I'm trying to remove bands from a raster I have. Basically I just want the first band as a new GeoTIFF file.

Here is my code so far:

from osgeo import gdal

# Load my GeoTIFF
tif = gdal.Open("mygeotif.tif")
tif.RasterCount == 3  # should return True

# Create a copy with same projection, etc.
gdal.GetDriverByName("GTiff").CreateCopy("mynewgeotif.tif", tif)
newtif = gdal.Open("mygeotif_band1.tif")

# remove bands 2 and 3 here

# Close files
newtif = None
tif = None

The biggest problem is that I can't find any documentation for writing out only one band or for removing bands.

  • I suppose that you need to start with GetRasterBand. Probably this cookbook helps you pcjericks.github.io/py-gdalogr-cookbook/raster_layers.html
    – user30184
    Apr 22, 2016 at 16:07
  • Doing something like newtif.GetRasterBand(1) will return a osgeo.gdal.Band vs a osgeo.gdal.Dataset and not have any of the geospatial information with it. Apr 22, 2016 at 16:25
  • Read the whole cookbook I suppose it includes advice for reading and copying geotransformation etc.
    – user30184
    Apr 22, 2016 at 16:30
  • Were you able to figure out a way to remove bands from a dataset? Mar 3, 2020 at 22:37

1 Answer 1


The next code only select the number 3 band (blue band) in a RGB raster and write it as blueband.tif.

from osgeo import gdal, osr
import os, struct
import numpy as np

layer = iface.activeLayer()

provider = layer.dataProvider()

path = provider.dataSourceUri()

fmttypes = {'Byte':'B', 'UInt16':'H', 'Int16':'h', 'UInt32':'I', 'Int32':'i', 'Float32':'f', 'Float64':'d'}

dataset = gdal.Open(path)

#Get projection
prj = dataset.GetProjection()

#setting number band (blue in this case)
number_band = 3

band = dataset.GetRasterBand(number_band)

geotransform = dataset.GetGeoTransform()

# Set name of output raster
if number_band == 3:
    output_file = "/home/zeito/pyqgis_data/blueband.tif"

# Create gtif file with rows and columns from parent raster 
driver = gdal.GetDriverByName("GTiff")

columns, rows = (band.XSize, band.YSize)

print "rows = %d columns = %d" % (columns, rows)

BandType = gdal.GetDataTypeName(band.DataType)

print "Band Type = ", BandType

raster = []

for y in range(band.YSize):

    scanline = band.ReadRaster(0, y, band.XSize, 1, band.XSize, 1, band.DataType)
    values = struct.unpack(fmttypes[BandType] * band.XSize, scanline)

dst_ds = driver.Create(output_file, 

#flattened list of raster values
raster = [ item for element in raster for item in element ]

#transforming list in array
raster = np.asarray(np.reshape(raster, (rows,columns)))

##writting output raster
dst_ds.GetRasterBand(1).WriteArray( raster )

#setting extension of output raster
# top left x, w-e pixel resolution, rotation, top left y, rotation, n-s pixel resolution

# setting spatial reference of output raster 
srs = osr.SpatialReference(wkt = prj)
dst_ds.SetProjection( srs.ExportToWkt() )

#Close output raster dataset 
dst_ds = None

#Close main raster dataset
dataset = None

I tried out the code with the RGB raster of the next image:

enter image description here

After running the code (at the Python Console of QGIS), I loaded the blue_band raster; as it can be observed at the next image:

enter image description here

You can adapt it for your particular purpose (customizing number bands, paths, etc).

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.