GDAL stands for Geospatial Data Abstraction Library,\nand is a veritable «Swiss army knife» of GIS data functionality. A subset\nof GDAL is the OGR Simple Features Library, which specializes\nin reading and writing vector geographic data in a variety of standard\nformats.
\nGeoDjango provides a high-level Python interface for some of the\ncapabilities of OGR, including the reading and coordinate transformation\nof vector spatial data and minimal support for GDAL’s features with respect\nto raster (image) data.
\n\nThe GDAL/OGR tools described here are designed to help you read in\nyour geospatial data, in order for most of them to be useful you have\nto have some data to work with. If you’re starting out and don’t yet\nhave any data of your own to use, GeoDjango tests contain a number of\ndata sets that you can use for testing. You can download them here:
\n$ wget https://raw.githubusercontent.com/django/django/main/tests/gis_tests/data/cities/cities.{shp,prj,shx,dbf}\n$ wget https://raw.githubusercontent.com/django/django/main/tests/gis_tests/data/rasters/raster.tif\nDataSourceLink to this heading#DataSource is a wrapper for the OGR data source object that\nsupports reading data from a variety of OGR-supported geospatial file\nformats and data sources using a consistent interface. Each\ndata source is represented by a DataSource object which contains\none or more layers of data. Each layer, represented by a Layer\nobject, contains some number of geographic features (Feature),\ninformation about the type of features contained in that layer (e.g.\npoints, polygons, etc.), as well as the names and types of any\nadditional fields (Field) of data that may be associated with\neach feature in that layer.
The constructor for DataSource only requires one parameter: the path of\nthe file you want to read. However, OGR also supports a variety of more\ncomplex data sources, including databases, that may be accessed by passing\na special name string instead of a path. For more information, see the\nOGR Vector Formats documentation. The name property of a\nDataSource instance gives the OGR name of the underlying data source\nthat it is using.
The optional encoding parameter allows you to specify a non-standard\nencoding of the strings in the source. This is typically useful when you\nobtain DjangoUnicodeDecodeError exceptions while reading field values.
Once you’ve created your DataSource, you can find out how many layers\nof data it contains by accessing the layer_count property, or\n(equivalently) by using the len() function. For information on\naccessing the layers of data themselves, see the next section:
>>> from django.contrib.gis.gdal import DataSource\n>>> ds = DataSource("/path/to/your/cities.shp")\n>>> ds.name\n'/path/to/your/cities.shp'\n>>> ds.layer_count # This file only contains one layer\n1\nReturns the number of layers in the data source.
\nReturns the name of the data source.
\nLayerLink to this heading#Layer is a wrapper for a layer of data in a DataSource object. You\nnever create a Layer object directly. Instead, you retrieve them from\na DataSource object, which is essentially a standard Python\ncontainer of Layer objects. For example, you can access a specific\nlayer by its index (e.g. ds[0] to access the first layer), or you can\niterate over all the layers in the container in a for loop. The\nLayer itself acts as a container for geometric features.
Typically, all the features in a given layer have the same geometry type.\nThe geom_type property of a layer is an OGRGeomType that\nidentifies the feature type. We can use it to print out some basic\ninformation about each layer in a DataSource:
>>> for layer in ds:\n... print('Layer "%s": %i %ss' % (layer.name, len(layer), layer.geom_type.name))\n...\nLayer "cities": 3 Points\nThe example output is from the cities data source, loaded above, which\nevidently contains one layer, called "cities", which contains three\npoint features. For simplicity, the examples below assume that you’ve\nstored that layer in the variable layer:
>>> layer = ds[0]\nReturns the name of this layer in the data source.
\n>>> layer.name\n'cities'\nReturns the number of features in the layer. Same as len(layer):
>>> layer.num_feat\n3\nReturns the geometry type of the layer, as an OGRGeomType object:
>>> layer.geom_type.name\n'Point'\nReturns the number of fields in the layer, i.e the number of fields of\ndata associated with each feature in the layer:
\n>>> layer.num_fields\n4\nReturns a list of the names of each of the fields in this layer:
\n>>> layer.fields\n['Name', 'Population', 'Density', 'Created']\nReturns a list of the data types of each of the fields in this layer. These\nare subclasses of Field, discussed below:
>>> [ft.__name__ for ft in layer.field_types]\n['OFTString', 'OFTReal', 'OFTReal', 'OFTDate']\nReturns a list of the maximum field widths for each of the fields in this\nlayer:
\n>>> layer.field_widths\n[80, 11, 24, 10]\nReturns a list of the numeric precisions for each of the fields in this\nlayer. This is meaningless (and set to zero) for non-numeric fields:
\n>>> layer.field_precisions\n[0, 0, 15, 0]\nReturns the spatial extent of this layer, as an Envelope object:
>>> layer.extent.tuple\n(-104.609252, 29.763374, -95.23506, 38.971823)\nProperty that returns the SpatialReference associated with this\nlayer:
>>> print(layer.srs)\nGEOGCS["GCS_WGS_1984",\n DATUM["WGS_1984",\n SPHEROID["WGS_1984",6378137,298.257223563]],\n PRIMEM["Greenwich",0],\n UNIT["Degree",0.017453292519943295]]\nIf the Layer has no spatial reference information associated\nwith it, None is returned.
