VP2+ UV Sensor Reading Dropoff - Opinions (Solved)

[aussiewmr]

Hi All,

I have installed my new VP2+ station and I feel something is amiss with the UV Sensor.

This sensor is mounted about 6 Meters AGL and receives sun from about 0630 through to about 1800 presently. The sensors are mounted on the sensor shelf on the ISS / SIM and plugged in direct with no cable extensions. Both have been cleaned as recommended by Davis (Ethol Alcohol).

The issue I have is during the day the sensor often drops to zero with only light cloud cover it seems. With light cloud cover I would expect there to be a UV level around 2 to 5 or 6 still. The Australian Government UV measurement for my area today never fell below 3 between about 0800 and 1200 today yet mine went to zero 5 times. And in fact my Fine Offset station, cheap as it may be, actually follows the Govt site much closer than the Davis.

The other thing is UV on the Govt site starts about 0600, yet I don't get a reading until about 0800, yet as I said above the sensor is receiving sun at about 0630. I understand Davis sensors seem to have a low UV cutoff of about 0.5, but going by that I should have a reading of 0.5 at about 0700.

Anyway my main concern is how it drops to zero during the day when UV levels should be measurable.

Using weather Link, when I overlay UV and Solar Radiation I can even see periods where the UV has gone to zero with no corresponding dip in Solar.

Does anyone else have this issue?

I am currently waiting for a cloud free day before I draw too many conclusions, but that may be a while here in Brisbane.

Cheers

Phil

[gemini06720]

Phil, although not the 'black & white' answer you might be looking for, have a look at this document: INTERPRETING UV READINGS

I understand that, to a certain extend, one feels the need to compare the results of one's weather station with other regional weather stations, but, unless the weather stations being compared are exactly identical, then all comparisons are less than useful...

Quite often, the UV index on my console will be at zero even when the solar radiation level is (way) above zero.

The Australian Government equipment measuring the UV index might have a wider viewing (collecting) angle than the Davis Instruments UV sensor or greater sensibility to the perceived UV index...

Remember that when the UV sensor is not aligned with the direct solar irradiance, measurements may understate the UV intensity at surfaces normal to the sun’s rays - in other words, if the sensor is aligned with the sun at solar noon, at times other than solar noon, the readings may be less than the actual intensity on the portions of an individual’s body that are normal to the sun’s rays...

[Gordon-Loomberah]

Hi from 4 or 500km SW of you
I suspect there is a problem with the sensor, because even when it's fairly heavy overcast (solar down to just over 100W/m^2) today, UV index has not dropped much below 2. There is a flat top at 16 most days though, due to the limitations of the Davis sensor. The lack of response below 0.5 is evident here, but it has to be extremely heavy overcast, generally under a storm, before it will drop to zero in the middle of the day, especially in summer.


Today:

Solar.gif

UV.gif

[Darknstormy]

Hi,
My weather station is the same as Gordon-Loomberah my UV and Solar Graphs look the same and drop off at the same time... so I agree its pointing towards your sensor being faulty.

[aussiewmr]

Thanks Ray - I understand what you are saying. But I do struggle as to why the sensor would go to zero during the day where the cloud is "scattered".

Today was the best day I have had all week with regard to UV readings with the exception that I never started reading UV til 0830 (when Solar was near 600 W/M2) at which point it leaped to 3 then 4, and then near the end of the day a strange spike to 4 - thats a bit weird.
And in the morning until about 9am (when the clouds start to form) for 8,9,10 Feb I get UV dropping off when solar is steady and increasing - when the sky is clear. I notice Gordon-Loomberah that yours does that too - it would be nice to understand why.

Thanks also Gordon-Loomberah and Darknstormy for your input - its (kinda) reassuring to see that others do not have their reading drop to zero frequently.

Trends of the last four days are below.

Today - Feb 10

Feb10.gif

Feb 9

Feb9.gif

Feb 8

Feb8.gif

Feb 7

Feb7.gif

[aussiewmr]

This post may or may not be related.

https://cumulus.hosiene.co.uk/viewtopic.php?f=6&t=7031

I am wondering if I could have a dodgy Sensor Interface Module at the rain gauge unit.

Cheers

Phil

[gemini06720]

Phil have you thought of directly contacting the Davis Instruments technical support at support@davisnet.com - would they not be the best people to ask all of your questions (rather than getting a lot of guesses from all of us) - they should know better, they have designed that weather station...

[aussiewmr]

Hi Ray,

As per my PM I must use the supplier and am happy to do so.

My "contract" is with him and not Davis. If there's donkey work to do with Davis he can do it. Presumably that's what part of his margins are for, and conversely Davis probably give him good prices in lieu of support he offers saving them time and money.

He should be an expert and to date he has been 110% there for help.

Thanks to you and all for advice received.

Cheers

Phil

[archae86]

Trends of the last four days are below.

Thanks for posting the trend graphs.

I have a rather new Davis VP2 UV and solar sensor, so have spent much time looking at graphs.

I am using a 5 minute update interval. Yours look higher resolution, so I'll hazard a guess they are one minute update interval?

If you compare to data posted by others at, for example, Weather Underground, I think you'll find that an interval so small as yours is rare, and all things being equal strongly increases the apparent short-term variation if there are any clouds or other intermittent shadow or reflection sources.

