Public Lab Research note


Fraunhofer

by cfastie | March 03, 2013 04:22 03 Mar 04:22 | #6168 | #6168

Image above: A spectrogram of the sky taken today by Ebert showing six Fraunhofer lines. The solar absorption lines are recorded in the intensity graphs of the blue, green, and red channels from the digital image (colored respectively) and less obvious in the graph of average intensity (black). A standard diagram of Fraunhofer lines is at the bottom.

When matter gets hot enough it appears to us to glow, like a red hot poker just removed from a fire. The color of the glow depends on the temperature of the material -- first invisible to us, then red, then blue, then white hot. Regardless of the temperature and the material from which the object is made, the glow will have a continuous spectrum. That is, all wavelengths will radiate from the material, although some regions of the spectrum will be brighter than others. Temperature determines which regions are brighter and therefore what color the glow appears to our eyes. Examples of hot objects that have continuous spectra include the tungsten filament of ordinary incandescent or halogen light bulbs, and the dense, incandescent gas inside the Sun.

The spectrum of sunlight viewed from Earth is continuous except for many very thin dark lines. These result from gases near the surface of the Sun or in Earth's atmosphere absorbing light of particular wavelengths. These were first described by William Wollaston in 1802, but Joseph von Fraunhofer started studying them carefully in 1814, so they are now called Fraunhofer lines. Each line is caused by an element absorbing a very narrow band of wavelengths, and together provide solid information about the chemical composition of the Sun. Fraunhofer didn't know that the absorption was caused by different elements, and named the lines with arbitrary letters and numbers.

This afternoon I set Ebert to capture the highest resolution spectrogram it could, and pointed it out the window into a heavily overcast and snowy sky. This spectrum appears to record six of the stronger Fraunhofer lines. Although they are pretty easy to see in the spectral image, they are more subtle in the graph. The small dips in the intensity graph of the individual color channels record the lines better than the graph of the three-channel average.

I first calibrated the spectrogram using a spectrogram of a compact fluorescent lamp. That allowed me to identify the Fraunhofer lines, which have well known wavelengths. I then used the macro function in Spectral Workbench to calibrate the sky spectrogram more precisely. The Fraunhofer lines are now within two or three nm of where they should be, more precise than the standard hand drawn Fraunhofer diagram in the image above.

Here are the settings I used because I thought they would produce the sharpest spectrogram: Canon Powershot S95, ISO 80, F/5.6, shutter speed 2.5 seconds, focal length 9.64 mm (lens range is 6.0-22.5), focus distance 2 feet, slit width about 0.01 mm. Ebert has a 1000 lines/mm grating. The spectrum above can be seen here at Spectral Workbench.

I was waiting for blue sky to do this observation, but that is rare around here these days, so I will repeat this exercise when the weather clears. Maybe then Ebert will be able to see more than just a third of the lines that Fraunhofer saw 200 years ago.


I did this Help out by offering feedback! Browse other activities for "spectrometry"


People who did this (1)



4 Comments

You know what we need? vertical averaging to smooth out noise. In live/waterfall mode, this'd be time averaging. For your spec, it'd just be averaging in the y-dimension. I bet we'd get much smoother spectra in either case. Proof of concept -- try blurring your spectrum image vertically in Photoshop?

Is this a question? Click here to post it to the Questions page.

Reply to this comment...


That sort of worked: https://spectralworkbench.org/analyze/spectrum/4476. I cut out a 50 pixel tall horizontal slice of the Snowy Sky spectrogram and applied a Photoshop filter: Filter/Blur/Motion blur/angle=90/distance=50 pixels. The intensity graphs are somewhat smoother, which highlights the vertical pattern in the image. I assume the corduroy effect is some artifact of the lines on the diffraction grating. More here: https://publiclab.org/notes/cfastie/3-4-2013/noise

Reply to this comment...


Better than the vertical blur would be to simply resize the image to be 1 pixel tall (most resizing algorithms do this by averaging) and then again resize it to be a bit taller (to make the lines more apparent).

Reply to this comment...


That's very clever.

Reply to this comment...


Login to comment.