I have received a GeoTiff file, whose pixel spacing is provided in degree (e.g. +1° to the right, -1° down). Of course, this means that the actual, metric pixel spacing changes in relation to the latitude, which is bad, if you want to relate pixel distances to actual, real-world distances. What is the straight-forward procedure to transform the image in such a way that the result is squared, metric pixels (e.g. +1m to the right, -1m down)?

For sake of completeness: The original image is provided in WGS84-latlon coordinates (EPSG:4326), the final image should be projected to the proper UTM zone (e.g. EPSG:32632).

I figure I'll probably first have to reproject the dataset. But then still the question remains about how to resample the resulting, reprojected image so that its pixel size can be set to an arbitrarily chosen, square size?

Using any distance measuring tool within QGIS is not an option, as I need to export the image again as a GeoTiff in order to process it externally with a different software.

bumped to the homepage by Community yesterday

This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.

  • This sounds like you have to transform the TIF from EPSG: 4326 to EPSG: 32632. Load the TIF with EPSG: 4326 in QGIS and save(as) in EPSG: 32632. – Mike May 23 '17 at 13:23
  • Sorry you do not reproject a raster layer using the save as option. You have to use raster/projections/warp (reproject) – Gerardo Jimenez May 23 '17 at 16:21
  • Yeah, this is what I meant with "I figure I'll probably have to reproject the dataset". But how will I achieve square pixels with a given pixel spacing? – Michael May 23 '17 at 16:22
  • 1
    according to this: gdal.org/gdalwarp.html the resampling option of the dialog box is used to interpolate between the values of the cells, not its size. Usually when you reproject a raster it will produce cells of equal size. When I perform this procedure on DEMs in 4326 with a spatial resolution of 1 second of arc it produces pixels of aproximatelly 30 meters when I reproject it to 32614...bear in mind that I work in with DEM from Mexico – Gerardo Jimenez May 23 '17 at 16:31
  • May be this can help gis.stackexchange.com/questions/132149/… – Gerardo Jimenez May 23 '17 at 16:35

As Gerardo Jimenez stated in the comments, you'll need to resample/warp your grid to a metric coordinate system like utm. If the cell size of your original geotiff is one degree, then your optimum warped cell size is 111 kilometer, do not use a cell size of one meter as you will not get a better image and this would eat up an enormous amount of disk space. For most cases a rapid nearest neighbour interpolation will yield acceptabele results. If the cell size of the original geotiff is 0.5 degrees, use 111000*0.5 meters for reprojected cell size.

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.