Reflection from bright sources in the field of view can cause optical ghosts in NIRISS imaging and WFSS observations.

On this page

See also: NIRISS GR150 Grisms

For the NIRISS imaging and wide field slitless spectroscopy (WFSS) modes, there is a chance that optical ghosts may appear in the field of view (FOV) and significantly affect some sources. These optical ghosts are caused by internal reflections of bright sources between the detector and the filter in the light path

Ghosts in NIRISS are characterized by extended morphology. The exact shape may vary over the FOV and the optical element (e.g., a filter or grism) in the light path. The integrated intensity of ghosts is proportional to the original source, ranging ~0.1%–4%, also depending on the optical element and the position of the source. The characterization of ghosts originating from extended sources is currently unknown. The intersection of lines between bright sources and ghosts, as shown in Figures 1 and 2, is the ghost axis point (GAP). Knowing the coordinates where these intersections occur can help predict and identify the locations of ghosts.  

The current knowledge of GAPs and other features of ghosts has been acquired from ground testing campaigns and on-sky commissioning data. More features will be evaluated using available on-sky data.

Figure 1. Example of ghosts in NIRISS direct image (ground testing)

NIRISS F150W direct image taken in the Optical Telescope Element/Integrated Science Module campaign. The position of each ghost can be related to the position of the original bright source through the GAP position that is specific to the optical element. A zoom-in image of one of the ghosts is shown, revealing the extended morphology.

Figure 2. An example of NIRISS ghosts (on-sky)

Left: An example of bright stars that produce ghosts in a NIRISS image. The magenta X marks the ghost axis point (GAP), the point around which ghosts are reflected from the sources. The stars producing the ghosts are circled in cyan and the objects circled in magenta are the ghosts. The red Xs mark sources that would be expected to produce ghosts, but not are identified indicating that only certain regions in the detector are susceptible to producing ghosts.

Right: A zoom-in of the image to illustrate the blobby morphology of ghosts, and the double ghosts produced in this particular filter (F150W). These images are from observation 116 in APT program 1063 with the F150W filter.

Mitigation strategies

The impact of ghosts on observations can be assessed by simulating the observing scene with MirageDithering may help to suppress some ghosts in the final image product from the pipeline IMAGE3 step.

Words in bold are GUI menus/
panels or data software packages; 
bold italics are buttons in GUI
tools or package parameters.

Figure 3 shows examples comparing 3 dither sizes that are available for the NIRISS primary observing mode (SMALL/MEDIUM/LARGE). In these examples, a cutout image of a ghost identified in one of the ground testing campaigns (the one shown in the right panel of Figure 1) was used. While the result may depend on the actual morphology of ghosts (i.e., positions of the source and the choice of filter), the example here suggests that a SMALL dither may not be large enough, and may let outer parts of the ghost overlap, which results in a significant residual (~3.6% of the total flux of the input ghost) in the final stacked image. Therefore, a dither size of MEDIUM or LARGE is generally recommended.
Figure 3. Comparison of ghosts after the final stack

Comparison of simulated ghosts with SMALL (top panel), MEDIUM (middle), and LARGE (bottom) dither sizes, with the F150W filter. Each panel shows data products from the pipeline calwebb_image2 step at the same reference sky coordinates (red cross) so the movement of the ghost can be monitored, and the final product of the calwebb_image3 step.
A caveat is that having a large dither size does not necessarily clean the footprint of ghosts completely. As seen in the examples above (the middle and bottom panels of Figure 3), there are some positive pixels left in the final combined images, possibly because the sigma clipping could not exclude faint parts of the original ghosts. For more complete suppression of the ghost footprint, extra steps to identify and then mask such ghosts (for example, as documented in the niriss_ghost repository) would be required before the final stack. 


Martel, A. 2019 JWST-STScI-004877
The Ghosts of NIRISS: Imaging

Latest updates
    Updated with in-flight information
Originally published