- Language: en
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 data sets that you can use for testing. You can download them here:
$ wget https://raw.githubusercontent.com/django/django/main/tests/gis_tests/data/cities/cities.{shp,prj,shx,dbf}
$ wget https://raw.githubusercontent.com/django/django/main/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 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')[source]¶ 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. Thename
property of aDataSource
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 obtainDjangoUnicodeDecodeError
exceptions while reading field values.Once you’ve created your
DataSource
, you can find out how many layers of data it contains by accessing thelayer_count
property, or (equivalently) by using thelen()
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
Returns the number of layers in the data source.
Returns the name of the data source.
Layer
¶
-
class
Layer
¶ Layer
is a wrapper for a layer of data in aDataSource
object. You never create aLayer
object directly. Instead, you retrieve them from aDataSource
object, which is essentially a standard Python container ofLayer
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 afor
loop. TheLayer
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 anOGRGeomType
that identifies the feature type. We can use it to print out some basic information about each layer in aDataSource
:>>> 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 variablelayer
:>>> 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, orNone
. When set with something other thanNone
, 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 toTrue
then the geometries are converted toGEOSGeometry
objects. Otherwise, they are returned asOGRGeometry
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 aFeature
object directly. Instead, you retrieve them from aLayer
object. Each feature consists of a geometry and a set of fields containing additional properties. The geometry of a field is accessible via itsgeom
property, which returns anOGRGeometry
object. AFeature
behaves like a standard Python container for its fields, which it returns asField
objects: you can access a field directly by its index or name, or you can iterate over a feature’s fields, e.g. in afor
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 theLayer.geom_type
property of theLayer
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 theLayer
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 theLayer
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 ofField
:>>> 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))
-
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)[source]¶ 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, abuffer
containing WKB data, or anOGRGeomType
object. These objects are also returned from theFeature.geom
attribute, when reading vector data fromLayer
(which is in turn a part of aDataSource
).Constructs an
OGRGeometry
from the given GML string.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 aPolygon
, or the number of geometries in aGeometryCollection
. Not applicable to other geometry types.-
__iter__
()¶
Iterates over the points in a
LineString
, the rings in aPolygon
, or the geometries in aGeometryCollection
. 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 aPolygon
, or the geometry at the specified index in aGeometryCollection
. Not applicable to other geometry types.Returns the number of coordinated dimensions of the geometry, i.e. 0 for points, 1 for lines, and so forth:
>>> polygon.dimension 2
Returns the coordinate dimension of this geometry. For example, the value would be 2 for two-dimensional geometries.
Deprecated since version 5.1: The
coord_dim
setter is deprecated. Useset_3d()
instead.New in Django 5.1.A boolean indicating if this geometry has Z coordinates.
New in Django 5.1.A method that adds or removes the Z coordinate dimension.
>>> p = OGRGeometry("POINT (1 2 3)") >>> p.is_3d True >>> p.set_3d(False) >>> p.wkt "POINT (1 2)"
New in Django 5.1.A boolean indicating if this geometry has M coordinates.
New in Django 5.1.A method to add or remove the M coordinate dimension.
>>> p = OGRGeometry("POINT (1 2)") >>> p.is_measured False >>> p.set_measured(True) >>> p.wkt "POINT M (1 2 0)"
Returns the number of elements in this geometry:
>>> polygon.geom_count 1
Returns the number of points used to describe this geometry:
>>> polygon.point_count 4
Alias for
point_count
.Alias for
point_count
.Returns the type of this geometry, as an
OGRGeomType
object.Returns the name of the type of this geometry:
>>> polygon.geom_name 'POLYGON'
Returns the area of this geometry, or 0 for geometries that do not contain an area:
>>> polygon.area 25.0
Returns the envelope of this geometry, as an
Envelope
object.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 aSpatialReference
object. It may be set with anotherSpatialReference
object, or any input thatSpatialReference
accepts. Example:>>> city.geom.srs.name 'GCS_WGS_1984'
-
srid
¶
Returns or sets the spatial reference identifier corresponding to
SpatialReference
of this geometry. ReturnsNone
if there is no spatial reference information associated with this geometry, or if an SRID cannot be determined.Returns a
GEOSGeometry
object corresponding to this geometry.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>'
Returns a string representation of this geometry in HEX WKB format:
>>> OGRGeometry("POINT(1 2)").hex '0101000000000000000000F03F0000000000000040'
Returns a string representation of this geometry in JSON format:
>>> OGRGeometry("POINT(1 2)").json '{ "type": "Point", "coordinates": [ 1.000000, 2.000000 ] }'
Returns a string representation of this geometry in KML format.
Returns the size of the WKB buffer needed to hold a WKB representation of this geometry:
>>> OGRGeometry("POINT(1 2)").wkb_size 21
Returns a
buffer
containing a WKB representation of this geometry.Returns a string representation of this geometry in WKT format.
Returns the EWKT representation of this geometry.
