jacobianAddPolyfit

Workspace.jacobianAddPolyfit(self: pyarts.arts._Workspace, jacobian_quantities: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.ArrayOfRetrievalQuantity]] = self.jacobian_quantities, jacobian_agenda: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Agenda]] = self.jacobian_agenda, sensor_response_pol_grid: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.ArrayOfIndex]] = self.sensor_response_pol_grid, sensor_response_dlos_grid: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Matrix]] = self.sensor_response_dlos_grid, sensor_pos: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Matrix]] = self.sensor_pos, poly_order: Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Index], no_pol_variation: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Index]] = 0, no_los_variation: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Index]] = 0, no_mblock_variation: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Index]] = 0, verbosity: Optional[Union[pyarts.arts.WorkspaceVariable, pyarts.arts.Verbosity]] = self.verbosity) → None

Includes polynomial baseline fit in the Jacobian.

This method deals with retrieval of disturbances of the spectra that can be described by an additive term, a baseline off-set.

The baseline off-set is here modelled as a polynomial. The polynomial spans the complete frequency range spanned by sensor_response_f_grid and the method should only of interest for cases with no frequency gap in the spectra. The default assumption is that the off-set differs between all spectra, but it can also be assumed that the off-set is common for all e.g. line-of-sights.

If the simulation/retrieval deals with a single spectrum, the number of elements added to the state vector (x) is poly_order+1. The coefficient for polynomial order 0 comes first etc. The same is true if no_pol_variation, no_los_variation and no_mblock_variation all are set to 1, even if several spectra are involved. Otherwise thenumber of elements added to x depends on the number of spectra and the settings of no_pol_variation, no_los_variation and no_mblock_variation. The coefficients of the different polynomial orders are treated as separate retrieval quantities. That is, the the elements associated with polynomial order 0 are grouped and form together a retrieval quantity. The coefficients for higher polynomial orders are treated in the same way.

Author(s): Patrick Eriksson

Parameters:

  • jacobian_quantities (ArrayOfRetrievalQuantity, optional) – The retrieval quantities in the Jacobian matrix. See jacobian_quantities, defaults to self.jacobian_quantities [INOUT]

  • jacobian_agenda (Agenda, optional) – Pure numerical Jacobian calculations. See jacobian_agenda, defaults to self.jacobian_agenda [INOUT]

  • sensor_response_pol_grid (ArrayOfIndex, optional) – The “polarisation grid” associated with sensor_response. See sensor_response_pol_grid, defaults to self.sensor_response_pol_grid [IN]

  • sensor_response_dlos_grid (Matrix, optional) – The zenith and azimuth angles associated with sensor_response. See sensor_response_dlos_grid, defaults to self.sensor_response_dlos_grid [IN]

  • sensor_pos (Matrix, optional) – The sensor position for each measurement block. See sensor_pos, defaults to self.sensor_pos [IN]

  • poly_order (Index) – Polynomial order to use for the fit. [IN]

  • no_pol_variation (Index, optional) – Set to 1 if the baseline off-set is the same for all Stokes components. Defaults to 0 [IN]

  • no_los_variation (Index, optional) – Set to 1 if the baseline off-set is the same for all line-of-sights (inside each measurement block). Defaults to 0 [IN]

  • no_mblock_variation (Index, optional) – Set to 1 if the baseline off-set is the same for all measurement blocks. Defaults to 0 [IN]

  • verbosity (Verbosity) – ARTS verbosity. See verbosity, defaults to self.verbosity [IN]