4.4. Filter Variation Correction

The filter transmission of a detector can be slightly different for objects observed at different positions on the detector. The variation of filter transmissions affects photometric measurements and, consequently, it can also have an effect on the estimation of photometric redshifts.

Phosphoros is able to take into account the filter variation in the redshift computation by applying a correction in modeled fluxes. It assumes that the filter variation corresponds to a translation along the wavelength of filter transmission curves. The amplitude of the shift is defined as the difference between the actual and the nominal filter central wavelength, \(\Delta\lambda=\lambda_c^{actual}-\lambda_c^{nominal}\), where the actual one is the measured central wavelength of the filter trasmission associated with a specific detected source. This information is expected to be found in the input catalog, as additional column, one for each photometric band in which filter variation is relevant or known.

Before estimating photometric redshifts, Phosphoros computes the correction due to filter variation that must be applied to modeled fluxes. In particular, a filter variation correction grid is generated through the following steps:

  • users define a set of values of \(\Delta\lambda\) for which flux corrections are computed (e.g., N points between -100 to 100 \(\mathring{\rm A}\));

  • for each model of the grid the flux correction is computed for each (selected) photometric band;

  • for each model and band the computed flux corrections are fitted as a function of the filter shift;

  • the coefficients of the fit are saved in the filter variation correction grid.

Consequently, through the filter variation correction grid, Phosphoros is able to compute in a fast way the filter variation flux corrections for each source of the input catalog.

4.4.1. Filter Variation Correction in the GUI

In the GUI, users have firstly to map filters to the catalog columns that contain the filter shift. Go to the Catalog Setup panel and click on the Filter Shift control (see the left panel of Fig. 4.3): a new column will appear in the Filter Mapping table where users can insert the name of the filter shift column. After saving, the Filter Shift Column is included in the filter_mapping.txt file (see Filter Mapping Format).

The filter variation correction grid is then computed at the step 2 (Grids Generation) of the Compute Redshifts panel (see the right panel of Fig. 4.3), and stored in the directory:

> $PHOSPHOROS_ROOT/IntermediateProducts/<Catalog Type>/FilterVariationCoefficientGrids

Users can select the range of the filter shift (in \(\mathring{\rm A}\)) and the number of sampling points. Clicking on Save Config. File, users can save the corresponding configuration file.

At this point, in the redshift computation, the flux correction due to filter variation will be applied in all the photometric bands for which the filter shift column has been provided.

../../_images/filter_var.png

Fig. 4.3 Filter variation correction with the GUI

4.4.2. Filter Variation Correction in the CLI

In the CLI, the filter variation correction grid is generated by the Compute Filter Variation Coefficient Grid (or CFVCG) action. The range of the filter variation (i.e., the maximum negative/positive shift of the filter central wavelength) and the number of sampling points can be defined by the following options:

--filter-variation-min-shift=-100 (default value)
--filter-variation-max-shift=100 (default value)
--filter-variation-shift-samples=200 (default value)

where the shift is in Angstrom. The output grid name and the format can be choosen through:

--output-filter-variation-coefficient-grid=<name>
--output-filter-variation-coefficient-grid-format=BINARY or TEXT

More options can be found with the --help parameter, including model normalization options, Galactic extiction option, etc. (similarly to the compute_model_grid action).

Before running Phosphoros, users should ensure that the filter_mapping.txt file contains the filter variation column name of each band for which filter variation correction must be computed (see CLI: Mapping filters to column names). This operation can be easily done by the GUI (see above). Otherwise, users have to edit the file and include the information by hand (see Filter Mapping Format).

Finally, the path and the file name of the filter variation correction grid must be specified in the configuration file of the compute_redshift action, through the option:

--filter-variation-coefficient-grid-file=<path>/<name>