Deconvolution of Spectra for Near-Infrared Astronomical Instruments

[Deckard]

Hi,

I am working on long slit spectroscopy for an astronomical instrument, and I think I need deconvolution, but I am not sure, so maybe you will be able to answer me :) Currently those are only simulations. I work in near-infrared (0.950 -> 2.2 micron) with a low resolution (R=30).

My problem is that the spectrum I am supposed to observe can be very steep (ie I can have a quick flux variation between $$\lambda$$ and $$\lambda+\delta\lambda$$), but when I watch them through my instrument, this spectrum is convolved with the slit LSF (line spread function), which is a sinc function ($$sinc(x)=sin(\pi x)/\pi x$$). Of course the width of the LSF varies acording to wavelength, and then is much larger at 0.95 micron than at 2.2 micron. And when I observe a steep change in the spectrum through the spectrograph, the change is much less steeper because it has been convolved with the LSF which is less steeper.

Are there some deconvolution methods that can correct, at least partialy what I describe?

I hpe what I say is making sense. If it's not I can post some picture to illustrate this. Thanks a lot in advance.

-- Arthur;

 

[leptonsoliton]

Deconvolution. With applications in spectroscopy, PA Jansson - New York: Academic Press, 1984, edited by Jansson, Peter A., 1984

... and thousands of other references. Use Google.

 

[sir-pinski]

Hi Arthur, thank you for reaching out with your question about deconvolution of spectra for near-infrared astronomical instruments. Deconvolution is a common technique used in spectroscopy to remove the effects of instrumental broadening, such as the slit LSF you mentioned, and reveal the intrinsic spectral features of the object being observed. It can also help improve the spectral resolution of the instrument.

There are several deconvolution methods that can be used, such as the Richardson-Lucy algorithm, maximum entropy method, and Fourier deconvolution. These methods work by mathematically reversing the convolution process, essentially "un-blurring" the spectrum. However, deconvolution can be a complex and challenging process, especially in the near-infrared where the signal-to-noise ratio is typically lower compared to visible wavelengths.

In your case, it may be helpful to first try simulating your data with a known spectral shape, and then applying different deconvolution methods to see which one gives the best results. Additionally, you may want to consider using a higher resolution instrument, if possible, to reduce the effects of the slit LSF on your observations.

I hope this information helps and good luck with your research!