Many observers (with any Ge data) may want to apply techniques discussed here to their data. For bright sources, you can use the unfiltered mosaic. For fainter sources, consider the fbcd product as a quick-look at your data and do your own filtering. If in doubt, err on the side of caution and redo the filtering and the mosaics yourself.
8.2.1 High-level summary
The basic processing steps are the following. Note that items 1 and 2 are done for you in our pipelines; you will most likely only have to spend time with items 3 and 4:
1. Fit slopes to ramps to make slope images.
2. Calibrate slope images. (See section 4.1.1 for more discussion.)
3. Correct for residual instrumental effects.
a. High-pass median time filter per pixel to remove residual variations of the slow response as a function of time. (This means: subtract the median value of the neighboring BCDs on a per pixel basis as a function of time.)
b. Column filter at 70 microns to remove residual stim latents. (This means: subtract median value of each column from every column since stim latents are correlated by column.)
c. Since the automatically-produced (''online'') filtered mosaics are optimized for point sources, not for extended regions (i.e., the background is removed by filtering), bright point sources can bias the filtering. To maintain the calibration of bright point sources, do the filtering offline in two passes. Identify bright sources (>0.2 Jy) in the first pass, then mask and re-calculate the filtering corrections.
4. Co-add images to make mosaics. (Can be done entirely with MOPEX.)
a. Correct for array distortion.
b. Interpolation of frames onto fiducial sky frame.