As a visualization engineer at Planet OS, I learn new things everyday. My job is to get just enough domain specific knowledge to come up with visual solutions. This week I learned what is a Temperature-Salinity (T-S) plot.
Here’s a definition of what you can use a T-S plot for (from the same textbook as the image above):
“Plots of salinity as a function of temperature, called T-S plots, are used to delineate water masses and their geographical distribution, to describe mixing among water masses, and to infer motion of water in the deep ocean. Here’s why the plots are so useful: water properties, such as temperature and salinity, are formed only when the water is at the surface or in the mixed layer. Heating, cooling, rain, and evaporation all contribute. Once the water sinks below the mixed layer, temperature and salinity can change only by mixing with adjacent water masses. Thus water from a particular region has a particular temperature associated with a particular salinity, and the relationship changes little as the water moves through the deep ocean.”
We had the need to plot temperature and salinity as a function of pressure (which could also be used to infer depth). As Ryan Abernathey kindly mentioned, this chart is different from a T-S plot. The image above shows a T-S plot on the right and a plot of temperature and salinity over pressure on the left. My goal was to reproduce the left one, inspired by the chart from this paper starting from this dataset.
From the visualization standpoint, we can see two lines: one for salinity and one for temperature, both on the X axis, while pressure is on the Y axis. I like to see them as sensors dropped from a boat, coming down the water column, then moving back up, getting measurements along the way. That’s why we can see these two lines making a round-trip from top to bottom and back.
The chart implemented can be zoomed and panned which is useful to better see some intricate patterns. One interesting feature I don’t get to implement often is multiple axes. The creator of ggplot2 (the best charts library IMO), Hadley Wickham, has some strong arguments against multiple axes. We could certainly discuss alternatives, like small multiples or some linked scatterplots, but T-S plots are the tool these experts are used to use. And my opinion is that the best tool in the world is less useful than the tool we already master.
Go try it out here! This T-S plot and many others are part of some upcoming features we will announce soon. So stay in touch!
Data visualization engineer at Planet OS