Thanks @Jens Kleb for your advice. Sadly the problem comes from a constraint on the available data. To my knowledge we are assuming everything is over a flat plain, but I understand this still has its complications. Maybe we'll reconsider our approach.

Hi Ellis Thompson ,

I could advice you to consider an off-shore application for your study, because the interpolation of vertical profile of wind speed is easier to model in open sea, respect to a realistic land topography. Moreover, you could use a spatial and time resolution lower than you selected. In this way, as starting point you could achieve an easier model, that you can make more precise (with increased resolution) by means of subsequent steps.

About the issue of availability of data, you could use the NASA database as source for data

Best regards

Claudio

Thanks Claudio F. Nicolosi for your advice. The project we are working on is in part with NASA so we are using a forecasting tool designed specifically for this project. I will however look into your suggestions, thanks!

So this has plagued me for a while. The short answer for anyone following this is to use linear interpolation or something like a 2-point COSINE interpolation (What we settled on).

The problem is, as intuitive as it may be, is that this assumes that the position you are interpolating will lie exactly on this line between point A and B. Secondly, interpolating wind speed might be easy but wind direction is not. A somewhat simple solution to this is to change it from direction and magnitude to just a vector and interpolate it that way.

The better interpolation method that we finally rested on was just to hit our wind server with more traffic in an effort to do 4 point interpolation.