Positive Spikes in Solar Radiation Measurement not in UV Index Measurement

For several weeks now I have been seeing random positive spikes in Solar Radiation and Brightness plots that do not show up in the UV Index plot (which I assume is derived from the same light sensor). Some days there are none and on other days several. In the sample plot you can see this. The negative spikes are due to bird landings on the sensor and are observed in the Solar, Brightness and UV plots. Both of the days shown were sunny, cloudless days with humidity between 50% and 60%. Initially I thought dew drops on the sensor might be focussing light on the sensor but the fact that the spikes do not show up in the UV Index plot and can occur in the afternoon of a warm, dry day argue against that possibility. Has anyone else seen this, or is there something wrong with my unit?

I’ve had the same problem, opened an issue with support… first response was “the spike isn’t in the data so it’s likely a graphing problem”. I sent pictures of what had been sent to wunderground (with the same spike) and they’re taking another look. Since now I know I’m not alone, I’m hoping this is something that can be fixed—although it’s a lot less bothersome than the low air temperature readings when in full blazing sun. I was wondering if the solar radiation sensor was reading high causing overcompensation of the computed air temperature. Who knows.

I’m collecting UPD data from my unit and see the same spikes. Did you get an answer?

Yes here is the response I received on August 18th:

Thanks for writing in. Our team is aware of these rare solar radiation spikes and are looking into a fix. The overall issue is fairly rare across all Tempest’s and we don’t think the device is defective. Hang tight for now and our goal is that a future firmware update will contain a fix for this. Thanks for the report!

Kind Regards,
William Gidley
WeatherFlow Support

1 Like

To update I still see days where positive spikes occur and days when none occur. I haven’t been able to correlate the presence of positive spikes with anything obvious like battery charging, temperature, humidity or any other condition I could think of. The only other observation I have is that these single point positive spikes are roughly 1.85x of the points to either side of the spike except near the peak of the solar radiation curve where the ratio becomes somewhat less than this. I suspect this condition is more widespread than WeatherFlow Support thinks as only people logging data from the Tempest are likely to notice these positive spikes.

On partly cloudy days like yesterday (2020-08-13) the Solar Radiation measurement is unrealistically high. Typical peak Solar Irradiance measured on a clear sunny day is around 1175 W/m2. On this partly cloudy day the peak value was 1620 W/m2 with many readings above 1200W/m2 (not including the asterisked anomalous positive spikes I also see in the Solar Radiation that are not present in the UV Index curve). Is the Solar Radiation/Brightness sensor in this unit broken?

It needs to be calibrated by CL. Typically UV is reading too low and SR is reading too high when it first comes out of the box.

My unit has been in continuous use since May so Continuous Learning should be well established by now. On clear sunny days the results are reasonable (except for the random positive spikes which are usually 2x the data points on either side of the spike) it’s only on partly cloudy days that the optical sensors return impossibly high values as shown in the example above.

Oh…I would send a support ticket to WF and see what they say.

To a degree, elevated UV on a cloudy day is actually quite realistic. Due to cloud edge reflection, the UV index can be elevated by up to 25%.



What about the nearly 40% increase in maximum solar radiation (from 1175 W/m2 to 1620 W/m2)?

That I don’t know I’m afraid

i’m not so much worried about the elevated values, but I do wonder about the spikes as these are unexpected. Are there some highly reflective surfaces around? or was there water on the sensor? If not there might be something wrong.

If you try to draw an imaginary smooth line through the partially clouded data, ignoring the data influenced by the clouds, the curve almost matches the data from clear days. the cloud data itself seems to be around the 25% increase you can expect, but the spikes are not.

No water droplets on sensor. Sunny days temperatures in high 70Fs, humidity around 50%. As you can see in site photo there are no reflective surfaces in sight of the sensor. Spikes occurred before the Crown of Thorns was added to the head as shown in this 3-Aug-2020 data set (negative spikes are mockingbird landings). The interesting thing is the negative bird spikes show in both the VIS and UV sensor data but the positive spikes (which are typically 2x neighboring points except near the peak) only show in the VIS sensor data.

a white butterfly (provided it’s wings are not reflecting the uv light)?? perhaps temporarily point a webcam at it.

The other funny thing is that the positive spikes do not occur every day. Here is yesterday’s Solar Radiation and UV Index plots (there were some thin clouds early in the morning).

so that might indicate that the device is working fine (at least on some days). A butterfly might not be a weird suggestion after all. Those wings have some weird spectral characteristics. next time you see a spike, run outside.

Butterfly wings are most likely Lambertian scatterers (i.e. they scatter light equally into all angles of a hemisphere) so they will not increase the irradiance measured by the VIS detector. Even if they were perfect mirrors they would not increase the irradiance recorded by the detector. The butterfly wings would have to have some curvature (optical power) to focus light on the detector to increase the W/m2 recorded by the detector. Typical Solar irradiance at ground level is around 1120 W/m2, and at the top of the atmosphere it averages about 1361 W/m2, so readings of 1620 W/m2 are highly unlikely.

Actually, in thinking about it some more butterfly wings (or other scatterers) could direct additional light into the sensor that could add to the baseline radiation from the sun. I would bet more on the mockingjays flying by, instead of perching on the sensor head as a possible cause of this but it’s still odd to see these single point (1 minute) spikes in the VIS but not UV sensor.

The 3-Aug-2020 data above might support the mockingjay fly-by explanation for the positive spikes. There were mockingjay landings on the sensor head only before noon and positive spikes also only before noon. I’ll have to go back and look at some other data sets from before I added the Crown of Thorns to see if there really is a correlation between bird landings and the appearance of these positive spikes.

for sure if you were to put a vertical piece of white paper just next to the sensor’s window and both are illuminated by the sun, that bright white paper as seen from the sensor is way more brighter than the blue sky it is blocking. If it was a perfect mirror, the sensor might see two suns. No curvature or lens is needed. I’m not saying it are butterflies, but it might be possible.

you probably have noticed at some time in your life, that on a sunny day with some partial clouds, you want to put on your sunglasses much more then with just blue sky. The sky might be brighter with some clouds.

if that is the bird, it is pretty white chested. I would think that depending on how it stands on your unit, it might indeed reflect more light into the unit without covering the sensor. Only helpful when their wings just as the butterflies don’t reflect UV as much as visible light. But then again I thought the positive spikes were still there after you installed the anti bird crown, I assume the bird isn’t sitting on the unit any more, or is it??
(flying over is pretty unlikely to go together with the exact time of measurement so that doesn’t cause the amount of positive spikes the you see)