bjweiner / mmt Goto Github PK

Need better grey scale stretch in 2d display, especially when there are a few pixels with extreme values.

The templates included in Specpro are lower resolution than Binospec data. Better results could be obtained with templates that have higher resolution and cover a wider wavelength range. SDSS composite spectra or PEGASE models are possibilities.

Redshift fitting by fitting a linear combination of 2 or 3 templates (eg as done by DEEP2) would likely yield better results and perhaps allow the fitting to be automatic rather than making the user choose a template from the menu. This is a more substantial code architecture change than simply adding / replacing templates.

Redshift fitting often fails on Binospec data because of poorly fluxed data at the ends of the spectrum. The redshift fitting works if the 1-d is zoomed in a little to clip off the ends, but the ends/bad data should be clipped or ignored by default. I tried an initial stab at this and couldn't get it to work.

Steve Willner pointed out that the conversion of wavelength from the Binospec format spec1d and spec2d files with wavelength CRVAL, etc in the headers, to specpro format with an array of wavelength, may be off by 1 pixel because the first pixel is 1-based in FITS. The problem lines would be addressed by this change:

; lambda = (dindgen(npix) - crpix1) * cdelt1 + crval1
lambda = (dindgen(npix)+1 - crpix1) * cdelt1 + crval1
; lambda1 = (dindgen(nx) - crpix1) * cdelt1 + crval1
lambda1 = (dindgen(nx)+1 - crpix1) * cdelt1 + crval1

Need to compare a feature in original and converted spectrum to verify the wavelength conversion.

A to-do item in specpro_bino.pro was to check air vs vacuum wavelengths. Steve Willner says that the output of the Binospec pipeline is in vacuum. The linelists such as emlines.dat are in air. The template spectra used appear to be in air according to Steve and Sean. Air vs vacuum wavelengths will cause a shift of about 2.7e-4 on average (~81 km/sec).

The DEIMOS pipeline that Specpro was written for appears to have written output files in air but converted to vacuum internally when redshift fitting.

The simplest fix is likely to convert either the spectrum to air, or templates to vacuum, before redshift fitting, using vactoair.pro or airtovac.pro. I think converting the spectrum to air is easier but converting the templates to vacuum is more technically correct (since the conversion is not exactly linear, and the redshift occurs in vacuum). The difference is likely undetectable, though.