Rasterio reads and writes geospatial raster datasets.

Rasterio employs GDAL under the hood for file I/O and raster formatting. Its functions typically accept and return Numpy ndarrays. Rasterio is designed to make working with geospatial raster data more productive and more fun.

Rasterio is pronounced raw-STIER-ee-oh.

Here's a simple example of the basic features rasterio provides. Three bands are read from an image and summed to produce something like a panchromatic band. This new band is then written to a new single band TIFF.

import numpy
import rasterio
import subprocess

# Register GDAL format drivers and configuration options with a
# context manager.

with rasterio.drivers(CPL_DEBUG=True):

    # Read raster bands directly to Numpy arrays.
    #
    with rasterio.open('tests/data/RGB.byte.tif') as src:
        b, g, r = src.read()

    # Combine arrays in place. Expecting that the sum will 
    # temporarily exceed the 8-bit integer range, initialize it as
    # 16-bit. Adding other arrays to it in-place converts those
    # arrays "up" and preserves the type of the total array.

    total = numpy.zeros(r.shape, dtype=rasterio.uint16)
    for band in r, g, b:
        total += band
    total /= 3

    # Write the product as a raster band to a new 8-bit file. For
    # keyword arguments, we start with the meta attributes of the
    # source file, but then change the band count to 1, set the
    # dtype to uint8, and specify LZW compression.

    kwargs = src.meta
    kwargs.update(
        dtype=rasterio.uint8,
        count=1,
        compress='lzw')

    with rasterio.open('example-total.tif', 'w', **kwargs) as dst:
        dst.write_band(1, total.astype(rasterio.uint8))

# At the end of the ``with rasterio.drivers()`` block, context
# manager exits and all drivers are de-registered.

# Dump out gdalinfo's report card and open the image.

info = subprocess.check_output(
    ['gdalinfo', '-stats', 'example-total.tif'])
print(info)
subprocess.call(['open', 'example-total.tif'])

The rasterio.drivers() function and context manager are new in 0.5. The example above shows the way to use it to register and de-register drivers in a deterministic and efficient way. Code written for rasterio 0.4 will continue to work: opened raster datasets may manage the global driver registry if no other manager is present.

Simple access is provided to properties of a geospatial raster file.

with rasterio.drivers():

    with rasterio.open('tests/data/RGB.byte.tif') as src:
        print(src.width, src.height)
        print(src.crs)
        print(src.affine)
        print(src.count)
        print(src.indexes)

# Output:
# (791, 718)
# {u'units': u'm', u'no_defs': True, u'ellps': u'WGS84', u'proj': u'utm', u'zone': 18}
# Affine(300.0379266750948, 0.0, 101985.0,
#        0.0, -300.041782729805, 2826915.0)
# 3
# [1, 2, 3]

Rasterio also affords conversion of GeoTIFFs to other formats.

with rasterio.drivers():

    rasterio.copy(
        'example-total.tif',
        'example-total.jpg', 
        driver='JPEG')

subprocess.call(['open', 'example-total.jpg'])

Rasterio's command line interface, named "rio", is documented at cli.rst. Its rio insp command opens the hood of any raster dataset so you can poke around using Python.

$ rio insp tests/data/RGB.byte.tif
Rasterio 0.10 Interactive Inspector (Python 3.4.1)
Type "src.meta", "src.read_band(1)", or "help(src)" for more information.
>>> src.name
'tests/data/RGB.byte.tif'
>>> src.closed
False
>>> src.shape
(718, 791)
>>> src.crs
{'init': 'epsg:32618'}
>>> b, g, r = src.read()
>>> b
masked_array(data =
 [[-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]
 ...,
 [-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]
 [-- -- -- ..., -- -- --]],
             mask =
 [[ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]
 ...,
 [ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]
 [ True  True  True ...,  True  True  True]],
       fill_value = 0)

>>> b.min(), b.max(), b.mean()
(1, 255, 44.434478650699106)

C library dependecies:

• GDAL 1.9+

Python package dependencies (see also requirements.txt):

• affine
• Numpy
• setuptools

Development also requires (see requirements-dev.txt)

• Cython
• pytest

Rasterio is a C extension and to install on Linux or OS X you'll need a working compiler (XCode on OS X etc). You'll also need Numpy preinstalled; the Numpy headers are required to run the rasterio setup script. Numpy has to be installed (via the indicated requirements file) before rasterio can be installed. See rasterio's Travis configuration for more guidance.

The following commands are adapted from Rasterio's Travis-CI configuration.

$ sudo add-apt-repository ppa:ubuntugis/ppa
$ sudo apt-get update -qq
$ sudo apt-get install python-numpy libgdal1h gdal-bin libgdal-dev
$ pip install -r https://raw.githubusercontent.com/mapbox/rasterio/master/requirements.txt
$ pip install rasterio

Adapt them as necessary for your Linux system.

Wheels are available on PyPI for Homebrew based Python environments.

$ brew install gdal
$ pip install -r https://raw.githubusercontent.com/mapbox/rasterio/master/requirements.txt
$ pip install rasterio

The wheels are incompatible with MacPorts. MacPorts users will need to specify a source installation instead: pip install --no-use-wheel.

Windows binary packages created by Christoph Gohlke are available here.

From the repo directory, run py.test

$ py.test

See https://github.com/mapbox/rasterio/tree/master/docs.

See LICENSE.txt

See AUTHORS.txt

See CHANGES.txt