aggregate

aggregate

quark.aggregate

Core aggregation engine: reproject and accumulate N-dimensional xarray Datasets onto a target projection grid.

Iteration 1 (current): - In-memory mode only (vars_batch_size must be None) - No supersampling (supersampling must be 1) - naive sum_method only (“kahan” accepted by interface, raises NotImplementedError) - Full NDIMS support: variables may be 2-D, 3-D, 4-D … (lat/lon rasters are always 2-D; extra dims are preserved in output)

Classes

Name	Description
Aggregator	Stateful aggregation pipeline.

Aggregator

aggregate.Aggregator(
    datasets,
    projection,
    lat_name='latitude',
    lon_name='longitude',
    variables=None,
    fail_on_schema_mismatch=True,
    sum_method='simple',
    skipna=True,
    supersampler=None,
    return_counts=False,
    return_sums=False,
    dtype=None,
)

Stateful aggregation pipeline.

Encapsulates projection, configuration, variable metadata, and accumulators. Eliminates the need to pass dozens of arguments through helper functions.

Methods

Name	Description
compute	Process all datasets and return the aggregated result.
process_dataset	Process one dataset: project and accumulate all variables.

compute

aggregate.Aggregator.compute()

Process all datasets and return the aggregated result.

Returns

Name	Type	Description
	xr.Dataset	Aggregated dataset with coordinates and attributes

process_dataset

aggregate.Aggregator.process_dataset(ds)

Process one dataset: project and accumulate all variables.

Parameters

Name	Type	Description	Default
ds	xr.Dataset	Source dataset to process	required