My own reaction to your graphs is that the thing which seems oddest is the long delay before UV comes off the peg in the morning. The easiest way to do that would be if your site shadowed the sensor. Mine has shadow from a neighbor's house, so instead of hopping up to 0.5 UV index (the first increment from zero, as you know), on clear days it was waiting until the sun cleared neighbor's house, at which point it would hop straight up to 1.1. On somewhat cloudy days, the UV was more scattered, so it would actually see enough to clear the 0.5 minimum EARLIER.

But examining your curve for solar vs. UV don't seem to make this quite the issue. Still, a careful review of any shadowing issues is always in order, and a comparison of clear and light or scattered cloudy days my be informative.

[aussiewmr]

Hi archae86,

Thanks for your thoughts and yes the logging interval is 1 minute. And that does make the issue more prominent. Still the UV should not go to zero given the light cloud cover. Mind you the last two days have not seen this issue but the cloud cover has been very scattered also.

Regarding the shadowing, I have been up my ladder at (or perhaps just before) 0630 and I can confirm that sun is on the UV sensor then. The sensors are aligned East - West in full sky view (so no pole in the way). Our house is higher up than our eastern neighbor and the nearest trees are a long way off (perhaps 200 meters).

The other give away that its not "permanent shading" is that varying time at which it comes "off peg". On the existing graphs above two are around 0830 and two are around 0730. Another observation is that 9th and 10th Feb were cloud free until about 0930 yet the UV starts about 1 hr apart.

Even todays was early with a couple of spikes before 0730. In fact if it were not for the spikes I would have been happy with today's trend - much like yesterday where it was good except for the spike at the end.

Here is the screen grab of today to support my observations.

Feb11.gif

Cheers
Phil

PS My site is also on WU (IQUEENSL165) using a 5 minute interval and the trends look quite good in comparison to the ones above. So you are correct in suggesting the 1 minute interval effects the short term variation. But the zeros or the spikes early and late in the day should not happen IMO.

[aussiewmr]

Hi all,

My last screen shot (i promise ) unless something different and unusual happens.

Here you see, in relative terms, a huge spike in UV this morning. The Brisbane UV forecast for 0745 was approx 2 it actually got to approx 1.5. Also note the drop to zero following the spike.

I think its over to my supplier now.

Cheers

Phil

Feb12 Morning.gif

[archae86]

Here you see, in relative terms, a huge spike in UV this morning.

While I agree that event looks suspect, as do some other aspects of things you have posted, I'd caution others not to be surprised to see temporary elevations of both solar and UV above their "maximum" trend line in the presence of broken clouds. I've seen that frequently here in New Mexico, and think that there is a genuine mechanism in partial cloud cover in which the full direct sun is hitting the sensor unobstructed, but a fortunate configuration of clouds reflects more UV or solar onto the sensor in addition than is blocked in transmission of scattered light by other clouds. At the 5-minute sampling (I hope that means averaging) I use, I hazard a guess that I've seen points 20% above "possible" that way. I don't claim to have seen points so high as yours, and agree that for the sort of clouds implied by your curve a minute to minute drop to zero (really <0.5) seems unlikely.

[aussiewmr]

At the 5-minute sampling (I hope that means averaging) I use

I am not 100% sure, but I get the impression that the only data that is averaged over the data log interval in the VP2 (can comment for vue and VP1) is the temperature.
I say this because this is the only data type that has a tick box like "average temp over archive period" in Weatherlink setup.

With respect to UV and Solar (and others like humidity, wind speed) I believe that they are not averaged but admit I do not know that for sure.

I do understand your comment re the possible spiking of the UV and its important for other readers to know that this can happen. Its just in my case that there were no clouds at the time the spike occurred (there were some about 15 mins prior but they had departed the scene) - the solar trend shows that.

Its also interesting to note, from our government web site that does real time UV (and I have noticed this with my sensor too) is that UV levels can vary from data to day in clear sky conditions. They could be a max of 9 or 10 one day and another they can be 12 or 13. I guess combinations of things like moisture, dust, ozone thickness even all attribute to this.

Cheers

Phil

[aussiewmr]

Well something different did happen so I had this urge to post again

Today the UV seemed to follow a somewht attenuated profile in comparison to Solar. I can confirm that there were no clouds until about 0900 so the UV should have been higher than it was recorded at. Compare to earlier screen shots.

After midday, the UV leaped to where it should have been - almost double. Trend screen shot below - following is the Government measurement for comparison. The interesting thing here is the first half of the days UV seems to be the correct one where as the 2nd half is almost double what it should be. Most of the other days have been more like the 2nd half.

This is new behaviour. Replacement parts on the way so hopefully I can sort out where the issues lie.

Phil

Feb15.gif

Feb15 Govt UV.gif

[aussiewmr]

Well good news it seems.

The new UV sensor arrived last week and I installed it last weekend. Since then all has been well.

Below are some screen shots of trends, the first two, Mar1 and Mar2 are prior to the weekend (really good sunny days too as you can see from the solar trend) and they either are "spikey" or start late in the morning. You can see one starts at about 6:45am the other at 8:30am yet the light levels are much the same. Also note the end time of 1530 and 1630 Hrs. On the first screen shot note also the abrupt decline in UV despite the solar maintaining the expected trend. This was all pretty typical behavior.

In the last two screen shots, Mar8 and Mar9, you will see UV starting at about 0645 and stopping at 1715 hrs - much more realistic for Brisbane at this time of year. Also in the last screenie you can see it was a perfect sun day like the first but the UV was also perfect. All in all consistent expected response.

Problem solved I think, turns out to be a faulty UV sensor it seems - and full credit to the supplier (Archer trading) for sticking by me.

Cheers

Phil