GDAL API

GDAL stands for Geospatial Data Abstraction Library, and is a veritable “Swiss army knife” of GIS data functionality. A subset of GDAL is the OGR Simple Features Library, which specializes in reading and writing vector geographic data in a variety of standard formats.

GeoDjango provides a high-level Python interface for some of the capabilities of OGR, including the reading and coordinate transformation of vector spatial data and minimal support for GDAL’s features with respect to raster (image) data.

Note

Although the module is named gdal, GeoDjango only supports some of the capabilities of OGR and GDAL’s raster features at this time.

Overview

Sample Data

The GDAL/OGR tools described here are designed to help you read in your geospatial data, in order for most of them to be useful you have to have some data to work with. If you’re starting out and don’t yet have any data of your own to use, GeoDjango tests contain a number of simple data sets that you can use for testing. You can download them here:

$ wget https://raw.githubusercontent.com/django/django/master/tests/gis_tests/data/cities/cities.{shp,prj,shx,dbf}
$ wget https://raw.githubusercontent.com/django/django/master/tests/gis_tests/data/rasters/raster.tif

Vector Data Source Objects

`DataSource`

DataSource is a wrapper for the OGR data source object that supports reading data from a variety of OGR-supported geospatial file formats and data sources using a simple, consistent interface. Each data source is represented by a DataSource object which contains one or more layers of data. Each layer, represented by a Layer object, contains some number of geographic features (Feature), information about the type of features contained in that layer (e.g. points, polygons, etc.), as well as the names and types of any additional fields (Field) of data that may be associated with each feature in that layer.

class DataSource(ds_input, encoding='utf-8')

The constructor for DataSource only requires one parameter: the path of the file you want to read. However, OGR also supports a variety of more complex data sources, including databases, that may be accessed by passing a special name string instead of a path. For more information, see the OGR Vector Formats documentation. The name property of a DataSource instance gives the OGR name of the underlying data source that it is using.

The optional encoding parameter allows you to specify a non-standard encoding of the strings in the source. This is typically useful when you obtain DjangoUnicodeDecodeError exceptions while reading field values.

Once you’ve created your DataSource, you can find out how many layers of data it contains by accessing the layer_count property, or (equivalently) by using the len() function. For information on accessing the layers of data themselves, see the next section:

>>> from django.contrib.gis.gdal import DataSource
>>> ds = DataSource('/path/to/your/cities.shp')
>>> ds.name
'/path/to/your/cities.shp'
>>> ds.layer_count                  # This file only contains one layer
1

layer_count

Returns the number of layers in the data source.

name

Returns the name of the data source.

`Layer`

class Layer

Layer is a wrapper for a layer of data in a DataSource object. You never create a Layer object directly. Instead, you retrieve them from a DataSource object, which is essentially a standard Python container of Layer objects. For example, you can access a specific layer by its index (e.g. ds[0] to access the first layer), or you can iterate over all the layers in the container in a for loop. The Layer itself acts as a container for geometric features.

Typically, all the features in a given layer have the same geometry type. The geom_type property of a layer is an OGRGeomType that identifies the feature type. We can use it to print out some basic information about each layer in a DataSource:

>>> for layer in ds:
...     print('Layer "%s": %i %ss' % (layer.name, len(layer), layer.geom_type.name))
...
Layer "cities": 3 Points

The example output is from the cities data source, loaded above, which evidently contains one layer, called "cities", which contains three point features. For simplicity, the examples below assume that you’ve stored that layer in the variable layer:

>>> layer = ds[0]

name

Returns the name of this layer in the data source.

>>> layer.name
'cities'

num_feat

Returns the number of features in the layer. Same as len(layer):

>>> layer.num_feat
3

geom_type

Returns the geometry type of the layer, as an OGRGeomType object:

>>> layer.geom_type.name
'Point'

num_fields

Returns the number of fields in the layer, i.e the number of fields of data associated with each feature in the layer:

>>> layer.num_fields
4

fields

Returns a list of the names of each of the fields in this layer:

>>> layer.fields
['Name', 'Population', 'Density', 'Created']

Returns a list of the data types of each of the fields in this layer. These are subclasses of Field, discussed below:

>>> [ft.__name__ for ft in layer.field_types]
['OFTString', 'OFTReal', 'OFTReal', 'OFTDate']

field_widths

Returns a list of the maximum field widths for each of the fields in this layer:

>>> layer.field_widths
[80, 11, 24, 10]

field_precisions

Returns a list of the numeric precisions for each of the fields in this layer. This is meaningless (and set to zero) for non-numeric fields:

>>> layer.field_precisions
[0, 0, 15, 0]

extent

Returns the spatial extent of this layer, as an Envelope object:

>>> layer.extent.tuple
(-104.609252, 29.763374, -95.23506, 38.971823)

srs

Property that returns the SpatialReference associated with this layer:

>>> print(layer.srs)
GEOGCS["GCS_WGS_1984",
    DATUM["WGS_1984",
        SPHEROID["WGS_1984",6378137,298.257223563]],
    PRIMEM["Greenwich",0],
    UNIT["Degree",0.017453292519943295]]

If the Layer has no spatial reference information associated with it, None is returned.

spatial_filter

Property that may be used to retrieve or set a spatial filter for this layer. A spatial filter can only be set with an OGRGeometry instance, a 4-tuple extent, or None. When set with something other than None, only features that intersect the filter will be returned when iterating over the layer:

>>> print(layer.spatial_filter)
None
>>> print(len(layer))
3
>>> [feat.get('Name') for feat in layer]
['Pueblo', 'Lawrence', 'Houston']
>>> ks_extent = (-102.051, 36.99, -94.59, 40.00) # Extent for state of Kansas
>>> layer.spatial_filter = ks_extent
>>> len(layer)
1
>>> [feat.get('Name') for feat in layer]
['Lawrence']
>>> layer.spatial_filter = None
>>> len(layer)
3

get_fields()

A method that returns a list of the values of a given field for each feature in the layer:

>>> layer.get_fields('Name')
['Pueblo', 'Lawrence', 'Houston']

get_geoms(geos=False)

A method that returns a list containing the geometry of each feature in the layer. If the optional argument geos is set to True then the geometries are converted to GEOSGeometry objects. Otherwise, they are returned as OGRGeometry objects:

>>> [pt.tuple for pt in layer.get_geoms()]
[(-104.609252, 38.255001), (-95.23506, 38.971823), (-95.363151, 29.763374)]

test_capability(capability)

Returns a boolean indicating whether this layer supports the given capability (a string). Examples of valid capability strings include: 'RandomRead', 'SequentialWrite', 'RandomWrite', 'FastSpatialFilter', 'FastFeatureCount', 'FastGetExtent', 'CreateField', 'Transactions', 'DeleteFeature', and 'FastSetNextByIndex'.

`Feature`

class Feature

Feature wraps an OGR feature. You never create a Feature object directly. Instead, you retrieve them from a Layer object. Each feature consists of a geometry and a set of fields containing additional properties. The geometry of a field is accessible via its geom property, which returns an OGRGeometry object. A Feature behaves like a standard Python container for its fields, which it returns as Field objects: you can access a field directly by its index or name, or you can iterate over a feature’s fields, e.g. in a for loop.

geom

Returns the geometry for this feature, as an OGRGeometry object:

>>> city.geom.tuple
(-104.609252, 38.255001)

get

A method that returns the value of the given field (specified by name) for this feature, not a Field wrapper object:

>>> city.get('Population')
102121

geom_type

Returns the type of geometry for this feature, as an OGRGeomType object. This will be the same for all features in a given layer and is equivalent to the Layer.geom_type property of the Layer object the feature came from.

num_fields

Returns the number of fields of data associated with the feature. This will be the same for all features in a given layer and is equivalent to the Layer.num_fields property of the Layer object the feature came from.

fields

Returns a list of the names of the fields of data associated with the feature. This will be the same for all features in a given layer and is equivalent to the Layer.fields property of the Layer object the feature came from.

fid

Returns the feature identifier within the layer:

>>> city.fid
0

layer_name

Returns the name of the Layer that the feature came from. This will be the same for all features in a given layer:

>>> city.layer_name
'cities'

index

A method that returns the index of the given field name. This will be the same for all features in a given layer:

>>> city.index('Population')
1

`Field`

class Field

name

Returns the name of this field:

>>> city['Name'].name
'Name'

type

Returns the OGR type of this field, as an integer. The FIELD_CLASSES dictionary maps these values onto subclasses of Field:

>>> city['Density'].type
2

type_name

Returns a string with the name of the data type of this field:

>>> city['Name'].type_name
'String'

value

Returns the value of this field. The Field class itself returns the value as a string, but each subclass returns the value in the most appropriate form:

>>> city['Population'].value
102121

width

Returns the width of this field:

>>> city['Name'].width
80

precision

Returns the numeric precision of this field. This is meaningless (and set to zero) for non-numeric fields:

>>> city['Density'].precision
15

as_double()

Returns the value of the field as a double (float):

>>> city['Density'].as_double()
874.7

as_int()

Returns the value of the field as an integer:

>>> city['Population'].as_int()
102121

as_string()

Returns the value of the field as a string:

>>> city['Name'].as_string()
'Pueblo'

as_datetime()

Returns the value of the field as a tuple of date and time components:

>>> city['Created'].as_datetime()
(c_long(1999), c_long(5), c_long(23), c_long(0), c_long(0), c_long(0), c_long(0))

`Driver`

class Driver(dr_input)

The Driver class is used internally to wrap an OGR DataSource driver.

driver_count

Returns the number of OGR vector drivers currently registered.

OGR Geometries

`OGRGeometry`

OGRGeometry objects share similar functionality with GEOSGeometry objects and are thin wrappers around OGR’s internal geometry representation. Thus, they allow for more efficient access to data when using DataSource. Unlike its GEOS counterpart, OGRGeometry supports spatial reference systems and coordinate transformation:

>>> from django.contrib.gis.gdal import OGRGeometry
>>> polygon = OGRGeometry('POLYGON((0 0, 5 0, 5 5, 0 5))')

class OGRGeometry(geom_input, srs=None)

This object is a wrapper for the OGR Geometry class. These objects are instantiated directly from the given geom_input parameter, which may be a string containing WKT, HEX, GeoJSON, a buffer containing WKB data, or an OGRGeomType object. These objects are also returned from the Feature.geom attribute, when reading vector data from Layer (which is in turn a part of a DataSource).

classmethod from_gml(gml_string)

Constructs an OGRGeometry from the given GML string.

classmethod from_bbox(bbox)

Constructs a Polygon from the given bounding-box (a 4-tuple).

__len__()

Returns the number of points in a LineString, the number of rings in a Polygon, or the number of geometries in a GeometryCollection. Not applicable to other geometry types.

__iter__()

Iterates over the points in a LineString, the rings in a Polygon, or the geometries in a GeometryCollection. Not applicable to other geometry types.

__getitem__()

Returns the point at the specified index for a LineString, the interior ring at the specified index for a Polygon, or the geometry at the specified index in a GeometryCollection. Not applicable to other geometry types.

dimension

Returns the number of coordinated dimensions of the geometry, i.e. 0 for points, 1 for lines, and so forth:

>> polygon.dimension
2

coord_dim

Returns or sets the coordinate dimension of this geometry. For example, the value would be 2 for two-dimensional geometries.

geom_count

Returns the number of elements in this geometry:

>>> polygon.geom_count
1

point_count

Returns the number of points used to describe this geometry:

>>> polygon.point_count
4

num_points

Alias for point_count.

num_coords

Alias for point_count.

geom_type

Returns the type of this geometry, as an OGRGeomType object.

geom_name

Returns the name of the type of this geometry:

>>> polygon.geom_name
'POLYGON'

area

Returns the area of this geometry, or 0 for geometries that do not contain an area:

>>> polygon.area
25.0

envelope

Returns the envelope of this geometry, as an Envelope object.

extent

Returns the envelope of this geometry as a 4-tuple, instead of as an Envelope object:

>>> point.extent
(0.0, 0.0, 5.0, 5.0)

srs

This property controls the spatial reference for this geometry, or None if no spatial reference system has been assigned to it. If assigned, accessing this property returns a SpatialReference object. It may be set with another SpatialReference object, or any input that SpatialReference accepts. Example:

>>> city.geom.srs.name
'GCS_WGS_1984'

srid

Returns or sets the spatial reference identifier corresponding to SpatialReference of this geometry. Returns None if there is no spatial reference information associated with this geometry, or if an SRID cannot be determined.

geos

Returns a GEOSGeometry object corresponding to this geometry.

gml

Returns a string representation of this geometry in GML format:

>>> OGRGeometry('POINT(1 2)').gml
'<gml:Point><gml:coordinates>1,2</gml:coordinates></gml:Point>'

hex

Returns a string representation of this geometry in HEX WKB format:

>>> OGRGeometry('POINT(1 2)').hex
'0101000000000000000000F03F0000000000000040'

json

Returns a string representation of this geometry in JSON format:

>>> OGRGeometry('POINT(1 2)').json
'{ "type": "Point", "coordinates": [ 1.000000, 2.000000 ] }'

kml

Returns a string representation of this geometry in KML format.

wkb_size

Returns the size of the WKB buffer needed to hold a WKB representation of this geometry:

>>> OGRGeometry('POINT(1 2)').wkb_size
21

wkb

Returns a buffer containing a WKB representation of this geometry.

wkt

Returns a string representation of this geometry in WKT format.

ewkt

Returns the EWKT representation of this geometry.

clone()

Returns a new OGRGeometry clone of this geometry object.

close_rings()

If there are any rings within this geometry that have not been closed, this routine will do so by adding the starting point to the end:

>>> triangle = OGRGeometry('LINEARRING (0 0,0 1,1 0)')
>>> triangle.close_rings()
>>> triangle.wkt
'LINEARRING (0 0,0 1,1 0,0 0)'

transform(coord_trans, clone=False)

Transforms this geometry to a different spatial reference system. May take a CoordTransform object, a SpatialReference object, or any other input accepted by SpatialReference (including spatial reference WKT and PROJ.4 strings, or an integer SRID).

By default nothing is returned and the geometry is transformed in-place. However, if the clone keyword is set to True then a transformed clone of this geometry is returned instead.

intersects(other)

Returns True if this geometry intersects the other, otherwise returns False.

equals(other)

Returns True if this geometry is equivalent to the other, otherwise returns False.

disjoint(other)

Returns True if this geometry is spatially disjoint to (i.e. does not intersect) the other, otherwise returns False.

touches(other)

Returns True if this geometry touches the other, otherwise returns False.

crosses(other)

Returns True if this geometry crosses the other, otherwise returns False.

within(other)

Returns True if this geometry is contained within the other, otherwise returns False.

contains(other)

Returns True if this geometry contains the other, otherwise returns False.

overlaps(other)

Returns True if this geometry overlaps the other, otherwise returns False.

boundary()

The boundary of this geometry, as a new OGRGeometry object.

convex_hull

The smallest convex polygon that contains this geometry, as a new OGRGeometry object.

difference()

Returns the region consisting of the difference of this geometry and the other, as a new OGRGeometry object.

intersection()

Returns the region consisting of the intersection of this geometry and the other, as a new OGRGeometry object.

sym_difference()

Returns the region consisting of the symmetric difference of this geometry and the other, as a new OGRGeometry object.

union()

Returns the region consisting of the union of this geometry and the other, as a new OGRGeometry object.

tuple

Returns the coordinates of a point geometry as a tuple, the coordinates of a line geometry as a tuple of tuples, and so forth:

>>> OGRGeometry('POINT (1 2)').tuple
(1.0, 2.0)
>>> OGRGeometry('LINESTRING (1 2,3 4)').tuple
((1.0, 2.0), (3.0, 4.0))

coords

An alias for tuple.

class Point

x

Returns the X coordinate of this point:

>>> OGRGeometry('POINT (1 2)').x
1.0

y

Returns the Y coordinate of this point:

>>> OGRGeometry('POINT (1 2)').y
2.0

z

Returns the Z coordinate of this point, or None if the point does not have a Z coordinate:

>>> OGRGeometry('POINT (1 2 3)').z
3.0

class LineString

x

Returns a list of X coordinates in this line:

>>> OGRGeometry('LINESTRING (1 2,3 4)').x
[1.0, 3.0]

y

Returns a list of Y coordinates in this line:

>>> OGRGeometry('LINESTRING (1 2,3 4)').y
[2.0, 4.0]

z

Returns a list of Z coordinates in this line, or None if the line does not have Z coordinates:

>>> OGRGeometry('LINESTRING (1 2 3,4 5 6)').z
[3.0, 6.0]

class Polygon

shell

Returns the shell or exterior ring of this polygon, as a LinearRing geometry.

exterior_ring

An alias for shell.

centroid

Returns a Point representing the centroid of this polygon.

class GeometryCollection

add(geom)

Adds a geometry to this geometry collection. Not applicable to other geometry types.

`OGRGeomType`

class OGRGeomType(type_input)

This class allows for the representation of an OGR geometry type in any of several ways:

>>> from django.contrib.gis.gdal import OGRGeomType
>>> gt1 = OGRGeomType(3)             # Using an integer for the type
>>> gt2 = OGRGeomType('Polygon')     # Using a string
>>> gt3 = OGRGeomType('POLYGON')     # It's case-insensitive
>>> print(gt1 == 3, gt1 == 'Polygon') # Equivalence works w/non-OGRGeomType objects
True True

name

Returns a short-hand string form of the OGR Geometry type:

>>> gt1.name
'Polygon'

num

Returns the number corresponding to the OGR geometry type:

>>> gt1.num
3

django

Returns the Django field type (a subclass of GeometryField) to use for storing this OGR type, or None if there is no appropriate Django type:

>>> gt1.django
'PolygonField'

`Envelope`

class Envelope(*args)

Represents an OGR Envelope structure that contains the minimum and maximum X, Y coordinates for a rectangle bounding box. The naming of the variables is compatible with the OGR Envelope C structure.

min_x

The value of the minimum X coordinate.

min_y

The value of the maximum X coordinate.

max_x

The value of the minimum Y coordinate.

max_y

The value of the maximum Y coordinate.

ur

The upper-right coordinate, as a tuple.

ll

The lower-left coordinate, as a tuple.

tuple

A tuple representing the envelope.

wkt

A string representing this envelope as a polygon in WKT format.

expand_to_include(*args)

Coordinate System Objects

`SpatialReference`

class SpatialReference(srs_input)

Spatial reference objects are initialized on the given srs_input, which may be one of the following:

OGC Well Known Text (WKT) (a string)
EPSG code (integer or string)
PROJ.4 string
A shorthand string for well-known standards ('WGS84', 'WGS72', 'NAD27', 'NAD83')

Example:

>>> wgs84 = SpatialReference('WGS84') # shorthand string
>>> wgs84 = SpatialReference(4326) # EPSG code
>>> wgs84 = SpatialReference('EPSG:4326') # EPSG string
>>> proj4 = '+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs '
>>> wgs84 = SpatialReference(proj4) # PROJ.4 string
>>> wgs84 = SpatialReference("""GEOGCS["WGS 84",
DATUM["WGS_1984",
     SPHEROID["WGS 84",6378137,298.257223563,
         AUTHORITY["EPSG","7030"]],
     AUTHORITY["EPSG","6326"]],
 PRIMEM["Greenwich",0,
     AUTHORITY["EPSG","8901"]],
 UNIT["degree",0.01745329251994328,
     AUTHORITY["EPSG","9122"]],
 AUTHORITY["EPSG","4326"]]""") # OGC WKT

__getitem__(target)

Returns the value of the given string attribute node, None if the node doesn’t exist. Can also take a tuple as a parameter, (target, child), where child is the index of the attribute in the WKT. For example:

>>> wkt = 'GEOGCS["WGS 84", DATUM["WGS_1984, ... AUTHORITY["EPSG","4326"]]')
>>> srs = SpatialReference(wkt) # could also use 'WGS84', or 4326
>>> print(srs['GEOGCS'])
WGS 84
>>> print(srs['DATUM'])
WGS_1984
>>> print(srs['AUTHORITY'])
EPSG
>>> print(srs['AUTHORITY', 1]) # The authority value
4326
>>> print(srs['TOWGS84', 4]) # the fourth value in this wkt
0
>>> print(srs['UNIT|AUTHORITY']) # For the units authority, have to use the pipe symbol.
EPSG
>>> print(srs['UNIT|AUTHORITY', 1]) # The authority value for the units
9122

attr_value(target, index=0)

The attribute value for the given target node (e.g. 'PROJCS'). The index keyword specifies an index of the child node to return.

auth_name(target)

Returns the authority name for the given string target node.

auth_code(target)

Returns the authority code for the given string target node.

clone()

Returns a clone of this spatial reference object.

identify_epsg()

This method inspects the WKT of this SpatialReference and will add EPSG authority nodes where an EPSG identifier is applicable.

from_esri()

Morphs this SpatialReference from ESRI’s format to EPSG

to_esri()

Morphs this SpatialReference to ESRI’s format.

validate()

Checks to see if the given spatial reference is valid, if not an exception will be raised.

import_epsg(epsg)

Import spatial reference from EPSG code.

import_proj(proj)

Import spatial reference from PROJ.4 string.

import_user_input(user_input)

import_wkt(wkt)

Import spatial reference from WKT.

import_xml(xml)

Import spatial reference from XML.

name

Returns the name of this Spatial Reference.

srid

Returns the SRID of top-level authority, or None if undefined.

linear_name

Returns the name of the linear units.

linear_units

Returns the value of the linear units.

angular_name

Returns the name of the angular units.”

angular_units

Returns the value of the angular units.

units

Returns a 2-tuple of the units value and the units name and will automatically determines whether to return the linear or angular units.

ellipsoid

Returns a tuple of the ellipsoid parameters for this spatial reference: (semimajor axis, semiminor axis, and inverse flattening).

semi_major

Returns the semi major axis of the ellipsoid for this spatial reference.

semi_minor

Returns the semi minor axis of the ellipsoid for this spatial reference.

inverse_flattening

Returns the inverse flattening of the ellipsoid for this spatial reference.

geographic

Returns True if this spatial reference is geographic (root node is GEOGCS).

local

Returns True if this spatial reference is local (root node is LOCAL_CS).

projected

Returns True if this spatial reference is a projected coordinate system (root node is PROJCS).

wkt

Returns the WKT representation of this spatial reference.

pretty_wkt

Returns the ‘pretty’ representation of the WKT.

proj

Returns the PROJ.4 representation for this spatial reference.

proj4

Alias for SpatialReference.proj.

xml

Returns the XML representation of this spatial reference.

`CoordTransform`

class CoordTransform(source, target)

Represents a coordinate system transform. It is initialized with two SpatialReference, representing the source and target coordinate systems, respectively. These objects should be used when performing the same coordinate transformation repeatedly on different geometries:

>>> ct = CoordTransform(SpatialReference('WGS84'), SpatialReference('NAD83'))
>>> for feat in layer:
...     geom = feat.geom # getting clone of feature geometry
...     geom.transform(ct) # transforming

Raster Data Objects

`GDALRaster`

GDALRaster is a wrapper for the GDAL raster source object that supports reading data from a variety of GDAL-supported geospatial file formats and data sources using a simple, consistent interface. Each data source is represented by a GDALRaster object which contains one or more layers of data named bands. Each band, represented by a GDALBand object, contains georeferenced image data. For example, an RGB image is represented as three bands: one for red, one for green, and one for blue.

Note

For raster data there is no difference between a raster instance and its data source. Unlike for the Geometry objects, GDALRaster objects are always a data source. Temporary rasters can be instantiated in memory using the corresponding driver, but they will be of the same class as file-based raster sources.

class GDALRaster(ds_input, write=False)

The constructor for GDALRaster accepts two parameters. The first parameter defines the raster source, and the second parameter defines if a raster should be opened in write mode. For newly-created rasters, the second parameter is ignored and the new raster is always created in write mode.

The first parameter can take three forms: a string representing a file path, a dictionary with values defining a new raster, or a bytes object representing a raster file.

If the input is a file path, the raster is opened from there. If the input is raw data in a dictionary, the parameters width, height, and srid are required. If the input is a bytes object, it will be opened using a GDAL virtual filesystem.

For a detailed description of how to create rasters using dictionary input, see Creating rasters from data. For a detailed description of how to create rasters in the virtual filesystem, see Using GDAL’s Virtual Filesystem.

The following example shows how rasters can be created from different input sources (using the sample data from the GeoDjango tests; see also the Sample Data section).

>>> from django.contrib.gis.gdal import GDALRaster
>>> rst = GDALRaster('/path/to/your/raster.tif', write=False)
>>> rst.name
'/path/to/your/raster.tif'
>>> rst.width, rst.height  # This file has 163 x 174 pixels
(163, 174)
>>> rst = GDALRaster({  # Creates an in-memory raster
...     'srid': 4326,
...     'width': 4,
...     'height': 4,
...     'datatype': 1,
...     'bands': [{
...         'data': (2, 3),
...         'offset': (1, 1),
...         'size': (2, 2),
...         'shape': (2, 1),
...         'nodata_value': 5,
...     }]
... })
>>> rst.srs.srid
4326
>>> rst.width, rst.height
(4, 4)
>>> rst.bands[0].data()
array([[5, 5, 5, 5],
       [5, 2, 3, 5],
       [5, 2, 3, 5],
       [5, 5, 5, 5]], dtype=uint8)
>>> rst_file = open('/path/to/your/raster.tif', 'rb')
>>> rst_bytes = rst_file.read()
>>> rst = GDALRaster(rst_bytes)
>>> rst.is_vsi_based
True
>>> rst.name  # Stored in a random path in the vsimem filesystem.
'/vsimem/da300bdb-129d-49a8-b336-e410a9428dad'

Changed in Django 2.0:

Added the ability to read and write rasters in GDAL’s memory-based virtual filesystem. GDALRaster objects can now be converted to and from binary data in-memory.

name

The name of the source which is equivalent to the input file path or the name provided upon instantiation.

>>> GDALRaster({'width': 10, 'height': 10, 'name': 'myraster', 'srid': 4326}).name
'myraster'

driver

The name of the GDAL driver used to handle the input file. For GDALRasters created from a file, the driver type is detected automatically. The creation of rasters from scratch is a in-memory raster by default ('MEM'), but can be altered as needed. For instance, use GTiff for a GeoTiff file. For a list of file types, see also the GDAL Raster Formats list.

An in-memory raster is created through the following example:

>>> GDALRaster({'width': 10, 'height': 10, 'srid': 4326}).driver.name
'MEM'

A file based GeoTiff raster is created through the following example:

>>> import tempfile
>>> rstfile = tempfile.NamedTemporaryFile(suffix='.tif')
>>> rst = GDALRaster({'driver': 'GTiff', 'name': rstfile.name, 'srid': 4326,
...                   'width': 255, 'height': 255, 'nr_of_bands': 1})
>>> rst.name
'/tmp/tmp7x9H4J.tif'           # The exact filename will be different on your computer
>>> rst.driver.name
'GTiff'

width

The width of the source in pixels (X-axis).

>>> GDALRaster({'width': 10, 'height': 20, 'srid': 4326}).width
10

height

The height of the source in pixels (Y-axis).

>>> GDALRaster({'width': 10, 'height': 20, 'srid': 4326}).height
20

srs

The spatial reference system of the raster, as a SpatialReference instance. The SRS can be changed by setting it to an other SpatialReference or providing any input that is accepted by the SpatialReference constructor.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.srs.srid
4326
>>> rst.srs = 3086
>>> rst.srs.srid
3086

srid

The Spatial Reference System Identifier (SRID) of the raster. This property is a shortcut to getting or setting the SRID through the srs attribute.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.srid
4326
>>> rst.srid = 3086
>>> rst.srid
3086
>>> rst.srs.srid  # This is equivalent
3086

geotransform

The affine transformation matrix used to georeference the source, as a tuple of six coefficients which map pixel/line coordinates into georeferenced space using the following relationship:

Xgeo = GT(0) + Xpixel*GT(1) + Yline*GT(2)
Ygeo = GT(3) + Xpixel*GT(4) + Yline*GT(5)

The same values can be retrieved by accessing the origin (indices 0 and 3), scale (indices 1 and 5) and skew (indices 2 and 4) properties.

The default is [0.0, 1.0, 0.0, 0.0, 0.0, -1.0].

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.geotransform
[0.0, 1.0, 0.0, 0.0, 0.0, -1.0]

origin

Coordinates of the top left origin of the raster in the spatial reference system of the source, as a point object with x and y members.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.origin
[0.0, 0.0]
>>> rst.origin.x = 1
>>> rst.origin
[1.0, 0.0]

scale

Pixel width and height used for georeferencing the raster, as a as a point object with x and y members. See geotransform for more information.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.scale
[1.0, -1.0]
>>> rst.scale.x = 2
>>> rst.scale
[2.0, -1.0]

skew

Skew coefficients used to georeference the raster, as a point object with x and y members. In case of north up images, these coefficients are both 0.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.skew
[0.0, 0.0]
>>> rst.skew.x = 3
>>> rst.skew
[3.0, 0.0]

extent

Extent (boundary values) of the raster source, as a 4-tuple (xmin, ymin, xmax, ymax) in the spatial reference system of the source.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.extent
(0.0, -20.0, 10.0, 0.0)
>>> rst.origin.x = 100
>>> rst.extent
(100.0, -20.0, 110.0, 0.0)

bands

List of all bands of the source, as GDALBand instances.