Property that may be used to retrieve or set a spatial filter for this\nlayer. A spatial filter can only be set with an OGRGeometry\ninstance, a 4-tuple extent, or None. When set with something other than\nNone, only features that intersect the filter will be returned when\niterating over the layer:
>>> print(layer.spatial_filter)\nNone\n>>> print(len(layer))\n3\n>>> [feat.get("Name") for feat in layer]\n['Pueblo', 'Lawrence', 'Houston']\n>>> ks_extent = (-102.051, 36.99, -94.59, 40.00) # Extent for state of Kansas\n>>> layer.spatial_filter = ks_extent\n>>> len(layer)\n1\n>>> [feat.get("Name") for feat in layer]\n['Lawrence']\n>>> layer.spatial_filter = None\n>>> len(layer)\n3\nA method that returns a list of the values of a given field for each\nfeature in the layer:
\n>>> layer.get_fields("Name")\n['Pueblo', 'Lawrence', 'Houston']\nA method that returns a list containing the geometry of each feature in the\nlayer. If the optional argument geos is set to True then the\ngeometries are converted to GEOSGeometry\nobjects. Otherwise, they are returned as OGRGeometry objects:
>>> [pt.tuple for pt in layer.get_geoms()]\n[(-104.609252, 38.255001), (-95.23506, 38.971823), (-95.363151, 29.763374)]\nReturns a boolean indicating whether this layer supports the given\ncapability (a string). Examples of valid capability strings include:\n'RandomRead', 'SequentialWrite', 'RandomWrite',\n'FastSpatialFilter', 'FastFeatureCount', 'FastGetExtent',\n'CreateField', 'Transactions', 'DeleteFeature', and\n'FastSetNextByIndex'.
FeatureLink to this heading#Feature wraps an OGR feature. You never create a Feature object\ndirectly. Instead, you retrieve them from a Layer object. Each\nfeature consists of a geometry and a set of fields containing additional\nproperties. The geometry of a field is accessible via its geom property,\nwhich returns an OGRGeometry object. A Feature behaves like a\nstandard Python container for its fields, which it returns as Field\nobjects: you can access a field directly by its index or name, or you can\niterate over a feature’s fields, e.g. in a for loop.
Returns the geometry for this feature, as an OGRGeometry object:
>>> city.geom.tuple\n(-104.609252, 38.255001)\nA method that returns the value of the given field (specified by name)\nfor this feature, not a Field wrapper object:
>>> city.get("Population")\n102121\nReturns the type of geometry for this feature, as an OGRGeomType\nobject. This will be the same for all features in a given layer and is\nequivalent to the Layer.geom_type property of the Layer\nobject the feature came from.
Returns the number of fields of data associated with the feature. This will\nbe the same for all features in a given layer and is equivalent to the\nLayer.num_fields property of the Layer object the feature\ncame from.
Returns a list of the names of the fields of data associated with the\nfeature. This will be the same for all features in a given layer and is\nequivalent to the Layer.fields property of the Layer\nobject the feature came from.
Returns the feature identifier within the layer:
\n>>> city.fid\n0\nReturns the name of the Layer that the feature came from. This\nwill be the same for all features in a given layer:
>>> city.layer_name\n'cities'\nA method that returns the index of the given field name. This will be the\nsame for all features in a given layer:
\n>>> city.index("Population")\n1\nFieldLink to this heading#Returns the name of this field:
\n>>> city["Name"].name\n'Name'\nReturns the OGR type of this field, as an integer. The FIELD_CLASSES\ndictionary maps these values onto subclasses of Field:
>>> city["Density"].type\n2\nReturns a string with the name of the data type of this field:
\n>>> city["Name"].type_name\n'String'\nReturns the value of this field. The Field class itself returns the\nvalue as a string, but each subclass returns the value in the most\nappropriate form:
>>> city["Population"].value\n102121\nReturns the width of this field:
\n>>> city["Name"].width\n80\nReturns the numeric precision of this field. This is meaningless (and set\nto zero) for non-numeric fields:
\n>>> city["Density"].precision\n15\nReturns the value of the field as a double (float):
\n>>> city["Density"].as_double()\n874.7\nReturns the value of the field as an integer:
\n>>> city["Population"].as_int()\n102121\nReturns the value of the field as a string:
\n>>> city["Name"].as_string()\n'Pueblo'\nReturns the value of the field as a tuple of date and time components:
\n>>> city["Created"].as_datetime()\n(c_long(1999), c_long(5), c_long(23), c_long(0), c_long(0), c_long(0), c_long(0))\nDriverLink to this heading#The Driver class is used internally to wrap an OGR DataSource\ndriver.
Returns the number of OGR vector drivers currently registered.
\nOGRGeometryLink to this heading#OGRGeometry objects share similar functionality with\nGEOSGeometry objects and are thin wrappers\naround OGR’s internal geometry representation. Thus, they allow for more\nefficient access to data when using DataSource. Unlike its GEOS\ncounterpart, OGRGeometry supports spatial reference systems and\ncoordinate transformation:
>>> from django.contrib.gis.gdal import OGRGeometry\n>>> polygon = OGRGeometry("POLYGON((0 0, 5 0, 5 5, 0 5))")\nThis object is a wrapper for the OGR Geometry class. These objects are\ninstantiated directly from the given geom_input parameter, which may be\na string containing WKT, HEX, GeoJSON, a buffer containing WKB data, or\nan OGRGeomType object. These objects are also returned from the\nFeature.geom attribute, when reading vector data from\nLayer (which is in turn a part of a DataSource).
Constructs an OGRGeometry from the given GML string.
Constructs a Polygon from the given bounding-box (a 4-tuple).
Returns the number of points in a LineString, the number of rings\nin a Polygon, or the number of geometries in a\nGeometryCollection. Not applicable to other geometry types.
Iterates over the points in a LineString, the rings in a\nPolygon, or the geometries in a GeometryCollection.\nNot applicable to other geometry types.
Returns the point at the specified index for a LineString, the\ninterior ring at the specified index for a Polygon, or the geometry\nat the specified index in a GeometryCollection. Not applicable to\nother geometry types.
Returns the number of coordinated dimensions of the geometry, i.e. 0\nfor points, 1 for lines, and so forth:
\n>>> polygon.dimension\n2\nReturns or sets the coordinate dimension of this geometry. For example, the\nvalue would be 2 for two-dimensional geometries.
\nReturns the number of elements in this geometry:
\n>>> polygon.geom_count\n1\nReturns the number of points used to describe this geometry:
\n>>> polygon.point_count\n4\nAlias 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:
\n>>> polygon.geom_name\n'POLYGON'\nReturns the area of this geometry, or 0 for geometries that do not contain\nan area:
\n>>> polygon.area\n25.0\nReturns the envelope of this geometry, as an Envelope object.
