Convolution

rrdecarvalho2009 wrote on Jan 15, 2009

Does anyone know how to convolve (in iraf) a spectrum with a gaussian where the FWHM varies with lambda ?

Reinaldo

---

Mike Fitzpatrick wrote on Jan 15, 2009

If I understand your question, then no, I can't think of any task that does specifically that. The FCONVOLVE task in the STSDAS.FOURIER package will allow you to convolve two images together. If you're asking because the spectra have a non-linear dispersion then you might look at REBIN or SCOPY as a way to linearize the dispersion (probably a log-linear is what you want) to see if a fixed-width gaussian convolution could then be used.

Please post more info about why you need this if you still have questions, although I'll probably punt to Frank for the expert answer at this point.

Cheers,
-Mike

---

rrdecarvalho2009 wrote on Jan 15, 2009

Hi Mke,

Thanks for replying!

The problem is that I am processing SDSS spectra and they provide also a resolution versus lambda map and the difference between the blue and red
channels are significant. Since they provide the mapping we know exactly
how FWHM varies with lambda. Now, I am comparing stellar population models
and observed spectra. I am not touching the data, I want to bring the models
to the same resolution x lambda mapping of the data. That's why I was thinking
of convolving the models with a gaussian but where FWHM varies with lambda.

Cheers,

Reinaldo

---

Francisco Valdes wrote on Jan 15, 2009

I'm sorry but I'm not aware of any task to do this. I think you will need to write your own program. To read the spectra you could use imfort (convert first to imh) or cfitsio. However, you could dump the spectrum to a file with wspectext and simply read in the text.

Yours,
Frank Valdes

---

Last post on Jan 15, 2009