I was wondering about this as well. For example my brightness and uv values were way off as in getting uv values of 15 where 4 or 5 was expected. They ‘fixed’ it remotely so now it gives a value that isn’t obviously wrong. They use remote calibration. If they calibrate so my brightness matches the nearest airport (or other official measurements) it might be wrong as my 180 degree view on the sky is for at least a few percent blocked by a high apartment flat.
If they calibrate the temperature so my temperature is the same as the nearest airport, my temperatures would be wrong, as I’m living in the city where temps are easily a few degrees higher.
I would have trusted it more when factory calibration would be almost spot on.
I still love my weatherflow station, but I don’t trust it.
This, combined with your phantom/crazy wind spikes suggest a hardware fault. Please open a ticket at http://got.wf/contact or support@weatherflow.com and - may need to send you a new one.
This will improve when the next phase is rolled out. Thanks for your patience!
The CL process does not simply use the nearest airport reading. The process, which varies by parameter and location, is more sophisticated than that. It attempts to account for local variability using a thorough offline analysis of your data vs other datasets (observed and modeled). And since it’s constantly being updated, it is much better than a factory calibration alone, which starts to drift as soon as it leaves the factory.
Note, the CL system is also pretty conservative, only suggesting an updated calibration when it’s very confident that it should. That’s one reason it’s taking us so long to roll out the process to everyone: we’re following the (meteorological version of the) Hippocratic Oath: First, do no harm (to your weather data)!
Thanks for your honesty. It’s our goal to win both your love and your trust!
I’ve had a couple of rainfall events since v94. All in all, it does indeed seem much better now, even if the rainfall calibration hasn’t begun yet. It tends to underreport slightly from my WS-1001 WIFI, but MUCH closer to reality than before.
I am noticing that I’m still getting quite a few rainfall notifications, which I assume is from vibration from wind or perhaps small birds. In almost every case, no rain accumulation is detected; I’m wondering if it would be better to change the notification to when at least 0.01" of rain has been detected in a distinct rainfall event?
If my understanding is correct, I think it is worth bearing in mind that despite Weather Flow knowing that the precised installation of the SKY unit has a large impact on its response to rainfall, all SKY units are currently using the same calibration. As a result some units are going to be reasonably spot on, some units are still going to be over reporting, and some (like mine) are going to be under reporting. This spread doesn’t mean all is lost, it is just the result of using the same calibration for everyone.
I am sure once the CL process is honed and ready to be rolled out, we’ll all see marked improvements to the performance of the SKY units, as the calibration used will be specific to your station and your precise installation.
As I follow the rain issue the main root of divergence from the basic factory calibration (as one followed the backer campaigns updates, WF spend a lot of time and thoughts on this ) seems to be the type of mounting and the influence of vibrations…
Does someone has put an micro- 3D- accelerometer to the Sky and take a look on parallel data ? @WFmarketing: might there be space inside the had of a sky unit for a sensor (I would suggest next to the ultrasonic transducers)?
I would have put one to each of my installations but I still have no vibrating issues as my rain reports are underestimating…
To be honest , as soon as the SKY is “triggered” - it reports rainfall- the rates and accumulation that are excellent- it is just that initial confirmation . After that it is very good.
My WF is back to over reporting again, after under reporting for a few rain events. It is raining now and it appears to be about double the actual rain.
We had a rain event here a couple of days ago that flooded my house from a neighbouring construction site. The readings from that construction site were 85mm in 15 minutes. The WF station on my roof recorded 15.5mm in 90 mins for the same period. Quite a difference.
Davis VantageVue: 8.6mm (my trusted other station)
Weatherflow (id 4979): 51.2mm
I have moved the SKY unit to a different location in an attempt to prevent wind gusts causing excessive vibrations. Will monitor for a couple of weeks or so. I previously had two other SKY units - neither of which detected rain (or me drumming with drum sticks) ! Weatherflow were very responsive in replacing the units.
FWIW it would appear that the SKY unit has similar issues to my VantageVue in that to get accurate wind speed you need to be off the ground (10metres is the recommendation) but rainfall should be measured at ground level (less chance of wind causing errors in rain fall detection/measurement).
Thanks for that. I checked with the Met Office here in the UK and their guidelines are very similar. For rain they recommend no more than 1.25 m above ground level. For wind it should be 10 m above ground (if in clear space) or 2.5-3 m above highest point (e.g. ridge of a roof) .
I’m finding the same - ie much lower from the Weather Flow, just put a calibration factor of 150% into weather display to compensate in the short term.
Rain doesn’t depend a lot on the height. The recommendation of 1.25 m above ground for the rain sensor, is provably historic, as it makes reading the instrument better. At 10m it rains the same as at 1.25m.
Having said that, the rain measurements should be correct. If you get false rain from your weatherflow, when there isn’t any, you should check your setup for vibrations.
Thanks for the info. For “calibration” purposes the unit is only a couple of metres off the ground at present. Still over registering rain by a factor between 6 and 8 ! And it isn’t subject to vibrations. The logged raining times are virtually identical to those of my Davis system. It is the quantities that are incorrect.
Looks like the Sky got confused and Rain Rate has been stuck in “Heavy” for 12 hours.
Actual rain is closer to 1/2 inch - not 4 inches.
It has been raining - heavy at a few times but not for 12 hours straight.
) seems to be the type of mounting and the influence of vibrations…