Positive Spikes in Solar Radiation Measurement not in UV Index Measurement

When the mockingjay stands on the sensor head it blocks sunlight getting to both the VIS and UV sensors. I agree that a flyby exactly corresponding to a once a minute measurement is unlikely. Do we know that this is how sensor readings are taken by the Tempest? If instead the sensor is queried multiple times in a minute and an average value for these measurements is reported each minute a flyby would be more likely to show up in the data.

indeed if it is standing on the window, it’s feet would block both uv and vis. if it’s feet are not on the window, the white chest might reflect a lot of extra light onto the window.

even if the values would be averaged during a minute. the time for a fly-over would be very short and hardly influence the average. Also if it is flying at any distance from the unit, its contribution will be pretty small as it covers only a tiny part of the sky.

do your bird landings introduce false rain events? if so, you could find correlation between false rain and spikes (both positive and negative)

(bedtime, hope you will find the cause of it)

Bird landings (negative spikes) induced a lot of false rain, especially during nesting season in June and early July. On July 2nd I recorded 0.29" of false rain lasting a total of 166 minutes. I never see false rain associated with the positive spikes. I think this July 2nd data shows conclusively that positive spikes are not connected with the Mockingbirds as not a single positive spike was observed at any time during this entire bird landing packed day.

you’re not alone to see this and I’ve been looking at it recently trying to understand

I have the Tempest on a 9 meter pole and the sky is just about at it’s base about 1 meter from ground

this is what the server gets for the same day

Not sure if there is some ‘smoothing’ done server side with values going way out of range or if our Udp sniffer has more ‘luck’ to catch these …

I need to have a few more days where i can correlate these and from there ask the devs to have a detailed look. No luck we have no open sunny days for now, often clouds passing by and I don’t see spikes …
example of value from Udp

1596713160, 961, 962
1596713220, 963, 962
1596713280, 1508, 963
1596713340, 964, 963
1596713400, 964, 963

i still think the positive spikes are suspicious and are in need of some explanation .
@eric Please make sure that the spikes are not in the zoomed in data as the zoomed out view might just show the nearest sample.
If it is in the udp data but not on the servers, it might indicate some strange software/firmware problem. Let’s hope it is, because that means it might be fixable.

Interesting enough is the fact that the little bumps in the bottom curve around 08:00, 15:40 and 16:00 aren’t visible in the top green/yellow graph. Is this perhaps a graph created on the server and is based on a combination of your data and other sources, where the other sources happen to see a cloud at those times? That would also explain why the spikes might get filtered out.

Here is a zoomed in sample from the Tempest Smart Weather app on an iPad showing positive spikes in the Solar Radiation plot that are not present in the UV Index plot. An interesting observation is that the spikes tend to be roughly 2x the data points to either side of the peak, except when they occur near the peak of the Solar Radiation curve.

Regarding the primary reason for this post, the anomalously high Solar Radiation readings, I think I now understand what is happening. On days where the sun is shining between surrounding white clouds the light sensor in the Tempest is seeing the base irradiance from the sun plus additional scattered sunlight from the surrounding white clouds. This produces a total irradiance on the sensor that is greater than the irradiance produced by the sun alone in a clear sky.


One way to identify positive spikes even on cloudy days is to compare corresponding Solar Radiation and UV Index curves. The positive spikes occur only in the Solar Radiation (and Brightness) curve and never in the UV Index curve. The strangest thing is the positive spikes are not there every day. Some days there are lots of them and on other days none. Not sure what is triggering them.

I think it would be really interesting to put a camera up.

I’m happy that at least that came across. Of course it depends on the type of cloud and the amount of clouds.

Here’s a spike in my UDP data. This was collected during our smoke from wildfires (AQI above 500!) in the Portland area. There were NO clouds (you could hardy see the sun) with a consistently brown sky:
9/12/20 3:18:53 PM 169
9/12/20 3:19:53 PM 170
9/12/20 3:20:53 PM 171
9/12/20 3:21:53 PM 171
9/12/20 3:22:53 PM 322
9/12/20 3:23:53 PM 170
9/12/20 3:24:53 PM 169
9/12/20 3:25:53 PM 168
9/12/20 3:26:53 PM 166
9/12/20 3:27:53 PM 165
I see these spikes most every day. Some days none, some one, some several. There’s never a corresponding spike in UV. There are corresponding spikes in Illuminance.

Support took a look at my data (after prodding) and they agreed that there were spikes in SR that did not show up in the UV readings so they sent a replacement unit. The replacement has not been spiking at all (up or down), even with the occasional bird-faux-rain events.

I received the following update from WeatherFlow Support on September 17th at 8:35AM:

Thanks for following up with us. Sorry for any confusion with ecommerce’s forum comment. That user wrote in to us on July 29th and after a few weeks of troubleshooting and communicating with our team, we decided to replace the Tempest unit. It was shortly after that our team decided that we didn’t need to replace these units anymore as a firmware fix would solve it. Once we got to your original support ticket about this, that decision for the firmware fix was already set.

To confirm, the firmware fix is coming and I checked with the team and they’ll update your Tempest to this new firmware update shortly. The new Tempest firmware version should be v142 - keep an eye on it for us!

Kind Regards,
William Gidley
WeatherFlow Support

My Tempest sensor head received the Firmware Revision 142 update between 8:34AM-8:43AM (at the same time as this e-mail was sent).
One other change made by this update was in the Sensor Status entry which changed from OK to LIGHTNING_DISTURBER. I’m not sure what the significance is of this change. I will update this post after monitoring the Tempest output for positive spikes for a week or so.

Not much significance - it’s perfectly normal. It indicates that the lightning sensor is basically identifying and rejecting interference signals from non-lightning sources. It is nothing to worry about, and the status may switch back to OK if the interference stops


After a little more than a week with Firmware Revision 142 I can say I have not seen any more positive spikes in the Solar Radiation or Brightness curves. However I have noticed new small correlated negative spikes in the Solar Radiation and Brightness curves and uncorrelated negative spikes in the UV Index curve. Also, Sensor Status has continued to show LIGHTNING_DISTURBER instead of OK since the update.

You can ignore the Lightning Disturber I have an Air that is constantly showing that error. I went through this with @djs three years ago and satisfied myself the unit is fine.

It’s perfectly normal and nothing to be concerned about.