Ok, this blog is about the most specialised, nerdy and technical paper we've published in ages... but I love a challenge, so let's try and see if we can make sense of one very dense analysis... *gulp!*
We humans live in a boundary layer - where the atmosphere meets the land. High up above us, wind is moving very fast. Right at the point where air meets dirt, the air isn't moving at all. In between, friction between land and air causes the wind to slow down from that fast motion up high, so that it eventually reaches that no movement situation at the ground.

How much friction we get and over what kind of distance depends partly on what is present on the land. For example, short grasslands don't slow the wind down much and it can stays fast quite close to the ground. A forest, on the other hand, is pretty good at slowing wind down and doing it gradually, so the tree tops might be shaking while you can walk along without feeling much breeze at all at ground level.
Other than causing you to feel sheltered or wind blasted, these conditions also influence processes like evaporation, gas exchange between ecosystems and the atmosphere (i.e. carbon cycling, nitrogen cycling etc), and weather.
So to predict and understand these important things, we have to be able to describe the effects of friction from the land on the air in models.
Typically, we use 2 numbers to do this.
One is called the zero plane displacement, and it describes the point where the wind speed drops to zero. It might be zero itself which just means that the wind continues to have some speed all the way to the ground level. But it might be higher - for example in a dense crop of tall wheat, it might be at about 0.2m, because below that height there's basically no more wind.
The other number measures how gradually the wind speed increases. It's called the momentum roughness length, and the longer it is, the more gradual the increase in wind speed as you move away from the ground.
You can estimate these numbers from data if you measure the wind speed at a bunch of different heights above the ground, and then back the numbers out from those measurements.
Ok so here's the problem.
Mostly you don't have measurements to help you estimate these numbers. So instead they are treated as being constant values that you instead estimate based on how tall the canopy of plants (or buildings in a city) is. If you measure these numbers over a grassland and check against the estimates, the estimates are ok and they don't change - great.
But we were looking at them over the Banksia Woodlands at Gingin Flux Tower... and they were all over the place and not really very similar to the estimates... not so great.
And it turned out that this wasn't just a problem at Gingin but all over the world. Forested sites had variable estimates of these numbers and often they weren't close to the "usual" estimates based on canopy height. Compare the blue ticks on the y axis (the usual estimates) against the PDFs (the variations we saw) below - this is for zero plane displacement. The grass site - Yanco - shows no variation and a mean of 0.05m which isn't too far from the estimated 0.2m. The other sites are much further off off.

Gah, this blog post is already crazy long, and we've only established the problem. Well, one thing you might notice in the figure above is that where we have tall deciduous forests, the measured zero plane displacement changes with the season, and is smaller when the trees have no leaves (this doesn't happen at Willow Creek, but is obvious at Bartlett and Charleston forests). So this suggested that to understand the variations in zero plane displacement and momentum roughness length, we might need to dive into models of how wind interacts with canopies.
PhD Graduate Dr. Ashvath Kunadi did that deep dive, and ended up using four different models that relate wind speed and canopy drag to explore variations in the zero plane displacement and momentum roughness length. I won't lie - it was a slog, particularly for some models that wouldn't allow analytical treatment. And as much as I love Ashvath's work here, I won't try to go through the models in this blog. But I will share what I think is my favourite picture from this paper, which shows how 2 of his models estimate the canopy drag (how strong the friction was) at Bartlett Forest during 2019, compared to phenocam pictures. Like the data above, we can see that the drag increases from the leaf-off deciduous condition and reaches its maximum during summer with a fully leafed out canopy. Neat!

Finally, while the canopy drag models had a lot of difficulty predicting the hourly wind speeds, they did a far better job of estimating zero plane displacement and momentum roughness length than did the standard models. For Gingin and Charleston forest, below, the shaded green shows the measured zero plane displacement, and the lines show different models - based only on velocity, on the friction velocity and then 2 of the canopy drag models. We'd do a lot better understanding these friction behaviours and their variations if we account for the canopy drag.

This is dense, this hasn't completely solved the problem, but we hope it's opened up some interesting research avenues in micrometeorology. Ashvath was a persistent CHAMPION in this work - going deeper, more theory based and with ever more comprehensive analyses. So if you are using a do or zo in your work - consider whether its variations matter, and whether Ashvath's models can help you zoom in on that variability.
Comments