STScI Logo

Calibration Database System

WFC3 Reference Files

Reference files used by the WFC3 calibration pipeline (calwf3 and PyDrizzle/MultiDrizzle) are updated on a regular basis. Current reference files files can be retrieved from the HST archive, or downloaded directly from the iref directory.



WFC3 Bias Images (BIASFILE)

Additive stationary pattern in the electronic zero point of the CCDs.

While the majority of the overall electronic zero point is removed via the overscan correction, the two-dimensional structure of the electronic bias is removed from science images by subtracting the bias reference image. For UVIS science data, this subtraction is done before the data are combined for cosmic-ray rejection.

The bias reference images have exposure times of zero, and are full-frame images, complete with serial and parallel overscan regions. The bias image size varies depending on the on-chip binning mode:

1 x 1 binning: 2070 x 4206 pixel image
2 x 2 binning: 1035 x 2102 pixel image
3 x 3 binning: 690 x 1402 pixel image

Bias reference images for each of the 2 CCDs that make up the full detector array are stored separately in 2 imsets in the bias file, like the UVIS science data (see Section 2.2 of the WFC3 Data Handbook). The bias reference file is in units of DN, i.e., it has not had the gain values applied.

The appropriate bias reference file is selected based upon the DETECTOR, CCDAMP, CCDGAIN, BINAXIS1, and BINAXIS2 keywords.




WFC3 Dark Images (DARKFILE)

Additive noise due to thermal effects.

The dark current reference image file consists of an image of the dark signal, i.e., the signal detected in the absence of photons from the sky. This signal, proportional to the science image dark time, is subtracted from science images to remove the dark signal.

For UVIS observations, the dark reference file is scaled to units of electrons/second; thus, calwf3 multiplies the dark image by the science image exposure time and divides it by the gain before subtracting it from the science image. The dark time is simply the exposure time; it does not include the idle time since the last flushing of the chip or the readout time (the correction for this dark contribution is contained within the bias file). The dark reference file is applied after the CCD overscan regions are trimmed from the input science image; therefore, the reference file must have its overscan regions trimmed off as well. The UVIS dark image size varies depending on the on-chip binning mode, as follows:

1 x 1 binning: 2051 x 4096 pixel image
2 x 2 binning: 1026 x 2046 pixel image
3 x 3 binning: 684 x 1364 pixel image

Dark images for each of the 2 CCDs that make up the full UVIS detector array are stored separately in 2 imsets in the dark file, like the UVIS science data (see Section 2.2 of the WFC3 Data Handbook). The appropriate dark reference file is selected by the DETECTOR, CCDAMP, BINAXIS1, BINAXIS2, and USEAFTER keywords.

For IR observations, the dark reference files contain full-frame 1024 x 1024 pixel images. There are a total of 16 imsets in the IR dark reference files; one for each readout of a full MultiAccum sample sequence. The IR dark images are in units of DN (i.e., the gain has not been applied) and they are not rescaled by exposure time. The appropriate dark reference file is selected by the DETECTOR, CCDAMP, CCDGAIN, SAMP_SEQ, and USEAFTER keywords.




WFC3 Flat-Field Images (PFLTFILE, DFLTFILE, LFLTFILE)

Variations in the sensitivity of the detectors.

WFC3 can utilize up to three different flat-field images during calibration: the pixel-to-pixel flat (PFLTFILE), the delta flat (DFLTFILE), and the low-order flat (LFLTFILE). If more than one type of flat is specified for a given science dataset, each flat is applied in turn to the science data by calwf3.

Pixel-to-Pixel Flat Image File (PFLTFILE)

This type of flat, commonly called P-flat, contains the wavelength-dependent, high spatial frequency information about the uniformity of the detector response. This image is divided into the science images during the course of calibration.

For UVIS observations, the P-flat reference file is applied after the CCD overscan regions are trimmed from the input science image and therefore must have its overscan regions trimmed off as well. The UVIS P-flat image size varies depending on the on-chip binning mode, as follows:

1 x 1 binning: 2051 x 4096 pixel image
2 x 2 binning: 1026 x 2046 pixel image
3 x 3 binning: 684 x 1364 pixel image

P-flat images for each of the 2 CCDs that make up the full UVIS detector array are stored separately in 2 imsets in the P-flat file, as for the UVIS science data (see Section 2.2 of the WFC3 Data Handbook). The appropriate P-flat reference file is selected by the DETECTOR, CCDAMP, FILTER, BINAXIS1, BINAXIS2, and USEAFTER keywords.

