Raster Dataset#

Opening a Raster#

Before opening a GDAL raster dataset, it is necessary to register drivers, then create a data object.

from osgeo import gdal

# Register drivers
gdal.AllRegister()

# TIF file
tif = r"C:\Temp\image.tif"

# Data object
dataset = gdal.Open(tif)

Raster Properties#

Dataset object properties: columns, rows, and band number.

# Get image size
cols = dataset.RasterXSize
rows = dataset.RasterYSize
bands = dataset.RasterCount

# Get projection
proj = dataset.GetProjection()

GetGeoTransform is a method on a dataset object to get projection info as a tuple with 6 numbers.

geotr = dataset.GetGeoTransform()  # returns (368000.0, 1000.0, 0.0, 8828000.0, 0.0, -1000.0)

Element	Description
`geotr[0]`	Top left X
`geotr[1]`	W-E pixel resolution
`geotr[2]`	Rotation, 0 if image is “north up”
`geotr[3]`	Top left Y
`geotr[4]`	Rotation, 0 if image is “north up”
`geotr[5]`	N-S pixel resolution

Value Properties#

Working with pixel values is limited to one TIF band.

band = dataset.GetRasterBand(1)

# Get NoData value
NoData = band.GetNoDataValue()

# Get min and max
vmin = band.GetMinimum()
vmax = band.GetMaximum()

# Get data type
vtype = band.DataType

# Get colour table
coltab = band.GetColorTable()

# Get statistics: min, max, mean, stddev
stats = band.GetStatistics(1, 1)

Raster data types:

Type	Description
`GDT_Byte`	Eight bit unsigned integer
`GDT_UInt16`	Sixteen bit unsigned integer
`GDT_Int16`	Sixteen bit signed integer
`GDT_UInt32`	Thirty two bit unsigned integer
`GDT_Int32`	Thirty two bit signed integer
`GDT_Float32`	Thirty two bit floating point
`GDT_Float64`	Sixty four bit floating point

Raster to Array#

# Convert to a 2D array
data = band.ReadAsArray()