Returns the envelope of this geometry as a 4-tuple, instead of as an\nEnvelope object:
>>> point.extent\n(0.0, 0.0, 5.0, 5.0)\nThis property controls the spatial reference for this geometry, or\nNone if no spatial reference system has been assigned to it.\nIf assigned, accessing this property returns a SpatialReference\nobject. It may be set with another SpatialReference object,\nor any input that SpatialReference accepts. Example:
>>> city.geom.srs.name\n'GCS_WGS_1984'\nReturns or sets the spatial reference identifier corresponding to\nSpatialReference of this geometry. Returns None if\nthere is no spatial reference information associated with this\ngeometry, or if an SRID cannot be determined.
Returns a GEOSGeometry object\ncorresponding to this geometry.
Returns a string representation of this geometry in GML format:
\n>>> OGRGeometry("POINT(1 2)").gml\n'<gml:Point><gml:coordinates>1,2</gml:coordinates></gml:Point>'\nReturns a string representation of this geometry in HEX WKB format:
\n>>> OGRGeometry("POINT(1 2)").hex\n'0101000000000000000000F03F0000000000000040'\nReturns a string representation of this geometry in JSON format:
\n>>> OGRGeometry("POINT(1 2)").json\n'{ "type": "Point", "coordinates": [ 1.000000, 2.000000 ] }'\nReturns a string representation of this geometry in KML format.
\nReturns the size of the WKB buffer needed to hold a WKB representation\nof this geometry:
\n>>> OGRGeometry("POINT(1 2)").wkb_size\n21\nReturns a buffer containing a WKB representation of this geometry.
Returns a string representation of this geometry in WKT format.
\nReturns the EWKT representation of this geometry.
\nReturns a new OGRGeometry clone of this geometry object.
If there are any rings within this geometry that have not been closed,\nthis routine will do so by adding the starting point to the end:
\n>>> triangle = OGRGeometry("LINEARRING (0 0,0 1,1 0)")\n>>> triangle.close_rings()\n>>> triangle.wkt\n'LINEARRING (0 0,0 1,1 0,0 0)'\nTransforms this geometry to a different spatial reference system. May take\na CoordTransform object, a SpatialReference object, or\nany other input accepted by SpatialReference (including spatial\nreference WKT and PROJ strings, or an integer SRID).
By default nothing is returned and the geometry is transformed in-place.\nHowever, if the clone keyword is set to True then a transformed\nclone of this geometry is returned instead.
Returns True if this geometry intersects the other, otherwise returns\nFalse.
Returns True if this geometry is equivalent to the other, otherwise\nreturns False.
Returns True if this geometry is spatially disjoint to (i.e. does\nnot intersect) the other, otherwise returns False.
Returns True if this geometry touches the other, otherwise returns\nFalse.
Returns True if this geometry crosses the other, otherwise returns\nFalse.
Returns True if this geometry is contained within the other, otherwise\nreturns False.
Returns True if this geometry contains the other, otherwise returns\nFalse.
Returns True if this geometry overlaps the other, otherwise returns\nFalse.
The boundary of this geometry, as a new OGRGeometry object.
The smallest convex polygon that contains this geometry, as a new\nOGRGeometry object.
Returns the region consisting of the difference of this geometry and\nthe other, as a new OGRGeometry object.
Returns the region consisting of the intersection of this geometry and\nthe other, as a new OGRGeometry object.
Returns the region consisting of the symmetric difference of this\ngeometry and the other, as a new OGRGeometry object.
Returns the region consisting of the union of this geometry and\nthe other, as a new OGRGeometry object.
Returns the coordinates of a point geometry as a tuple, the\ncoordinates of a line geometry as a tuple of tuples, and so forth:
\n>>> OGRGeometry("POINT (1 2)").tuple\n(1.0, 2.0)\n>>> OGRGeometry("LINESTRING (1 2,3 4)").tuple\n((1.0, 2.0), (3.0, 4.0))\nAn alias for tuple.
Returns the X coordinate of this point:
\n>>> OGRGeometry("POINT (1 2)").x\n1.0\nReturns the Y coordinate of this point:
\n>>> OGRGeometry("POINT (1 2)").y\n2.0\nReturns the Z coordinate of this point, or None if the point does not\nhave a Z coordinate:
>>> OGRGeometry("POINT (1 2 3)").z\n3.0\nReturns a list of X coordinates in this line:
\n>>> OGRGeometry("LINESTRING (1 2,3 4)").x\n[1.0, 3.0]\nReturns a list of Y coordinates in this line:
\n>>> OGRGeometry("LINESTRING (1 2,3 4)").y\n[2.0, 4.0]\nReturns a list of Z coordinates in this line, or None if the line does\nnot have Z coordinates:
>>> OGRGeometry("LINESTRING (1 2 3,4 5 6)").z\n[3.0, 6.0]\nReturns the shell or exterior ring of this polygon, as a LinearRing\ngeometry.
An alias for shell.
Returns a Point representing the centroid of this polygon.
Adds a geometry to this geometry collection. Not applicable to other\ngeometry types.
\nOGRGeomTypeLink to this heading#This class allows for the representation of an OGR geometry type\nin any of several ways:
\n>>> from django.contrib.gis.gdal import OGRGeomType\n>>> gt1 = OGRGeomType(3) # Using an integer for the type\n>>> gt2 = OGRGeomType("Polygon") # Using a string\n>>> gt3 = OGRGeomType("POLYGON") # It's case-insensitive\n>>> print(gt1 == 3, gt1 == "Polygon") # Equivalence works w/non-OGRGeomType objects\nTrue True\nReturns a short-hand string form of the OGR Geometry type:
\n>>> gt1.name\n'Polygon'\nReturns the number corresponding to the OGR geometry type:
\n>>> gt1.num\n3\nReturns the Django field type (a subclass of GeometryField) to use for\nstoring this OGR type, or None if there is no appropriate Django type:
>>> gt1.django\n'PolygonField'\nEnvelopeLink to this heading#Represents an OGR Envelope structure that contains the minimum and maximum\nX, Y coordinates for a rectangle bounding box. The naming of the variables\nis compatible with the OGR Envelope C structure.