For IR observations, the P-flat reference file is applied before the detector reference pixels are trimmed from the input science image; therefore, IR P-flats are full-frame images with a size of 1024 x 1024 pixels. The reference pixels are populated with a value of 1 in the IR P-flat file. There is only 1 imset in the IR P-flat file. The appropriate P-flat reference file is selected by the DETECTOR, CCDAMP, FILTER, and USEAFTER keywords.

WFC3 Linearity Correction Files (NLINFILE)

Non-linearity effects in the IR detector.

The linearity correction file contains a set of coefficients for each pixel that generate a linear correction over the non-linear range of the IR detector. The number of correction coefficients is tunable; as a result, the primary header of the non-linearity reference file must contain an NCOEF and an NERR keyword, indicating the number of COEF and ERR arrays, respectively, in the NLINFILE. Each coefficient must be at least 1. The file to be used is selected based upon the DETECTOR and USEAFTER keyword.

The observed response of the detector is represented by two regimes. At count levels below the saturation threshold the detector response deviates from the incident flux in a manner that can be adequately represented by a second-order polynomial. At high count levels (as saturation sets in) the response becomes highly nonlinear and is not correctable to sufficient scientific accuracy. The two regimes are defined and separated by the saturation threshold (the level beyond which no useful data can be extracted). For lower points, the following correction is used:

where I is the uncorrected SCI image value, I' is the corrected value, and c0, c1, and c2 are the linearity coefficients for a given pixel. The uncertainty for such a correction is:

where I is the input pixel value, σ is the input ERR image value, σ' is the updated ERR value, e0, e1, and e2 are the variances of coefficients c0, c1, and c2, respectively, and e01, e12, and e02 are the covariances of the coefficients.

This file departs from the standard format of SCI, ERR, DQ image extensions. The linearity coefficients are stored in floating-point image extensions with EXTNAME=COEF and EXTVER values ranging from 1 to 3. The variance, or error, values are stored in six floating point image extensions with EXTNAME=ERR and EXTVER values of 1 to 6. Data quality flags are stored, as usual, in a short-integer image extension with EXTNAME=DQ and EXTVER=1. The saturation threshold values (in units of DN) are stored in a floating-point image extension with EXTNAME=NODE and EXTVER=1. In addition to the above data, the LIN reference file also contains two floating-point image extensions that store a "super zero read" image and its associated statistical error image. These image extensions are designated by EXTNAME=ZSCI, EXTNAME=ZERR, and EXTVER=1. The ZSCI image is the average of a large number of MultiAccum zeroth read images and is used by calwf3 to estimate the amount of signal that may be present in the zeroth read of the science image that is being calibrated.

The format of this reference file is shown below:





WFC3 Bad Pixel Tables (BPIXTAB)

All known permanently bad pixels.

This reference file maintains a record of all known bad pixels for each WFC3 detector. The DQ values associated with each pixel listed in the BPIXTAB are written to the DQ image extensions of the science data being processed by calwf3 and are combined with any previously existing DQ values.

The positions of bad pixels are stored as pixel lists using the following columns:

Both UVIS and IR bad pixel tables list pixels in trimmed coordinates (i.e., overscan (UVIS) or reference (IR) pixels are not included). Bad pixel tables are selected by DETECTOR and USEAFTER.

Table 2.5 in Section 2.2 of the WFC3 Data Handbook summarizes the data quality flags as of Jan 2009. Note that many values may not appear in the bad pixel table as they are marked during calwf3 processing steps (such as cosmic-ray rejection).




WFC3 Detector Characteristics Tables (CCDTAB)

Detector parameters such as readnoise, gain, and bias levels.

Up to four detector amplifiers can be used for any given observation and each amplifier has its own read-out characteristics. However, only a single, nominal value for these characteristics can be commanded by the observer. The CCDTAB table provides the conversion from the commanded values to the calibrated values for each amplifier. These calibrated values are then used during processing by calwf3 to ensure that a pixel read out by an amplifier has been properly calibrated for that amplifier's readout characteristics. The characteristics affected are readout noise (READNSE), A-to-D gain (ATODGN), and bias level (CCDBIAS).

The table contains one row for each amplifier configuration used in the readout. This configuration is uniquely identified by the list of amplifiers used (CCDAMP), the particular chip being read out (CCDCHIP), the commanded gain (CCDGAIN), the commanded bias level offset (CCDOFST), and the bin sizes of the pixels read out (BINAXIS). Each amplifier can be used to read out a section of the chip or the entire chip depending on how many amplifiers are used to read out the observation. As a result, the values AMPX and AMPY specify the boundaries between amplifier read-out sections when used in concert to read out a chip.

