2.3. Spectroscopy

Here we give an overview of the science products contained within the levels of PACS spectroscopy observations.

2.3.1. Slicing in the pipeline

A single PACS spectroscopy observation contains, at each Level, red and blue camera data products, and each data product contains one or more "slices". Slicing is used by the pipeline to chunk the data into the separate wavelength ranges and pointings that the observing plan required, as it makes the pipeline processing more efficient. At each stage in the pipeline, for each new product created, all of the slices are held together in product contexts: for example, several individual Frames in a SlicedFrames, PacsCubes in a SlicedPacsCubes, etc.

The class of the containers of these slices is a type of ListContext: the ListContexts that contain the Frames, PacsCube, and PacsRebinnedCube slices are, respectively, SlicedFrames, SlicedPacsCube, SlicedPacsRebinnedCube, and for all the rest of the Level 2 products, the class is a ListContext.

To see a listing of the data ranges contained in each slice in a "slicedXXX", you can use the pipeline task slicedSummary to see something like this:

# For a SlicedFrames taken straight from the ObservationContext
HIPE> slicedFrames = obs.level0_5.blue.fitted.product
HIPE> slicedSummary(slicedFrames) 
noSlices: 5
noCalSlices: 1
noScienceSlices: 4
slice#  isScience  onSource  offSource  rasterId lineId band  dimensions  wavelengths 
0   false  no    no     0 0   [0,1]  ["B2A","B3A"]   [18,25,1680]  50.389 - 76.325  
1   true   yes   no     0 0   [2]    ["B2A"]         [18,25,1500]  71.932 - 73.922  
2   true   no    yes    0 0   [2]    ["B2A"]         [18,25,1500]  71.932 - 73.922  
3   true   no    yes    0 0   [2]    ["B2A"]         [18,25,1500]  71.932 - 73.922  
4   true   yes   no     0 0   [2]    ["B2A"]         [18,25,1500]  71.932 - 73.922  

The contents "slicedXXX" contexts also contain organisational and other information: a History, Meta data, and for some also a "MasterBlockTable".

[Note] Note
A single "BlockTable" is essentially a table of the data blocks, where a "block" is a group of datapoints taken while the instrument was in a particular setting: a particular wavelength range, or nod position, or position in a raster pointing pattern, or repetition number, for example. A BlockTable is the organisation table for a singleFrames or PacsCube, and the MasterBlockTable is the concatenation of all the individual BlockTables of the Frames orPacsCubes slices. The blocktable is useful for those who wish to interact with their data on a level even deeper than running an interactive pipeline, but is unlikely to be of interest to anyone else

For those interacting with PACS products via the HSA-download on disc, the effect of having a sliced pipeline is that the products are contained within directories that serve as the contexts: within each context-directory there will be one FITS file that serves as the context translator for HIPE, and a subdirectory that contains the actual cubes or tables with the science data. These science FITS files and will have, near the end of the filename, an "s_##" that indicates the slice number. Each slice can be a different wavelength range or pointing. If you are not working within HIPE (where these information can be found in the Observation viewer or via a typed command) then to find out which slice is what within the observation means looking at the keywords of the FITS file (first extension). The Meta data/keywords to look for can be found in Chapter 4.

2.3.2. Level 0 to 1

The lowest-level products are of interest only to the person re-reducing their data or wanting to check on the SPG processing that was done at the HSA. Level 0 data are unprocessed, Level 0.5 data have been corrected for all instrumental effects and have sky coordinates and wavelengths, and Level 1 products have been flatfielded, de-glitched, and spectrally regridded, but are not fully flux calibrated (this is done at Level 2). To see these levels from an observation loaded in the Observation viewer in HIPE, click on + next to the"level0/0_5/1" in the Data panel and you will get a listing of their contents, e.g.,

The level 0 and 0.5 contents for an ObservationContext produced by SPG 13 and higher

Figure 2.5. The level 0 and 0.5 contents for an ObservationContext produced by SPG 13 and higher


The level 0.5 and 1 contents for an ObservationContext produced by SPG 13 and higher

Figure 2.6. The level 0.5 and 1 contents for an ObservationContext produced by SPG 13 and higher


Each entry in these screenshots is a context containing other products. For example, many individual Herschel Pacs Spectroscopy Fitted Frames Red products are held in a context called HPSFITR, and Herschel Pacs Spectroscopy (3D) cubes Red (PacsCube) are in a context called HPS3DR. The individual Frames and PacsCube products held in these contexts are different parts of the same observation: these are the "slices" mentioned previously.

