Why the GWC is defined for several roughness and height classes?

[had]

Dear WAsP team,

I am a long-term user of WAsP, but, shamefully, there is one very fundamental feature of the model, which is not very clear to me.

As I understand the principle of WAsP, the model generalizes the site specific wind data in a way that the effects of local roughness, orography, measurement height and possible obstacles are deducted from the data, so that only the information of "general" windiness is preserved. I would expect then, that the wind climatology for just one standard height and roughness length would be a full description of the generalized wind climate, because any other information is site specific and can be calculated from the general data and local properties. Even more, I would expect that the input wind data, at least the ones provided by .tab observed wind climatology files, cannot provide more information than for one general roughness and height, because the file contains no further information that would enable to distinguish between the classes (like the wind shear or stability perhaps could be).

In spite of it, the classification of GWC in height and roughness classes is a prominent feature of WAsP application. Then my questions are:

1) Why are multiple height and roughness classes defined?

2) From where the information comes from, which makes the relations between wind parameters in different classes to be different in different wind atlases? For example, if the wind climatology for z0=0.1 and h=10m is the same in two GWCs, why it may differ in other height/stability classes? (May it?)

3) Could you advice some "rules of thumb", which circumstances and how do impact the differentiation as mentioned in 2) ?

4) Is there any literature describing this topic?

With regards

David Hanslian

[Rogier]

Hi David, these are some good questions. Since some of these decisions were already made when WAsP was developed for the first time (80s/90s) and I have been at DTU only since since 2010 I can only give you my best guesses. 

1) The generalized wind climate can be considered as a lookup table for the wind climate conditions. Typically, the workflow would be that you have one mast that you convert to a generalized wind climate and apply it to many turbines positions or a many points (resource grid). Therefore it computationally it make sense to do the transformation to a generalized wind climate only once. For all turbine positions or points in a resource grid we then only have to do a lookup in the generalized wind climate and not reprocess the whole histogram. Technically, the input histogram is actually first converted to a geostrophic wind climate (using the geostrophic drag law) and the the down transformation is continued (see page 585 here: https://backend.orbit.dtu.dk/ws/portalfiles/portal/112135732/European_Wind_Atlas.pdf ). But you could also continue from the geostrophic wind climate straight to the output locations, thereby removing some of the small interpolations errors that are made when interpolating the generalized wind climate. There is some code for doing this that we may introduce in the future, but it is not available in the WAsP GUI, also because the interpolations errors are typically very small. It is also easier to compare wind conditions if you have some standard heights and roughness you can compare with (as is done in the original european wind atlas tables). So in other words, the generalized heights and roughness are not strictly required but just a model choice that were made at some point.

2) The generalized wind climate is already 'cleaned' for effects of stability so these will not have an impact. So if your input wind climate, height, location (comes in via the coriolis parameter which depends on latitude) and stability for the generalization is the same and the wind climate at z0=0.1 and h=10 m is the same, all other heights and roughnesses should also be the same.

3) The main rules of thumb related to the generalized wind climate are to set the height of your observations and prediction close the heights of the generalized wind climate. Like that there will be minimal interpolation errors related to heights. Typically the generalized roughness classes interpolation errors are harder to avoid, unless the roughness length at your site is the same for all wind directions. If there are differences in roughness, you can set them so that the roughness length classes span the mesoscale roughnesses that are observed in your observed wind climate and your predicted wind climate. 

4) The only literature that described the generalized wind climate procedure (as far as I know) is the European wind atlas itself (mostly around page 585). Some details may be different in the code that is currently in WAsP though.

[had]

Hi Rogier, thank you for the explanation. An interesting information is that WAsP knows, beyond its GUI, to calculate directly from geostrophic wind to local wind, without generalization. It means that for such calculation, no height and roughness classes are applied? This is what WindPRO uses in its time series calculation (the scaler)?

Though I still struggle to fully understand the logic. I would expect that it would be sufficient do define the general wind climate for just one roughness and one height and the rest can be derived from it. And that the recalculation to different classes would be relatively simple, without need to store pre-calculated data, because this it is not site- and project- dependent. Am I wrong?

The most confusing thing to me is that different general wind climatologies sometimes provide different wind shears. You write that if my input wind climate, height, location and stability for the generalization is the same and the wind climate at z0=0.1 and h=10 m is the same, all other heights and roughnesses should also be the same. This is expectable. But why this need not apply if the wind climate at z0=0.1 and h=10 m is the same, but the wind climate, height, location and stability for generalization are different?

David

[Rogier]

Quote

It means that for such calculation, no height and roughness classes are applied? This is what WindPRO uses in its time series calculation (the scaler)?

Yes there is no height and roughness classes there, but I don't think this is what WindPRO is using. This code is in beta and has not left DTU as far as know. 

Regarding your other questions: the problem is that the observed wind climate is valid for a certain roughness and height, so if your output location is not for exactly that location you will need more information about what the wind climate is for other heights and roughnesses. 

[Duncan]

Another way to address this question is to by considering the data in the classic GWC files: the LIB format. 

Looking at the modelling workflow on page 584 of the EWA, we can see that there are four stages of calculation between geostrophic wind climate and the data in the generalised wind climates. To do these calculations, I think that we would need the latitude, friction velocities, surface land fractions, and heat flux parameters for stability corrections.

None of these are stored in the LIB files. Instead, we do these calculations in the generalisation and then store results for standard conditions. As Rogier says, this was convenient in the past, and in practice causes only negligible interpolation errors in most cases.

Note that the stability correction is applied at the very last stage of preparing the GWC. This is why the wind speed profiles in the GWC data set are not logarithmic.