The CCD table is selected by DETECTOR and USEAFTER. Table rows are selected by CCDCHIP, CCDAMP, CCDGAIN, CCDOFSTA, CCDOFSTB, CCDOFSTC, CCDOFSTD, BINAXIS1, and BINAXIS2.




WFC3 Overscan Region Tables (OSCNTAB)

Overscan (UVIS) or reference pixel (IR) areas.

This table describes the overscan regions for each detector along with the regions to be used for determining the actual bias level of the observation. Each row corresponds to a specific configuration as given by the amplifiers used (CCDAMP), the chip (CCDCHIP), and the on-chip binning mode (BINX, BINY).

The UVIS detector images can contain up to four serial overscan regions (1 physical pre-scan region and 1 virtual post-scan region associated with each of the 2 amplifiers on each chip), requiring the specification of 4 trim regions in the X-direction. Each UVIS detector chip also contains one parallel virtual overscan region. This requires the specification of at least 1 trim region in the Y-direction, either at the top or bottom of the image (depending on which chip is being processed).

The IR detector contains one physical overscan region on each of the four outer edges of the chip, requiring the specification of 2 trim regions in each of the X- and Y-directions of the images.

The columns TRIMX* give the number of columns of image data to trim off the beginning, middle, and end of each row, while the TRIMY* columns give the number of rows to trim off the top and bottom of each column. The columns NX and NY give the size of the image including the overscan sections.

The columns BIASSECTA1 through BIASSECTA2 provide the range of columns to be used for determining the bias level in the leading (physical) overscan region of the first amplifier used to read out the chip, while the BIASSECTB columns give the range of columns to be used to determine the bias level in the leading overscan region of the second amplifier. Similarly, the BIASSECTC and BIASSECTD columns give the range of columns to be used to determine the bias level in the trailing (virtual) overscan regions of the first and second amplifiers, respectively.

Finally, the parallel virtual overscan region for a single amplifier readout starts at pixel (VX1, VY1) and extends to pixel (VX2, VY2). Two separate parallel overscan regions are specified for two-amplifier readouts, using the (VX1,VY1) and (VX2,VY2) coordinate pairs for the first amplifier and the (VX3,VY3) and (VX4,VY4) coordinate pairs for the second amplifier.

The OSCNTAB is selected by DETECTOR and USEAFTER. Table rows are selected by CCDAMP, CCDCHIP, BINX, and BINY.




WFC3 Cosmic Ray Rejection Tables (CRREJTAB)

Parameters for performing image stacking.

This reference table contains all the basic parameters necessary for performing cosmic-ray (CR) rejection on images with the same pointing (e.g., CR-SPLIT, REPEAT-OBS). The CR-rejection process requires a number of input parameters to control how the cosmic rays are detected and removed. The process starts by creating a first guess for the CR-combined image either by using median or minimum value when combining the input exposures, as specified by INITGUES. Determination of the sky and noise values is controlled by the SKYSUB and SCALENSE values, respectively. Actual detection of the cosmic rays requires the specification of a threshold above which a pixel value is considered a cosmic ray (CRSIGMAS, CRTHRESH) and a radius around that pixel which of possibly affected pixels by the cosmic ray (CRRADIUS). Once a pixel is determined to be affected by a cosmic ray, that pixel will be marked in the input image DQ array, if CRMASK is set to yes.




WFC3 MultiDrizzle Parameters Table (MDRIZTAB)

Input parameters for combining associated exposures.

The MDZTAB contains input parameters for MultiDrizzle, optimized for the widest range of science cases. These serve as default parameters for the instrument calibration pipeline to process any given single image or association. Observers should inspect these values and determine whether any adjustments or fine-tuning may be necessary for their science data.

For a full listing and description of the parameters, please see the MultiDrizzle Handbook.




WFC3 Image Photometry Table (IMPHTTAB)

Parameters for performing photometry correction.

This reference table is used in the PHOTCORR step of the HSTCAL (IRAF-indepndent) version of the calwf3 pipeline (calwf3 version 3.0). The file (one for the IR channel and one for UVIS) contains the PHOTFLAM values for all configurations of that channel in the WFC3 instrument. PHOTFLAM is a keyword present in the main header of HST data. The value of this keyword for a particular instrument configuration is a measure of the inverse sensitivity (ergs/cm2/Ang/electron) for that configuration, and can be used to translate data from units of electrons to calibrated fluxes. In previous versions of the calwf3 pipeline, the PHOTFLAM value was calculated on the fly via a call to synphot. In the new HSTCAL version of the pipeline, this call to synphot has been replaced with the IMPHTTAB, which functions as a look-up table. For more details on the PHOTFLAM header keyword, see the WFC3 Data Handbook.




