Some of my stations update the web server every minute with current data. That can make a scattery polar plot of directions. I've already sampled the data to plot fewer dots on the polar graph, but it's still not as clear as I'd like it to be.
VWS seems to apply some smoothing to wind direction data.
I thought about doing a 10 minute degrees average, but that has a problem. If you have 5 samples of data at around 315 degrees each (NW), and 5 samples at around 45 degrees (NE), the average will be 180 degrees (South), when the average is clearly 0 degrees (north).
I have a couple of thoughts, but I wanted to see what other ideas people have.