Returns a new
OGRGeometry
clone of this geometry object.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)'
Transforms this geometry to a different spatial reference system. May take a
CoordTransform
object, aSpatialReference
object, or any other input accepted bySpatialReference
(including spatial reference WKT and PROJ strings, or an integer SRID).By default nothing is returned and the geometry is transformed in-place. However, if the
clone
keyword is set toTrue
then a transformed clone of this geometry is returned instead.Returns
True
if this geometry intersects the other, otherwise returnsFalse
.Returns
True
if this geometry is equivalent to the other, otherwise returnsFalse
.Returns
True
if this geometry is spatially disjoint to (i.e. does not intersect) the other, otherwise returnsFalse
.Returns
True
if this geometry touches the other, otherwise returnsFalse
.Returns
True
if this geometry crosses the other, otherwise returnsFalse
.Returns
True
if this geometry is contained within the other, otherwise returnsFalse
.Returns
True
if this geometry contains the other, otherwise returnsFalse
.Returns
True
if this geometry overlaps the other, otherwise returnsFalse
.The boundary of this geometry, as a new
OGRGeometry
object.The smallest convex polygon that contains this geometry, as a new
OGRGeometry
object.Returns the region consisting of the difference of this geometry and the other, as a new
OGRGeometry
object.Returns the region consisting of the intersection of this geometry and the other, as a new
OGRGeometry
object.Returns the region consisting of the symmetric difference of this geometry and the other, as a new
OGRGeometry
object.Returns the region consisting of the union of this geometry and the other, as a new
OGRGeometry
object.Returns a
Point
representing the centroid of this geometry.Changed in Django 5.1:centroid
was promoted from aPolygon
only attribute to being available on all geometry types.-
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
-
m
¶
New in Django 5.1.Returns the M coordinate of this point, or
None
if the Point does not have an M coordinate:>>> OGRGeometry("POINT ZM (1 2 3 4)").m 4.0
-
-
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]
-
m
¶
New in Django 5.1.Returns a list of M coordinates in this line or
None
if the line does not have M coordinates:>>> OGRGeometry("LINESTRING(0 1 2 10, 1 2 3 11, 2 3 4 12)").m [10.0, 11.0, 12.0]
-
OGRGeomType
¶
-
class
OGRGeomType
(type_input)[source]¶ 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
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
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)[source]¶ 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.
The value of the minimum X coordinate.
The value of the maximum X coordinate.
The value of the minimum Y coordinate.
The value of the maximum Y coordinate.
The upper-right coordinate, as a tuple.
The lower-left coordinate, as a tuple.
A tuple representing the envelope.
A string representing this envelope as a polygon in WKT format.
Coordinate System Objects¶
SpatialReference
¶
-
class
SpatialReference
(srs_input)[source]¶ 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 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 >>> proj = "+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs " >>> wgs84 = SpatialReference(proj) # PROJ 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
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
The attribute value for the given target node (e.g.
'PROJCS'
). The index keyword specifies an index of the child node to return.Returns the authority name for the given string target node.
Returns the authority code for the given string target node.
Returns a clone of this spatial reference object.
This method inspects the WKT of this
SpatialReference
and will add EPSG authority nodes where an EPSG identifier is applicable.Morphs this SpatialReference from ESRI’s format to EPSG
Morphs this SpatialReference to ESRI’s format.
Checks to see if the given spatial reference is valid, if not an exception will be raised.
Import spatial reference from EPSG code.
Import spatial reference from PROJ string.
Import spatial reference from WKT.
Import spatial reference from XML.
Returns the name of this Spatial Reference.
Returns the SRID of top-level authority, or
None
if undefined.Returns the name of the linear units.
Returns the value of the linear units.
Returns the name of the angular units.”
Returns the value of the angular 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.
Returns a tuple of the ellipsoid parameters for this spatial reference: (semimajor axis, semiminor axis, and inverse flattening).
Returns the semi major axis of the ellipsoid for this spatial reference.
Returns the semi minor axis of the ellipsoid for this spatial reference.
Returns the inverse flattening of the ellipsoid for this spatial reference.
Returns
True
if this spatial reference is geographic (root node isGEOGCS
).Returns
True
if this spatial reference is local (root node isLOCAL_CS
).Returns
True
if this spatial reference is a projected coordinate system (root node isPROJCS
).Returns the WKT representation of this spatial reference.
Returns the ‘pretty’ representation of the WKT.
Returns the PROJ representation for this spatial reference.
Alias for
SpatialReference.proj
.Returns the XML representation of this spatial reference.
CoordTransform
¶
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 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)[source]¶ 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 or
Path
representing a file path (filesystem or GDAL virtual filesystem), 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
, andsrid
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'
-
name
[source]¶ 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
[source]¶ The name of the GDAL driver used to handle the input file. For
GDALRaster
s created from a file, the driver type is detected automatically. The creation of rasters from scratch is an in-memory raster by default ('MEM'
), but can be altered as needed. For instance, useGTiff
for aGeoTiff
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
[source]¶ The width of the source in pixels (X-axis).
>>> GDALRaster({"width": 10, "height": 20, "srid": 4326}).width 10
-
height
[source]¶ The height of the source in pixels (Y-axis).
