Page 1 of 1

UV sensor sensitivity

Posted: Tue 26 Aug 2014 4:35 pm
by Tony_J
I have been observing the behaviour of this sensor since attaching it to the system, and it seems to be very sensitive to the angle of the light impinging on the sensor. For example, today has been bright and clear all day with the UV index getting up to 3 at mid-day, but the fall-off before 10:00 AM and after 4:00 PM is dramatic. Current reading is zero despite blazing sunshine outside. I suspect that at least part of the problem is that the UV-sensitive cell on top of the unit has a limited angle of acceptance (viewing angle) and therefore reads artificially low when the angle of the sun is outside the viewing angle. Yes, it is also a question of how much atmosphere is between the sun and the sensor, but I don't think that is the whole story.

So, it occurs to me that it might be worth experimenting with a crude "fisheye" lens on top of the UV cell - for example a small hemispherical glass/plastic lens - that would tend to increase the angle of view. Has anyone attempted to do this? With what result?

Re: UV sensor sensitivity

Posted: Wed 27 Aug 2014 9:58 pm
by AllyCat
Hi Tony,

Sorry, I don't have any experience of the WMR88, however I have examined a data sheet, for the (very expensive) Davis UV sensor (6490), which is readily available on the web.

But firstly, how do you know that your readings are incorrect? If the UV level simply correlated with the subjective "visible" light level (Lux or watts/m2) then there would be no point in using a separate sensor.

Secondly, the Davis sensor has a "Cosine" response which means that the sensitivity is supposed to reduce at lower elevations, i.e. to half at 30 degrees and to zero at the horizon.

Thirdly, how do you propose to determine if your additional "lens" is transparent to UV? In general, glass is opaque to the UV wavelengths specified for the "UV Index".

But, as I have said several times on this forum, I must admit that I consider the "amateur" measurement of the UV index to be completely pointless, with respect to either meteorology or skin protection. For what purpose do you intend to use the data ? Also, does the WMR88 report only integer UV values or decimals as well?

Cheers, Alan.

Re: UV sensor sensitivity

Posted: Thu 28 Aug 2014 2:51 pm
by geoffp
Very interesting Alan, I am awaiting delivery of my UV sensor for the WMR88, it will be interesting to see what my readings come out at :groan:

Re: UV sensor sensitivity

Posted: Thu 28 Aug 2014 10:28 pm
by Tony_J
AllyCat wrote:Hi Tony,

Sorry, I don't have any experience of the WMR88, however I have examined a data sheet, for the (very expensive) Davis UV sensor (6490), which is readily available on the web.

But firstly, how do you know that your readings are incorrect? If the UV level simply correlated with the subjective "visible" light level (Lux or watts/m2) then there would be no point in using a separate sensor.

Secondly, the Davis sensor has a "Cosine" response which means that the sensitivity is supposed to reduce at lower elevations, i.e. to half at 30 degrees and to zero at the horizon.

Thirdly, how do you propose to determine if your additional "lens" is transparent to UV? In general, glass is opaque to the UV wavelengths specified for the "UV Index".

But, as I have said several times on this forum, I must admit that I consider the "amateur" measurement of the UV index to be completely pointless, with respect to either meteorology or skin protection. For what purpose do you intend to use the data ? Also, does the WMR88 report only integer UV values or decimals as well?

Cheers, Alan.
Hi Alan -

I think you have the cosine response thing bass ackwards - see the explanation of cosine response here: http://www.apogeeinstruments.co.uk/apog ... formation/ In other words, they correct for the fact that at low incident light angles, the reading would otherwise be artificially low. That is exactly what I suspect is happening with the Oregon sensor and it isn't being (cosine) corrected.

Yes, you are right, I have no way of telling whether or not the Oregon sensor is accurate; however, I seriously doubt that on an otherwise clear day that UV index levels only register above zero between 10 AM and 4PM. I guess the only way to nail it would be to compare with the readings given by a known "good" UV sensor. The Oregon system only reports integer values of UV index.

Fair point about the UV transparency of the lens; I will have to look for one that is appropriately transparent.

What am I going to use the data for? Nothing other than personal interest.

Re: UV sensor sensitivity

Posted: Fri 29 Aug 2014 9:51 am
by mcrossley
Tony_J wrote:I think you have the cosine response thing bass ackwards - see the explanation of cosine response here: http://www.apogeeinstruments.co.uk/apog ... formation/ In other words, they correct for the fact that at low incident light angles, the reading would otherwise be artificially low.
What that site is saying is that they have corrected for reflectivity at low angles of light incidence to give a 'truer' cosine response, not that they correct by applying an inverse cosine correction to get a more linear response.

Re: UV sensor sensitivity

Posted: Fri 29 Aug 2014 9:56 am
by Tony_J
mcrossley wrote:
Tony_J wrote:I think you have the cosine response thing bass ackwards - see the explanation of cosine response here: http://www.apogeeinstruments.co.uk/apog ... formation/ In other words, they correct for the fact that at low incident light angles, the reading would otherwise be artificially low.
What that site is saying is that they have corrected for reflectivity at low angles of light incidence to give a 'truer' cosine response, not that they correct by applying an inverse cosine correction to get a more linear response.
OK - thanks.

