The UV data from my weatherflow (4979) have been reasonably consistent, if slightly high, for several weeks; but last weekend (Sat/Sunday onwards) it has been reading very high - peaking at 16 !!!
Even now with overcast skies it is reading 4!
I don’t know what has changed. The solar power is giving a reasonable 177 W/sqM, it reads slightly high most of the time.
Any suggestions anyone?
Cheers
Phil
Hi @phil, thanks for the report. We took a look at the data and it turns out that on UV CL mistakenly considered June 15th as a “clear day” for you when it shouldn’t have. This threw off the calibration. That issue has been corrected, and your data should start looking sane again. Sorry for the inconvenience!
Thanks - I will keep an eye on it!
Phil
As I suggested before, it would be better if correction would be done in tiny steps. It would be totally weird if some sensor is still working ok and from one day to another suddenly needs a big change in callibration. Taking small steps would probably have hidden this problem and it should have corrected itself at the next callibration.
Alternatively, if the calibration change is greater than some threshold, don’t change the calibration until the next calibration would normally take place while flagging the out of range calibration for further inspection. This way the issue wouldn’t be masked but wouldn’t mess up a station’s calibration either.
depends on what you want. Do you want to see a big jump when the second calibration is also ‘out-of-range’, or is that jump ignored as well, or do you want to see the system slowly move to calibrated values. Big jumps are fine for the initial calibration where the system corrects possible bad factory calibration, or adapts to the setup environment. Other then that, calibration should perhaps only correct for drift in sensors, which hopefully is pretty small, so once it is calibrated, no big jumps should be allowed. ((but to allow for changing setups, the steps could be as large as say max 5% (way more then the sensor drift) so it can change a lot, given time))
Flagging for further inspection as you suggest, might get complicated. Who does the further inspection? the user? he or she doesn’t get to see all that is going on with calibration. Weatherflow staff? they might be unaware that you moved the unit to a different location.
But I would like to see in the graphs when a calibration is applied to the data (a little exclamation mark on the horizontal axis perhaps)
It could but doesn’t have to be. If the second calibration is close to what the first one was then move quickly to it but maybe not in one big jump, but if it is back to close to prior to the first calibration, flag the first calibration attempt and if another appears frequently then notify a person, otherwise ignore it and drop it after some period of time. This could allow for the unit environment changing or something like this happening and later getting cleaned off:
