Source code for pygmt.sampling

"""
GMT modules for Sampling of 1-D and 2-D Data
"""
import pandas as pd

from .clib import Session
from .helpers import (
    build_arg_string,
    fmt_docstring,
    GMTTempFile,
    data_kind,
    dummy_context,
    use_alias,
)
from .exceptions import GMTInvalidInput


[docs]@fmt_docstring @use_alias(n="interpolation", V="verbose") def grdtrack(points, grid, newcolname=None, outfile=None, **kwargs): """ Sample grids at specified (x,y) locations. Grdtrack reads one or more grid files and a table with (x,y) [or (lon,lat)] positions in the first two columns (more columns may be present). It interpolates the grid(s) at the positions in the table and writes out the table with the interpolated values added as (one or more) new columns. A bicubic [Default], bilinear, B-spline or nearest-neighbor interpolation is used, requiring boundary conditions at the limits of the region (see *interpolation*; Default uses “natural” conditions (second partial derivative normal to edge is zero) unless the grid is automatically recognized as periodic.) Full option list at :gmt-docs:`grdtrack.html` {aliases} Parameters ---------- points : pandas.DataFrame or str Either a table with (x, y) or (lon, lat) values in the first two columns, or a filename (e.g. csv, txt format). More columns may be present. grid : xarray.DataArray or str Gridded array from which to sample values from, or a filename (netcdf format). newcolname : str Required if 'points' is a pandas.DataFrame. The name for the new column in the track pandas.DataFrame table where the sampled values will be placed. outfile : str Required if 'points' is a file. The file name for the output ASCII file. {V} {n} Returns ------- track: pandas.DataFrame or None Return type depends on whether the outfile parameter is set: - pandas.DataFrame table with (x, y, ..., newcolname) if outfile is not set - None if outfile is set (track output will be stored in outfile) """ with GMTTempFile(suffix=".csv") as tmpfile: with Session() as lib: # Store the pandas.DataFrame points table in virtualfile if data_kind(points) == "matrix": if newcolname is None: raise GMTInvalidInput("Please pass in a str to 'newcolname'") table_context = lib.virtualfile_from_matrix(points.values) elif data_kind(points) == "file": if outfile is None: raise GMTInvalidInput("Please pass in a str to 'outfile'") table_context = dummy_context(points) else: raise GMTInvalidInput(f"Unrecognized data type {type(points)}") # Store the xarray.DataArray grid in virtualfile if data_kind(grid) == "grid": grid_context = lib.virtualfile_from_grid(grid) elif data_kind(grid) == "file": grid_context = dummy_context(grid) else: raise GMTInvalidInput(f"Unrecognized data type {type(grid)}") # Run grdtrack on the temporary (csv) points table # and (netcdf) grid virtualfile with table_context as csvfile: with grid_context as grdfile: kwargs.update({"G": grdfile}) if outfile is None: # Output to tmpfile if outfile is not set outfile = tmpfile.name arg_str = " ".join( [csvfile, build_arg_string(kwargs), "->" + outfile] ) lib.call_module(module="grdtrack", args=arg_str) # Read temporary csv output to a pandas table if outfile == tmpfile.name: # if user did not set outfile, return pd.DataFrame column_names = points.columns.to_list() + [newcolname] result = pd.read_csv(tmpfile.name, sep="\t", names=column_names) elif outfile != tmpfile.name: # return None if outfile set, output in outfile result = None return result