In the Data tab, each of these Frames and PacsCubes has a title that indicates what is unique about it: in the screenshot above you will see the several Level 0.5 Frames listed under HPSFITR have:

  • a sequence digit (0—6)

  • a spectral line sequencer ("L 101/102")

  • a nod-sequence number (N1,2,3)

  • a nodding position A and B: for chopNod observations these refer to the chopper position, for unchopped observations they refer on-source (B) or off-source (A)

  • a raster position, here R(0 0) which is raster row 0 column 0

The differences between levels are (i) at Level 1 the calibration block slice has been removed, hence there is one more slice at Level 0.5 than at Level 1 (ii) towards the end of the pipeline the nods and then the individual raster pointings (for mapping observations) are combined, so the number of slices is again reduced. In the screenshot above there are 6 science Frames: 1 line id, 1 pointing, 3 nod sequences, and 2 nods in each. Hover with the cursor over any of the objects shown in the screenshot, and a banner comes up telling you what class they are and what type of information are contained therein.

These Frames and PacsCube at Level 1 have units of microns along the wavelength grid, and the RA and Dec have been assigned to each of the 16 pixels of the 25 modules/each of the 25 spaxels. The data are, however, not yet flux calibrated. The calibration to Jy is applied at the end of Level 2.

The other contexts in the screenshot are more science contexts: HPSFITB are the blue equivalent to the HPSFITR, and HPS3DB are blue equivalent to the HPS3DR.

The other layers contain engineering data (HPSENG), satellite housekeeping data (HPSGENK and HPSHK), DecMec data (HPSDMC[R|B]) which are used in some of the masking pipeline tasks, and data of the central detector pixel which is downlinked in raw format (HPSRAW).

In Table 2.1, the products at the Levels 0 to 1 are given.

Table 2.1. The contents of Levels 0, 0.5, and 1 of an ObservationContext

Level Context Class Contains Provided for what type of observation?
ObservationContext, 0, HPSXXX of level 0.5, 1, 2, 2.5 History HistoryDataset A history of tasks that were run on the particular object the History is attached to All
0 HPENG a context A table of engineering data All
0 HPGENHK a context A table of satellite housekeeping data All
0, 0.5 HPSDMC[R|B] a context A table of DecMec data, used in the pipeline masking tasks All
0, 0.5, 1 HPSFIT[R|B] SlicedFrames Individual Frames, which contain all the signals gathered from the astronomical source and the two calibration sources. At level 0.5 it is "sliced" by nod position, nod cycle, wavelength, pointing: one Frames becomes many SlicedFrames, each containing one or more Frames. All
0 HPSHK a context Herschel PACS Spectroscopy housekeeping information table All
0 HPSRAW[R|B] SlicedRawFrames Raw data Frames, which contain data only in one spatial pixel, and only used in the saturation pipeline task All
1, 2 HPS3D[R|B] SlicedPacsCube The first cubes produced by the pipeline; all 3 axes are non-equidistant and spectra axis unique to each spaxel All

2.3.3. Level 2 and higher

The final pipeline products for the end-user are found in Level 2, 2.5, and 3. The science-use products are found here: see Chapter 6 for a guide to which particular science-use product(s) is the most useful to focus on, as well as the prerequisites for their use.

The products in these higher levels are various types of cubes and spectral tables. To see the products in the level, click on + next to the "level2":

The Level 2 contents for a spectroscopy ObservationContext produced by SPG 13 (taken from the same observation as in the figures above). The PacsCubes (HPS3D[R|B]) and the PacsRebinnedCubes (HPS3DR[R|B]) are present for both cameras. The mosaic cubes (drizzled, interpolated, projected) provided for the red and blue differ, because for this observation the mapping was oversampled in the red (4.5" steps in a 2x2 raster) but is considered undersampled in the blue (where steps of 3" in a raster of 3x3 is required). Therefore, projected cubes (HPS3DP[R|B]) are present in both cameras and additionally there is a blue interpolated cube (HPS3DI[R|B]) and a red drizzled cube (HPS3DE[R|B]).

