1.37. doFilterLoads

Full Name: herschel.hifi.pipeline.generic.DoFilterLoadsTask
Alias: doFilterLoads
Type: Java Task - Java Task
Import: from herschel.hifi.pipeline.generic import DoFilterLoadsTask
Category

HIFI/Pipeline/Level 1 Pipeline

Description

Task for the level1 pipeline for removing the standing waves from the hot/cold load measurements.

It should be included as additional step in the level1 pipeline algo before applying the mkFluxHotCold-task. It effectively removes standing waves from the loads counts in level 0.5 hot cold datasets (modifies the original htp).

The following processing steps are applied:

  • Compute an average OFF spectrum per LO frequency setting.

  • Divide hot cold load spectra (for the given LO frequency setting) by this average OFF spectrum.

  • Apply a suitable filter method to this ratio spectrum (filterMethod=cubic_splines or filterMethod=fft). Depending on the scheme chosen,

    • all hot and cold spectra (scheme1)

    • all cold load spectra but no hot load spectra (scheme2) or

    • all cold spectra but only central subbands of the hot load spectra (scheme3)

    are filtered.

  • Then the filtered ratios are multiplied by the average OFF spectrum which defines the modified load spectra.

This method proves to be successful depending on the detector band:

  • For Bands 1 and 2: Here, the scheme proves to be specially useful. It removes noticeably the 90-100 MHz standing waves seen when observing strong continuum sources. It also improves the intensity calibration for spectral lines for all observations.

  • For bands 3 and 4: Different scheme are available (using the scheme-parameter:

    • scheme=scheme1 (default): Removes the 90-100 MHz standing waves seen when observing strong continuum sources but only in the central subbands. The standing wave in the outer subbands will be degraded.

    • scheme=scheme2 : Removes the 90-100 MHz standing waves only in the cold load.

    • scheme=scheme3 : Removes the 90-100 MHz standing waves in the Cold load and in the central subbands of the hot load measurements.

  • For bands 5, 6 and 7: The three methods can be tried out but no real improvements are to be expected - on the contrary, even some degradation is expected.

Not optimized yet for FSW modes, but works. Will not work with LoadChopNoRef, FSwitchNoRef modes since no OFF measurements are available there. Could in principle be used with HRS, however scheme3 is not applicable there since the concept of 'central' bands does not exist for HRS.

Currently two methods to filter the ratio spectra are available:

  • fft

  • cubic_splines

These methods have limitations at the edges of the subbands and whenever there are strong IF features. (For example in band 1-2 at IF 4100 MHz).

Example

Example 1: In HIPE:
htp = doFilterLoads(htp=htp, scheme="scheme2", filterMethod="fft")

API details

Properties

HifiTimelineProduct htp [INOUT, MANDATORY, default=no default value]

The timeline product (observation) to be processed.

PipelineConfiguration params [INPUT, OPTIONAL, default=no default value]

Pipeline configuration parameters that can be passed to the task.

Boolean ignore [INPUT, OPTIONAL, default=no default value]

Flag to indicate whether the execution of the module should be ignored.

String scheme [INPUT, OPTIONAL, default=scheme1]

The scheme to be applied - indicating for which load spectra to apply the filtering. Available are "scheme1", "scheme2", "scheme3" - see details above.

String filterMethod [INPUT, OPTIONAL, default=cubic_splines]

The filter method to be applied. Available are "fft" and "cubic_splines".

Double splinePointsPerKnot [INPUT, OPTIONAL, default=140 / resolution]

The number of points per knot used in the definition for the spline interpolation scheme. In case the parameter relativeFilterParams is set to True the number of points is computed by the formula floor( splinePointsPerKnot / resolution ) where resolution is the instrument resolution obtained from the meta data.

Double fftFilterMin [INPUT, OPTIONAL, default=resolution * 0.7 / 0.03]

The lower bound of the range that is cut out of the Fourier spectrum when using the 'fft' as filter method. In case the parameter relativeFilterParams is set to True this bound is computed by floor( fftFilterMin * resolution ) where resolution is the instrument resolution obtained from the meta data.

Double fftFilterMax [INPUT, OPTIONAL, default=resolution * 1.3 / 0.03]

The upper bound of the range that is cut out of the Fourier spectrum when using the 'fft' as filter method. In case the parameter relativeFilterParams is set to True this bound is computed by floor( fftFilterMax * resolution ) where resolution is the instrument resolution obtained from the meta data.

Boolean relativeFilterParams [INPUT, OPTIONAL, default=True]

If set to True the parameters splinePointsPerKnot, fftFilterMin or fftFilterMax are expressed relative to the instrument resolution that is looked from the meta data: splinePointsPerKnot inversely proportional to resolution and fftFilterMin or fftFilterMax proportional to resolution.

MapContext calibration [INPUT, OPTIONAL, default=no default value]

Parameter to pass a calibration context from which all the calibration input, here the splinePointsPerKnot, fftFilterMin, fftFilterMax and relativeFilterParamscan be retrieved.

See also

History

  • 2011-07-01 - melchior: : initial version.
  • 2012-03-14 - melchior: : added parameters to configure the smoothing filter.