>>> rst = GDALRaster({"width": 1, "height": 2, 'srid': 4326,
...                   "bands": [{"data": [0, 1]}, {"data": [2, 3]}]})
>>> len(rst.bands)
2
>>> rst.bands[1].data()
array([[ 2.,  3.]], dtype=float32)

warp(ds_input, resampling='NearestNeighbour', max_error=0.0)

Returns a warped version of this raster.

The warping parameters can be specified through the ds_input argument. The use of ds_input is analogous to the corresponding argument of the class constructor. It is a dictionary with the characteristics of the target raster. Allowed dictionary key values are width, height, SRID, origin, scale, skew, datatype, driver, and name (filename).

By default, the warp functions keeps most parameters equal to the values of the original source raster, so only parameters that should be changed need to be specified. Note that this includes the driver, so for file-based rasters the warp function will create a new raster on disk.

The only parameter that is set differently from the source raster is the name. The default value of the the raster name is the name of the source raster appended with '_copy' + source_driver_name. For file-based rasters it is recommended to provide the file path of the target raster.

The resampling algorithm used for warping can be specified with the resampling argument. The default is NearestNeighbor, and the other allowed values are Bilinear, Cubic, CubicSpline, Lanczos, Average, and Mode.

The max_error argument can be used to specify the maximum error measured in input pixels that is allowed in approximating the transformation. The default is 0.0 for exact calculations.

For users familiar with GDAL, this function has a similar functionality to the gdalwarp command-line utility.

For example, the warp function can be used for aggregating a raster to the double of its original pixel scale:

>>> rst = GDALRaster({
...     "width": 6, "height": 6, "srid": 3086,
...     "origin": [500000, 400000],
...     "scale": [100, -100],
...     "bands": [{"data": range(36), "nodata_value": 99}]
... })
>>> target = rst.warp({"scale": [200, -200], "width": 3, "height": 3})
>>> target.bands[0].data()
array([[  7.,   9.,  11.],
       [ 19.,  21.,  23.],
       [ 31.,  33.,  35.]], dtype=float32)

transform(srid, driver=None, name=None, resampling='NearestNeighbour', max_error=0.0)

Returns a transformed version of this raster with the specified SRID.

This function transforms the current raster into a new spatial reference system that can be specified with an srid. It calculates the bounds and scale of the current raster in the new spatial reference system and warps the raster using the warp function.

By default, the driver of the source raster is used and the name of the raster is the original name appended with '_copy' + source_driver_name. A different driver or name can be specified with the driver and name arguments.

The default resampling algorithm is NearestNeighbour but can be changed using the resampling argument. The default maximum allowed error for resampling is 0.0 and can be changed using the max_error argument. Consult the warp documentation for detail on those arguments.

>>> rst = GDALRaster({
...     "width": 6, "height": 6, "srid": 3086,
...     "origin": [500000, 400000],
...     "scale": [100, -100],
...     "bands": [{"data": range(36), "nodata_value": 99}]
... })
>>> target = rst.transform(4326)
>>> target.origin
[-82.98492744885776, 27.601924753080144]

info: New in Django 2.0:

Returns a string with a summary of the raster. This is equivalent to the gdalinfo command line utility.

metadata

New in Django 2.0:

The metadata of this raster, represented as a nested dictionary. The first-level key is the metadata domain. The second-level contains the metadata item names and values from each domain.

To set or update a metadata item, pass the corresponding metadata item to the method using the nested structure described above. Only keys that are in the specified dictionary are updated; the rest of the metadata remains unchanged.

To remove a metadata item, use None as the metadata value.

>>> rst = GDALRaster({'width': 10, 'height': 20, 'srid': 4326})
>>> rst.metadata
{}
>>> rst.metadata = {'DEFAULT': {'OWNER': 'Django', 'VERSION': '1.0'}}
>>> rst.metadata
{'DEFAULT': {'OWNER': 'Django', 'VERSION': '1.0'}}
>>> rst.metadata = {'DEFAULT': {'OWNER': None, 'VERSION': '2.0'}}
>>> rst.metadata
{'DEFAULT': {'VERSION': '2.0'}}

vsi_buffer: New in Django 2.0:
A bytes representation of this raster. Returns None for rasters that are not stored in GDAL’s virtual filesystem.

is_vsi_based: New in Django 2.0:
A boolean indicating if this raster is stored in GDAL’s virtual filesystem.

`GDALBand`

class GDALBand

GDALBand instances are not created explicitly, but rather obtained from a GDALRaster object, through its bands attribute. The GDALBands contain the actual pixel values of the raster.

description: The name or description of the band, if any.

width: The width of the band in pixels (X-axis).

height: The height of the band in pixels (Y-axis).

pixel_count: The total number of pixels in this band. Is equal to width * height.

statistics(refresh=False, approximate=False)

Compute statistics on the pixel values of this band. The return value is a tuple with the following structure: (minimum, maximum, mean, standard deviation).

If the approximate argument is set to True, the statistics may be computed based on overviews or a subset of image tiles.

If the refresh argument is set to True, the statistics will be computed from the data directly, and the cache will be updated with the result.

If a persistent cache value is found, that value is returned. For raster formats using Persistent Auxiliary Metadata (PAM) services, the statistics might be cached in an auxiliary file. In some cases this metadata might be out of sync with the pixel values or cause values from a previous call to be returned which don’t reflect the value of the approximate argument. In such cases, use the refresh argument to get updated values and store them in the cache.

