The rasterstats
python module provides a fast and flexible tool to summarize geospatial raster datasets based on vector geometries (i.e. zonal statistics).
- Raster data support
- Any raster data source supported by GDAL
- Support for continuous and categorical
- Respects null/no-data metadata or takes argument
- Vector data support
- Points, Lines, Polygon and Multi-* geometries
- Flexible input formats
- Any vector data source supported by OGR
- Python objects that are geojson-like mappings or support the geo_interface
- Well-Known Text/Binary (WKT/WKB) geometries
- Depends on GDAL, Shapely and numpy
Using ubuntu 12.04:
sudo apt-get install python-numpy python-gdal
pip install rasterstats
Given a polygon vector layer and a digitial elevation model (DEM) raster, calculate the mean elevation of each polygon:
>>> from rasterstats import zonal_stats
>>> stats = zonal_stats("tests/data/polygons.shp", "tests/data/elevation.tif")
>>> stats[1].keys()
['__fid__', 'count', 'min', 'max', 'mean']
>>> [(f['__fid__'], f['mean']) for f in stats]
[(1, 756.6057470703125), (2, 114.660084635416666)]
The zonalstats
command line utility is useful for interoperating with other software that speaks GeoJSON. The input and output of zonalstats
are both GeoJSON FeatureCollections with the output containing additional fields for the aggregate raster statistics overlapping the geometry:
# Find mean rainfall of each state
zonalstats states.geojson states_w_rainfall.geojson --stats="mean" -r rainfall.tif
Rather than deal with with files, you can use the default input and output (stdin and stdout) to pipe data:
ogr2ogr -f GeoJSON /vsistdout/ states.shp | zonalstats -r rainfall.tif > states_w_rainfall.geojson
By default, the zonal_stats
function will return the following statistics
- min
- max
- mean
- count
Optionally, these statistics are also available
- sum
- std
- median
- majority
- minority
- unique
- range
- percentile (see note below for details)
You can specify the statistics to calculate using the stats
argument:
>>> stats = zonal_stats("tests/data/polygons.shp",
"tests/data/elevation.tif",
stats=['min', 'max', 'median', 'majority', 'sum'])
>>> # also takes space-delimited string
>>> stats = zonal_stats("tests/data/polygons.shp",
"tests/data/elevation.tif",
stats="min max median majority sum")
Note that certain statistics (majority, minority, and unique) require significantly more processing due to expensive counting of unique occurences for each pixel value.
You can also use a percentile statistic by specifying percentile_<q>
where <q>
can be a floating point number between 0 and 100.
You can define your own aggregate functions using the add_stats
argument. This is a dictionary with the name(s) of your statistic as keys and the function(s) as values. For example, to reimplement the mean statistic:
from __future__ import division
import numpy as np
def mymean(x):
return np.ma.mean(x)
then use it in your zonal_stats
call like so:
stats = zonal_stats(vector, raster, add_stats={'mymean':mymean})
In addition to the basic usage above, rasterstats supports other mechanisms of specifying vector geometries.
It integrates with other python objects that support the geo_interface (e.g. Fiona, Shapely, ArcPy, PyShp, GeoDjango):
>>> import fiona
>>> # an iterable of objects with geo_interface
>>> lyr = fiona.open('/path/to/vector.shp')
>>> features = (x for x in lyr if x['properties']['state'] == 'CT')
>>> zonal_stats(features, '/path/to/elevation.tif')
...
>>> # a single object with a geo_interface
>>> lyr = fiona.open('/path/to/vector.shp')
>>> zonal_stats(lyr.next(), '/path/to/elevation.tif')
...
Or by using with geometries in "Well-Known" formats:
>>> zonal_stats('POINT(-124 42)', '/path/to/elevation.tif')
...
By default, an __fid__ property is added to each feature's results. None of the other feature attributes/proprties are copied over unless copy_properties
is set to True:
>>> stats = zonal_stats("tests/data/polygons.shp",
"tests/data/elevation.tif"
copy_properties=True)
>>> stats[0].has_key('name') # name field from original shapefile is retained
True
There is no right or wrong way to rasterize a vector. The default strategy is to include all pixels along the line render path (for lines), or cells where the center point is within the polygon (for polygons).
Alternatively, you can opt for the ALL_TOUCHED
strategy which rasterizes the geometry by including all pixels that it touches. You can enable this specifying:
>>> zonal_stats(..., all_touched=True)
Both approaches are valid and there are tradeoffs to consider. Using the default rasterizer may miss polygons that are smaller than your cell size resulting in None
stats for those geometries. Using the ALL_TOUCHED
strategy includes many cells along the edges that may not be representative of the geometry and may give severly biased results in some cases.
You can treat rasters as categorical (i.e. raster values represent discrete classes) if you're only interested in the counts of unique pixel values.
For example, you may have a raster vegetation dataset and want to summarize vegetation by polygon. Statistics such as mean, median, sum, etc. don't make much sense in this context (What's the sum of oak + grassland
?).
The polygon below is comprised of 12 pixels of oak (raster value 32) and 78 pixels of grassland (raster value 33):
>>> zonal_stats(lyr.next(), '/path/to/vegetation.tif', categorical=True)
>>> [{'__fid__': 1, 32: 12, 33: 78}]
Keep in mind that rasterstats just reports on the pixel values as keys; It is up to the programmer to associate the pixel values with their appropriate meaning (e.g. oak
is key 32
) for reporting.
Internally, we create a masked raster dataset for each feature in order to calculate statistics. Optionally, we can include these data in the output of zonal_stats
using the raster_out
argument:
stats = zonal_stats(vector, raster, raster_out=True)
Which gives us three additional keys for each feature:
``mini_raster`` : Numpy ndarray
``mini_raster_GT`` : Six-tuple defining the geotransform (GDAL ordering)
``mini_raster_NDV`` : Nodata value in the returned array
Keep in mind that having ndarrays in your stats dictionary means it is more difficult to serialize to json and other text formats.
Find a bug? Report it via github issues by providing
- a link to download the smallest possible raster and vector dataset necessary to reproduce the error
- python code or command to reproduce the error
- information on your environment: versions of python, gdal and numpy and system memory