Post-Flash Image File (FLSFILE)

Application of a post-flash image.

This reference file corrects a UVIS image for the signal added to an exposure by the post-flash procedure. The scaled reference image (see below) is subtracted from the science exposure by calwf3. The post-flash reference images are full-frame, including physical and virtual overscan regions. The calibration software, calwf3, multiplies the post-flash image by the science image flash time FLASHDUR and divides it by the gain before subtracting it from the science image. This requires the post-flash image to be scaled to an exposure time of 1 second and a gain of 1. There are separate post-flash images for each on-chip CCD binning mode, with sizes as follows:

1 x 1 binning: 2070 x 4206 pixel image
2 x 2 binning: 1035 x 2102 pixel image
3 x 3 binning: 690 x 1402 pixel image

There are also separate reference images for each shutter (keyword SHUTRPOS), since light reflected from the closed shutter is used to perform the post-flash, and the two different shutters result in slightly different light patterns. There are also separate files for each lamp current setting FLASHCUR since these result in different count rates.

Normalized post-flash pattern in jpg and fits formats.

Ratio of post-flash illumination pattern in shutter blade A to shutter blade B jpg

Post-flash images for each of the 2 CCDs that make up the full UVIS detector array are stored separately in 2 imsets in the post-flash file, like the UVIS science data. This reference file is selected by the DETECTOR, CCDAMP, BINAXIS1, BINAXIS2, SHUTRPOS, FLASHCUR, and USEAFTER keywords.




Detector-to-Image Correction File (D2IMFILE)

2-D look-up table of astrometric corrections for shifts caused by the lithographic-mask pattern.

The D2IMFILE reference file is a 2-D look-up table that contains astrometric corrections for shifts induced by the lithographic-mask pattern that was imprinted on the UVIS detector during the manufacturing process. This correction table is used in coordinate transformations prior to applying the large-scale polynomial geometric solution (IDCTAB). The look-up table is incorporated by running updatewcs (via AstroDrizzle) with the filename specified in the D2IMFILE keyword in the primary image header, which will then be attached to the image as a fits extension of type 'D2IMARR' (four extensions total, two for each chip and dimension). Each element of the 32x17 table gives the astrometric xy shift in pixels to apply at a particular location on the detector. The latest IRAFX release (May 20, 2013) was updated to allow AstroDrizzle to interpolate across a 2-D table grid, and hence only this IRAFX version (and later) can apply the D2IMFILE corrections.




WFC3 Image Distortion Coefficients Tables (IDCTAB)

Geometric distortion models.

This reference table contains a description of the geometric distortion models for the UVIS and IR detectors as a function of filter. The IDCTAB contains the coefficients of a polynomial fit that is used to transform image coordinates from raw (distorted) space to an undistorted space. Earlier versions of MultiDrizzle used the IDCTAB directly to perform the distortion correction. Now, the PyRAF routine makewcs uses the IDCTAB to populate the "simple image polynomial" (SIP) header keywords, which are used by MultiDrizzle to perform the distortion correction (see the MultiDrizzle Handbook).

The IDCTAB contains information such as the calibrated image size in X and Y (XSIZE, YSIZE), the reference point position in X, Y (XREF, YREF) and in V2, V3 (V2REF, V3REF), the angle THETA from V3 to Y, the chosen output pixel scale (SCALE), and the distortion coefficients in X and Y.

The table contains one row for each part of the detector that has its own distortion correction (e.g., CCD detector chip). The primary header keyword NORDER is used to indicate the order of the polynomial that is represented in the table and the corresponding number of coefficients.

The IDCTAB is selected by DETECTOR and USEAFTER. Rows are selected by DETCHIP and FILTER.




Analog-to-Digital Table (ATODTAB) (Currently not in use)

Irregularities in A-to-D conversion.

This table provides the actual number of counts for each detected count in the image and allows for possible irregularities that might occur in the A-to-D conversion, such as were seen on the original WF/PC-1. The conversion takes into account the gain setting, the amplifiers used, and, typically, the exposure time of the observation. This reference table is selected by the DETECTOR keyword, and rows within the table are selected by CCDCHIP, CCDAMP, and CCDGAIN.