Thanks for the great side-by-side data, Peter! Having a Davis VP2 co-located with your WeatherFlow station produces some great data. Wish you had been part of the field test!
Your rain graph actually highlights something interesting: the SKY is a lot more sensitive than a tipping bucket. Note how, even in this relatively strong rain event, the start of rain happens significantly earlier by SKY. In lighter rain this is even more extreme, since the Davis tipper needs 0.1 mm (or 0.01 inch) before you know it’s raining, but the SKY will report much lower rain amounts - it will actually let you know (via the “rain start” event) within 5 seconds of rain starting. This higher sensitivity is also what makes the SKY’s curve so much smoother than the Davis’s. It looks like the total accumulation is a bit low with the out-of-the box calibration, but that will improve when we roll out the rain component of our CL system (real soon now!).
Your wind graphs provide a great case study too: At these low speeds it’s easy to see how the Davis reports very discrete gust values: 0.0, 1.6, 3.2, 4.8… km/h (0, 1, 2, 3… mph). That’s because the gusts are measured by counting the number of times the cups spin over a fixed time period. The resolution of the Davis is 1.6 km/h (1.0 mph) - the Davis can’t measure gusts anywhere between multiples of that value. The SKY, on the other hand (thanks to the fact that sound waves move much faster than spinning cups) has 10 times higher resolution: 0.16 km/h (0.1 mph).
As for the SKY reporting wind when the Davis did not, there are two possibilities. First, as @weather34 alludes to, when there’s rain (or anything that’s not air) on the reflective plate, that can cause wind readings to be less accurate. SKY has filter algorithms that attempt to correct wind data when it thinks there may be rain (or any other object) in the path of the ultrasonic sensor signal. These filters are not perfect, but they do a pretty good job. However, after looking over your SKY’s debug data during this period, we can see that these filters were NOT in effect. That is, the filters had no effect on the wind values from your SKY during this period.
The other possibility has to do with physics. A new Davis anemometer has a cut-in speed of 3 km/h, thanks to the inertia and static friction inherent in the spinning cups. That cut-in speed increases with age as the bearings wear and corrode. On the other hand, the ultrasonic anemometer in the SKY does not have any inertia or bearings that need to be overcome in order to detect wind (or bearings to wear and corrode). They do have a noise floor that defines the cut-in speed which varies by device, but this is in the neighborhood of 1-2 km/h.
The wind speeds shown in your analysis period are very light. In fact, many of the 3-sec readings report 0, but it’s very rare for outdoor wind to be truly zero for a full minute at a time. There may have been some very light wind - enough that the SKY was picking up, but not enough to get the Davis cups going. Next time the Davis reports zero and it seems like there is virtually no wind, experiment by lightning a match and watching the smoke.
PS: Note, your Davis will still report speeds lower than 3 km/h, but only as the already-spinning cups slow down.
PPS: I’m really looking forward to @weather34’s sub-$1 solution to the rain on the plate issue (even though it doesn’t look like that was the issue in this case)!!
Personally, I don’t want that my station simply takes (or compars) data from “neighboring” stations since our territory differs from site to site vastly, especially the wind.