Figure 2.7. The Level 2 contents for a spectroscopy ObservationContext produced by SPG 13 (taken from the same observation as in the figures above). The PacsCubes (HPS3D[R|B]) and the PacsRebinnedCubes (HPS3DR[R|B]) are present for both cameras. The mosaic cubes (drizzled, interpolated, projected) provided for the red and blue differ, because for this observation the mapping was oversampled in the red (4.5" steps in a 2x2 raster) but is considered undersampled in the blue (where steps of 3" in a raster of 3x3 is required). Therefore, projected cubes (HPS3DP[R|B]) are present in both cameras and additionally there is a blue interpolated cube (HPS3DI[R|B]) and a red drizzled cube (HPS3DE[R|B]).


The Level 2 and 3 contents for a spectroscopy pointed observation produced by SPG 14.

Figure 2.8. The Level 2 and 3 contents for a spectroscopy pointed observation produced by SPG 14.


In Table 2.2, the products at the Levels 2, 2.5, and 3 are given. At Level 2 are the science-quality end-of-pipeline cubes and tables for most observations. For most unchopped range scan observations these products are found in In Level 2.5 instead. Why? In the AOT for unchopped range scan observations, the on-source observation (the target) and off-source observation (the "background") are two separate obsids. They are each individually processed and their Level 2s are filled. The off-source data are then subtracted from the on-source data and the resulting science-grade products placed in the Level 2.5 of the on-source observation. These products have the same name as the Level 2 cubes, but with a "BS" (background subtracted) appended. If the observer did not request an off-source observation, then only Level 2 is present in both the on- and off-source obsids. Finally, there is a Level 3 for all chopNod pointed observations that were taken as SEDs (i.e. include the full spectral range of PACS).

Table 2.2. The contents of Levels 2, 2.5, and 3 of an ObservationContext

Level Context Class Contains Provided in SPG 14 for what type of observation?
ObservationContext, 0, HPSXXX of level 0.5, 1, 2, 2.5 History HistoryDataset A history of tasks that were run on the particular object the History is attached to All
1, 2 HPS3D[R|B] SlicedPacsCube The first cubes produced by the pipeline; all 3 axes are non-equidistant and the wavelengths in each spaxel are not gridded All
2 HPS3DR[R|B] SicedPacsRebinnedCube The second cubes produced by the pipeline, and one possible pipeline end-product to adopt; all 3 axes are non-equidistant and the spectral axis now gridded and the same for each spaxel. Spaxels are 9.4" and the spatial footprint is that of the native instrument All
2.5 HPS3DRBS[R|B] SicedPacsRebinnedCube The same as above but these cubes have been background-subtracted (see main text) All unchopped range scan AOTs
2 HPS3DP[R|B] and HPS3DEQP[R|B] SlicedPacsProjectedCube, also ListContext of SpectralSimpleCube Projected cubes, a type of "mosaic" cube, produced primarily for mapping observations. These are created from the rebinned cubes; are one possible pipeline end-product to adopt. Spatial axes are regular but spectral grid still non-equidistant. "EQ": the same cubes resampled to a an equidistant spectral grid For all mapping (Nyquist, oversampled, tiling) observations. Also for pointed observations, but not recommended for science measurements. The equidistant version is provided for the Nyquist and oversampled range scan observations
2.5 HPS3DPBS[R|B] and HPS3DEQPBS[R|B] SlicedPacsProjectedCube, ListContext of SpectralSimpleCube The same as above but these cubes have been background-subtracted (see main text) All unchopped range scan mapping (Nyquist, oversampled, tiling) observations. The equidistant version is provided for the Nyquist and oversampled range scan observations
2 HPS3DD[R|B] and HPS3DEQD[R|B] ListContext of SpectralSimpleCubes Drizzled cubes, a type of "mosaic" cube produced only for mapping observations. Are created from the Level 2 PacsCubes; are one possible pipeline end-product to adopt. Spatial axes are regular but spectral grid still non-equidistant. "EQ": the same cubes resampled to a an equidistant spectral grid For Nyquist and oversampled mapping observations for line scans. The equidistant version is provided for the Nyquist and oversampled line scan observations
2 HPS3DI[R|B] and HPS3DEQI[R|B] ListContext of SpectralSimpleCubes Interpolated cubes, a type of "mosaic" cube produced for pointed and mapping observations. They are created from the rebinned cubes; are one possible pipeline end-product to adopt. Spatial axes are regular but spectral grid still non-equidistant. "EQ": the same cubes resampled to a an equidistant spectral grid For pointed, Nyquist mapping and tiling observations. The equidistant version is provided for the tiling and pointed observations
2.5 HPS3DIBS[R|B] and HPS3DEQIBS[R|B] ListContext of SpectralSimpleCubes The same as above but these cubes have been background-subtracted (see main text) For all unchopped range scan pointed, Nyquist mapping and tiling observations. The equidistant version is provided for the tiling and pointed observations
2 HPSTBR[R|B] SlicedPacsSpecTable The data in the SlicedPacsRebinnedCubes presented as a table all
2.5 HPSTBRBS[R|B] SlicedPacsSpecTable The same as above but containing the background-subtracted data All unchopped range scan AOTs
2 HPSSPEC[R|B] SlicedPacsCentralSpectrum A table containing the spectrum of the central spaxel and point-source calibrated spectra, the output of the task extractCentralSpectrum: c1, c9, c129 for chopNod observations and c1, c9 for unchopped observations (see chp. 8 of the PDRG) For all pointed observations
2.5 HPSSPECBS[R|B] SlicedPacsCentralSpectrum The same as above but containing the background-subtracted data (see main text) For all unchopped range scan pointed observations
3 HPSSPEC SlicedPacsCentralSpectrum The data from the HPSSPECR and HPSSPECB joined into a single table, and moreover including the data from all observations that were taken to cover the entire PACS SED For all chopNod pointed observations observed in SED mode

