Using a Digital Camera to Measure Skyglow

[deleted-353326]

Hi - I have a couple of questions about the calibration stage and the comparison to calibration as well as a question about comparing the colour channels.

I understand that we take the pictures of the white in order to get a baseline with which to compare skyglow PVs to because the PV is not a linear function at the low and high ends. I don't understand why it is necessary to do this at the high ends if we are taking pictures in the dark? None of my 30 second exposures at ISO 200 or f2.8 got anywhere close to saturated or even beyond 125. Also, I took my calibration pictures on a sunny day in the shade as suggested, and my average pixel value for 1/1000th of a second was 121 and was saturated (255) at all speeds from 1/30th to 30 seconds. The graph in the method suggests it would be about 25 so there's quite a big difference. (I also note that the graph in the handout goes down to 1/8000th which I also needed to do and not stop at 1/1000th).

Of course if this was done in darker conditions then this would result in different results, but given the method suggests in the shade on a sunny day I am surprised my results vary so much. Given (if I understand correctly) that this part of the experiment is to understand how the PV varies by exposure length, why is it necessary to do in daylight at all? Couldn't this be done in artificial light starting at the fastest shutter speed and working up to 255?

Onto the part where the skyglow photos are compared with the calibration curve to calculate an equivalent exposure time - if all the PVs for skyglow pictures fall within the linear part of the curve (lets say between 50 and 120) - can we not just fit a line of best fit for the calibration values and calculate an EET exactly from this gradient function?

Finally, I will be comparing the luminescence of the RGB colour channels to compare and see if they drop off separately the same amount as the RGB combined - I presume I can just do calibration curves for each of these channels separately from the same original calibration photographs (I don't know if/how separate channels behave) and graph the results of EET for each channel.

Many thanks

Andrew

[norman40]

Hello Andrew,

A good calibration should include the range of data anticipated from the experimental results. And I think the calibration procedure for this project was designed to accommodate a very wide range of night sky light levels. The pixel values for your night sky photos didn’t approach saturation. But higher pixel values could be found for photos taken in areas with high levels of light pollution.

The differences between your calibration results and the example given in the project procedure could be due to any of several factors. Differences in lighting conditions, the reflectance of the paper being photographed, or camera sensor sensitivity could result in different pixel values at the same camera settings. The difficulties with controlling all the factors that influence pixel values make it necessary to include the calibration experiment in the project. And you get the opportunity to understand how exposure time is related to pixel values. As you point out, the calibration could be done with artificial light as long as exposures up to a pixel value of 255 are included.

It should be possible use a “best fit” line from your calibration data to calculate your equivalent exposure time values. You might be able to find a non-linear equation that will fit the entire data set (or most of it). Fitting a subset of the calibration data as you suggest should work too.

I hope this helps. Please ask again if you have more questions.

A. Norman

[deleted-353326]

Thanks for your helpful response.