Visualising the Gulf Stream

My PhD research focuses on the turbulent fluid dynamics within the Gulf Stream, which involves analysing data from observations and models. There is a rich source of observational data on the worlds oceans, and this is primarily due to introduction of satellites to collect data. We now live in a golden age of satellite observations, where data sets on various quantities (e.g. temperature, salinity, chlorophyll) have become very accessible; tools now exist to select data from any region of the globe, from any time, from any particular mission. The data is also very clean and requires virtually no pre-processing, meaning you can jump straight into the analysis.

The aim of this post is simply to provide simple illustrations to the sort of data that is available, and what you can do with it. The primary region of focus will be the Gulf Stream, which is a strong current which hugs the east coast of the US before separating into the open ocean at Cape Hatteras. The Gulf Stream is a highly turbulent and complex system, making it great for data visualisations. All figures are created using the Matplotlib library for Python.

Figure 1: A snapshot of sea-surface temperature in the North Atlantic on 26th October 2017. Data obtained from the Group for High Resolution Sea Surface Temperature (GHRSST), distributed by NASA Jet Propulsion Laboratory (

Let’s start with the big picture of how the Gulf Stream fits in with the rest of the North Atlantic. Figure 1 shows a snapshot of sea-surface temperature (SST) on 26th October 2017. These data sets combines raw data from multiple satellites into a single gridded product – this particular product has an impressive spatial resolution of ~5 km.

It can be seen from Figure 1 that warm waters (red) are located close to the equator while waters become cooler (blue) nearer the Arctic Circle. The Gulf Stream transports large amounts of heat northeast towards western Europe, drawing up waters from the south along the east coast of the US.

This however is just a snapshot; the wiggly meanders of the Gulf Stream vary from week-to-week and year-to-year. This is because the underlying fluid dynamics are turbulent, which makes the system hard to predict, but also interesting to study.

The video above shows the ocean currents – with the brighter colours indicating a higher speed – on each day for the period 2014-16. This velocity data came from satellite observations, and was produced by Ssalto/Duacs and distributed by Aviso, with support from Cnes (

There are many ways to visualise such ocean currents, and whichever way you choose mainly depends on personal preference. Figure 2 shows various methods for displaying velocity data from 27th September 1995. The contourf method (top right) produces a nice, clean, display of the ocean current speed, while the quiver method (bottom right) includes information on the direction of the flow. One potential issue with the quiver and streamplot methods is that they can lead to confusing plots which have too much going on.

Figure 2: Visualising a snapshot of the ocean current velocity on September 27th 1995, using four different methods: pcolor (top left), contourf (top right), quiver (bottom left) and streamplot (bottom right).

Plots can be improved by adding information on the continents and countries within view, in addition to visualising the ocean data. This can be achieved by the brilliant Matplotlib Basemap Toolkit. Not only does this library draw continents but you can also customise the 3D projection used to visualise geographical data on a sphere – both the continents and the 3D projection in Figure 1 were produced using the Basemap package. These projections can be a bit fiddly to get right, but once you do they look great.

The Basemap library can also be used to add state boundaries for North America – this can provide a sense of spatial scale to the observational data being visualised. For example, the video above shows Aviso sea-surface height (SSH) data for the period 2014-2016 (full-screen is recommended). The state boundaries, in combination with annotating major cities, provides a better sense of the scale of the Gulf Stream – it’s difficult to comprehend the scale of a 100 km wide turbulent eddy in your head.

Another Matplotlib Toolkit that is useful is the 3D plotting library. This is particularly useful SSH data which naturally lends itself to three spatial coordinate axes. A 3D surface plot is shown in Figure 3; a snapshot of SSH from 27th September 1995 (the same day as Figure 2) is used.

Figure 3: A 3D surface plot of surface height using the 3D plotting toolkit within Matplotlib. The data is a snapshot of SSH on 27th September (land points were made brown and were given a negative value simply to differentiate them from ocean points).

(The python scripts used to make these illustrations can be found at Visualising the Gulf Stream on GitHub)

Author: tombolton7

Physics PhD student at the University of Oxford. Data science and machine learning enthusiast.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s