General PACS related questions:

• What does the acronym PACS stand for?
• The answer I'm looking for is not here ... where to next?
• I have 10s or 100s (a large number) of AORs, which employ the same technique and require only small or minor bit of information to be changed. For example, the observing strategy is identical but the object/position is different, or the exposure time depends on the object. Is there a tool that will automatically do this for me?

• What is the saturation limit of the PACS bolometers?
• How is the PACS confusion noise calculated?
• What is the best way in which to observe a small point-like source as deep as possible?
• How do I translate the PACS 70um point source detection limits into the pixel-to-pixel noise of the instrument?
• What is the best way to map a large field (of order degrees) to detect sources at very faint levels and with high position certainty with the PACS bolometer?
• PACS extended source noise question: I am planning to make an observation (below) and would like to confirm that the HSpot extended source noise is correct (1-sigma in MJy/st). It seems a bit high. Can you explain how it is calculated? My hypothetical observation is made in chopper raster mode with 2x2 points with a repetition factor of 8 (2080 on source sec.). This gives a 1 sigma point source noise of 1.9 mJy, that translate in 2.2 MJy/sr for extended emission. The 2.2MJy/sr noise seems quite large--is it correct?

• If I am interested in a line in the short blue region, what wavelengths do I get for "free" in the red band?
• Which technique is better for baseline subtraction for my science in PACS spectrometer: chopping or frequency switching?
• What is the recommended minimum number of repetitions for the Nyquist SED sampling mode for PACS spectroscopy?

Question: The answer I'm looking for is not here ... where to next?
Answer: If you cannot find the answer here, please use the NHSC helpdesk.

Question: I have 10s or 100s (a large number) of AORs, which employ the same technique and require only small or minor bit of information to be changed. For example, the observing strategy is identical but the object/position is different, or the exposure time depends on the object. Is there a tool that will automatically do this for me?

Comments: This is certainly a very tedious chore and many of us have been there. AOR duplication can only help so far. For 10s or 100s of objects this is still a lot of interactivity and a potential pitfall for mistakes.

Suggestion: Automating this task is not straightforward. The best scenario involves creation of a tool that can edit certain fields in the AOR files. Since the AOR files are written in XML, it should be relatively easy to identify the fields and substitute a replacement value. We envision a tool, which takes as input:

- AOR file to be operated on (may contain multiple AORs).
- A tagged multi-column text file containing the key that requires changing and the change that needs to be made. For example, this can be the object position and the new position (one per AOR in the file).

The first file is simply the output of HSpot. The second file is supplied by the user. It identifies the field to be edited and in the following columns, the replacement values. The tags present identify the start and end of the number of fields to edit. For each AOR, the program will make a copy of the original AOR and apply the requested change and append it to an automatically generated output file, which will become the user's new AOR file. What this will save the user is the time invested in interactively duplicating/copying the original AOR and manually editing the fields to be modified. The user will still need to create a list of the replacement values and make sure that the visualization is correct after the changes. This is a recommendation for an AOR duplication tool but this tool does not yet exist.

Question: What is the saturation limit of the PACS bolometers?

Answer: By design bolometers do not saturate; rather it is the electronics behind the detector that saturate. For PACS the nominal observing mode uses amplifier gains tuned towards providing well sampled signal for faint sources. The nominal gain setting is useful up to 1600, 2600, and 2300 Jy in the blue, green, and red bands, respectively. If the user wishes to observe brighter sources, they can input the source flux in HSpot, which then automatically switches the gain setting (with a warning message to the user) to allow for observations up to a factor ~4x brighter. Please note that faint source photometry is significantly degraded in the low-gain/bright source setting.

Question: How is the PACS confusion noise calculated?

Answer: For information on the Herschel confusion model see:

• http://herschel.esac.esa.int/Docs/HCNE/pdf/HCNE_v2007Jan31a.pdf
• http://kisag.konkoly.hu/confnoise/

Question: What is the best way in which to observe a small point-like source as deep as possible?

