Zonal statistics

Draft: written, not yet checked by a person Operation · zonal-statistics

Summary statistics of raster cell values inside each zone polygon.

Pitfall. Zones smaller than a cell may get no cells at all unless all_touched is set.

Pitfall. Overlapping zones are handled differently by different engines; some rasterize them and lose the overlap.

About this operation
Also called	Zonal Statistics as Table, zonal stats, summarize raster within polygons
Group	`raster/statistics`
Checks	`crs.agreement, raster.nodata, raster.stats_sampled`
Pitfalls	2 listed above
Columbus tool	`raster_op · zonal_stats`
Notes	Not stated
Status	active
Broader	Not stated
Narrower	Not stated
Related	Not stated

Parameters

Name	Type	Required	Default	Values	Note
`stats`	list	no	Not stated	`count, mean, min, max, sum, std`	Not stated
`band`	integer	no	Not stated	Not stated	Not stated
`all_touched`	boolean	no	Not stated	Not stated	Count every cell a zone touches, not only cells whose centre is inside.

Inputs and outputs

Direction	Name	Type	Geometry	Note
Input	`raster`	raster	Not stated	Not stated
Input	`zones`	vector	polygon	Not stated
Output	`output`	vector	Not stated	Not stated

In each engine

Engine	Tool	Parameters	Differences
`gdal`	`gdal raster zonal-stats`	`stats` → `stat` : count = count; mean = mean; min = min; max = max; sum = sum; std = stdev `band` → `band` `all_touched` → `pixels` : true = all-touched; false = default `raster` → `input` `zones` → `zones` `output` → `output`	Also offers fractional cell coverage (pixels fractional) and weighted statistics.
`grass`	`v.rast.stats`	`stats` → `method` : count = number; mean = average; min = minimum; max = maximum; sum = sum; std = stddev `raster` → `raster` `zones` → `map`	Writes the statistics as new columns in the zone map's own attribute table; there is no separate output. Preset: `{"column_prefix":"zonal"}`
`qgis`	`native:zonalstatisticsfb`	`stats` → `STATISTICS` : count = 0; sum = 1; mean = 2; std = 4; min = 5; max = 6 `band` → `RASTER_BAND` `raster` → `INPUT_RASTER` `zones` → `INPUT` `output` → `OUTPUT`	Output columns are named with COLUMN_PREFIX (default "_"), e.g. _mean.
`rasterstats`	`rasterstats.gen_zonal_stats`	`stats` → `stats` : count = count; mean = mean; min = min; max = max; sum = sum; std = std `band` → `band` `all_touched` → `all_touched` `raster` → `raster` `zones` → `vectors`	A generator of one dict per zone; rasterstats.zonal_stats takes the same arguments and returns a list.
`terra`	`terra::zonal`	`stats` → `fun` : count = notNA; mean = mean; min = min; max = max; sum = sum; std = sd `all_touched` → `touches` `raster` → `x` `zones` → `z`	Zones can be a raster of zone codes or polygons (SpatVector); results come back as a data frame. std is computed by R's sd (sample standard deviation).
`whiteboxtools`	`ZonalStatistics`	`stats` → `stat` : mean = mean; min = minimum; max = maximum; std = standard deviation; sum = total. not supported: count `raster` → `input` `zones` → `features` : A raster of zone ids, not polygons: rasterize the zones onto the input's grid first. `output` → `output`	Zones are a raster, not polygons, and it computes one statistic per run.
`xarray-spatial`	`xrspatial.zonal_stats`	`stats` → `stats_funcs` : count = count; mean = mean; min = min; max = max; sum = sum; std = std `raster` → `values` `zones` → `zones`	Zones are a raster of zone ids on the same grid as values (polygons need rasterizing first); returns a pandas DataFrame per zone id.