### B Processing JCMT Legacy Survey data

Currently, three of the JCMT Legacy Surveys have recipes optimized for the goals of the surveys. Support for the others will be added in as timely manner as possible in response to survey input. Picard recipes also exist which replicate the steps performed on the processed data.

#### B.1 Cosmology Legacy Survey (CLS)

The CLS recipe employs a “jack-knife” approach using independent halves of the data in order to estimate and remove residual noise on large spatial scales.

• Maps are made with a modified blank-field config file and coadded into a single map.
• An artificial gaussian source is inserted into the data and the maps are remade (and coadded to make a PSF map).
• The signal-only maps are divided into two groups that are coadded separately and subtracted to form a jack-knife map.
• The central portion of the jack-knife map is used to estimate the spatial power spectrum which is applied to the signal coadd to remove residual low-spatial frequency noise (“whitening”).
• A matched filter is applied to highlight compact sources using a whitened version of the PSF map as the input PSF. A signal-to-noise ratio image is also calculated.

Currently, this recipe works best on single scanned fields (i.e. not a mosaic of multiple fields).

#### B.2 Survey Of Nearby Stars (SONS)

This recipe is very similar to that for CLS with an additional step at the beginning. This step makes maps for each 30-second subscan and calculates the noise level in those maps to determine the noisiest subscans which are ignored by the map-making step. (Note the option exists to use the time-series noise instead.)

For this recipe, it is recommended that the artificial source used to determine the FCF correction be offset from the centre to avoid contamination of the signal.

#### B.3 SCUBA-2 “Ambitious-Sky” Survey (SASSy)

Processing of SASSy data is focused on detecting compact sources with the aim of following up previously-unknown or interesting-looking detections with more sensitive observations to probe the detailed structure.

• A peak-detection task (findclumps) is run to identify 5-$\sigma$ peaks, which are written to a catalogue file.