WIND

FROM CYCLINGPOWERLAB


Readers who have found this article having read Yaw, Drag & Component Choice  may be wondering if we at CPM aren't a little obsessed with the subleties of wind. The truth is, we have to be, because when a rider is spending the majority of his power output moving through the air, the relative speed and direction of that air matters - a lot. It matters in terms of his speed over the ground and it matters whenever we want to accurately model that speed or the power required to produce it. 


This article appears in response to an important question...What is the best or correct wind input to a model of cycling power? Is it forecast wind, is it something that should be measured locally, or can we use forecast wind tempered by some correction factor that suits our purposes? 


We recently did some field testing on quite a windy (read "typically windy" if you're in the UK) winters morning and found that the metorological wind implied by our validated power model was significantly different than the number we read from an airport weather station just a few miles away. An initial conclusion was that we might achieve better results measuring the wind with a hand held anemometer, at rider height, but what when that isn't possible? We then worked through a number of the criteria that may affect wind and eventually came up with a model based correction factor.

Wind speed in particular is affected by surface friction attributed to the local terrain. This means that the closer we get to the ground, the slower the wind. Have you ever stood on a roof terrace and noted that the wind feels significantly stronger than it did in the street? Pilots are familiar with this concept of wind gradient because it means they have to apply different wind correction factors to their navigation plans, depending on crusing altitude, and because they have to be very aware of extreme wind gradients close to the ground, aka "windshear". It also means that the use of the surrounding ground can have a real impact on surface windspeed. Ride through a field of sunflowers on the Tour de France and the wind gradient may be steeper than were you to ride across the desert in the Tour of Qatar.

Perhaps obviously, wind speed is affected by nearby obstructions such as buildings and trees. In fact wind gradients differ between urban and rural environments - a town centre prologue time trial is not the same as a longer, more windswept test through the fields of Normandy. International norms for the siting of weather station equipment state that wind speed and direction sensors should be located 10 metres above any obstructions (usually the ground) and clear of obstacles.


The speed of the wind itself, as measured at a "surface" weather station, affects the magnitude of wind gradient, greater speeds being associated with steeper gradients. As does the stability of the air, in some way related to the amount of solar radiation it receives - the wind gradient genuinely differs between day and night.


A Model for Wind Gradient


To find a model for wind gradient we have to look to an unusual field...smoke dispersion models used in the regulatory control of chimney stack siting! In this domain it is normal to use a model of wind gradient to relate wind speed at the top of a chimney (height X) to speed at a nearby weather station (height 10 metres). There is a standardised model of wind gradient, applicable to other engineering applications, that is said to hold at all heights AGL (above ground level) and we can apply it to the problem of calculating a reasonable wind at rider height, say 1 metre from the ground. Really interested? Really?! There is an outline of this particular model here. 


We are now offering this method of wind gradient correction in our main power models. Simply input the wind speed from a local weather station, as normal, check the "Gradient" option next to the wind speed input box and the model will convert what is assumed to be a windspeed observed at 10 metres AGL into a slightly lower speed more like that experienced by a rider moving 1 metre from the ground. You can try this simple model here:

Just as an awareness of expected weather conditions helps cyclists match clothing to the demands of an event, attention to the winds that might be expected on race day is an important consideration to ensure optimal equipment (especially wheel) choice and can greatly inform the development of an optimal pacing strategy. Even without the weather forecast for the riders time-on-course at a particular event savvy racers, PB hunting time triallists, or even just “Strava Tailwind Segment Cheats” can gain an advantage by studying the expected wind conditions on a particular course. But what exactly is the expected wind? 


Weather forecasts provide one interpretation of the expected on-course winds but what if you dont trust their accuracy, want to consider all eventualities, or need to plan for a performance some way in the future? Choosing an event that could maximise your PB or buying the optimal wheels for an goal event are examples of decisions that need to rely on a longer range approach. The problem becomes one of identifying the most probable weather conditions or range of conditions and optimising based on the assumption that these will come to pass, a form of “probability weighted” view on the weather.

Wind Roses


Wind roses are a graphical representation of the history of winds experienced at a particular location. And because the worlds environment is relatively stable (in spite of what some global warming naysayers may like to claim) history is a great predictor of future exprience. In other words the distribution of winds observed at a location throughout the previous 10 years (for example) is equivalent to the probability distribution of winds likely to occur in the future. Now imagine how useful it would be to study a wind rose that illustrates the probability distribution of winds on a particular event course, in the month of the event, filtered down to the time of day in which the event takes place. That is exactly what our windrose builder allows you to do.

Inputs


Outputs