The range of products provided depends on the observing mode (the AOT). Some products are provided for all observations, but the mosaic cubes and some of the tables are provided only for some observations.

  • All observations have PacsCubes (HPS3D[R|B]), PacsRebinnedCubes (HPS3DR[R|B]), and PacsSpecTables (HPSTBR[R|B]) at Level 2; unchopped range scan on-source observations usually also have the rebinned cubes, with "BS" appended to the name, at Level 2.5.

  • All pointed observations have the PacsCentralSpectrum tables (HPSSPEC[R|B]) at provided at Level 2, and unchopped range scans usually have the "BS" version at Level 2.5.

  • All chop-nod pointed observations containing the full spectral range of PACS ("SED" AOT) have a single Level 3 PacsCentralSpectrum table (HPSSPEC), created from concatenated Level 2 tables from multiple obsids.

  • Of the three "mosaic" cubes listed in Table 2.2 (projected, drizzled, and interpolated), which ones are provided for any observation depends on the pointing mode (pointed or mapping) and the type of raster if the mode is mapping. In Table 2.3 the types of cubes provided for the different pointing modes is given. In Table 2.4 the raster settings which were recommended to the observers for the various mapping modes are given, and observations that followed these recommendations will have the related mosaic cubes provided in the ObservationContext.

    The algorithms used to create the projected, drizzled, and interpolated mosaic cubes are documented in their individual PACS URM entries and in the PDRG chp. 1 and chp. 9. Both references are recommended as further reading since they contain important information about how the cubes are built and how the flags, NaNs, and errors are propagated. The drizzled cubes are the most sophisticated, but are provided only for lineScan observations. The projected cubes are the next most sophisticated and are provided for all observations. The interpolated cubes are the least sophisticated but are good to use for pointed and tiling observations.

Two of the three mosaic cubes are provided for most observations, and additionally one of the cubes is provided with an equidistant wavelength grid (which are easier to read into external software than the standard cubes).

[Note] Note

By default, the dispersion of the wavelength grid in any PACS cube depends on the resolution, which changes with wavelength. To create cubes with an "equidistant" wavelength grid, one mosaic cube per observation is spectrally resampled on a linear ("equidistant") grid. These equidistant cubes are also provided as the standalone browse products (Chapter 5, and in more detail in the PDRG chp. 7).

