Trace Straightening Tricks

The package allows the user to adjust the data image as he or she needs prior to profile fitting. For example, if the profile's trace is curved or changes quickly between rows, some sort of staightening may be used.

The polynomial and median fitting functions assume that the profile is slowly varying in terms of wavelength. If this is not the case a large number of pixels can be erroneously masked off during the profile fitting and optimal extraction.

The basic optimal spectrum extraction functions are designed to work without knowledge of what the adjustment functions do to the data. The package calls the function passed in through ADJFUNC with the data frames it needs. For convince, any options to the adjustment function are passed through ADJOPTIONS, which if more than one option is needed, can be a structure. Likewise, ADJPARMS, which contains the parameters of the data array can also be a structure. For simplicity, since not all adjustment functions need all of the information optimal extraction uses, that info is also passed through a structure.

Included in the optimal extraction package is ADJGAUSS, which uses Gaussian fitting to line up the centers and widths of the profile. To use this technique on the data, call optspecextr with the ADJFUNC parmater equal to ADJGAUSS. The options to adjgauss are held in the keyword ADJOPTIONS, which takes a structure. Some cases to use adjgauss include:

Correcting for Changing Profile Center or Width

Quickly varying center: If the center of the profile changes quickly with wavelength, it might be possible to line up the centers using adjgauss.pro. First set the ADJFUNC keyword to adjguass. Then pass in a structure with the tag CENTER = 1 to the ADJOPTIONS keyword:
	IDL> OptSpec = Optspecextr(Data, varim, rn, Q, x1, x2, adjfunc=adjgauss, $
		adjoptions = {center:1})
Quickly varying width: If the width of the profile changes quickly with wavelength then set the tag WIDTH = 1 in the ADJOPTIONS keyword.
	IDL> OptSpec = Optspecextr(Data, varim, rn, Q, x1, x2, adjfunc=adjgauss, $
		adjoptions = {width:1})
Slowly varying center: If the center does move between the top and bottom of the CCD, it might be best to fit a polynomial to the centers and used that as the estimate of shift. Set CENTERFIT = 1 in the ADJOPTIONS keyword to perform a polynomial fit to each row's shift amount. Set CENTERDEG to the appropriate polynomial fitting degree (usually 2-4) and CENTER tag to 1 
	IDL> OptSpec = Optspecextr(Data, varim, rn, Q, x1, x2, adjfunc=adjgauss, $
		adjoptions = {center:1, centerfit:1, centerdeg:2})

Writing Your Own Adjustment Function

If your data requires a different adjustment to the data before fitting the profile you can write your own function. The specs of your driver function:
OutArray = ADJFUNC(InArray, OptInfo, AdjParms=AdjParms, AdjOptions=AdjOptions, Revert=Revert)
OutArray: A data cube where each slice is a modifed image
InArray: A data cube where each slice is an image to be modified.
OptInfo: A structure with all the Optimal Extraction Package's information on the data array up to the begining of profile fitting
AdjParms: A varaible where internal data can be saved in order to be used when it is necessary to convert the fitted array back into the origional shape.
AdjOptions: A variable where options needed for your function can be passed through.
Revert: If set, InArray contains modified images and the function should revert them to the origional size and shape.

To use your function set the ADJFUNC keyword to the name of your function and the ADJOPTIONS keyword to any options your function accepts. Prior to profile fitting and after background fitting the optimal extraction package with create a data cube by stacking the data image, variance image, background image, and sky variance image. It will pass this data through InArray. It then expects to find in OutArray a stack of the data image, varaince image, background image, and sky variance image in the same order passed in. The user defined function does not need to worry about what is in the arrays, only that they are stacked through the Z dimension.

Any information about the data array is saved in OptInfo:
.Dataim: The data image
.Varim: The variance image
.Bgim: The fitted background image
.Skyvar: The sky variance image
.Q: The gain of the array
.V0: The readnoise of the array
.X1: The start pixel for the profile image
.X2: The end pixel for the profile image
.Spec: The standard extracted spectrum