\nThe value of the minimum X coordinate.
\nThe value of the maximum X coordinate.
\nThe value of the minimum Y coordinate.
\nThe value of the maximum Y coordinate.
\nThe upper-right coordinate, as a tuple.
\nThe lower-left coordinate, as a tuple.
\nA tuple representing the envelope.
\nA string representing this envelope as a polygon in WKT format.
\nSpatialReferenceLink to this heading#Spatial reference objects are initialized on the given srs_input,\nwhich 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',\n'NAD27', 'NAD83')
Example:
\n>>> wgs84 = SpatialReference("WGS84") # shorthand string\n>>> wgs84 = SpatialReference(4326) # EPSG code\n>>> wgs84 = SpatialReference("EPSG:4326") # EPSG string\n>>> proj = "+proj=longlat +ellps=WGS84 +datum=WGS84 +no_defs "\n>>> wgs84 = SpatialReference(proj) # PROJ string\n>>> wgs84 = SpatialReference(\n... """GEOGCS["WGS 84",\n... DATUM["WGS_1984",\n... SPHEROID["WGS 84",6378137,298.257223563,\n... AUTHORITY["EPSG","7030"]],\n... AUTHORITY["EPSG","6326"]],\n... PRIMEM["Greenwich",0,\n... AUTHORITY["EPSG","8901"]],\n... UNIT["degree",0.01745329251994328,\n... AUTHORITY["EPSG","9122"]],\n... AUTHORITY["EPSG","4326"]]"""\n... ) # OGC WKT\nReturns the value of the given string attribute node, None if the node\ndoesn’t exist. Can also take a tuple as a parameter, (target, child), where\nchild is the index of the attribute in the WKT. For example:
>>> wkt = 'GEOGCS["WGS 84", DATUM["WGS_1984, ... AUTHORITY["EPSG","4326"]]'\n>>> srs = SpatialReference(wkt) # could also use 'WGS84', or 4326\n>>> print(srs["GEOGCS"])\nWGS 84\n>>> print(srs["DATUM"])\nWGS_1984\n>>> print(srs["AUTHORITY"])\nEPSG\n>>> print(srs["AUTHORITY", 1]) # The authority value\n4326\n>>> print(srs["TOWGS84", 4]) # the fourth value in this wkt\n0\n>>> print(srs["UNIT|AUTHORITY"]) # For the units authority, have to use the pipe symbol.\nEPSG\n>>> print(srs["UNIT|AUTHORITY", 1]) # The authority value for the units\n9122\nThe attribute value for the given target node (e.g. 'PROJCS').\nThe index keyword specifies an index of the child node to return.
Returns the authority name for the given string target node.
\nReturns the authority code for the given string target node.
\nReturns a clone of this spatial reference object.
\nThis method inspects the WKT of this SpatialReference and will add EPSG\nauthority nodes where an EPSG identifier is applicable.
Morphs this SpatialReference from ESRI’s format to EPSG
\nMorphs this SpatialReference to ESRI’s format.
\nChecks to see if the given spatial reference is valid, if not\nan exception will be raised.
\nImport spatial reference from EPSG code.
\nImport spatial reference from PROJ string.
\nImport spatial reference from WKT.
\nImport spatial reference from XML.
\nReturns the name of this Spatial Reference.
\nReturns the SRID of top-level authority, or None if undefined.
Returns the name of the linear units.
\nReturns the value of the linear units.
\nReturns the name of the angular units.»
\nReturns the value of the angular units.
\nReturns a 2-tuple of the units value and the units name and will\nautomatically determines whether to return the linear or angular units.
\nReturns a tuple of the ellipsoid parameters for this spatial reference:\n(semimajor axis, semiminor axis, and inverse flattening).
\nReturns the semi major axis of the ellipsoid for this spatial reference.
\nReturns the semi minor axis of the ellipsoid for this spatial reference.
\nReturns the inverse flattening of the ellipsoid for this spatial reference.
\nReturns True if this spatial reference is geographic (root node is\nGEOGCS).
Returns True if this spatial reference is local (root node is\nLOCAL_CS).
Returns True if this spatial reference is a projected coordinate system\n(root node is PROJCS).
Returns the WKT representation of this spatial reference.
\nReturns the “pretty” representation of the WKT.
\nReturns the PROJ representation for this spatial reference.
\nAlias for SpatialReference.proj.
Returns the XML representation of this spatial reference.
\nCoordTransformLink to this heading#Represents a coordinate system transform. It is initialized with two\nSpatialReference, representing the source and target coordinate\nsystems, respectively. These objects should be used when performing the same\ncoordinate transformation repeatedly on different geometries:
>>> ct = CoordTransform(SpatialReference("WGS84"), SpatialReference("NAD83"))\n>>> for feat in layer:\n... geom = feat.geom # getting clone of feature geometry\n... geom.transform(ct) # transforming\n...\nGDALRasterLink to this heading#GDALRaster is a wrapper for the GDAL raster source object that\nsupports reading data from a variety of GDAL-supported geospatial file\nformats and data sources using a consistent interface. Each\ndata source is represented by a GDALRaster object which contains\none or more layers of data named bands. Each band, represented by a\nGDALBand object, contains georeferenced image data. For example, an RGB\nimage is represented as three bands: one for red, one for green, and one for\nblue.
The constructor for GDALRaster accepts two parameters. The first\nparameter defines the raster source, and the second parameter defines if a\nraster should be opened in write mode. For newly-created rasters, the second\nparameter is ignored and the new raster is always created in write mode.
The first parameter can take three forms: a string or\nPath representing a file path (filesystem or GDAL virtual\nfilesystem), a dictionary with values defining a new raster, or a bytes\nobject representing a raster file.
If the input is a file path, the raster is opened from there. If the input\nis raw data in a dictionary, the parameters width, height, and\nsrid are required. If the input is a bytes object, it will be opened\nusing a GDAL virtual filesystem.