Comments: We considered all of the available observing modes and determined point source photometry to be the most efficient; however, it only provides 4 independent positions on the array. The dithering (~1 pixel) option places the object within the PSF, which helps with the position determination but does not add another independent location on the array. Raster and small source photometry use ½ of the time off the point source and hence are only ½ as efficient (there is no real gain in using these modes in this scenario).

Suggestion: For the deepest exposures on a point source, we recommend using the point source photometry pending resolution of scanmap option questions. The NHSC, however, highly recommends that 3-5 concatenated AORs on the same source are employed to actually perform the deep observations. This will lead to 3-5 times 4, or 12-20 independent measurements on the object rather than the default 4 from the photometry. For each of these concatenated AORs, the position of the target is slightly offset by ~4 pixels from the nominal target position. The Table below lists the recommended offsets to use. The Figures below illustrate the 3x4 position scenario.

Table 1: The recommended offsets (from object position) to use between AORs for concatenated AORS on deep point-source photometry observations

Figure 1: The nominal positions of the object, denoted as the star symbol, in the point source photometry

Figure 2: The resulting positions on the array from the recommended 3x4 position concatenated AOR scenario for deep observations

Question: How do I translate the PACS 70um point source detection limits into the pixel-to-pixel noise of the instrument?

Answer: The point-source sensitivity values are based on a model of PACS. The sensitivities per integration time for each mode are listed in the observer's manual. The details of the conversion between pixel noise and point-source noise can be inferred from HSPOT. HSPOT returns both surface brightness noise and point-source noise and this ratio is constant for all modes. In a SCAN-mode example at 70um, point-source rms = 10.1mJy extended-source rms = 11.45 MJy/sr (see PACS HSpot Time Estimation and messages)

1MJy/sr = 0.0235 mJy/sq-arcsec

1 blue pix =~ 3.2"

11.45 MJy/sr *0.0235 [mJy/sq-arcsc/(MJy/sr)] * (3.2")^2/pixel = 2.76 mJy/pixel ==> Conversion between point-source noise to per pixel noise == (NP)

PACS-70 NP = 10.1 mJy/(2.76mJy/pixel) = 3.7 "noise" pixels

rms(point-source)[mJy] =~ 3.7 rms(mJy/pixel) (for PACS 70um)

Sanity check based on the number of pixels per beam (i.e., point-source noise is really noise per beam):

FWHM~5.2" for PACS-70

Gaussian-Beam = 1.133*FWHM^2 = 30.63 sq-arcsec = 3.0 pixels

Typically find NP~1.1--1.4*(pixels/Gaussian-Beam) [e.g., Spitzer], so you should expect NP ~ 3.3 -- 4.2 for PACS-70. This is independent of aperture size; one should find similar effective NP values via aperture photometry and aperture corrections.

Question: What is the best way to map a large field (of order degrees) to detect sources at very faint levels and with high position certainty with the PACS bolometer?

Comments: Scan map is the only viable and efficient option for this or any area larger than ~12' (see Figure below). Even so a few steps can be taken to ensure easier data sampling and reduction to meet the faint source detection and position reconstruction requirements.

Suggestion: The following points are offered as suggestions:

• Dither manually between repetitions. This is best accomplished by adding an offset to the target position and concatenating/chaining AORs. Dithering spreads the coverage on to different pixels for multiply repeated scans and ensures that the same dead pixel does not cover the same region.
• Consider changing the scan angles between dithers/multiply concatenated AORs. This minimizes "streaking" a-la-IRAS.
• For better mapping, allow for a two-pass reduction scheme.
• Pass 1 to get the relative offsets between each scan, and
• Pass 2 to put the final map/mosaic together.
• Look for any possible alignment between inter-matrix gaps and scan direction. This angle is a constant when scanning in array coordinates, and varies with observing date when scanning in sky coordinates.

Figure: The observing efficiency of various PACS AOTs

Question: PACS extended source noise question: I am planning to make an observation (below) and would like to confirm that the HSpot extended source noise is correct (1-sigma in MJy/st). It seems a bit high. Can you explain how it is calculated? My hypothetical observation is made in chopper raster mode with 2x2 points with a repetition factor of 8 (2080 on source sec.). This gives a 1 sigma point source noise of 1.9 mJy, that translate in 2.2 MJy/sr for extended emission. The 2.2MJy/sr noise seems quite large--is it correct?