For empty bands (where all pixel values are “no data”), all statistics are returned as None.

The statistics can also be retrieved directly by accessing the min, max, mean, and std properties.

min: The minimum pixel value of the band (excluding the “no data” value).

max: The maximum pixel value of the band (excluding the “no data” value).

mean: The mean of all pixel values of the band (excluding the “no data” value).

std: The standard deviation of all pixel values of the band (excluding the “no data” value).

nodata_value

The “no data” value for a band is generally a special marker value used to mark pixels that are not valid data. Such pixels should generally not be displayed, nor contribute to analysis operations.

To delete an existing “no data” value, set this property to None (requires GDAL ≥ 2.1).

datatype(as_string=False): The data type contained in the band, as an integer constant between 0 (Unknown) and 11. If as_string is True, the data type is returned as a string with the following possible values: GDT_Unknown, GDT_Byte, GDT_UInt16, GDT_Int16, GDT_UInt32, GDT_Int32, GDT_Float32, GDT_Float64, GDT_CInt16, GDT_CInt32, GDT_CFloat32, and GDT_CFloat64.

color_interp(as_string=False): New in Django 2.0:

The color interpretation for the band, as an integer between 0and 16. If as_string is True, the data type is returned as a string with the following possible values: GCI_Undefined, GCI_GrayIndex, GCI_PaletteIndex, GCI_RedBand, GCI_GreenBand, GCI_BlueBand, GCI_AlphaBand, GCI_HueBand, GCI_SaturationBand, GCI_LightnessBand, GCI_CyanBand, GCI_MagentaBand, GCI_YellowBand, GCI_BlackBand, GCI_YCbCr_YBand, GCI_YCbCr_CbBand, and GCI_YCbCr_CrBand. GCI_YCbCr_CrBand also represents GCI_Max because both correspond to the integer 16, but only GCI_YCbCr_CrBand is returned as a string.

data(data=None, offset=None, size=None, shape=None)

The accessor to the pixel values of the GDALBand. Returns the complete data array if no parameters are provided. A subset of the pixel array can be requested by specifying an offset and block size as tuples.

If NumPy is available, the data is returned as NumPy array. For performance reasons, it is highly recommended to use NumPy.

Data is written to the GDALBand if the data parameter is provided. The input can be of one of the following types - packed string, buffer, list, array, and NumPy array. The number of items in the input should normally correspond to the total number of pixels in the band, or to the number of pixels for a specific block of pixel values if the offset and size parameters are provided.

If the number of items in the input is different from the target pixel block, the shape parameter must be specified. The shape is a tuple that specifies the width and height of the input data in pixels. The data is then replicated to update the pixel values of the selected block. This is useful to fill an entire band with a single value, for instance.

For example:

>>> rst = GDALRaster({'width': 4, 'height': 4, 'srid': 4326, 'datatype': 1, 'nr_of_bands': 1})
>>> bnd = rst.bands[0]
>>> bnd.data(range(16))
>>> bnd.data()
array([[ 0,  1,  2,  3],
       [ 4,  5,  6,  7],
       [ 8,  9, 10, 11],
       [12, 13, 14, 15]], dtype=int8)
>>> bnd.data(offset=(1, 1), size=(2, 2))
array([[ 5,  6],
       [ 9, 10]], dtype=int8)
>>> bnd.data(data=[-1, -2, -3, -4], offset=(1, 1), size=(2, 2))
>>> bnd.data()
array([[ 0,  1,  2,  3],
       [ 4, -1, -2,  7],
       [ 8, -3, -4, 11],
       [12, 13, 14, 15]], dtype=int8)
>>> bnd.data(data='\x9d\xa8\xb3\xbe', offset=(1, 1), size=(2, 2))
>>> bnd.data()
array([[  0,   1,   2,   3],
       [  4, -99, -88,   7],
       [  8, -77, -66,  11],
       [ 12,  13,  14,  15]], dtype=int8)
>>> bnd.data([1], shape=(1, 1))
>>> bnd.data()
array([[1, 1, 1, 1],
       [1, 1, 1, 1],
       [1, 1, 1, 1],
       [1, 1, 1, 1]], dtype=uint8)
>>> bnd.data(range(4), shape=(1, 4))
array([[0, 0, 0, 0],
       [1, 1, 1, 1],
       [2, 2, 2, 2],
       [3, 3, 3, 3]], dtype=uint8)

metadata: New in Django 2.0:

The metadata of this band. The functionality is identical to GDALRaster.metadata.

Creating rasters from data

This section describes how to create rasters from scratch using the ds_input parameter.

A new raster is created when a dict is passed to the GDALRaster constructor. The dictionary contains defining parameters of the new raster, such as the origin, size, or spatial reference system. The dictionary can also contain pixel data and information about the format of the new raster. The resulting raster can therefore be file-based or memory-based, depending on the driver specified.

There’s no standard for describing raster data in a dictionary or JSON flavor. The definition of the dictionary input to the GDALRaster class is therefore specific to Django. It’s inspired by the geojson format, but the geojson standard is currently limited to vector formats.

Examples of using the different keys when creating rasters can be found in the documentation of the corresponding attributes and methods of the GDALRaster and GDALBand classes.

The `ds_input` dictionary

Only a few keys are required in the ds_input dictionary to create a raster: width, height, and srid. All other parameters have default values (see the table below). The list of keys that can be passed in the ds_input dictionary is closely related but not identical to the GDALRaster properties. Many of the parameters are mapped directly to those properties; the others are described below.

The following table describes all keys that can be set in the ds_input dictionary.

