pygmt.filter1d

pygmt.filter1d(data, output_type='pandas', outfile=None, *, end=None, filter_type=None, time_col=None, **kwargs)[source]

Time domain filtering of 1-D data tables.

A general time domain filter for multiple column time series data. The user specifies which column is the time (i.e., the independent variable) via time_col. The fastest operation occurs when the input time series are equally spaced and have no gaps or outliers and the special options are not needed. Read a table and output as a numpy.ndarray, pandas.DataFrame, or ASCII file.

Full option list at https://docs.generic-mapping-tools.org/6.5/filter1d.html

Aliases:

  • E = end

  • F = filter_type

  • N = time_col

Parameters:
  • filter_type (str) –

    typewidth[+h]. Set the filter type. Choose among convolution and non-convolution filters. Append the filter code followed by the full filter width in same units as time column. By default, this performs a low-pass filtering; append +h to select high-pass filtering. Some filters allow for optional arguments and a modifier.

    Available convolution filter types are:

    • (b) Boxcar: All weights are equal.

    • (c) Cosine Arch: Weights follow a cosine arch curve.

    • (g) Gaussian: Weights are given by the Gaussian function.

    • (f) Custom: Instead of width give name of a one-column file with your own weight coefficients.

    Non-convolution filter types are:

    • (m) Median: Returns median value.

    • (p) Maximum likelihood probability (a mode estimator): Return modal value. If more than one mode is found we return their average value. Append +l or +u if you rather want to return the lowermost or uppermost of the modal values.

    • (l) Lower: Return the minimum of all values.

    • (L) Lower: Return minimum of all positive values only.

    • (u) Upper: Return maximum of all values.

    • (U) Upper: Return maximum of all negative values only.

    Upper case type B, C, G, M, P, F will use robust filter versions: i.e., replace outliers (2.5 L1 scale off median, using 1.4826 * median absolute deviation [MAD]) with median during filtering.

    In the case of L|U it is possible that no data passes the initial sign test; in that case the filter will return 0.0. Apart from custom coefficients (f), the other filters may accept variable filter widths by passing width as a two-column time-series file with filter widths in the second column. The filter-width file does not need to be co-registered with the data as we obtain the required filter width at each output location via interpolation. For multi-segment data files the filter file must either have the same number of segments or just a single segment to be used for all data segments.

  • end (bool) – Include ends of time series in output. The default [False] loses half the filter-width of data at each end.

  • time_col (int) – Indicate which column contains the independent variable (time). The left-most column is 0, while the right-most is (n_cols - 1) [Default is 0].

  • output_type (str) –

    Determine the format the xyz data will be returned in [Default is pandas]:

  • outfile (str) – The file name for the output ASCII file.

Returns:

ret (pandas.DataFrame or numpy.ndarray or None) – Return type depends on outfile and output_type: