Like essentially al IR arrays, the WFC3/IR detector exhibits
after-images, known as persistence, following exposure to light
that approaches or exceeds the saturation level of individual
pixels of the detector. In extreme cases, this
persistence can be seen for several hours in WFC3. As a result,
persistence can affect science exposures not only within a single
visit, but also in exposures of later visit. A description
of persistence in the WFC3/IR detector can be found here.
Normally, persistence due to images within a visit is not a
problem, because most observers use small dithers and the
after-glow images are faint. As a result, afterglows within
a visit are usually undetectable since they are found in the wings
of the point spread function of the follow-on images.
Persistence due to earlier visits, while less common, can be
more of a problem scientifically, since the after-images from
earlier visits can be anywhere in the science images of the
current visit. The numbers of programs which have been
adversely affected are small, but observers need to be aware of
In order to help observers assess the degree to which a their
exposures are affected, STScI has developed a model
for persistence as it exists in WFC3. We have used this model
and the time-history of earlier exposures to generate estimates of
the amount of persistence in each "flt" file and to produce fits
files that contain estimates for each pixel in each "flt"
file. We also produce a persistence-subtracted
Since persistence due to images within a visit and those from
previous visits reflect different issues, we distinguish between
the two. We call persistence within a visit "internal
persistence", and persistence from earlier visits "external
Access to Persistence Information for WFC3/IR
One can use this
search form to locate persistence information about WFC3/IR
The output of the query contains the percentage of pixels
estimated to have persistence > 0.1, 0.03, and 0.01
electrons/s. It provides these for the external
persistence (EXT1, EXT2, and EXT3, respectively) and for the total
of internal + external persistence (TOT1, TOT2,TOT3).
Generally speaking, the primary columns to be concerned about are
EXT1 and EXT2, the percentage of pixels with external
persistence in excess of 0.1 and 0.03 electrons/s,
respectively. For reference, the median dark current in WFC3/IR is
about 0.015 electrons/s.
In any event, more details of the persistence estimates as well
as evaluation images showing the regions of the detector where
persistence is likely to be greatest can be seen by following the
link associated with the dataset name.
In addition, by clicking on the visit link, one can retrieve the
fits files that are produced to as part of the persistence
estimate for this visit. When you uncompress the tar.gz file
there you will find files of the following types:
rootname_extper.fits - estimate of the external persistence
rootname_persist.fits - estimate of the total (internal +
Datasets marked with a yellow background and followed by an "@" sign are still proprietary
and are not available to the general user. You will be prompted to enter your
archive user name and password and if that account is authorized to retrieve the data, then
you will be able to see the web page and to download the tarfile.
These data are available through the browser at http://archive.stsci.edu/pub/wfc3_persist/ and are also
available via ftp from archive.stsci.edu (logon with archive user name and password to obtain authorized proprietary data).
As with most standard MAST search interfaces, one may run the search interface has an HTTP GET request and embed the search in a script.
will find observations associated with Proposal ID 11937 Visit 01 and return the results as a JSoN format. For more information about
options see http://archive.stsci.edu/vo/mast_services.html.
See the field descriptions for the column names and data types.
Observers should be cautioned that the persistence
model is only an approximation to the actual persistence.
Therefore, one should not "blindly" replace the standard 'flt"
files with the persistence-corrected flt files in downstream
analysis. A better strategy in some cases, may be to use the
external persistence file to flag bad pixels for downstream
analysis. In other cases scaling the models up or down by a small
factor may result in a cleaner corrected image.
Observers and archive users are encouraged to contact the help
(firstname.lastname@example.org) if the
persistence subtraction substantially limits their ability to
extract science from an image or set of images.
Users who have questions about the interface or encounter problems with the interface or downloading the data should contact
the archive help desk email@example.com.