Table 2.3. The mosaic cubes provided for the different pointing modes of PACS spectroscopy AOTs

Pointing mode (spectral mode) Cubes: HPS3Dxxx Details
Pointed ..I[R|B] @L2; ..IBS[R|B] @L2.5 Interpolated cubes, 3" spaxels on a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
  ..EQI[R|B] @L2; ..EQIBS[R|B] @L2.5 Interpolated cubes, 3" spaxels on a regular spatial grid; a regular and equidistant spectral grid
  ..P[R|B] @L2; ..PBS[R|B] @L2.5 Projected cubes, 0.5" spaxels on a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
Mapping: undersampled/tiling ..I[R|B] @L2; ..IBS[R|B] @L2.5 Interpolated cubes, 3" spaxels on a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
  ..EQI[R|B] @L2; ..EQIBS[R|B] @L2.5 Interpolated cubes, 3" spaxels on a regular spatial grid; a regular and equidistant spectral grid
  ..P[R|B] @L2; ..PBS[R|B] @L2.5 Projected cubes, 1.5" spaxels with a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
Mapping: Nyquist (range scans) ..P[R|B] @L2; ..PBS[R|B] @L2.5 Projected cubes, 3" spaxels with a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
  ..EQP[R|B] @L2; ..EQPBS[R|B] @L2.5 Projected cubes, 3" spaxels with a regular spatial grid; a regular and equidistant spectral grid
  ..I[R|B] @L2; ..IBS[R|B] @L2.5 Interpolated cubes, 3" spaxels on a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
Mapping: Nyquist (line scans) ..D[R|B] @L2 Drizzled cubes, small spaxels (size is optimised for the central wavelength) on a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
  ..EQD[R|B] @L2 Drizzled cubes, small spaxels (size is optimised for the central wavelength) on a regular spatial grid; a regular and equidistant spectral grid
  ..P[R|B] @L2; PBS[R|B] @L2.5 Projected cubes, same spatial grid as the drizzled cubes; a regular, but non-equidistant spectral grid (scales with wavelength)
Mapping: oversampled (range scans) ..P[R|B] @L2; ..PBS[R|B] @L2.5 Projected cubes, 3" spaxels with a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
  ..EQP[R|B] @L2; EQPBS[R|B] @L2.5 Projected cubes, 3" spaxels with a regular spatial grid; a regular and equidistant spectral grid
Mapping: oversampled (line scans) ..D[R|B] @L2 Drizzled cubes, small spaxels (size is optimised for the central wavelength) on a regular spatial grid; a regular, but non-equidistant spectral grid (scales with wavelength)
  ..EQD[R|B] @L2 Drizzled cubes, small spaxels (size is optimised for the central wavelength) on a regular spatial grid; A regular and equidistant spectral grid
  ..P[R|B] @L2; ..PBS[R|B] @L2.5 Projected cubes, same spatial grid as the drizzled cube a regular, but non-equidistant spectral grid (scales with wavelength)

The details defining the mapping modes are given in Table 2.4, and they are are inclusive: for example, to be considered a Nyquist map, the red camera data must have both a raster of 2x2 and steps sizes between 4.5" and 24". Any observation that has a smaller raster pattern or exceeds the step sizes is considered to be undersampled.

[Note] Note
A consequence of this is that observations that were taken to be Nyquist or oversampled in the red camera, and have a 2x2 raster pattern, will always be considered undersampled in the blue camera as the required raster pattern is less than 3x3. The types of cubes provided for the red and blue cameras may therefore be different (as indeed is the case in Figure 2.7 above).

Table 2.4. The observing parameters for the different mapping modes. The number and size of the steps need to be fulfilled for an observation to be considered to have the indicated mapping mode, and both keywords for each row must have the indicated value.

Undersampled/Tiling Nyquist Oversampled Keywords Meta data
Number of steps less than 2 (red) or 3 (blue) equal to or greater than 2 (red) or 3 (blue) equal to or greater than 2 (red) or 3 (blue) NRASTLIN and NRASTCOL numRasterLines and numRasterCol
Step size greater than 24" (red) or 16" (blue) 4.5"—24" (red) or 3"—16" (blue) up to 4.5" (red) or 3" (blue) RASSTEPL and RASSTEPC lineStep and pointStep

Finally, in Table 2.5 the Meta data and FITS keywords that can be found in any Level 2 or higher product which allows you to identify the AOT of the observation are given: these can be used to find out if an observation is: chop-nod or unchopped, line scan or range scan, pointed or mapping (all of these choices are the same for all cubes in any one observatoin), and in addition the spectral range, slice number, and raster Meta data are listed.