Key	Default	Usage
`srid`	required	Mapped to the `srid` attribute
`width`	required	Mapped to the `width` attribute
`height`	required	Mapped to the `height` attribute
`driver`	`MEM`	Mapped to the `driver` attribute
`name`	`''`	See below
`origin`	`0`	Mapped to the `origin` attribute
`scale`	`0`	Mapped to the `scale` attribute
`skew`	`0`	Mapped to the `width` attribute
`bands`	`[]`	See below
`nr_of_bands`	`0`	See below
`datatype`	`6`	See below
`papsz_options`	`{}`	See below

name: String representing the name of the raster. When creating a file-based raster, this parameter must be the file path for the new raster. If the name starts with /vsimem/, the raster is created in GDAL’s virtual filesystem.

datatype

Integer representing the data type for all the bands. Defaults to 6 (Float32). All bands of a new raster are required to have the same datatype. The value mapping is:

Value	GDAL Pixel Type	Description
1	GDT_Byte	Eight bit unsigned integer
2	GDT_UInt16	Sixteen bit unsigned integer
3	GDT_Int16	Sixteen bit signed integer
4	GDT_UInt32	Thirty-two bit unsigned integer
5	GDT_Int32	Thirty-two bit signed integer
6	GDT_Float32	Thirty-two bit floating point
7	GDT_Float64	Sixty-four bit floating point

nr_of_bands: Integer representing the number of bands of the raster. A raster can be created without passing band data upon creation. If the number of bands isn’t specified, it’s automatically calculated from the length of the bands input. The number of bands can’t be changed after creation.

bands: A list of band_input dictionaries with band input data. The resulting band indices are the same as in the list provided. The definition of the band input dictionary is given below. If band data isn’t provided, the raster bands values are instantiated as an array of zeros and the “no data” value is set to None.

papsz_options

New in Django 2.0:

A dictionary with raster creation options. The key-value pairs of the input dictionary are passed to the driver on creation of the raster.

The available options are driver-specific and are described in the documentation of each driver.

The values in the dictionary are not case-sensitive and are automatically converted to the correct string format upon creation.

The following example uses some of the options available for the GTiff driver. The result is a compressed signed byte raster with an internal tiling scheme. The internal tiles have a block size of 23 by 23:

>>> GDALRaster({
...    'driver': 'GTiff',
...    'name': '/path/to/new/file.tif',
...    'srid': 4326,
...    'width': 255,
...    'height': 255,
...    'nr_of_bands': 1,
...    'papsz_options': {
...        'compress': 'packbits',
...        'pixeltype': 'signedbyte',
...        'tiled': 'yes',
...        'blockxsize': 23,
...        'blockysize': 23,
...    }
... })

The `band_input` dictionary

The bands key in the ds_input dictionary is a list of band_input dictionaries. Each band_input dictionary can contain pixel values and the “no data” value to be set on the bands of the new raster. The data array can have the full size of the new raster or be smaller. For arrays that are smaller than the full raster, the size, shape, and offset keys control the pixel values. The corresponding keys are passed to the data() method. Their functionality is the same as setting the band data with that method. The following table describes the keys that can be used.

Key	Default	Usage
`nodata_value`	`None`	Mapped to the `nodata_value` attribute
`data`	Same as `nodata_value` or `0`	Passed to the `data()` method
`size`	`(with, height)` of raster	Passed to the `data()` method
`shape`	Same as size	Passed to the `data()` method
`offset`	`(0, 0)`	Passed to the `data()` method

Using GDAL’s Virtual Filesystem

GDAL has an internal memory-based filesystem, which allows treating blocks of memory as files. It can be used to read and write GDALRaster objects to and from binary file buffers.

This is useful in web contexts where rasters might be obtained as a buffer from a remote storage or returned from a view without being written to disk.

GDALRaster objects are created in the virtual filesystem when a bytes object is provided as input, or when the file path starts with /vsimem/.

Input provided as bytes has to be a full binary representation of a file. For instance:

# Read a raster as a file object from a remote source.
>>> from urllib.request import urlopen
>>> dat = urlopen('http://example.com/raster.tif').read()
# Instantiate a raster from the bytes object.
>>> rst = GDALRaster(dat)
# The name starts with /vsimem/, indicating that the raster lives in the
# virtual filesystem.
>>> rst.name
'/vsimem/da300bdb-129d-49a8-b336-e410a9428dad'

To create a new virtual file-based raster from scratch, use the ds_input dictionary representation and provide a name argument that starts with /vsimem/ (for detail of the dictionary representation, see Creating rasters from data). For virtual file-based rasters, the vsi_buffer attribute returns the bytes representation of the raster.

Here’s how to create a raster and return it as a file in an HttpResponse:

>>> from django.http import HttpResponse
>>> rst = GDALRaster({
...     'name': '/vsimem/temporarymemfile',
...     'driver': 'tif',
...     'width': 6, 'height': 6, 'srid': 3086,
...     'origin': [500000, 400000],
...     'scale': [100, -100],
...     'bands': [{'data': range(36), 'nodata_value': 99}]
... })
>>> HttpResponse(rast.vsi_buffer, 'image/tiff')

Settings

`GDAL_LIBRARY_PATH`

A string specifying the location of the GDAL library. Typically, this setting is only used if the GDAL library is in a non-standard location (e.g., /home/john/lib/libgdal.so).

Exceptions

exception GDALException: The base GDAL exception, indicating a GDAL-related error.

exception SRSException: An exception raised when an error occurs when constructing or using a spatial reference system object.

GDAL API

GDAL API

Overview

Sample Data

Vector Data Source Objects

DataSource

Layer

Feature

Field

Driver

OGR Geometries

OGRGeometry

OGRGeomType

Envelope

Coordinate System Objects

SpatialReference

CoordTransform

Raster Data Objects

GDALRaster

GDALBand

Creating rasters from data

The ds_input dictionary

The band_input dictionary

Using GDAL’s Virtual Filesystem

Settings

GDAL_LIBRARY_PATH

Exceptions

`DataSource`

`Layer`

`Feature`

`Field`

`Driver`

`OGRGeometry`

`OGRGeomType`

`Envelope`

`SpatialReference`

`CoordTransform`

`GDALRaster`

`GDALBand`

The `ds_input` dictionary

The `band_input` dictionary

`GDAL_LIBRARY_PATH`