29/05/2026
In metocean, we rely on complex computer models to predict wave heights, wind speeds, and ocean currents. But how do find the most accurate one without scrolling through millions of rows of data?
We use Taylor Diagram, the ultimate visual report card. It uses the geometry of a triangle to combine three vital statistics for Standard Deviation, Correlation, and RMSE into one single chart.
Here is how to read it using a real-world example:
🎯 The Bullseye (Reality): Let's say our actual offshore wave heights vary by 5 meters (Standard Deviation). Because this is reality, its Correlation is 1.0, and its error (RMSE) is 0. This is our target!
Now, let's test three models against this 5-meter reality:
❌ Model A. SD = 10m, Correlation = 0.2. This model predicts waves twice as big as reality AND its timing is completely out of sync. On the diagram, this dot is swung far away from our bullseye.
🥈 Model B. SD = 3m, Correlation = 0.95. Excellent timing! While it predicts slightly smaller waves (3m) than reality (5m), its patterns match perfectly. On the diagram, it sits remarkably close to the target.
⚠️ Model C. SD = 9.5m, Correlation = 0.95. While the timing is spot-on, it severely overestimates the wave heights, predicting 9.5m fluctuations instead of 5m. This pushes its dot far away from our 5m target.
Against a 5-meter standard, Model B takes the lead! 🏆
Taylor diagram are mathematically linked in the exact same way as the three sides of a triangle (using the Law of Cosines). Visualizing these metrics in one place is a total game-changer for evaluating metocean data quickly and accurately.