For a detailed description of how to create rasters using dictionary input,\nsee Creating rasters from data. For a detailed description of how to\ncreate rasters in the virtual filesystem, see Using GDAL’s Virtual Filesystem.
\nThe following example shows how rasters can be created from different input\nsources (using the sample data from the GeoDjango tests; see also the\nSample Data section).
\n>>> from django.contrib.gis.gdal import GDALRaster\n>>> rst = GDALRaster("/path/to/your/raster.tif", write=False)\n>>> rst.name\n'/path/to/your/raster.tif'\n>>> rst.width, rst.height # This file has 163 x 174 pixels\n(163, 174)\n>>> rst = GDALRaster(\n... { # Creates an in-memory raster\n... "srid": 4326,\n... "width": 4,\n... "height": 4,\n... "datatype": 1,\n... "bands": [\n... {\n... "data": (2, 3),\n... "offset": (1, 1),\n... "size": (2, 2),\n... "shape": (2, 1),\n... "nodata_value": 5,\n... }\n... ],\n... }\n... )\n>>> rst.srs.srid\n4326\n>>> rst.width, rst.height\n(4, 4)\n>>> rst.bands[0].data()\narray([[5, 5, 5, 5],\n [5, 2, 3, 5],\n [5, 2, 3, 5],\n [5, 5, 5, 5]], dtype=uint8)\n>>> rst_file = open("/path/to/your/raster.tif", "rb")\n>>> rst_bytes = rst_file.read()\n>>> rst = GDALRaster(rst_bytes)\n>>> rst.is_vsi_based\nTrue\n>>> rst.name # Stored in a random path in the vsimem filesystem.\n'/vsimem/da300bdb-129d-49a8-b336-e410a9428dad'\nThe name of the source which is equivalent to the input file path or the name\nprovided upon instantiation.
\n>>> GDALRaster({"width": 10, "height": 10, "name": "myraster", "srid": 4326}).name\n'myraster'\nThe name of the GDAL driver used to handle the input file. For GDALRasters created\nfrom a file, the driver type is detected automatically. The creation of rasters from\nscratch is an in-memory raster by default ('MEM'), but can be\naltered as needed. For instance, use GTiff for a GeoTiff file.\nFor a list of file types, see also the GDAL Raster Formats list.
An in-memory raster is created through the following example:
\n>>> GDALRaster({"width": 10, "height": 10, "srid": 4326}).driver.name\n'MEM'\nA file based GeoTiff raster is created through the following example:
\n>>> import tempfile\n>>> rstfile = tempfile.NamedTemporaryFile(suffix=".tif")\n>>> rst = GDALRaster(\n... {\n... "driver": "GTiff",\n... "name": rstfile.name,\n... "srid": 4326,\n... "width": 255,\n... "height": 255,\n... "nr_of_bands": 1,\n... }\n... )\n>>> rst.name\n'/tmp/tmp7x9H4J.tif' # The exact filename will be different on your computer\n>>> rst.driver.name\n'GTiff'\nThe width of the source in pixels (X-axis).
\n>>> GDALRaster({"width": 10, "height": 20, "srid": 4326}).width\n10\nThe height of the source in pixels (Y-axis).
\n>>> GDALRaster({"width": 10, "height": 20, "srid": 4326}).height\n20\nThe spatial reference system of the raster, as a\nSpatialReference instance. The SRS can be changed by\nsetting it to an other SpatialReference or providing any input\nthat is accepted by the SpatialReference constructor.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.srs.srid\n4326\n>>> rst.srs = 3086\n>>> rst.srs.srid\n3086\nThe Spatial Reference System Identifier (SRID) of the raster. This\nproperty is a shortcut to getting or setting the SRID through the\nsrs attribute.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.srid\n4326\n>>> rst.srid = 3086\n>>> rst.srid\n3086\n>>> rst.srs.srid # This is equivalent\n3086\nThe affine transformation matrix used to georeference the source, as a\ntuple of six coefficients which map pixel/line coordinates into\ngeoreferenced space using the following relationship:
\nXgeo = GT(0) + Xpixel * GT(1) + Yline * GT(2)\nYgeo = GT(3) + Xpixel * GT(4) + Yline * GT(5)\nThe same values can be retrieved by accessing the origin\n(indices 0 and 3), scale (indices 1 and 5) and skew\n(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})\n>>> rst.geotransform\n[0.0, 1.0, 0.0, 0.0, 0.0, -1.0]\nCoordinates of the top left origin of the raster in the spatial\nreference system of the source, as a point object with x and y\nmembers.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.origin\n[0.0, 0.0]\n>>> rst.origin.x = 1\n>>> rst.origin\n[1.0, 0.0]\nPixel width and height used for georeferencing the raster, as a point\nobject with x and y members. See geotransform for more\ninformation.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.scale\n[1.0, -1.0]\n>>> rst.scale.x = 2\n>>> rst.scale\n[2.0, -1.0]\nSkew coefficients used to georeference the raster, as a point object\nwith x and y members. In case of north up images, these\ncoefficients are both 0.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.skew\n[0.0, 0.0]\n>>> rst.skew.x = 3\n>>> rst.skew\n[3.0, 0.0]\nExtent (boundary values) of the raster source, as a 4-tuple\n(xmin, ymin, xmax, ymax) in the spatial reference system of the\nsource.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.extent\n(0.0, -20.0, 10.0, 0.0)\n>>> rst.origin.x = 100\n>>> rst.extent\n(100.0, -20.0, 110.0, 0.0)\nList of all bands of the source, as GDALBand instances.
>>> rst = GDALRaster(\n... {\n... "width": 1,\n... "height": 2,\n... "srid": 4326,\n... "bands": [{"data": [0, 1]}, {"data": [2, 3]}],\n... }\n... )\n>>> len(rst.bands)\n2\n>>> rst.bands[1].data()\narray([[ 2., 3.]], dtype=float32)\nReturns a warped version of this raster.
\nThe warping parameters can be specified through the ds_input\nargument. The use of ds_input is analogous to the corresponding\nargument of the class constructor. It is a dictionary with the\ncharacteristics of the target raster. Allowed dictionary key values are\nwidth, height, SRID, origin, scale, skew, datatype, driver, and name\n(filename).