Re: UV sensor sensitivity

Posted: Fri 29 Aug 2014 10:44 am
by Tony_J
There's an interesting post here (second one, from Nico) http://www.weather-watch.com/smf/index. ... ic=51393.0 that suggests the Oregon sensor is rather less than optimal. The response curve that he gives seems similar to what I am seeing on my system. How to correct its behaviour to get the "proper" response is another question...

Re: UV sensor sensitivity

Posted: Fri 29 Aug 2014 4:48 pm
by AllyCat
Hi,

The "UV Index" has a specific definition which gives a very heavy bias (around 100:1) towards the shorter UV (i.e. UVB/UVC) wavelengths. The apogee sensor does not incorporate such a filter (nor does it claim to be a sensor for the UV Index). IMHO the three Davis documents give an excellent background to (their) UV measurement sensors. In particular the "Interpreting UV Measurements" application note explains why even their ($350+) sensor may produce "misleading" results !

Personally, I'm doubtful if the cheaper UV sensors even attempt to achieve the "Erythemal Action Spectrum (EAS)" response of the "true" UV Index; I'm fairly sure that the Fine Offset (WH-308x) Solar Pod doesn't. But equally, I'm quite surprised that the UV Index specifies a Cosine response: That's fine for sunbathing (laying on the ground), but if, for example, you're walking through a beach, water or snow and wearing a sunhat (as I'm much more likley to be doing), then a Cosine response may give a seriously low estimate of the (reflected) UV level (and the potential for skin damage).

However, maybe the WMR88 sensor is indeed "sharper" than a cosine response (as have many LEDs and phototransistors, etc.). In this context, it's interesting that the Fine Offset "Lux" (daylight) sensor is mounted under a "white dome " (diffuser), yet its overall response seems to match quite closely with the theoretical "Max Solar Rad" graph created by Cumulus (which assumes a cosine response). So perhaps something like "half a ping-pong ball" could improve the WMR88 response shape. It will probably attenuate the overall level (which could be re-calibrated within Cumulus) but it's difficult to predict which materials are acceptably transparent to UV (B&C).

Finally, did I see somewhere that Oregon don't offer a sensor for the "visible spectrum" (watts/m2), but do for UV? If so, that sounds to me rather like a "cop out". A basic watts/m2 sensor is much easier and cheaper to implement than one for UV, but the "dynamic range" is quite large, making the data management and reporting a little tricky. The UV Index uses a very simple linear scale from 0 to 10+, and it's virtually impossible to easily validate (or disprove) its accuracy. ;)

Cheers, Alan.

Re: UV sensor sensitivity

Posted: Tue 02 Sep 2014 9:35 am
by Tony_J
AllyCat wrote:Hi,

The "UV Index" has a specific definition which gives a very heavy bias (around 100:1) towards the shorter UV (i.e. UVB/UVC) wavelengths. The apogee sensor does not incorporate such a filter (nor does it claim to be a sensor for the UV Index). IMHO the three Davis documents give an excellent background to (their) UV measurement sensors. In particular the "Interpreting UV Measurements" application note explains why even their ($350+) sensor may produce "misleading" results !

Personally, I'm doubtful if the cheaper UV sensors even attempt to achieve the "Erythemal Action Spectrum (EAS)" response of the "true" UV Index; I'm fairly sure that the Fine Offset (WH-308x) Solar Pod doesn't. But equally, I'm quite surprised that the UV Index specifies a Cosine response: That's fine for sunbathing (laying on the ground), but if, for example, you're walking through a beach, water or snow and wearing a sunhat (as I'm much more likley to be doing), then a Cosine response may give a seriously low estimate of the (reflected) UV level (and the potential for skin damage).

However, maybe the WMR88 sensor is indeed "sharper" than a cosine response (as have many LEDs and phototransistors, etc.). In this context, it's interesting that the Fine Offset "Lux" (daylight) sensor is mounted under a "white dome " (diffuser), yet its overall response seems to match quite closely with the theoretical "Max Solar Rad" graph created by Cumulus (which assumes a cosine response). So perhaps something like "half a ping-pong ball" could improve the WMR88 response shape. It will probably attenuate the overall level (which could be re-calibrated within Cumulus) but it's difficult to predict which materials are acceptably transparent to UV (B&C).

Finally, did I see somewhere that Oregon don't offer a sensor for the "visible spectrum" (watts/m2), but do for UV? If so, that sounds to me rather like a "cop out". A basic watts/m2 sensor is much easier and cheaper to implement than one for UV, but the "dynamic range" is quite large, making the data management and reporting a little tricky. The UV Index uses a very simple linear scale from 0 to 10+, and it's virtually impossible to easily validate (or disprove) its accuracy. ;)

Cheers, Alan.
As you say, UV transparency of materials is difficult to predict - I will do some experimentation - I have a few ping pong balls somewhere :)

I have a small (10mm diam) plano-convex lens that has about a 14mm focal length which may be worth a try too.

You are correct - Oregon don't do a watts/m2 sensor. They sell a small solar panel add-on for their wind sensor which would be a great starting point for one if they felt so moved. As you say, seems to be a bit of a cop out on their part.

Regards,
Tony

Re: UV sensor sensitivity

Posted: Mon 10 Oct 2016 11:01 am
by soundhill1
I am interested in health effects of vitamin D and looking for hours of sunlight times UV index that hour to get a sort of dose. That is for various NZ cities including Hamilton.