Box plot

The spread of many readings per category: stations, missions or depth layers compared by their middle and their range.

Depth layers

Temperature through each 2026 cast in three layers. The surface layer warms and spreads through the summer while the deep water hardly moves.

Medium
<BoxPlot
	categories={missions.map((m) => m.date)}
	series={[
		{ key: 'surface', label: '0–25 dbar', values: missions.map((m) => surfaceTemperatures(m)) },
		{ key: 'deep', label: '100 dbar and deeper', values: missions.map((m) => deepTemperatures(m)) },
	]}
	valueLabel="Temperature"
	unit="°C"
/>

One series

Oxygen through the whole cast, per mission: the dots below each box are the low readings near the bottom.

Rules

  • For many readings per category, where the spread matters as much as the middle. A handful of values belongs in a bar or a table.
  • Tukey’s box: the quartiles, the median across, whiskers to the furthest reading within 1.5 box-lengths, anything further a dot. Nothing is hidden: every reading is in the box, on a whisker or drawn.
  • The value axis fits the data, not zero: a box is read by where it sits, not how long it is.
  • Side by side for groups that are compared, like depth layers at each mission, a colour each with a key.
  • Hover reads the five numbers, how many readings and how many outliers.