By default, the warp functions keeps most parameters equal to the\nvalues of the original source raster, so only parameters that should be\nchanged need to be specified. Note that this includes the driver, so\nfor file-based rasters the warp function will create a new raster on\ndisk.
\nThe only parameter that is set differently from the source raster is the\nname. The default value of the raster name is the name of the source\nraster appended with '_copy' + source_driver_name. For file-based\nrasters it is recommended to provide the file path of the target raster.
The resampling algorithm used for warping can be specified with the\nresampling argument. The default is NearestNeighbor, and the\nother allowed values are Bilinear, Cubic, CubicSpline,\nLanczos, Average, and Mode.
The max_error argument can be used to specify the maximum error\nmeasured in input pixels that is allowed in approximating the\ntransformation. The default is 0.0 for exact calculations.
For users familiar with GDAL, this function has a similar\nfunctionality to the gdalwarp command-line utility.
For example, the warp function can be used for aggregating a raster to\nthe double of its original pixel scale:
\n>>> rst = GDALRaster(\n... {\n... "width": 6,\n... "height": 6,\n... "srid": 3086,\n... "origin": [500000, 400000],\n... "scale": [100, -100],\n... "bands": [{"data": range(36), "nodata_value": 99}],\n... }\n... )\n>>> target = rst.warp({"scale": [200, -200], "width": 3, "height": 3})\n>>> target.bands[0].data()\narray([[ 7., 9., 11.],\n [ 19., 21., 23.],\n [ 31., 33., 35.]], dtype=float32)\nTransforms this raster to a different spatial reference system\n(srs), which may be a SpatialReference object, or any\nother input accepted by SpatialReference (including spatial\nreference WKT and PROJ strings, or an integer SRID).
It calculates the bounds and scale of the current raster in the new\nspatial reference system and warps the raster using the\nwarp function.
By default, the driver of the source raster is used and the name of the\nraster is the original name appended with\n'_copy' + source_driver_name. A different driver or name can be\nspecified with the driver and name arguments.
The default resampling algorithm is NearestNeighbour but can be\nchanged using the resampling argument. The default maximum allowed\nerror for resampling is 0.0 and can be changed using the max_error\nargument. Consult the warp documentation for detail\non those arguments.
>>> rst = GDALRaster(\n... {\n... "width": 6,\n... "height": 6,\n... "srid": 3086,\n... "origin": [500000, 400000],\n... "scale": [100, -100],\n... "bands": [{"data": range(36), "nodata_value": 99}],\n... }\n... )\n>>> target_srs = SpatialReference(4326)\n>>> target = rst.transform(target_srs)\n>>> target.origin\n[-82.98492744885776, 27.601924753080144]\nReturns a string with a summary of the raster. This is equivalent to\nthe gdalinfo command line utility.
\nThe metadata of this raster, represented as a nested dictionary. The\nfirst-level key is the metadata domain. The second-level contains the\nmetadata item names and values from each domain.
\nTo set or update a metadata item, pass the corresponding metadata item\nto the method using the nested structure described above. Only keys\nthat are in the specified dictionary are updated; the rest of the\nmetadata remains unchanged.
\nTo remove a metadata item, use None as the metadata value.
>>> rst = GDALRaster({"width": 10, "height": 20, "srid": 4326})\n>>> rst.metadata\n{}\n>>> rst.metadata = {"DEFAULT": {"OWNER": "Django", "VERSION": "1.0"}}\n>>> rst.metadata\n{'DEFAULT': {'OWNER': 'Django', 'VERSION': '1.0'}}\n>>> rst.metadata = {"DEFAULT": {"OWNER": None, "VERSION": "2.0"}}\n>>> rst.metadata\n{'DEFAULT': {'VERSION': '2.0'}}\nA bytes representation of this raster. Returns None for rasters\nthat are not stored in GDAL’s virtual filesystem.
A boolean indicating if this raster is stored in GDAL’s virtual\nfilesystem.
\nGDALBandLink to this heading#GDALBand instances are not created explicitly, but rather obtained\nfrom a GDALRaster object, through its bands\nattribute. The GDALBands contain the actual pixel values of the raster.
The name or description of the band, if any.
\nThe width of the band in pixels (X-axis).
\nThe height of the band in pixels (Y-axis).
\nThe total number of pixels in this band. Is equal to width * height.
Compute statistics on the pixel values of this band. The return value\nis a tuple with the following structure:\n(minimum, maximum, mean, standard deviation).
If the approximate argument is set to True, the statistics may\nbe computed based on overviews or a subset of image tiles.
If the refresh argument is set to True, the statistics will be\ncomputed from the data directly, and the cache will be updated with the\nresult.
If a persistent cache value is found, that value is returned. For\nraster formats using Persistent Auxiliary Metadata (PAM) services, the\nstatistics might be cached in an auxiliary file. In some cases this\nmetadata might be out of sync with the pixel values or cause values\nfrom a previous call to be returned which don’t reflect the value of\nthe approximate argument. In such cases, use the refresh\nargument to get updated values and store them in the cache.
For empty bands (where all pixel values are «no data»), all statistics\nare returned as None.
The statistics can also be retrieved directly by accessing the\nmin, max, mean, and std properties.
The minimum pixel value of the band (excluding the «no data» value).
\nThe maximum pixel value of the band (excluding the «no data» value).
\nThe mean of all pixel values of the band (excluding the «no data»\nvalue).
\nThe standard deviation of all pixel values of the band (excluding the\n«no data» value).
\nThe «no data» value for a band is generally a special marker value used\nto mark pixels that are not valid data. Such pixels should generally not\nbe displayed, nor contribute to analysis operations.
\nTo delete an existing «no data» value, set this property to None.
The data type contained in the band, as an integer constant between 0\n(Unknown) and 14. If as_string is True, the data type is\nreturned as a string. Check out the «GDAL Pixel Type» column in the\ndatatype value table for possible values.
