The thing about wind

I’ve written before about my frustration/hatred of wind in daily life, but today I wanted to raise another complaint: using wind data in past climate research.

Wind direction is easy and cheap to monitor. All you really need is a damp finger and a decent sense of which way is north.

The strength of wind has also been observed for a long time, for its vital role in turning windmills and pushing ships.

But the thing about local wind information, is that it’s just not great for climate research.

The ease of watching the wind does not make up for that fact that it’s a pain in the arse to analyse when you are looking at long term changes in the climate.

And wind is predicted to change. A warmer earth means stronger weather systems and a shift in global wind patterns. This will in turn affect ocean currents and sea ice coverage and lots of other important things. So, we need to know how the wind is changing.

But historical wind records can be more like a drama queen than a reliable source of climate information:  irrational and highly sensitive to the smallest changes.

Wind data can be subjective

Wind observations have been taken for centuries, long before technical weather instruments like anemometers were used. Observing wind strength and direction was a vital part of any sailors’ duties, at land and sea. While direction could be recorded relatively accurately using a compass, converting what you see around you to a number is a subjective thing.

Here, for example, is the famous Beaufort scale (discussed in loving detail in this book):

Now these descriptions are poetic, but what if you aren’t near a chimney, or any small branches?

Or, what if your wind strength scale had slightly different descriptions?

Francis Beaufort was not the only guy trying to turn the chaotic experience of wind into a numerical scale. During my recent work with historical wind data in Europe, I came across the Beaufort scale, a French wind strength scale and a nine-point strength scale used in Turkey. These indices all use different definitions to approximate the strength of wind.

Even worse, today, we use the speed of the wind to study it, so these values need to be converted to something approximating metres per second, or kilometres per hour, or knots, or cyclone scales, or tornado strengths.

Do you see where I’m going with this? Combining all of these different methods of measuring how strong and/or fast the wind is it a tricky job. It also means that we are comparing data with say, 13 different levels (0–12 on the Beaufort scale), with data that have many more (an anemometer can measure at 0.05 m/s intervals).

This is almost, but not quite, an apples and oranges scenario.

Also, wind data are mighty sensitive

Even once the Beaufort scale was replaced with good ole’ technology, wind data can still easily be rubbish, because they are extremely sensitive to the local environment.

This is fine if you are recording the wind at an airport for safe landings, or to see if a wind tunnel has accidentally been made, but if you want to look at large wind patterns over a long period of time, you might be in trouble.

The growth of a tree nearby, the erection (teehee) of a building across the road, or a small change in the location or height of the instrument can have a big impact on wind data. Temperature, rainfall and other weather variables can also be affected by these things, but wind observations are particularly fussy.

It’s not all bad

I realise it sounds a lot like I’m saying observing wind is a waste of time. But things aren’t quite that bleak.

Understanding local wind is really important for many things, like renewable energy, aviation safety and urban planning.

And, to be fair, wind observations taken at sea are generally much better than on the land, due to the lack of trees, buildings and other terrestrial nonsense.

Climate models are also pretty good at deriving large-scale winds from air pressure observations, so most long-term wind studies looking at climate change use models instead.

Finally, there are methods of correcting the non-climate features in wind data, if you are careful and have good information about the weather stations (it’s been done in Canada for example, and the Iberian Pensinsula).

But be careful next time you explore some old wind data. There’s a hidden drama queen in there, who might be telling you more about the neighbour’s pine trees than climate change.