2.13.2.3. pycsamt.stratagem.process#

stratagem.process#

Data-processing classes for Stratagem AMT surveys.

StaticShiftCorrector

Estimates and removes the static-shift effect from apparent-resistivity curves using the Adaptive Moving-Average (AMA) spatial filter (Kouadio et al., 2024; Torres-Verdín & Bostick, 1992). Delegates to estimate_ss_ama() and apply_ss_factors().

NoiseRemover

Multi-stage noise-removal pipeline: powerline notch filter → Hampel outlier filter → optional log-frequency smoothing. Delegates to pycsamt.emtools.remove_noise.

Both classes follow the fit() → out() pattern: fit() applies corrections in-place on the Z arrays of the supplied EDIFile objects, stores the result in edi_objects_, and returns self for chaining. out() either returns the processed objects or writes them to a directory.

Note

Because emtools processing functions modify impedance tensors in-place, passing the same EDIFile list through multiple processors sequentially is safe. If you need to preserve the originals, pass copy=True to fit().

Classes

NoiseRemover(*[, mains_hz, n_harm, tol_hz, ...])

Multi-stage noise-removal pipeline for Stratagem AMT data.

StaticShiftCorrector(*[, sort_by, ...])

Estimate and remove static-shift from Stratagem AMT impedance data.
class pycsamt.stratagem.process.NoiseRemover(*, mains_hz=50.0, n_harm=30, tol_hz=0.08, notch_mode='interp', hampel_win=3, hampel_nsig=3.0, smooth=False, smooth_win=3, verbose=0)[source]

Bases: PyCSAMTObject

Multi-stage noise-removal pipeline for Stratagem AMT data.

Applies three sequential filters to the impedance tensor:

  1. Powerline notch — masks (interpolates) the mains frequency and its harmonics. Controlled by mains_hz and n_harm.

  2. Hampel outlier filter — identifies and replaces frequency-domain spike outliers using a median-absolute-deviation test.

  3. Log-frequency smoothing (optional) — applies a triangular or Gaussian kernel along the log-frequency axis.

All corrections are applied in-place on EDIFile.Z.z.

Parameters:

  • mains_hz (float, default 50.0) – Mains frequency (Hz). Use 60.0 for North American data.

  • n_harm (int, default 30) – Number of powerline harmonics to notch.

  • tol_hz (float, default 0.08) – Frequency tolerance (Hz) around each harmonic for the notch filter.

  • notch_mode ({‘interp’, ‘zero’, ‘nan’}, default 'interp') – How to handle notched bins: 'interp' interpolates across them (recommended), 'nan' flags them as missing.

  • hampel_win (int, default 3) – Half-window size for the Hampel outlier filter (in frequency bins).

  • hampel_nsig (float, default 3.0) – Outlier threshold in units of median absolute deviation.

  • smooth (bool, default False) – Enable log-frequency smoothing (stage 3).

  • smooth_win (int, default 3) – Smoothing half-window. Values above 4 may trigger a known shape issue in smooth_logfreq() for short frequency vectors; keep ≤ 3 unless you have verified your data.

  • verbose (int, default 0)

Variables:

edi_objects (list of EDIFile) – Denoised EDI objects (in-place modified unless copy=True).

Examples

>>> from pycsamt.stratagem.process import NoiseRemover
>>> nr = NoiseRemover(mains_hz=50.0, smooth=True, smooth_win=3)
>>> nr.fit(edis)
>>> paths = nr.out("2/2EDID")
fit(edi_objects, *, copy=False)[source]

Apply the noise-removal pipeline.

Parameters:

  • edi_objects (list of EDIFile)

  • copy (bool, default False) – Deep-copy Z data before processing.

Return type:

self

out(savepath=None, *, overwrite=False)[source]

Return denoised EDI objects or write them to savepath.

Parameters:

  • savepath (path-like, optional)

  • overwrite (bool, default False)

Return type:

list of EDIFile or list of Path

class pycsamt.stratagem.process.StaticShiftCorrector(*, sort_by='lon', half_window=3, weights='tri', pband=None, max_skew=6.0, verbose=0)[source]

Bases: PyCSAMTObject, MetadataMixin

Estimate and remove static-shift from Stratagem AMT impedance data.

Implements the AMA (Adaptive Moving-Average) spatial filter to estimate per-station static-shift factors and correct the impedance tensor amplitudes accordingly.

The correction is applied in-place on EDIFile.Z.z. Use copy=True in fit() to preserve originals.

Parameters:

  • sort_by ({‘lon’, ‘lat’, ‘name’}, default 'lon') – Spatial ordering of stations for the AMA spatial average. Use 'lon' for E-W profiles, 'lat' for N-S profiles.

  • half_window (int, default 3) – Number of neighbour stations on each side used in the AMA spatial average.

  • weights ({‘tri’, ‘gauss’, ‘uniform’}, default 'tri') – Distance-weighting scheme for AMA neighbours.

  • pband (tuple of (float, float), optional) – Period range (T_min, T_max) in seconds used when estimating the shift factor. Useful for restricting the estimation to a band free of near-surface distortion.

  • max_skew (float or None, default 6.0) – Phase-tensor skew threshold: stations with median |β| above this are excluded from the spatial average (strong 3-D distortion). Set to None to disable.

  • verbose (int, default 0)

Variables:

  • factors (pandas.DataFrame) – Per-station shift factors with columns station, delta_log10_rho, fac_rho, fac_z, n_used.

  • edi_objects (list of EDIFile) – Corrected EDI objects (in-place modified unless copy=True).

Examples

>>> from pycsamt.stratagem.process import StaticShiftCorrector
>>> sc = StaticShiftCorrector(sort_by="lon", half_window=3).fit(edis)
>>> sc.factors_.head()
>>> paths = sc.out("2/2EDISS")
fit(edi_objects, *, copy=False)[source]

Estimate and apply static-shift corrections.

Parameters:

  • edi_objects (list of EDIFile)

  • copy (bool, default False) – Deep-copy Z data before correcting.

Return type:

self

out(savepath=None, *, overwrite=False)[source]

Return corrected EDI objects or write them to savepath.

Parameters:

  • savepath (path-like, optional) – When None returns the list of EDIFile objects; otherwise writes files and returns a list of pathlib.Path.

  • overwrite (bool, default False)

Return type:

list of EDIFile or list of Path

On this page