The color interpretation for the band, as an integer between 0and 16.\nIf as_string is True, the data type is returned as a string\nwith the following possible values:\nGCI_Undefined, GCI_GrayIndex, GCI_PaletteIndex,\nGCI_RedBand, GCI_GreenBand, GCI_BlueBand, GCI_AlphaBand,\nGCI_HueBand, GCI_SaturationBand, GCI_LightnessBand,\nGCI_CyanBand, GCI_MagentaBand, GCI_YellowBand,\nGCI_BlackBand, GCI_YCbCr_YBand, GCI_YCbCr_CbBand, and\nGCI_YCbCr_CrBand. GCI_YCbCr_CrBand also represents GCI_Max\nbecause both correspond to the integer 16, but only GCI_YCbCr_CrBand\nis returned as a string.
The accessor to the pixel values of the GDALBand. Returns the complete\ndata array if no parameters are provided. A subset of the pixel array can\nbe requested by specifying an offset and block size as tuples.
If NumPy is available, the data is returned as NumPy array. For performance\nreasons, it is highly recommended to use NumPy.
\nData is written to the GDALBand if the data parameter is provided.\nThe input can be of one of the following types - packed string, buffer, list,\narray, and NumPy array. The number of items in the input should normally\ncorrespond to the total number of pixels in the band, or to the number\nof pixels for a specific block of pixel values if the offset and\nsize parameters are provided.
If the number of items in the input is different from the target pixel\nblock, the shape parameter must be specified. The shape is a tuple\nthat specifies the width and height of the input data in pixels. The\ndata is then replicated to update the pixel values of the selected\nblock. This is useful to fill an entire band with a single value, for\ninstance.
For example:
\n>>> rst = GDALRaster(\n... {"width": 4, "height": 4, "srid": 4326, "datatype": 1, "nr_of_bands": 1}\n... )\n>>> bnd = rst.bands[0]\n>>> bnd.data(range(16))\n>>> bnd.data()\narray([[ 0, 1, 2, 3],\n [ 4, 5, 6, 7],\n [ 8, 9, 10, 11],\n [12, 13, 14, 15]], dtype=int8)\n>>> bnd.data(offset=(1, 1), size=(2, 2))\narray([[ 5, 6],\n [ 9, 10]], dtype=int8)\n>>> bnd.data(data=[-1, -2, -3, -4], offset=(1, 1), size=(2, 2))\n>>> bnd.data()\narray([[ 0, 1, 2, 3],\n [ 4, -1, -2, 7],\n [ 8, -3, -4, 11],\n [12, 13, 14, 15]], dtype=int8)\n>>> bnd.data(data="\\x9d\\xa8\\xb3\\xbe", offset=(1, 1), size=(2, 2))\n>>> bnd.data()\narray([[ 0, 1, 2, 3],\n [ 4, -99, -88, 7],\n [ 8, -77, -66, 11],\n [ 12, 13, 14, 15]], dtype=int8)\n>>> bnd.data([1], shape=(1, 1))\n>>> bnd.data()\narray([[1, 1, 1, 1],\n [1, 1, 1, 1],\n [1, 1, 1, 1],\n [1, 1, 1, 1]], dtype=uint8)\n>>> bnd.data(range(4), shape=(1, 4))\narray([[0, 0, 0, 0],\n [1, 1, 1, 1],\n [2, 2, 2, 2],\n [3, 3, 3, 3]], dtype=uint8)\nThe metadata of this band. The functionality is identical to\nGDALRaster.metadata.
This section describes how to create rasters from scratch using the\nds_input parameter.
A new raster is created when a dict is passed to the GDALRaster\nconstructor. The dictionary contains defining parameters of the new raster,\nsuch as the origin, size, or spatial reference system. The dictionary can also\ncontain pixel data and information about the format of the new raster. The\nresulting raster can therefore be file-based or memory-based, depending on the\ndriver specified.
There’s no standard for describing raster data in a dictionary or JSON flavor.\nThe definition of the dictionary input to the GDALRaster class is\ntherefore specific to Django. It’s inspired by the geojson format, but the\ngeojson standard is currently limited to vector formats.
Examples of using the different keys when creating rasters can be found in the\ndocumentation of the corresponding attributes and methods of the\nGDALRaster and GDALBand classes.
ds_input dictionaryLink to this heading#Only a few keys are required in the ds_input dictionary to create a raster:\nwidth, height, and srid. All other parameters have default values\n(see the table below). The list of keys that can be passed in the ds_input\ndictionary is closely related but not identical to the GDALRaster\nproperties. Many of the parameters are mapped directly to those properties;\nthe others are described below.
The following table describes all keys that can be set in the ds_input\ndictionary.
Key | \nDefault | \nUsage | \n
|---|---|---|
| \nrequired | \nMapped to the | \n
| \nrequired | \nMapped to the | \n
| \nrequired | \nMapped to the | \n
| \n
| \nMapped to the | \n
| \n
| \nSee below | \n
| \n
| \nMapped to the | \n
| \n
| \nMapped to the | \n
| \n
| \nMapped to the | \n
| \n
| \nSee below | \n
| \n
| \nSee below | \n
| \n
| \nSee below | \n
| \n
| \nSee below | \n
String representing the name of the raster. When creating a file-based\nraster, this parameter must be the file path for the new raster. If the\nname starts with /vsimem/, the raster is created in GDAL’s virtual\nfilesystem.
Integer representing the data type for all the bands. Defaults to 6\n(Float32). All bands of a new raster are required to have the same datatype.\nThe value mapping is:
Value | \nGDAL Pixel Type | \nDescription | \n
|---|---|---|
1 | \nGDT_Byte | \n8 bit unsigned integer | \n
2 | \nGDT_UInt16 | \n16 bit unsigned integer | \n
3 | \nGDT_Int16 | \n16 bit signed integer | \n
4 | \nGDT_UInt32 | \n32 bit unsigned integer | \n
5 | \nGDT_Int32 | \n32 bit signed integer | \n
6 | \nGDT_Float32 | \n32 bit floating point | \n
7 | \nGDT_Float64 | \n64 bit floating point | \n
12 | \nGDT_UInt64 | \n64 bit unsigned integer (GDAL 3.5+) | \n
13 | \nGDT_Int64 | \n64 bit signed integer (GDAL 3.5+) | \n
14 | \nGDT_Int8 | \n8 bit signed integer (GDAL 3.7+) | \n
Integer representing the number of bands of the raster. A raster can be\ncreated without passing band data upon creation. If the number of bands\nisn’t specified, it’s automatically calculated from the length of the\nbands input. The number of bands can’t be changed after creation.
