I am using the following gdal shell command (gdaltindex) to create a raster tile index from a directory of raster (tif) files.

find /path/to/raster/directory -name "*.tif" -exec gdaltindex tile-index.shp {} \;

Is there an open source Python approach that can produce the same raster tile index shapefile?


I dont't know, but it's easy to do with osgeo.gdal, GeoPandas and shapely box ( shapely.geometry.box(minx, miny, maxx, maxy, ccw=True) = bounding box)¶

Original raster files

import os 
StartDir = "/Shared/scan_ign/68"    
for dir, subdir, files in os.walk(StartDir):
    for fname in files:
       if fname.endswith(".tif"):


With gdalindex

gdaltindex index.shp *.tif


import geopandas as gpd
index =  gpd.read_file('index.shp')
print index.head()
    location                geometry
0  68_1.tif  POLYGON ((226018.4754020752 58174.93482261647,...
1  68_2.tif  POLYGON ((234019.0396324735 58174.8250049785, ...
2  68_3.tif  POLYGON ((242020.0156871924 58175.55617328714,...
3  68_4.tif  POLYGON ((250020.4746237129 58176.34366682862,...
4  68_5.tif  POLYGON ((226019.5264973617 48175.2918232924, ...

With osgeo.gdal

from osgeo import gdal
import geopandas as gpd
from shapely.geometry import box
# compute the bounding box of a gdal raster file
def bounds_raster(path):
    raster = gdal.Open(path) 
    ulx, xres, xskew, uly, yskew, yres  = raster.GetGeoTransform()
    lrx = ulx + (raster.RasterXSize * xres) 
    lry = uly + (raster.RasterYSize * yres)
    return box(lrx,lry,ulx,uly) 
# creation of the index file
 df = gpd.GeoDataFrame(columns=['location','geometry'])
 # iterate through multiple tif files in a folder
 for dir, subdir, files in os.walk(StartDir):
     for fname in files:
         if fname.endswith(".tif"):
              df = df.append({'location':fname, 'geometry': bounds( os.path.join(dir+"/", fname))},ignore_index=True)
 print df.head()
      location                  geometry
 0  68_1.tif  POLYGON ((226018.4754020752 48173.94230128427,...
 1  68_2.tif  POLYGON ((234019.0396324735 48173.84503544428,...
 2  68_3.tif  POLYGON ((242020.0156871924 48174.31088528392,...
 3  68_4.tif  POLYGON ((250020.4746237129 48174.72574336932,...
 4  68_5.tif  POLYGON ((226019.5264973617 38173.0677178388, ...
 # save resulting shapefile

As Spacedman points out, you can also use rasterio

 import rasterio
 df = gpd.GeoDataFrame(columns=['location','geometry'])
 for dir, subdir, files in os.walk(StartDir):
       for fname in files:
         if fname.endswith(".tif"):
            bounds =rasterio.open(os.path.join(dir+"/", fname)).bounds
            df = df.append({'location':fname, 'geometry': box(bounds[0], bounds[1], bounds[2], bounds[3])},ignore_index=True)
  • 1
    Is there a typo in line 2 of your bounds_raster function? rast=gdal.Open(path) then raster.GetGeoTransform() – Spacedman May 6 '18 at 14:13
  • The rasterio approach is particularly nice. Thanks to @Spacedman too. FYI, the indentation is off on that portion;) – Aaron May 6 '18 at 22:18
  • Think I've fixed the indenting for you. – Spacedman May 7 '18 at 9:35

As I understand it, gdaltindex returns a feature for each raster input, as a rectangular polygon of the bounds of each raster. I don't know of a ready plug-in solution for doing gdaltindex in pure python (as opposed to shelling out to run gdaltindex) and I'll assume you've searched for it. The parts to build a solution are available though.

You can use rasterio to read rasters and get the bounds:

>>> r = rasterio.open("mwi_lc_1990.tif")
>>> r.bounds
BoundingBox(left=454965.0, bottom=8094361.0, right=823965.0, top=8974795.0)
>>> r.bounds.left

And then you can use the fiona package to create shapefiles.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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