# Description¶

## Assumptions¶

Beginning with the Build 5 pipeline, it is assumed that the input science exposure data from near-IR instruments have had the superbias subtraction applied, therefore the correction coefficients stored in the linearity reference files for those instruments must also have been derived from data that had the zero group subtracted.

It is also assumed that the saturation step has already been applied to the science data, so that saturation flags are set in the GROUPDQ array of the input science data.

## Algorithm¶

The linearity step applies the “classic” linearity correction adapted from the HST WFC3/IR linearity correction routine, correcting science data values for detector non-linearity. The correction is applied pixel-by-pixel, group-by-group, integration-by-integration within a science exposure. Pixels having at least one correction coefficient equal to NaN (not a number), or are flagged with “Linearity Correction not determined for pixel” (NO_LIN_CORR) in the PIXELDQ array will not have the linearity correction applied. Pixel values flagged as saturated in the GROUPDQ array for a given group will not have the linearity correction applied. All non-saturated groups for such a pixel will have the correction applied.

The correction is represented by an nth-order polynomial for each pixel in the detector, with n+1 arrays of coefficients read from the linearity reference file.

The algorithm for correcting the observed pixel value in each group of an integration is currently of the form

where is the observed counts (in DN), are the polynomial coefficients, and is the corrected counts. There is no limit to the order of the polynomial correction; all coefficients contained in the reference file will be applied.

The ERR array of the input science exposure is not modified.

The values from the linearity reference file DQ array are propagated into the PIXELDQ array of the input science exposure using a bitwise OR operation.

## Subarrays¶

This step handles input science exposures that were taken in subarray modes in a flexible way. If the reference data arrays are the same size as the science data, they will be applied directly. If there is a mismatch, the routine will extract a matching subarray from the reference file data arrays and apply them to the science data. Hence full-frame reference files can be used for both full-frame and subarray science exposures, or subarray-dependent reference files can be provided if necessary.