Level 1 spectra are both frequency and intensity calibrated, although some effects of both the observatory and the observing mode are still to be removed. Level 1 is considered by the HIFI ICC to be as far as automatic processing of data is "safe", it is expected that human interaction above level 1 will improve the results of the pipeline.

CheckDataStructure, CheckFreqGrid, CheckPhases. 

The initial three steps of the level 1 pipeline are checks that the data have the structure, content and form the patterns that are expected from the observing mode. The CheckFreqGrid step also groups datasets according to LO (local oscillator) tuning and checks for any possible frequency drifts in the data.

MkFluxHotCold. 

Hot/Cold load measurements are used to obtain both, the receiver (or system) temperature and the bandpass for each "hc" dataset in the original timeline product and included in a calibration product (of type CalFluxHotCold). The bandpass is used for the intensity calibration, the system temperature and, possibly, for the determination of the channel-dependent weights to be included in the spectra. If channel weights are not to be determined by the system temperature, then this step of the pipeline can be moved to anywhere before the DoFluxHotCold step.

DoChannelWeights. 

Fill values into the weights column, per subband. The default behaviour is to make use of the receiver temperature from MkFluxHotCold, other possibilities include weighting by integration time, or by the variance of the spectra within a given (moving) window.

DoRefSubtract. 

Reference measurements taken from blank sky (in DBS modes), from an internal load (in Load Chop modes) or taken at a different LO frequency (in Frequency Switch modes) are subtracted from the source measurements (science datasets) in order to eliminate instrumental drifts from the source measurements.

This step constitutes one half of the double subtraction scheme typical for HIFI and by the end of it, the science datasets are replaced by the differenced spectra. For the Frequency Switch modes, the shifted and the un-shifted spectra are overlayed (with opposite signs) by DoRefSubtract, these spectra need to be 'folded'.

MkOffSmooth. 

Average and smooth (on the frequency scale) the flux data from the OFF measurements. The calculation is processed on a per dataset basis so that a baseline is constructed for each OFF dataset. These baselines will be subtracted in the DoOffSubtract step.

The default operation of this step is to first take the average over all the spectra included in the OFF dataset and then to smooth the data using a Gaussian filter (a box filter is an alternative). Other options available are to apply a polynomial fit after averaging the OFF datasets, or to only take the average of the OFF datasets. There are many options available when taking the average and these are described in the 'MkOffSmooth' section in the HIFI Pipeline Specification document.

DoOffSubtract. 

The calibrated baseline(s) calculated in the MkOffSmooth step are subtracted from the ON measurements of load chop, frequency switch, and position switch modes (on a row-by-row basis for the position switch modes). In the case of DBS modes, the ON and OFF positions, which both contain science data, are averaged on a scan-by-scan basis.

DoFluxHotCold. 

The calibrated intensity scale obtained in the MkFluxHotCold task - the bandpass - is applied to the flux data. This transforms the intensity scale to Kelvin units (Ta'). All science data (dataset with type 'science') found in the given product are adjusted in this way.

For frequency switch modes, a bandpass is available at both LO frequencies (separated by the LO throw). This makes it possible to consider calibration schemes in which the division of the flux by the bandpass is carried through for each LO frequency separately, i.e. before the DoRefSubtract step. The current calibration scheme applies the DoFluxHotCold after the DoRefSubtract and the bandpass with the same LO frequency as the source phase is used.

DoVelocityCorrection. 

Corrects the frequency scale for the velocity of the spacecraft and possibly of the source. A relativistic approach is adopted when correcting for the motion of the spacecraft relative to SSB or LSR or, in the case of SSOs, relative to the SSO. For non-SSOs, the motion of the sources relative to the LSR or SSB is treated classically.

Possible target rest frames to transform to are "HSO" (Herschel Space Observatory), "GEOCENTRIC", "SSB", or "BARYCENTRIC", "LSR" or "SOURCE". By default, the task transforms to the "LSR" frame for non SSO's and "SOURCE" for SSO's. A description of the definitions of these frames and their relation to each other is Chapter 19.

An alternative task, which uses a non-relativistic approach and is not a default pipeline step, is DoRadialVelocity