A list of band_input dictionaries with band input data. The resulting\nband indices are the same as in the list provided. The definition of the\nband input dictionary is given below. If band data isn’t provided, the\nraster bands values are instantiated as an array of zeros and the «no\ndata» value is set to None.
A dictionary with raster creation options. The key-value pairs of the\ninput dictionary are passed to the driver on creation of the raster.
\nThe available options are driver-specific and are described in the\ndocumentation of each driver.
\nThe values in the dictionary are not case-sensitive and are automatically\nconverted to the correct string format upon creation.
\nThe following example uses some of the options available for the\nGTiff driver. The result is a compressed raster with an internal tiling\nscheme. The internal tiles have a block size of 23 by 23:
\n>>> GDALRaster(\n... {\n... "driver": "GTiff",\n... "name": "/path/to/new/file.tif",\n... "srid": 4326,\n... "width": 255,\n... "height": 255,\n... "nr_of_bands": 1,\n... "papsz_options": {\n... "compress": "packbits",\n... "tiled": "yes",\n... "blockxsize": 23,\n... "blockysize": 23,\n... },\n... }\n... )\nband_input dictionaryLink to this heading#The bands key in the ds_input dictionary is a list of band_input\ndictionaries. Each band_input dictionary can contain pixel values and the\n«no data» value to be set on the bands of the new raster. The data array can\nhave the full size of the new raster or be smaller. For arrays that are smaller\nthan the full raster, the size, shape, and offset keys control the\npixel values. The corresponding keys are passed to the data()\nmethod. Their functionality is the same as setting the band data with that\nmethod. The following table describes the keys that can be used.
Key | \nDefault | \nUsage | \n
|---|---|---|
| \n
| \nMapped to the | \n
| \nSame as | \nPassed to the | \n
| \n
| \nPassed to the | \n
| \nSame as size | \nPassed to the | \n
| \n
| \nPassed to the | \n
GDAL can access files stored in the filesystem, but also supports virtual\nfilesystems to abstract accessing other kind of files, such as compressed,\nencrypted, or remote files.
\nGDAL has an internal memory-based filesystem, which allows treating blocks of\nmemory as files. It can be used to read and write GDALRaster objects\nto and from binary file buffers.
This is useful in web contexts where rasters might be obtained as a buffer\nfrom a remote storage or returned from a view without being written to disk.
\nGDALRaster objects are created in the virtual filesystem when a\nbytes object is provided as input, or when the file path starts with\n/vsimem/.
Input provided as bytes has to be a full binary representation of a file.\nFor instance:
# Read a raster as a file object from a remote source.\n>>> from urllib.request import urlopen\n>>> dat = urlopen("http://example.com/raster.tif").read()\n# Instantiate a raster from the bytes object.\n>>> rst = GDALRaster(dat)\n# The name starts with /vsimem/, indicating that the raster lives in the\n# virtual filesystem.\n>>> rst.name\n'/vsimem/da300bdb-129d-49a8-b336-e410a9428dad'\nTo create a new virtual file-based raster from scratch, use the ds_input\ndictionary representation and provide a name argument that starts with\n/vsimem/ (for detail of the dictionary representation, see\nCreating rasters from data). For virtual file-based rasters, the\nvsi_buffer attribute returns the bytes representation\nof the raster.
Here’s how to create a raster and return it as a file in an\nHttpResponse:
>>> from django.http import HttpResponse\n>>> rst = GDALRaster(\n... {\n... "name": "/vsimem/temporarymemfile",\n... "driver": "tif",\n... "width": 6,\n... "height": 6,\n... "srid": 3086,\n... "origin": [500000, 400000],\n... "scale": [100, -100],\n... "bands": [{"data": range(36), "nodata_value": 99}],\n... }\n... )\n>>> HttpResponse(rast.vsi_buffer, "image/tiff")\nDepending on the local build of GDAL other virtual filesystems may be\nsupported. You can use them by prepending the provided path with the\nappropriate /vsi*/ prefix. See the GDAL Virtual Filesystems\ndocumentation for more details.
Instead decompressing the file and instantiating the resulting raster, GDAL can\ndirectly access compressed files using the /vsizip/, /vsigzip/, or\n/vsitar/ virtual filesystems:
>>> from django.contrib.gis.gdal import GDALRaster\n>>> rst = GDALRaster("/vsizip/path/to/your/file.zip/path/to/raster.tif")\n>>> rst = GDALRaster("/vsigzip/path/to/your/file.gz")\n>>> rst = GDALRaster("/vsitar/path/to/your/file.tar/path/to/raster.tif")\nGDAL can support online resources and storage providers transparently. As long\nas it’s built with such capabilities.
\nTo access a public raster file with no authentication, you can use\n/vsicurl/:
>>> from django.contrib.gis.gdal import GDALRaster\n>>> rst = GDALRaster("/vsicurl/https://example.com/raster.tif")\n>>> rst.name\n'/vsicurl/https://example.com/raster.tif'\nFor commercial storage providers (e.g. /vsis3/) the system should be\npreviously configured for authentication and possibly other settings (see the\nGDAL Virtual Filesystems documentation for available options).
GDAL_LIBRARY_PATHLink to this heading#A string specifying the location of the GDAL library. Typically,\nthis setting is only used if the GDAL library is in a non-standard\nlocation (e.g., /home/john/lib/libgdal.so).
The base GDAL exception, indicating a GDAL-related error.
\nAn exception raised when an error occurs when constructing or using a\nspatial reference system object.
\n