ConditionalTurningBandsParameters

evo.compute.tasks.geostatistics.conditional_turning_bands.ConditionalTurningBandsParameters

Parameters for the conditional turning-band simulation task.

Performs a conditional block turning-bands simulation using a pre-existing continuous distribution object. The task outputs an ensemble attribute on the target grid, containing one column per realization.

Example: >>> params = ConditionalTurningBandsParameters( ... source=pointset, ... source_attribute="locations.attributes[?name=='grade']", ... target=grid, ... distribution=distribution_object, ... variogram_model=variogram, ... neighborhood=SearchNeighborhood( ... ellipsoid=Ellipsoid(ranges=EllipsoidRanges(70, 70, 5)), ... max_samples=40, ... ), ... block_discretization=BlockDiscretization(nx=5, ny=5, nz=5), ... realizations=10, ... )

source

source: GeoscienceObjectReference

Reference to the pointset containing the source conditioning points.

source_attribute

source_attribute: str

Attribute reference for the source values (e.g. "locations.attributes[?name=='grade']").

target

target: GeoscienceObjectReference

Reference to the target 3-D grid or masked grid to simulate onto.

filter

filter: Filter | None = None

Optional filter restricting simulation to a subset of target-grid locations.

source_filter

source_filter: Filter | None = None

Optional filter restricting conditioning to a subset of the source data.

neighborhood

neighborhood: SearchNeighborhood

Search neighbourhood used both for simulation and for the conditioning kriging step.

distribution

distribution: GeoscienceObjectReference

Reference to a non-parametric continuous cumulative distribution object.

Typically created by the continuous-distribution task. The distribution is used for the normal-score back-transformation of simulation results.

variogram_model

variogram_model: GeoscienceObjectReference

Reference to the variogram model used to model spatial covariance.

kriging_method

kriging_method: Literal['simple', 'ordinary'] = 'simple'

The kriging method for the conditioning step.

"simple" (default) — assumes a known constant mean.
"ordinary" — estimates the local mean from nearby samples.

block_discretization

block_discretization: BlockDiscretization = Field(default_factory=BlockDiscretization)

Sub-block discretisation for support correction.

Each grid cell is subdivided into nx * ny * nz sub-cells, simulated individually, and then averaged to the block scale. Defaults to BlockDiscretization(nx=1, ny=1, nz=1) (point simulation at cell centres).

number_of_lines

number_of_lines: int = Field(500, ge=1, le=1000)

Number of turning-band lines.

Higher values produce more accurate results at the cost of runtime. Must be between 1 and 1000. Defaults to 500.

realizations

realizations: int = Field(1, ge=1, le=100)

Number of simulation realizations to produce.

All realizations are saved to the ensemble attribute on the target object. Must be between 1 and 100. Defaults to 1.

random_seed

random_seed: int = 38239342

Random seed for reproducible simulations.