There is a Herschel-wide convention on processing levels of the different instruments. There are different classes of products created at the Level 0, 0.5, 1 and 2 of the pipeline. A more detailed product description is given in the PACS Data Reduction Guide (PDRG) and also the Herschel Products Definitions Document (HPDD), both of which are available via the HIPE help. Here we give you a summary of these information. For all the products described here the red and blue ranges are separate products. They are identified by an R or a B at the end of the product name. The product names start with HPS, which stands for Herschel Pacs Spectroscopy.
Level 0 data: Level 0 products are complete sets of data as a starting point for scientific data reduction. Level 0 products may reach a data volume of 200 MB/hour. For all standard observing modes, the scientific information is contained in the structures called HPSFITB/R (objects of the Frames class), which constitute the starting point for the pipeline. These contain all measured slopes described in section 2.4.5, i.e. the 18 x 25 x time-line slopes for one camera. Correct interpretation of the signal requires knowledge of the instrument status at every time. That can be found in sub-structures of these Frames, but level0 products also contain many useful additional information, like spacecraft pointing, time correlation, and selected spacecraft housekeeping information, raw data (i.e. un-fitted ramps) for some selected pixels, etc. Finally, level 0 contain the calibration data needed for data analysis. The fit-ramps products (HPSFITB/R) are of dimensions 18 x 25 x time-line, the averaged ramps products (HPSAVGB/R) are of dimensions 18 x 25 x time-line x 4. A fit-ramps HPSFITB/R product contains, for each pixel (18x25), one value for each ramp taken over the course of the observation; these values are the slopes that have been fit to each and every ramp. The averaged-ramps product contains, for each pixel, 4 values per ramp taken over the course of the observation; these being the set-of-4 averages taken for each ramp. The pipeline starts on the HPSFIT products.
Level 0.5 data : Processing until this level is AOT independent. Additional information like processing flags and masks (saturation, damaged pixel, signals affected by chopper and grating transitions) is added, block selections generated, basic unit conversions applied (digital readouts to Volts/s) and for the spectrometer the wavelength calibration (inclusive velocity correction) is done. Also the center of field coordinates are computed for every frame and sky coordinates are assigned for every pixel. The only products here are the fit-ramps, the HPSFITB/R, of dimensions of 18 x 25 x nb of slopes.
Level 1 data: The automatic data generation of level 1 products is partly AOT dependent. Detector readouts ar flux calibrated and converted to physical units, in principle instrument and observatory independent. Level 1 processing includes the flux calibration and adds further status information to the product (e.g. chopper angle, masks, etc.). These are the biggest products in the processing chain and may reach 2GB/h in the case of spectroscopy. The spectroscopy Level 1 product (PacsCube - HPS3DB/R) contains fully calibrated 5 x 5 x nb of slopes cubes per pointing/spectral range.
Level 2 data: Further processed level-1 data to such a level that scientific analysis can be performed. Processing to this level contains actual spectra and is highly AOT dependent. For optimal results many of the processing steps involved to generate level-2 data may require human interaction, based both on instrument understanding as well as understanding of the scientific aims of the observation. These data products are at browse quality level and should be suitable for publishing following an interactive enhancement.
The level 2 data product contains noise filtered, regularly sampled data cubes HPS3DRB/R of dimensions lambda x 5 x 5; and a combined cube projected on the WCS: HPS3DPB/R of dimensions lambda x N x M, where 'N' and 'M' are the spatial dimensions of the projected map. This cube is constructed by rebinning the integral field cube (oversampled in wavelength, different projection per wavelength layer due to distortions) onto the same sky / wavelength grid of the instrumental resolution. The cube flux values are Jy/pixel, the wavelength is in microns.
In case of mapping observations (rasters), the pipeline loops over all raster positions and combines the rasters into a single grid by adding up for each new spaxel the fluxes of the contributing old spaxels, multiplied by their overlap weights. The HPS3DP product is worth using even for pointed observations (i.e. no rastering) because it does not just add together, or mosaic, multiple pointings, but also sets the correct spatial grid for each wavelength of your cube: for the PACS spectrometer, each wavelength sees a slightly different spatial position, even for spectra within a single spaxel.
Level 3 data: These are the publishable science products where level-2 data products are used as input. These products are not only from the specific instrument, but are usually combined with theoretical models, other observations, laboratory data, catalogues, etc. Their formats should be VO compatible and these data products should be suitable for VO access.