a region has the structure you expect | Browser |
| Look up one float and read its recent profiles | Browser |
| Subset the GDAC before downloading | Browser (or the index files) |
| Curate a collection from several cruises and formats | Desktop ODV |
| Build a publication-quality section figure | Desktop ODV |
| Grid data with DIVA or a custom derived variable | Desktop ODV |
| Run the same analysis over thousands of profiles | Scripted workflow |
| Combine Argo with another dataset | Scripted workflow |
The pattern is consistent: browsers are strongest when the question is which data is worth attention, desktop and scripted tools when the question is what exactly does this data say. Most projects need both, in that order.
Where OceanGraph Sits
OceanGraph covers the first row of that table for Argo data. It exists because the exploration stage is the one that is most often skipped and most often the reason a later analysis has to be redone — the profiles were not what you assumed, the feature was not there, the float had already drifted out of the region.
In practice that means searching profiles by region, time, and WMO ID, reading a selected vertical profile directly, comparing downloaded profiles side by side, and using derived indicators such as mixed layer depth to see how a property varies across many profiles at once.
It is not a replacement for ODV’s collection management, gridding, or figure production, and it is not a numerical analysis environment. It is the step before those, aimed at deciding what to spend the heavier tools on. The step-by-step workflow is in Visualizing Argo Float Data Without Python, and the App Guide documents Search and Bookmark and Analysis Lab: Vertical Profiles (JSON upload).
Knowing When to Move On
A few signals reliably mean the browser stage is over:
- You want a plot the tool does not offer, and you know exactly what it should look like
- You are repeating the same steps across dozens or hundreds of floats
- You need to compute on the values, not read them off a plot
- Someone has asked how you produced the figure
- Your dataset is not one the service hosts
None of these are failures of the browser tool. They are the point at which the question changed from exploratory to specific, which is when a desktop program or a script starts paying for its setup cost. Making a T-S Diagram: Python vs Interactive Tools walks through that transition for one concrete plot.
Frequently Asked Questions
Is there an official online version of Ocean Data View?
Yes. webODV is developed at AWI by the ODV team and offers a Data Extractor for subsetting and a Data Explorer that reproduces much of the desktop interface in a browser, against datasets hosted on the server.
Can I use a browser tool with my own data files?
It depends on the tool and is the main structural difference from desktop software. Most browser services analyse collections they host. Some accept uploads in a specific format — OceanGraph’s Analysis Lab, for example, takes Argo-format JSON profiles — but none offer the open format support of a desktop program.
Do browser tools give the same numbers as ODV?
For the same source data and the same operation, the underlying values are the same data. Differences appear in derived quantities, where the choice of algorithm, threshold, or interpolation matters. Compare the documented definitions rather than assuming two tools mean the same thing by the same label.
Which one should I open first?
Whichever answers “is this data relevant to my question” fastest. That is nearly always a browser tool, because the alternative — installing software or writing a loader — is a real cost to pay before you know whether the answer is yes.
Does using a browser tool mean I can skip quality control?
No. Visual screening is not a substitute for checking data mode and quality flags before you use a value scientifically. That step belongs in whatever workflow produces your final numbers, wherever the exploration happened.
