0

I currently have a 87Mb Compressed GeoTiff with a resolution of 50m. I need to interpolate this to 2.5m

I have used gdalwarp to achieve this using a cubicspline, but it took 12hrs and resulted in a 44Gb GeoTiff. I did use -co COMPRESS=LZW but wondering if gdal_warp uses these creation options?

I then tried -r average which only took a few hours but still resulted in a 22Gb GeoTiff.

If I look at the results the image still looks very course (granular) even though gdalinfo does say it now has a pixel size of 2.5

Am i doing something wrong or expecting too much from gdal

thanks

Improve this question
3
  • I know this isn't how quite how it works, but you're increasing the raster size roughly by 400 times (20x20 cells of 2.5 versus 1x1 of 50).
    – mkennedy
    Commented Apr 30, 2014 at 17:22
  • COMPRESS and gdalwarp may not always work, see: trac.osgeo.org/gdal/wiki/UserDocs/GdalWarp although I'm not sure this applies to your version. You can also use -wo NUM_THREADS=ALL_CPUS or -wo NUM_THREADS=N to (possibly) speed things up.
    – user10353
    Commented Apr 30, 2014 at 17:28
  • @kyle thanks for the tip about NUM_THREADS very useful. I was using --config GDAl_CACHEMAX 1024 and -wm 1024 which also help
    – tjmgis
    Commented Apr 30, 2014 at 22:20

1 Answer 1

Reset to default
2

Using gdalwarp to transfrom a raster from a coarse resolution to a finer resolution will increase the numbers of pixel (in your case by a factor of 400, like mkennedy stated in his comment above). If you choose an interpolation technique other from nearest neighbour (-r near) the raster will contain different pixel values based on the interpolation result, therefore the raster will look "smoother", but you can not create additional information to get a "sharper" result. If you need to get a "real" finer resolution you could use for example "pan sharpening" techniques, if you handling multispectral images and a panchromatic image with a higher geometric resolution. For example this post from Markus Neteler on planet.qgis offers a quick example for Landsat-OLI data using GRASS GIS. But i do not know if have access to a higher resolution image. So i do not think you are doing something wrong, but perhaps you are expecting to much from gdal in this case ;) please correct me if i am wrong!

Improve this answer

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.