PACS PHOTOMETER JScanam highly processed Mosaics

JScanam mosaics of the Galactic Plane around the Galactic Centre (|l| < ~10 deg). Credit: Javier Gracia-Carpio (MPE)

The main purpose of these Highly Processed Data Products (HPDP) is to improve the quality of the PACS JScanam maps of very extended regions. These regions could not be covered by a single PACS observation (scan + cross-scan) and required several pointings. Examples of these kind of maps are those of the Milky Way, the Orion star forming region and the Large and Small Magellanic Clouds.

The PACS Standard Product Generation (SPG) pipeline indeed processed each pointing separately. This leads to two main issues when one tries two combine several SPG maps into a final mosaic:

• The borders of each separate SPG map are generally more noisy than the rest of the map because the JScanam algorithm does not have enough information to clean them. If we do not remove these noisy borders before we do the mosaic they will be visible in the overlap areas.
• Due to the fact that the global (constant) background level of a given map cannot be derived from the map alone, mosaicing several SPG maps generally produces artificial signal jumps in the overlap regions. It is also not possible to relate the extended continuum flux levels from regions coming from different pointings (SPG maps).

On top of that, The Large and Small Magellanic Clouds maps constitute a special case. They were not observed using the standard PACS scan + cross-scan configuration, and as a result they did not have JScanam SPG maps in the Herschel Science Archive (HSA).

METHOD

The first issue is resolved by creating a mask during the data reduction process that keeps track of those spatial regions that were only visited either by the scan or the cross scan. These regions coincide with the noisy borders described above.

In that context two mosaics are then created with exactly the same WCS projection: one using the mask (i.e. it does not contain the noisy border data) and one that uses all the data. The final mosaic is a combination of the two mosaics on a pixel by pixel level. If a given pixel position contains data both in the masked and the unmasked mosaics, the masked mosaic value is used. If there are data only in one of the mosaics (by definition the unmasked mosaic), that value is then used. This effectively removes the noisy data from the overlap regions, still keeping the data from the non-overlapping borders (better noisy data than no data at all).

To take care of the second issue each map is projected using exactly the same WCS projection and one calculates the average difference between the two maps on the overlapping region. This difference is then subtracted from all the pixels of one of the maps to bring the two maps to the same continuum level. This process is applied iteratively when more that two maps were involved.

As a final correction, the region in the mosaic with the lowest continuum level is searched using a square box of typically 50x50 pixels. The median pixel continuum value in that region is calculated and subtracted from all the pixels in the mosaic. As a result, on average the lowest continuum level in the mosaics is always zero. This assumption is of course not necessarily true, and should always be kept in mind.

Due to the large size of some the maps (LMC, SMC and Virgo), a tiling strategy was followed, with each tile being reduced separately. At the end all the tiles were aligned and merged to create the final HPDP maps.

The scripts used to generate these products (in HIPE) are provided in a scripts directory. They are organised following the same structure used for the products, as indeed different processing methods were applied for different use cases. On top of that, a dedicated sub-directory provides scripts in order to combine Galactic plane tiles optimally with the Swarp program