Realtime Aurora Spectrometer Viby

About this spectrometer

Technical implementation:

The spectrometer design is inspired by Michael Theusner's aurora monitor based in Bremerhaven, Germany (52° North). The optical system is based on a blaze transmission grating. The 300 grooves/mm grating is illuminated by a single slit. It is observed by a ZWO ASI120MM CMOS camera with a 16mm f/1.2 CS-mount lens. The spectrometer has been calibrated using several low-power lasers, and a daytime/twilight Fraunhofer-calibration is in preparation.

Data acquisition and analysis is performed by a Raspberry Pi 3 Model B+ running Raspbian Stretch. The acquisition code is based on Thomas Jacquin's Wireless All Sky Camera project. The analysis code is self-developed and based on Mark Kness' Colorpy package. A night's worth of data amounts to 3.8 GiB (can be compressed losslessly to below 1 GiB), which are reduced on-line to around 120 MiB of spectral plots and 300 kiB of analytic spectral data.

Here is a raw picture obtained by the spectrometer, taken with low ambient temperature, clear skies and aurora present. See some explanations of the individual effects:

The spectrometer takes regular exposures of 45 seconds duration. Those are analyzed and evaluated individually online, integrating over the full inclination range. Later, offline analyses stack frames to exposures of 3 minutes. The spectra are also analyzed differentially in inclination to produce keograms.

These are two typical spectra, the upper one with the presence of aurora borealis shows a clear line at 557.7nm as well as at 630nm, 634nm, 391nm and 427nm. In addition, some H-alpha light at 656.3nm is present. The spectrum below has been obtained some time after the aurora faded away. A faint remainder of the 557.7nm line is still recognizable. Now the spectrum is dominated by artifical street lights and their sodium and mercury lines.

For convenience, the spectrometer is not regularly operated outdoors. In fact I let it obtain spectra from behind a tripe-glass insulating window. The plot below is taken through the window, and one can basically not see dramatic changes as compared to an outside operation.

When operating inside, the typical CMOS sensor temperature varies between 21°C and 24°C. A visible effect is the higher noise in the spectrum and a widening of the spectral lines; however, the ability to spot aurora is just marginally degraded.

Pictures and some first results:

The spectrometer saw its first aurora on 2017 September 22: Here is the first aurora borelis spectrum recorded with this device, the spectrum of a fluorescent lamp with its many spectral lines, and here is a picture of the prototype housing built in summer 2017. To calibrate the spectrometer, laser light at wavelenthgs of 405nm, 455nm, 532nm and 656nm has been used (Recently, also a "Na-Hg" calibration using artificial streetlight lines at 435nm and 819nm has been tried, as well as a 395-nm UV LED). The line at 810nm is the second order light of the 405nm laser:

Since October 2017 the spectrometer has been operated on regular basis. During an aurora display on the evening of 2017 December 4, some results beyond the online plots were produced, in particular a timelapse of the 2017 December 04 evening with aurora borealis and a keogram of that night. Note how aurora is setting in and gaining intensity to later fade off again. Also note that the ambient temperature of the spectrometer reduced from about 22°C to about 0°C, and how the noise in the overall spectrum decreases. However, the spectrometer delivers results in the whole temperature range, it can ever work from behind a closed window.

Keogram for the night of 2017-12-04, with aurora present from 19:00 until midnight:

Some pictures of the hardware setup:

The spectrometer seen from front. One can clearly see the entrance slit, which is bent at a radius of 12cm, so as to be focused on the CMOS. The entrance slit is about 4cm high.

Spectrometer observing the northern sky through the window. The device is supported such that it observers the sky between 20° and 35° inclination at roughly northern direction.

Robert Wagner, initial release: 2017-10-06, latest release: 2017-12-08. Imprint & Copyright