To correct persistent effects, including long-term latents, low-level jailbars, and gradients, you should use your own data to self-calibrate.
First, assemble a list of BCDs without the very bright sources but with the persistent effects. If the latents are caused by your own data, make sure to separate BCDs before and after the offending bright sources. For some data sets, you may want to include the non-prime BCDs.
Second, take the median of the list(s) of BCDs and normalize it. Then, divide all affected BCDs by the normalized median. Note that the S12 pipeline (and later) provided a non-normalized median of entire AOR as a data product, bmed.fits.
Finally, make a new mosaic from your modified BCDs. See Figure 8.1, Figure 8.2, and Figure 8.3 for example ''before'' and ''after'' images.
Figure 8.1: Example mosaic before (left), with long-term bright latents and weak jailbars, and after (right) self-calibration, where those effects are well-corrected. Note that the flat fields in this example did not match the spot positions in the data very well (upper right).
Figure 8.2: Example mosaic before (left), with long-term dark latents, and after (right) self calibration of photometry data using non-prime data.
Figure 8.3: Another example of self-calibration removing dark latents, this time in a large scan map. Left is before, right is after.
8.1.3 Sky Background Removal
The BCDs contain in their headers an estimate of the sky background using the SSC model (Reach 2000). The model reports three values for the Cosmic Infrared Background (CIB), ISM, and Zodiacal Light. We have found that in Scan Maps where one suspects Zodi gradient to be important, one can improve the quality of the mosiac by using these sky background estimates. To remove it, scale the measured background in a BCD free of large extended sources to the reported Zodi and then subtract that constant.
A simple example is that of NGC300; the corresponding science paper is Helou et al. 2004, ApJS, 154, 253. We compare the standard SSC post-BCD mosaic and a ''zodi'' free mosaic, where the model zodiacal light has been removed in Figure 8.4. In both mosaics the dynamic range of the gray scale is 1.0 MJy/sr and with a similar histogram distribution. In this NGC 300 observation the Zodiacal light varies from 23.6 (bottom of the image) to 23.9, i.e. a gradient of a bit more than 1%. Even this small variation is noticeable in the MIPS 24um mosaics.
Figure 8.4: NGC 300 in 24 microns, rotated to have north up. The left image is the standard post-BCD mosaic.