opendrift.readers.reader_schism_native

Attributes

logger

Classes

Reader	A reader for unstructured (irregularily gridded) SCHISM files.
ReaderBlockUnstruct	Class to store and interpolate the data from an unstructured reader.

Module Contents

opendrift.readers.reader_schism_native.logger

class opendrift.readers.reader_schism_native.Reader(filename=None, name=None, proj4=None, use_3d=None)

Bases: opendrift.readers.basereader.BaseReader, opendrift.readers.basereader.UnstructuredReader

A reader for unstructured (irregularily gridded) SCHISM files.

Args:

param filename:

a single SCHISM netcdf output file, or a pattern of files. The netCDF file can also be an URL to an OPeNDAP server.

type filename:

string, required.

param name:

Name of reader

type name:

string, optional

param proj4:

PROJ.4 string describing projection of data.

type proj4:

string, optional

param use_3d:

switch to use 3d flows (if available)

type use_3d:

boolean, optional

See also

py:mod:opendrift.readers.basereader.unstructured.

Common constructor for all readers

interpolation = 'linearNDFast'

convolve = None

return_block = True

use_3d = None

x

y

reader_KDtree

boundary

variable_mapping

variables = []

xmin

xmax

ymin

ymax

var_block_before

var_block_after

get_variables(requested_variables, time=None, x=None, y=None, z=None, block=False)

The function extracts ‘requested_variables’ from the native SCHISM files which will then be used in _get_variables_interpolated_() to initialise the ReaderBlockUnstruct objects used to interpolate data in space and time

For now the function will extract the entire slice of data of ‘requested_variables’ at given ‘time’

There is an option to extract only a subset of data around particles clouds to have less data but it means we need to recompute the KDtree of the subset nodes every time in ReaderBlockUnstruct. (Speed gain to be tested)

convert_3d_to_array(id_time, data, variable_dict)

The function reshapes a data matrix of dimensions = [node,vertical_levels] (i.e. data at vertical levels, at given time step) into a one-column array and works out corresponding 3d coordinates [lon,lat,z] using the time-varying ‘zcor’ variable in SCHISM files (i.e. vertical level positions).

These 3D coordinates will be used to build the 3D KDtree for data interpolation and will be added to the ‘variable_dict’ which is eventually passed to get_variables_interpolated().

args:

-id_time -data -variable_dict

out :

-flattened ‘data’ array -addition of [‘x_3d’,’y_3d’,’z_3d’] items to variable_dict if needed.

_get_variables_interpolated_(variables, profiles, profiles_depth, time, reader_x, reader_y, z)

This method _must_ be implemented by every reader. Usually by subclassing one of the reader types (e.g. structured.StructuredReader).

Arguments are in _native projection_ of reader.

set_convolution_kernel(convolve)

Set a convolution kernel or kernel size (of array of ones) used by get_variables on read variables.

__convolve_block__(env)

Convolve arrays with a kernel, if reader.convolve is set

covers_positions_xy(x, y, z=0)

Check which points are within boundary of mesh. Wrapper function of covers_positions() from unstructured.py which is called in get_variables_interpolated_xy() function from variables.py It returns indices of in-mesh points, and in-mesh point coordinates rather than a boolean array (inside/outside) Within get_variables_interpolated_xy() from variables.py, data is queried for these in-mesh points only and the full array (incl. out of mesh positions) is re-generated with correct masking

plot_mesh(variable=None, vmin=None, vmax=None, filename=None, title=None, buffer=1, lscale='auto', plot_time=None)

Plot geographical coverage of reader.

class opendrift.readers.reader_schism_native.ReaderBlockUnstruct(data_dict, KDtree=None, interpolation_horizontal='linearNDFast', interpolation_vertical='linear')

Class to store and interpolate the data from an unstructured reader. This is the equivalent of ReaderBlock (regular grid) for unstructured grids.

arguments: (in addition to ReaderBlock)

KDtreefor nearest-neighbor search (initialized using SCHISM nodes in reader’s _init_() )

This is read from reader object, so that it is not recomputed every time

logger

x

y

time

data_dict

block_KDtree = None

_initialize_interpolator(x, y, z=None)

interpolate(x, y, z=None, variables=None, profiles=[], profiles_depth=None)

_interpolate_horizontal_layers(data, nearest=False)

Interpolate all layers of 3d (or 2d) array.

covers_positions(x, y, z=None)

Check if given positions are covered by this reader block.