Table 2.5. The Meta data and FITS keywords in the Level 2 and higher cubes and tables to inspect to find the observing mode of any observation. The value of the keywords indicate what the specific mode is, i.e. the value of "aot" tells you whether the observation is a line or a range scan

Line or range scan? Chop-nod or unchopped? Pointed or mapping? Full SED? Mean wavelength Slice number Raster line,column sequence
Meta datum aot, cusMode aotMode aotMode, obsMode rangeSPOT (is "SED" in the value?) meanWavelength / lineDescription sliceNumber rasterId
FITS keyword AOT, CUSMODE AOTMODE AOTMODE, OBS_MODE RANGDESC (is "SED" in the value?) MEANWAVE / LINEINFO SLICENUM RASID

2.3.4. Missing products

Sometimes one or both cameras are missing from the Levels 0.5 and higher, or some of the slices in several of the products contexts are missing. If the observation is not classified as FAILED (see the quality/qualitySummary, and this information is also given in the HSA search-results panel), then this usually occurs when entire spectral ranges have been masked out. The affected slices cannot be processed to the final pipeline level and are therefore disgarded. If all the slices are so-affected, then the entire context for that camera is disgarded. This occurs first in the Level 0 to 0.5 processing, when the wavelength information is added and during which the OutOfBand (OOB) mask is created: spectral ranges that fall outside of the spectral band are masked as bad, this typically happening when the prime range was defined in one camera and the associated parallel range in the other camera lies (fully or partially) outside of a band. The second task in which data can be disgarded is during the flatfielding (Level 1 to 2 pipeline), when the data in the light leak regions are masked out (the NOTFFD mask): again, if an entire slice is masked out then that entire slice will be disgarded. Hence, if your observation only has products from one camera, or you cannot find the wavelength slice you expected, first check whether there is a quality comment (in the qualitySummary and also in the pacsObsSummary) about these data having been masked out.

Table 2.6. Spectral range limits: outside of these ranges the data are masked out

Band OutOfBand (OOB) Light leak regions (NOTFFD)
R1 101.6 - 220.2 102.0 - 190.0
B2A 51.5 - 72.0 55.0 - 73.0
B3A 53.0 - 72.0 55.0 - 70.0
B2B 68.0 - 105.0 70.0 - 95.0

2.3.5. Spectroscopy FITS files on disc

In Section 2.2.2 is a screenshot of a tarball gotten from the HSA and unpacked on disc. Here we will link the names of the FITS files with the name of the products as seen in HIPE. In general the FITS file names contain the obsid, the name of the context (e.g. HPS3DB for blue PacsCube), the pipeline level they are from, and a sequence number. Level 3 FITS files are slightly different, in that they include the RA and Dec rather than the obsid, and include also a version number.

In the screengrab below you can see the organisation of an unpacked tarball of a spectroscopy observation gotten from the HSA. The layout is similar to that when viewing an observation in HIPE: the top level is the obsid; then you have the various contexts containing the browseProduct, level0..3, quality, etc; within the various "level#" are the product contexts (HPSXXX), from within which, one level lower, lie the cubes and tables.

Since the spectroscopy pipeline uses slicing (Section 2.3.1), there is one FITS file per context that contains the slice-organsation information and one or more FITS files that are the actual cubes or tables. If you are grabbing data directly from disc, it is naturally these latter than you want. You need to go to the lowest directory level, which name is that of the product class (e.g. PacsRebinnedCube, SpectralSimpleCube, PacsSpecTable...), as shown in the screengrab.

All FITS files have a slice index in the name ("s_##"). Where only one slice is present in a context, it is clear that is the file you want. But where more than one slice is present (which is the case for all cubes from observations with multiple requested wavelength ranges, and for all rebinned cubes for the mapping observations), the only way you can tell which wavelength/raster position a FITS file is from is to open the file and inspect the FITS keywords: see Table 4.2 and Table 2.5.

In Table 2.7 we link the levels (2, 2.5, and 3) and product context names as seen in HIPE to the directories and filenames as seen on disc, so you can locate the cubes and tables discussed in this manual on disc as well as in HIPE.

