Abstract base class for heterogeneous ambient states that are based on data from NetCDF files or an xarray Dataset.

Attributes¶

data_source: str or xarray.Dataset

The data or the file search pattern, should end with suffix ‘.nc’. One or many files.

ovars: list of str

The output variables

var2ncvar: dict

Mapping from variable names to variable names in the nc file

fixed_vars: dict

Uniform values for output variables, instead of reading from data

load_mode: str

The load mode, choices: preload, lazy, fly. preload loads all data during initialization, lazy lazy-loads the data using dask, and fly reads only states index and weights during initialization and then opens the relevant files again within the chunk calculation

time_format: str

The datetime parsing format string

sel: dict, optional

Subset selection via xr.Dataset.sel()

isel: dict, optional

Subset selection via xr.Dataset.isel()

weight_factor: float

The factor to multiply the weights with

check_times: bool

Whether to check the time coordinates for consistency

check_input_nans: bool

Whether to check input data for NaNs

preprocess_nc: callable, optional

A function to preprocess the netcdf Dataset before use

interp_pars: dict

Additional parameters the interpolation

Public members¶

DatasetStates(data_source, output_vars, var2ncvar={}, ...)[source]

Constructor.

property data_source

The data source

preproc_first(algo, data, cmap, vars, bounds_extra_space, ...)[source]

Preprocesses the first file.

gen_states_split_size()[source]

Generator for suggested states split sizes for output writing.

load_data(algo, cmap, variables, bounds_extra_space=None, ...)[source]

Load and/or create all model data that is subject to chunking.

set_running(algo, data_stash, sel=None, isel=None, verbosity=0)[source]

Sets this model status to running, and moves all large data to stash.

unset_running(algo, data_stash, sel=None, isel=None, verbosity=0)[source]

Sets this model status to not running, recovering large data from stash

output_point_vars(algo)[source]

The variables which are being modified by the model.

size()[source]

The total number of states.

index()[source]

The index list

get_calc_data(mdata, cmap, variables)[source]

Gathers data for calculations.

interpolate_data(idims, icrds, d, pts, vrs, times)[source]

Interpolates data to points.

calculate(algo, mdata, fdata, tdata)[source]

The main model calculation.

reset(algo=None, states_sel=None, states_loc=None, verbosity=0)[source]

Reset the states, optionally select states

classmethod new(states_type, *args, **kwargs)[source]

Run-time states factory.

output_coords()[source]

Gets the coordinates of all output arrays

ensure_output_vars(algo, tdata)[source]

Ensures that the output variables are present in the target data.

run_calculation(algo, *data, out_vars, **calc_pars)[source]

Starts the model calculation in parallel.

__repr__()[source]

Return repr(self).

property model_id

Unique id based on the model type.

var(v)[source]

Creates a model specific variable name.

unvar(vnm)[source]

Translates model specific variable name to origninal variable name.

property initialized

Initialization flag.

sub_models()[source]

List of all sub-models

initialize(algo, verbosity=0, force=False)[source]

Initializes the model.

property running

Flag for currently running models

finalize(algo, verbosity=0)[source]

Finalizes the model.

get_data(variable, target, lookup='smfp', mdata=None, ...)[source]

Getter for a data entry in the model object or provided data sources