Key Differences for JWST Moving Target Observations

The JWST calibration pipeline has the ability to co-add moving target exposures. While many steps in the pipeline are identical for moving and fixed targets, steps specific to moving targets are outlined below.

On this page

Stage 1

Moving target header keywords

As for all JWST science data, stage 1 processing includes data formatting, science frame re-orientation, calculation of WCS keyword values, and "ramps-to-slopes" fitting. For moving target data, additional WCS entries are created to enable co-addition of data in the co-moving frame of the target. These keywords are created at the exposure level in stage 1 (these are carried over to the stage 2 products), and at the observation level in stage 3. See the stage 3 discussion below for further details about the header keywords for those products.

The right ascension and declination values of the target pixel at the mid-time of each exposure (stage 1 and 2 products) are recorded with the following keywords present in the SCI header:

  • MT_RA
  • MT_DEC

Note that there are no comparable keywords giving the target position at the mid-time of each integration within an exposure at a given dither position. Such information can be reconstructed by retrieving the "uncal" products and accessing the "MOVING_TARGET_POSITION" table extension, and the MT_APPARENT_RA and MT_APPARENT_DEC values. That extension also contains the heliocentric and observer-centric distances as a function of time during the exposure, along with the X, Y, and Z positions of the target and JWST.

Also, if multiple exposures are specified with the same spectral elements, the stage 1 and stage 2 products will have different values of MT_RA and MT_DEC for the separate exposures.

It is worth noting that dither and mosaic offsets are preserved by a correction applied to the CRVAL keywords in the headers of moving target products. In this case the standard keyword values are replaced by values that are corrected for the target motion during the exposure (or observation, at stage 3). In this case the RA and DEC of the reference pixel at the exposure (or observation) mid-time, RA_REF0, DEC_REF0, are calculated, and new values for CRVAL1 and CRVAL2 calculated as:

  • CRVAL1 = average(CRVAL1  + (RA_REF0  - RA_REF) )
  • CRVAL2 = average(CRVAL2  + (DEC_REF0  - DEC_REF) )

Rejection of signal from fixed sources in moving target observations

Another unique consideration for moving targets is that the stars and other fixed objects trail through the scene. Due to the small pixels of the JWST instruments and PSF, fixed objects can cross one or many pixels during a single integration ramp, causing transient count rate increases or decreases as they enter and leave specific pixels. These transient signals are treated as cosmic rays in the jump detection step (see Figure 1).

Figure 1. Star trails in a moving target stage 2 (rate.fits) file

Click on the image for a larger view.

NIRCam rate image from PID 1021, observation 7, showing a fast moving near-Earth asteroid (NEA), (464798) 2004 JX20. The moving target is stationary on the detector, causing background stars to trail. Due to this trailing, the flux from the stars is treated like cosmic rays in the stage 1 jump-detection step, removing the majority of the flux from the resulting rate image. Note that moving targets in fixed target observations are subject to the same type of rejection by jump detection.

Stage 2

As with fixed target observations, stage 2 processing, or single-exposure calibration processing, outputs calibrated single exposure images (e.g., cal, i2d). These steps are identical to those for fixed targets, since the moving target is "fixed" to the detector frame through telescope tracking.

For some applications users may be interested in stage 2 products for each integration within an exposure. This may be the case for observations containing multiple targets or features of interest that are moving at different rates (e.g., cloud features on Jupiter and/or Jupiter's satellites). At this time users have to generate these products off-line by installing the jwst pipeline package and processing the "rateints" files to produce "calints" files.

Stage 3

Stage 3 processing combines the stage 2 products through co-adding the single exposures that result from dithers and/or mosaic offsets. For moving targets, co-addition of exposures occurs in the target's co-moving frame. The pipeline accomplishes this by implementing existing fixed target algorithms, and instead of the normal WCS, uses the unique moving target's WCS data, an approach pioneered by Hershel and Spitzer. Note that this has not been implemented for the HST pipeline. For JWST, both imaging and spectroscopy are supported.

The re-projection of the data into the co-moving frame is achieved using the additional and altered keywords noted under stage 1, above. In the stage 3 products, the right ascension and declination values of the target at the mid-time of each observation, limited to a specific instrument configuration (e.g. filter) are recorded as:


There are similar updates to the CRVAL values in the stage 3 headers, such that the final image is centered on the target location at the mid-time of the observation.

Stage 3 processing also includes an outlier detection step with a rejection threshold that may be too aggressive for some moving target observations. This is especially the case for observations with few dithers and/or complex scenes; observations of point-like sources, or observations with a large number of dithers, are less likely to be affected. For observations that are subject to this effect, the stage 3 products will have otherwise acceptable pixels rejected, resulting in those pixel values being replaced with 0 or NaN (see Figure 2). For this reason, it is highly recommended to make use of the stage 2 products or to run the pipeline on your local machine with the rejection threshold set to a higher sigma threshold.

The stage 3 outlier rejection step will also further erode signal from trailed stars in moving target observations (and of moving targets in fixed target observations).

Figure 2. Outlier rejection in a stage 3 "s3d" file for an extended target

Click on figure for a larger view.

Demonstration of the default outlier rejection threshold in stage 3 of the pipeline. The 4 images on the left are Ganymede at 3.15 μm at each of 4 dither positions in NIRSpec IFU stage 2 "s3d" files (PID 1373, observation 28). The image on the right is Ganymede in the stage 3, dither-combined "s3d" file at the same wavelength. Note that some pixels were set to 0 by the outlier rejection step in the center of the disk and along the limb of Ganymede, even though those same pixels do not appear to be outliers in the stage 2 data. Point source observations are less likely to be affected in the same way.

Notable updates


  • Updated extensively using in-flight knowledge.


  • Minor wording and formatting changes.
Originally published