In Table 2.8 the HIPE and FITS file names of the standalone browse products (SBPs) are given. These products are also available from the Levels 2, 2.5, or 3, so you can get them from both locations in the unpacked directory structure on disc. The Level 2.5 is provided for unchopped range scan observations only and the SBPs for the on-source obsid include the Level 2.5 and the Level 2 products, i.e. the background-subtracted and the pre-background-subtracted versions.

As pointed out in Section 2.3.3 and the tables therein (particularly Table 2.3), the range of products (the cubes and tables) that can be found in an observation depends on the AOT of the observation: line or range scan, if range scan is it SED or short range, pointed or mapping, chop-nod or unchopped? A certain amount of information about the AOT can be gleaned from the HSA: if you performed an advanced search you may have e.g. limited the search to mapping or pointed observations; the HSA search results panel includes information about whether the observation is a line or range scan. The postcard that is displayed in the HSA search results panel has text above the plot that indicates if the observation is unchopped or chop-nod, line or range, pointed or mapping. One can also open any Level 2 cube and inspect the FITS keywords that are indicated in Table 2.5.

Screengrabs of parts of a spectroscopy ObservationContext stored on disc, having been downloaded from the HSA as a tarball
Screengrabs of parts of a spectroscopy ObservationContext stored on disc, having been downloaded from the HSA as a tarball
Screengrabs of parts of a spectroscopy ObservationContext stored on disc, having been downloaded from the HSA as a tarball

Figure 2.9. Screengrabs of parts of a spectroscopy ObservationContext stored on disc, having been downloaded from the HSA as a tarball


Table 2.7. The science-grade PACS spectroscopy products as found in an unpacked tarball gotten from the HSA, for Levels 2, 2.5, and 3. [R|B] and [r|b] indicate a choice between the red and blue camera product/file. The "Directory" is the one in which the FITS file can be found, and in the unpacked directory structure this is located in [obsid]/levelX/HPSXXX, where level"X" and HPS"XXX" are the given in the product context column. The ## in the FITS filenames are a sequence indicator (slice number).

Level; Product context Directory FITS filename Description
2; HPS3DR[R|B] herschel.pacs.signal.PacsRebinnedCube hpacs[obsid]_20hps3dr[r|b]s_##_[long number].fits The PacsRebinnedCubes, one FITS file for each slice (each wavelength range, and raster position if a mapping observation)
2.5; HPS3DRBS[R|B] herschel.pacs.signal.PacsRebinnedCube hpacs[obsid]_25hps3drbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) PacsRebinnedCubes
2; HPS3DP[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_20hps3dp[r|b]s_##_[long number].fits The projected cubes, one FITS file for each slice (each wavelength range); provided for pointed and mapping observations
2.5; HPS3DPBS[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_25hps3dpbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) projected cubes
2; HPS3DI[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_20hps3di[r|b]s_##_[long number].fits The interpolated cubes, one FITS file for each slice (each wavelength range); provided for all pointed and undersampled mapping observations and for Nyquist-sampled mapping range scan observations
2.5; HPS3DIBS[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_25hps3dibs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) interpolated cubes
2; HPS3DD[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_20hps3dd[r|b]s_##_[long number].fits The drizzled cubes, one FITS file for each slice (each wavelength range); provided for line scan mapping observations
2.5; HPS3DDBS[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_25hps3ddbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) drizzled cubes
2; HPSTBR[R|B] herschel.pacs.signal.PacsSpecTable hpacs[obsid]_20hpstbr[r|b]s_##_[long number].fits The rebinned cubes but in tabular format, one FITS file for each slice (each wavelength range) but all raster pointings are in a single table
2.5; HPSTBRBS[R|B] herschel.pacs.signal.PacsSpecTable hpacs[obsid]_25hpstbrbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) rebinned cube tables
2; HPSSPEC[R|B] herschel.pacs.signal.PacsCentralSpectrum hpacs[obsid]_20hpsspec[r|b]s_##_[long number].fits Tables of the central and different point-source calibrated spectra (all taken from the central spaxel spectrum), one FITS file for each slice (each wavelength range); provided for all pointed observations
2.5; HPSSPECBS[R|B] herschel.pacs.signal.PacsCentralSpectrum hpacs[obsid]_25hpsspecbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) central and point-source spectra tables
3.0; HPSSPEC is located directly off level3/HPSSPEC hpacs[obsid]_30HPSSPEC_(RA)_[p|m](DEC)_v[##]_[long number].fits The combined central+point source spectrum tables from Level 2 from multiple obsids; provided for pointed chop-nod range scans taken in "SED" mode

