Solar Radiation, Illuminance, and UV Index, is there any difference?

OK, I can scale to those numbers, but here’s the three readings for the last day scaled to 1000 W/M^2, 120,000 Lux, and UV Index of 8:


I can see slight differences between the two light levels and the UV index, but there’s no apparent different between the two light levels…

I’ve wondered the same thing. Is there only one sensor and two of the values are merely calculated from one? That is what I think @wpns is asking, too. If there is only one sensor value then it doesn’t make sense to store three values when the other two can just be recalculated if they are needed.

My guess is yes. Probably one physical sensor that is reported with 3 different scales. But that’s just a guess. Good question for @WFmarketing

I got the impression that there are two sensors, but I can’t find where I read it.
However in their page with formulas, there is no formula for any of the three, which might suggest three sensors. But my assumption, which matches your observation, is that the Solar Radiation and brightness are measured by the same sensor, and the page of formulas, is missing the conversion formula.

btw uv can be way higher then 12 (depending on your location. it never happens in the netherlands, but in australia the value goes at least to 14)


interesting test, to convince yourself could be to hold a piece of thick clear glass against the senor. It shouldn’t reduce the brightness much, but should show significant less uv.

good sunglasses might also work, they do reduce the light, but should reduce the uv relatively more.

Or, one could simply ask David.

There are actually two sensors, each tuned to a different part of the light spectrum, in the part we use to measure sunlight (the APDS-9200 from Broadcom). The APDS-9200 reports “lux” (aka “brightness”) and “uv index” directly (which must be calibrated), and we estimate solar radiation from the brightness value assuming it’s light from the sun (ie., we know the makeup of the full spectrum, more of less, even though we’re only measuring radiation from a small portion of the spectrum).

It’s not on the page because it is not “derived” in the same sense that the other quantities on that page are derived. That is, the variables shown on that page are things the smart weather system provides (via the app or the API), but the sensor devices (AIR or SKY) do not provide directly. However, based on the information in this forum through, you can probably figure out the formula yourself. “The proof is left as an exercise to the reader,” as they say… :smile:

Note: While SR is directly related to LUX, each parameter can be individually calibrated, so the “formula” for calculating one from one to the other is not necessarily constant (though in practice it usually is).


In this case, is Solar Radiation a derived value?

Based on the graphs above:
sr = illumination / 120 .
looks like a very, very easy formula, but alas, in my opinion it is a derived value and kind of should be in the list of such formulas.
But apparently this 120 could change tomorrow to 119, if it were any better, so it isn’t a physics based derivation, but a pratical one. I can understand david’s reasoning.

1 Like

Say whaaaaaaaaa ???

That’s an unusually non-answer answer.

Can we rewind a little and see if we deciphered your explanation correctly ?

  • there are two sensors tuned to different parts of the spectrum. Are both APDS-9200 ? Unclear from your wording
  • is lux calibrated (?) - Unclear based on your wording
  • uv index ‘is’ calibrated based on what ? Lat/lon/elevation ? Surrounding stations ?
  • solar radiation is estimated but you’re not saying how, using which values, just that magic occurs. Is it estimated based on direct values from one or more sensors, before or after calibration ? Or phase of the moon or stock price or approximate airspeed velocity of an unladen swallow ???
1 Like

The APDS-9200 has two internal sensors.

Lux is measured by one of the sensors

UV is measured by the other sensor.

By-the-way, I understand David’s answer. He gave several hints.


Here is a hint.

UV is measured at ~320nm.
Illumination is measure at ~ 560nm.

Per the spec sheet, the sensor is packaged with a light filter that emulates the human eye spectral response (V-Lambda characteristics). My head started hurting before I made it through the second paragraph of this Wikipedia article


Thank you for those links. Very informative.

Two answers:

  1. No, Solar Radiation is not a derived value in the way all other values on our “Derived Metrics” page are. The equations and relations on that page only apply to values that are calculated on our back-end or in our apps. They do not apply to any values produced by the sensor hardware itself.

  2. Yes, Solar Radiation is a derived value in the same sense that ALL values from the sensor hardware are derived. For example, the relative humidity sensor doesn’t measure humidity directly - it measures capacitance and uses that value along with others to estimate humidity. And the wind sensor doesn’t measure wind speed and direction directly, it measures the time it takes for sounds to travel between the ultrasonic transducers and converts those values into speed an direction.

Let me restate the question. How is Solar Radiation derived from the Broadcom sensor?

1 Like

@garyfunk there are graphs in the docs of the sensor that tell you how brightness is measured in the sensor but because in the case of weatherflow the light passes through a diffuse window, calibration is needed.
The only relevant question could be how brightness and solar radiation are related, but that was left as an exercise to the reader. If the above graphs of @wpns are correct, it is just a ratio of 1000 / 120.000. As David mentioned that ratio is fixed in practice, but doesn’t need to be.

I’m a little puzzled. In reply #13, it states UV is measured at 320 nm, which is in the UVA spectrum and, I believe, has relatively little physiological effect. states that the erythemal (sunburn) effect of UV at ± 325 nm is 0.3% as damaging as at 295 nm. I understand (possibly incorrectly) that the Diffey-weighted UV irradiance , or DUV, integrates the erythemal effect over the UVA and, essentially, the UVB parts of the spectrum. I suggest that expressing only the measured UVA values as a numeric index (eg 0-20) may be a cause for confusion because the erythemal effect may be totally different from the value shown in a similar index. In practical terms, if WF shows UV as 8, does the layman plaster him/herself with cream, even though the sunburn effect may be much lower due to high absorption of UVB in the ozone layer?

As I understand it, as a layman, the effective UVA/B is a function of the ‘ozone layer’ density and is measured in Dobson Units which is the thickness of the stratospheric ozone expressed in mm, when it is compressed down to Normal Pressure.

Obviously, it is ridiculous to measure both UVA and UVB and translate them to ‘sunburn effect’ in an economical device such as WF, but can it be made clearer that the UV measured is just an arbitrary value, not necessarily related to sunburn risk, please?

1 Like

the docs of the sensor tells us that it measures both uva and uvb. it returns values on a scale from 0 to 11. It has a graph in figure 3 with the uv response. It has a linear scale. This looks close to the blue line in the second figure of the wikipedia article you referred to (note that that graph is on a log scale, so the values on the tail to the right are more then 10 times lower then the maximum, which matches the graph in the docs of the sensor).
So I think the intention is to return a value that IS related to sunburn and the UV index you see in other weather reports.

Thank you for your reply. My intervention to this thread is mainly because, in practice, I’ve not been able to correlate the WF UV figures with the official ones. I hope the following figs will give you an idea of my concern.
Although it’s not very clear, the deep violet line at the bottom shows the measured UV in WF, showing a UV index peak of 5.3 at 11:37 and generally following the solar light curve (orange) including the peaks and valleys.
This shows the official measured UV covering the same period with an index peak of 3.9. You may wonder why the curve is relatively smooth, compared with WF; this is because light cloud, visible in the WD graph, is transparent to UVB, less so to UVA and even less to visible. The intermittency of visible light in my graph is reflected in the UV curve of WF, whereas there is relatively little intermittency in the official UV one, related to visible light.

I’ll continue to study this and I’ll try to obtain a better UV/solar graphs to explain my concerns better.