Answer: Assuming that you are using the Blue-band and adopting 200" and 100" offsets (~full offsets), I was able to reproduce your quoted values of 1.9mJy point-source noise and 2.2 MJy/sr surface brightness noise. The quoted extended noise is the surface brightness noise as measured per pixel. So HSpot is reporting the correct values. They do make sense.

The ratio between point source noise and extended noise for PACS-70 is about: surface-brightness_noise/point-source_noise ~ 1.13 [MJy/sr]/mJy. In your case, you have 1.9mJy point-source noise which would correspond to an expected surface-brightness noise of about 2.2 MJy/sr. Below is an example showing the conversions between point-source noise, noise per pixel, and extended surface brightness noise (previous response to a different user) --- perhaps this is useful for you as well.

Example:
point-source rms = 10.1mJy
extended-source rms = 11.45 MJy/sr

1MJy/sr = 0.0235 mJy/sq-arcsec
1 blue pix =~ 3.2"

11.45 MJy/sr *0.0235 [mJy/sq-arcsc/(MJy/sr)] * (3.2")^2/pixel = 2.76 mJy/pixel

==> Conversion between point-source noise to per pixel noise == (NP)
PACS-70 NP = 10.1 mJy/(2.76mJy/pixel) = 3.7 "noise" pixels (in SSC APEX/MOPEX jargon)

rms(point-source)[mJy] =~ 3.7 rms(mJy/pixel) (for PACS 70um)

Sanity check based on the number of pixels per beam (i.e., point-source noise is effectively noise per beam):
FWHM~5.2" for PACS-70
Gaussian-Beam = 1.133*FWHM^2 = 30.63 sq-arcsec = 3.0 pixels
Typically find NP~1.1--1.4*(pixels/Gaussian-Beam) [e.g., Spitzer], so would expect NP ~ 3.3 -- 4.2 for PACS-70, consistent with HSPOT.

Question: If I am interested in a line in the short blue region, what wavelengths do I get for "free" in the red band?

Answer: The grating equation can be applied here to obtain the wavelengths asimultaneously covered by the blue and red bands. The relationship is 3*blue_wavelength = 1*red_wavelength for the short blue filter (order 3). Similarly, for the "green" or order 2 filter, the relationship will be 2*green_wavelength = 1*red_wavelength.

Question: Which technique is better for baseline subtraction for my science in PACS spectrometer: chopping or frequency switching?

Comments: The best thing to do is to consider the relative merits of each mode convolved with the detector limitation. The relevant points are offered below:

• Consider the transients. If you are worried about transients, you are probably better off using wavelength switching. The flux step involved in chopped measurements means a change in line+continuum flux. The flux step involved in frequency switching is line flux only. However, note that the very high telescope background may make this point mute. Thus, if you have sources where the continuum dominates (ie weak lines) you may be better off using frequency switching.
• Consider the line width. Frequency switching is useless for lines that are broader than 2 x the resolution element.
• Consider mapping or off-position options. If neither of the two options seems feasible (ie chopping does not reach far enough and line is broad or confused), consider mapping or concatenated AORs to get a larger offset. A 2x1 raster or an off position may be the only way to acquire a useful baseline.
• In frequency switching mode, one completely looses the continuum information. Thus, this mode is only good for line detections/fluxes.

Suggestion: NHSC is planning to provide a flowchart diagram that helps guide them to the recommended option.

Question: What is the recommended minimum number of repetitions for the Nyquist SED sampling mode for PACS spectroscopy?

Answer: Section 4.2.2.3 of the PACS observer manual http://herschel.esac.esa.int/Docs/PACS/html/ch04s02.html#sec-rangespec) states "It has been realized in instrument ground tests, that the current implementation in HSpot of this mode is not satisfactory : the time spent per grating position is too low. As a consequence the user is asked to enter a repetition factor of at least two for a proper measurement. Hence the total duration of an SED observation with two concatenated AORs (SED red and SED blue) shall take slighty more than one hour."

The NHSC, therefore, advises observers interested in the SED Nyquist mode to use a minimum repetition factor of 2.