vibespatial.raster.zonal¶

Zonal statistics via CCCL segmented reduce.

Computes per-zone aggregate statistics (count, sum, mean, min, max, std, median) over raster values using CCCL segmented reduce primitives.

GPU path: sort by zone_id (CCCL radix_sort), find segment boundaries, then CCCL segmented_reduce for each requested statistic. All computation stays on-device until the final DataFrame transfer.

CPU path: numpy sort + segment loop (baseline).

Functions¶

`zonal_stats`(→ pandas.DataFrame)	Compute zonal statistics of values raster over zone labels.
`zonal_stats_gdf`(→ pandas.DataFrame)	Compute zonal statistics of a raster over vector zone boundaries.

Module Contents¶

vibespatial.raster.zonal.zonal_stats(zones: vibespatial.raster.buffers.OwnedRasterArray, values: vibespatial.raster.buffers.OwnedRasterArray, stats: vibespatial.raster.buffers.ZonalSpec | list[str] | None = None, *, use_gpu: bool | None = None) → pandas.DataFrame¶

Compute zonal statistics of values raster over zone labels.

Parameters¶

zonesOwnedRasterArray: Integer label raster defining zones (must be single-band). Each unique nonzero value is a zone.
valuesOwnedRasterArray: Value raster to aggregate. Must have same height/width as zones. May be single-band or multiband. For multiband values, per-band statistics are computed and returned with band_N_ prefixed column names (e.g. band_1_mean, band_2_mean).
statsZonalSpec or list[str] or None: Statistics to compute. Default: count, sum, mean, min, max.
use_gpubool or None: Force GPU (True), force CPU (False), or auto-dispatch (None). Auto uses GPU when CuPy is available and pixel count >= 100k.

Returns¶

pd.DataFrame: One row per zone. For single-band values, columns are zone plus each requested statistic name. For multiband values, columns are zone plus band_N_<stat> for each band and statistic.

vibespatial.raster.zonal.zonal_stats_gdf(gdf, values: vibespatial.raster.buffers.OwnedRasterArray, stats: vibespatial.raster.buffers.ZonalSpec | list[str] | None = None, *, grid_spec: vibespatial.raster.buffers.GridSpec | None = None, use_gpu: bool | None = None) → pandas.DataFrame¶

Compute zonal statistics of a raster over vector zone boundaries.

Rasterizes the GeoDataFrame zones to a label grid matching the values raster, then runs zonal_stats on the label grid.

Parameters¶

gdfGeoDataFrame: Vector zones. Each row is a zone.
valuesOwnedRasterArray: Value raster to aggregate.
statsZonalSpec or list[str] or None: Statistics to compute.
grid_specGridSpec or None: Grid to rasterize zones onto. If None, matches the values raster grid.
use_gpubool or None: Force GPU (True), force CPU (False), or auto-dispatch (None).

Returns¶

pd.DataFrame: One row per zone with zone index and requested statistics.