In a paper titled "Deadly Ultraviolet UV-C and UV-B Penetration to Earth’s Surface:" J. Marvin Herndon, Raymond D. Hoisington, and Mark Whiteside write:
They get values for UV-C (their readings are shown in the graph above in red and black) that are much higher than found in space (the green line). The graph above is a comparison they make in their paper. I've corrected for the rather misleading log scale they used:
Their readings obviously go wrong below 280 nm.
Herndon's answer to this problem is that his cheap device is right and NASA's satellite measurements for the last couple of decades are wrong.
I asked the manufacturers of the instrumente used, International Light Technologies, about this:
So the rise in UV at the bottom of the range is due to the multiplication of noise from stray sunlight multiplied by a function that ramps up for low values of UV.
i.e. it's a graph of the correction function made visible by not filtering out all the other light.
Follow up from ILT:
ILT say the only way to measure UV-C with this instrument is monochromator. This is a type of specialized filter that blocks out all stray light outside of one narrowly defined part of the spectrum. This was not done by Herndon, et. al, which is why they have the peak in UV-C. It's simply the calibration curve added in by the software. No UV-C was detected.
It is unfortunate that the authors of the paper were seemingly unfamiliar with the problems of stray light in measuring UV-C. This is actually a well know problem in the field, and was in fact raised as an issue by of the peer-reviewers of the paper:
The problem is well known to the manufacturers and experienced users of these instruments. Here's discussion from Instrument Systems, a division of Konica Minolta:
They include this explanatory chart:
Notice the most accurate readings (green) with the scanning double monochromator (as suggested by ILT) are around 0.001. Basically zero.
Herndon, et. al, do mention "stray light"
What they fail to realize is that 0.3% of very bright sunlight is still a lot of stray light falling on the UV-C portion of the sensor.
[Note: this is a summary post of the thread below, original discussion follows]