WCS Distortion Conventions
snd22 wrote on Jan 26, 2010
Hello,
I have some distortion correction parameters for a particular instrument that are in the SIP format (stored in the headers as "A_m_n", etc.). I would like to convert these parameters to the TNX format (header parameters named "WAT*"). From what I understand, this involves some non-trivial recalculation. Do you know of any task that can do this automatically, or any simple way to do it?
Thanks for any help.</ul>
I have some distortion correction parameters for a particular instrument that are in the SIP format (stored in the headers as "A_m_n", etc.). I would like to convert these parameters to the TNX format (header parameters named "WAT*"). From what I understand, this involves some non-trivial recalculation. Do you know of any task that can do this automatically, or any simple way to do it?
Thanks for any help.</ul>
Francisco Valdes wrote on Jan 26, 2010
Hello,
The approach I can suggest is to use a file of (x,y,ra,dec) as the intermediate glue. By this I mean some tool dumps some fitting points based on the SIP WCS and then another tool fits the TNX WCS.
The pointers I can give to this are as follows. WCSTOOLS and DS9 I believe can provide a way to create RA/DEC at particular pixel locations. In IRAF the likely tools and packages are CCMAP and MSCTPEAK.
While you might think that you have to tie this to actual stars, this kind of transformation can just be done on a nice grid of points. Some of my WCS tools basically just do this. So the recipe is
1. Use something in WCSTOOLS to make a list of RA/DEC values at a grid of points across the image. Say 20x20 with endpoints at the first and last row/column.
2. Then these 400 values can be put into CCMAP with parameters saying to fit a TNX function of some order.
Later you could use something like MSCTPEAK to tweak a solution based on an actual list of star positions. But the the recipe above will preserve the basic accuracy of the SIP and TNX at the appropriate level of degrees of freedom as long as the grid is of sufficient size.
Yours,
Frank Valdes
The approach I can suggest is to use a file of (x,y,ra,dec) as the intermediate glue. By this I mean some tool dumps some fitting points based on the SIP WCS and then another tool fits the TNX WCS.
The pointers I can give to this are as follows. WCSTOOLS and DS9 I believe can provide a way to create RA/DEC at particular pixel locations. In IRAF the likely tools and packages are CCMAP and MSCTPEAK.
While you might think that you have to tie this to actual stars, this kind of transformation can just be done on a nice grid of points. Some of my WCS tools basically just do this. So the recipe is
1. Use something in WCSTOOLS to make a list of RA/DEC values at a grid of points across the image. Say 20x20 with endpoints at the first and last row/column.
2. Then these 400 values can be put into CCMAP with parameters saying to fit a TNX function of some order.
Later you could use something like MSCTPEAK to tweak a solution based on an actual list of star positions. But the the recipe above will preserve the basic accuracy of the SIP and TNX at the appropriate level of degrees of freedom as long as the grid is of sufficient size.
Yours,
Frank Valdes
Last post on Jan 26, 2010