Potential Problems and suggested solutions

Most problems will appear as insufficient, poor or simply incorrect corrections and unfortunately this can have many reasons, each of which has it's own set of solutions.

Window size

The 2D correlation is very sensitive to noise. This means that we want the iage size used for correlation to be as small as possible while still containing the main galaxy structure.

In the case of large offsets one may need to apply manual offsets in order to bring the images einto approximate alignment before running the code. This is described here.

Normalization

Before the crosscorrelation is done the images need to be normalized. This can be done in several different ways and certain datasets are more or less suitable for a given method, depending on the pixel-statistics of a given individual image.

CR Rejection

If there are cosmic rays present in an image this will naturally make the cross correlation difficult. This is not an issue for ACS/SBC since the mama detector does not show cosmic rays.

CRs need to be masked separately before crosscorrelation and making sure that they are masked properly is crucial. We recommend using astroscrappy.

However, when the drizzling is done in CROCOA it is done with a very small pixel scale which can lead to issues where some final pixels only have contributions from masked pixels.