>>> GDALRaster({"width": 10, "height": 20, "srid": 4326}).height 20
-
srs
[source]¶ The spatial reference system of the raster, as a
SpatialReference
instance. The SRS can be changed by setting it to an otherSpatialReference
or providing any input that is accepted by theSpatialReference
constructor.>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326}) >>> rst.srs.srid 4326 >>> rst.srs = 3086 >>> rst.srs.srid 3086
-
srid
[source]¶ 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
[source]¶ 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) andskew
(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
[source]¶ Coordinates of the top left origin of the raster in the spatial reference system of the source, as a point object with
x
andy
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
[source]¶ Pixel width and height used for georeferencing the raster, as a point object with
x
andy
members. Seegeotransform
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
[source]¶ Skew coefficients used to georeference the raster, as a point object with
x
andy
members. In case of north up images, these coefficients are both0
.>>> 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
[source]¶ 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
[source]¶ 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)[source]¶ Returns a warped version of this raster.
The warping parameters can be specified through the
ds_input
argument. The use ofds_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 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 isNearestNeighbor
, and the other allowed values areBilinear
,Cubic
,CubicSpline
,Lanczos
,Average
, andMode
.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 thegdalwarp
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
(srs, driver=None, name=None, resampling='NearestNeighbour', max_error=0.0)[source]¶ Transforms this raster to a different spatial reference system (
srs
), which may be aSpatialReference
object, or any other input accepted bySpatialReference
(including spatial reference WKT and PROJ strings, or an integer 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 thedriver
andname
arguments.The default resampling algorithm is
NearestNeighbour
but can be changed using theresampling
argument. The default maximum allowed error for resampling is 0.0 and can be changed using themax_error
argument. Consult thewarp
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_srs = SpatialReference(4326) >>> target = rst.transform(target_srs) >>> target.origin [-82.98492744885776, 27.601924753080144]
-
info
[source]¶ Returns a string with a summary of the raster. This is equivalent to the gdalinfo command line utility.
-
metadata
¶ 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'}}
-
GDALBand
¶
-
class
GDALBand
¶ GDALBand
instances are not created explicitly, but rather obtained from aGDALRaster
object, through itsbands
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 toTrue
, the statistics may be computed based on overviews or a subset of image tiles.If the
refresh
argument is set toTrue
, 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 therefresh
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
, andstd
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
.
-
datatype
(as_string=False)¶ The data type contained in the band, as an integer constant between 0 (Unknown) and 14. If
as_string
isTrue
, the data type is returned as a string. Check out the “GDAL Pixel Type” column in the datatype value table for possible values.
-
color_interp
(as_string=False)¶ The color interpretation for the band, as an integer between 0and 16. If
as_string
isTrue
, 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
, andGCI_YCbCr_CrBand
.GCI_YCbCr_CrBand
also representsGCI_Max
because both correspond to the integer 16, but onlyGCI_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 thedata
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 theoffset
andsize
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
¶ 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 8 bit unsigned integer 2 GDT_UInt16 16 bit unsigned integer 3 GDT_Int16 16 bit signed integer 4 GDT_UInt32 32 bit unsigned integer 5 GDT_Int32 32 bit signed integer 6 GDT_Float32 32 bit floating point 7 GDT_Float64 64 bit floating point 12 GDT_UInt64 64 bit unsigned integer (GDAL 3.5+) 13 GDT_Int64 64 bit signed integer (GDAL 3.5+) 14 GDT_Int8 8 bit signed integer (GDAL 3.7+)
-
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 toNone
.
-
papsz_options
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 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", ... "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 can access files stored in the filesystem, but also supports virtual filesystems to abstract accessing other kind of files, such as compressed, encrypted, or remote files.
Using memory-based 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("https://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")
Using other Virtual Filesystems¶
Depending on the local build of GDAL other virtual filesystems may be
supported. You can use them by prepending the provided path with the
appropriate /vsi*/
prefix. See the GDAL Virtual Filesystems
documentation for more details.
Compressed rasters¶
Instead decompressing the file and instantiating the resulting raster, GDAL can
directly access compressed files using the /vsizip/
, /vsigzip/
, or
/vsitar/
virtual filesystems:
>>> from django.contrib.gis.gdal import GDALRaster
>>> rst = GDALRaster("/vsizip/path/to/your/file.zip/path/to/raster.tif")
>>> rst = GDALRaster("/vsigzip/path/to/your/file.gz")
>>> rst = GDALRaster("/vsitar/path/to/your/file.tar/path/to/raster.tif")
Network rasters¶
GDAL can support online resources and storage providers transparently. As long as it’s built with such capabilities.
To access a public raster file with no authentication, you can use
/vsicurl/
:
>>> from django.contrib.gis.gdal import GDALRaster
>>> rst = GDALRaster("/vsicurl/https://example.com/raster.tif")
>>> rst.name
'/vsicurl/https://example.com/raster.tif'
For commercial storage providers (e.g. /vsis3/
) the system should be
previously configured for authentication and possibly other settings (see the
GDAL Virtual Filesystems documentation for available options).