Table 2.8. The PACS spectroscopy standalone browse products as found in an unpacked tarball gotten from the HSA. [R|B] and [r|b] indicate a choice between the red and blue camera product/file. The "Directory" is the one in which the FITS file can be found, and in the unpacked directory structure this is located in [obsid]/browseProduct/HPSXXX, where HPS"XXX" is given in the product context column. The ## in the FITS filenames are a sequence indicator (slice number). Note that the same products are also located in the Level 2/2.5/3 directories.

Originating Level; Product context Directory FITS filename Description
2; HPS3DEQP[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_20hps3dp[r|b]s_##_[long number].fits The projected cubes with an equidistant wavelength grid, one FITS file for each slice (each wavelength range); provided for pointed and mapping observations
2.5; HPS3DEQPBS[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_25hps3dpbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) equidistant projected cubes
2; HPS3DEQI[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_20hps3di[r|b]s_##_[long number].fits The interpolated cubes with an equidistant wavelength grid, one FITS file for each slice (each wavelength range); provided for all pointed and undersampled mapping observations and for Nyquist-sampled mapping range scan observations
2.5; HPS3DEQIBS[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_25hps3dibs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) equidistant interpolated cubes
2; HPS3DEQD[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_20hps3dd[r|b]s_##_[long number].fits The drizzled cubes with an equidistant wavelength grid, one FITS file for each slice (each wavelength range); provided for line scan mapping observations
2.5; HPS3DEQDBS[R|B] herschel.ia.dataset.spectrum.SpectralSimpleCube hpacs[obsid]_25hps3ddbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) equidistant drizzled cubes
2; HPSTBR[R|B] herschel.pacs.signal.PacsSpecTable hpacs[obsid]_20hpstbr[r|b]s_##_[long number].fits The rebinned cubes but in tabular format, one FITS file for each slice (each wavelength range)
2.5; HPSTBRBS[R|B] herschel.pacs.signal.PacsSpecTable hpacs[obsid]_25hpstbrbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) rebinned cube tables
2; HPSSPEC[R|B] herschel.pacs.signal.PacsCentralSpectrum hpacs[obsid]_20hpspec[r|b]s_##_[long number].fits Tables of the central and different point-source calibrated spectra (all taken from the central spaxel spectrum), one FITS file for each slice (each wavelength range); provided for all pointed observations
2.5; HPSSPECBS[R|B] herschel.pacs.signal.PacsCentralSpectrum hpacs[obsid]_25hpsspecbs[r|b]s_##_[long number].fits The background subtracted (for unchopped range scans) central and point-source spectra tables
3.0; HPSSPEC is located directly in HPSSPEC hpacs[obsid]_30HPSSPEC_(RA)_[p|m](DEC)_v[##]_[long number].fits The combined central+point source spectrum tables from Level 2 from multiple obsids; provided for pointed chop-nod range scans taken in "SED" mode

2.3.6. Unchopped range scan obsids: the on- and the off-source obsid

Unchopped range scan observations were usually done in pairs: an on-source obsid (pointed or mapping) and an off-source obsid (usually pointed). As mentioned before, where an off-source obsid is provided, this is subtracted from the on-source obsid and the "background subtracted" products are placed in the Level 2.5 of the on-source obsid. However, not all observers requested an off-source obsid. For AORs that were created in the latter part of the mission, observers could indicate in HSPOT whether the observation was to be on-source or off-source, and if the button was ticked, the Meta datum "OnOffPosition" would be filled in correctly—however, many observers ignored this button and hence it cannot be relied on. There is also a Meta datum "onOffSource", however it is usually not set correctly and can neither be relied on to identify the on- or off-source nature of an observation. The most simple way to identify whether any particular unchopped range scan observation is an on-source or off-source observation, other than the name of the AOR (which is provided in the HSA search results panel), is to see whether it is included on the list that was used to set which observations should be processed to Level 2.5: see the Herschel Data Products page. In addition, in the qualitySummary you will find a comment informing on the obsid of the off-source observation that was used by the SPG to subtract the background from an on-source observation.