Extraction of Arcturus/Solar/Transmission Spectrum

Extraction of Arcturus/Solar/Transmission Spectrum

Note: Wavelengths are entered in microns, wavenumbers are entered in inverse cm just below on the form depending upon which option you select.

Start Wavelength/Wavenumber: End Wavelength/Wavenumber:


Output format?




Resolution? (plots only)

Brief Description of The Form:

The form above allows one to plot the data from Ken Hinkle's optical/IR spectral atlases over a specified range between 0.3727 microns and 5.359 microns (although there gaps in the near-infrared where the atmosphere blocks all radiation). Output can be a raw text dump of the spectrum values or as a plot in gif/jpeg/postscript/xmgr format as selected.

One can plot the "raw" arcturus spectrum (with atmospheric lines), the solar spectrum (optical range only), the atmospheric tranmittance, or the corrected arcturus spectrum. These are the "Spectrum" options above. For the infrared, wavelength 0.93 microns or more, there is no solar spectrum available here. In that case one can plot either the "summer" or "winter" Arcturus spectrum.

Four resolutions are available: full resolution, and the GNIRS resolutions of 2000, 6000, and 18000. The raw text case always returns full resolution information. For the other cases, the spectrum is convolved with a gaussian if lower resolution is requested.

There are regions in the original IR data with no value (values of 1.0 or -0.05) which are places of heavy atmospheric absorption. When the plots are smoothed these regions tend to have gaussian or triangular edges which of course as not real. One should take care in the regions adjacent to such parts of the spectrum. Occasional data points of more than 1.0 are also in the original data, and these may cause peaks in the smoothed data which also are not real. So some care needs to be taken in interpreting the plots for regions of strong atmospheric lines.

You can request that line be displayed on the plot--or prefixed to the text file in the "raw text" option. If so vertical lines are shown on the plots marking spectral lines contained in a few line ID files. This line list is not exhaustive. The line ID is written above each line marker in the plots.

Depending on your browser, the plot may initially appear below the visible region in the browser window. If after requesting a plot you see only a white screen, try to scroll down.

For those who use xmgr (specificly version 4.1.2) it is possible to download the plot as an xmgr save file, which is what is used internally to generate the plot and put on the labels. If you choose that option, the resulting text file can be downloaded and can then be saved for direct examination.


If you request raw text and wavelength values the data values that are returned will be in descending wavelength order in the IR, beyond 9300 Angstroms = 0.93 microns, since the driving routine reads the values in wavenumbers from the original Arcturus IR spectral atlas files. Similarly the optical spectrum is originally in Angstroms and so if you request part of the optical spectrum in raw text as wavenumbers the values will be in decreasing order. The various plots are not affected by whether the points are in ascending or descending order.

It may be possible to "overload" the program with too many lines or too many points, if you insist on asking for a high resolution plot of a large wavelength range. If you ask for a large range and get an error message back about the server having an internal error this is probably the reason. It make no sense to try to plot too large a wavelength range with lines marked and at full resolution. You will not see anything in the resulting plot anyway.

In the R=2000 plots in particular the edges of the individual sections are not handled properly by the gaussian convolution. These parts of the convolved spectrum are excluded. When the wavelength interval requested is small, you can see these gaps in the plot.

Another defect of the automatic plots is that the edges of the individual sections may go to zero when they should not. This is due to NOT trying to merge the original sections before doing the convolution. It is best, if the plot is not in quite the right form, to get the actual values as raw text and plot them oneself. This page uses only a rather crude method to automaticly generate the plots from the data.

This page was last modified April 1, 2005.

Kevin Volk (kvolk@gemini.edu)