On Google Earth Engine, I created a classified land cover map based on a composite of images (spanning 2 years). It's a time series study so increasing the range of images (i.e. increasing to more than 2 years) would be detrimental.

The missing pixels are very minimal only about 3-5% of my study area but they're a hindrance to simulation modeling (my end goal).

I understand that QGIS has "Fill NoData cells" in its Processing Toolbox which works to an extent but also creates pixels outside of my study area. I only want to fill in the "holes" caused by the missing pixels.

Is there another way to interpolate the missing pixels? Using GEE or QGIS? and is interpolating these missing pixels generally acceptable?

  • for QGIS see: spatialthoughts.com/2020/06/17/fix-nodata-values
    – eurojam
    Commented Mar 18, 2023 at 6:08
  • Hi! Yes, I have seen that (and I've mentioned that in my 3rd paragraph). I'd like to know though if it's a generally acceptable practice to do so.
    – Linus AG
    Commented Mar 18, 2023 at 6:30
  • Are you trying to interpolate between imagery in your time-series or based on adjacent pixels? Commented Mar 18, 2023 at 12:44
  • Hi! Just based on adjacent pixels. (context: all of the images in my time-series range seem to have missing pixels in a similar area). I'm thinking the FillNoData cells tools is acceptable? It uses IDW interpolation.
    – Linus AG
    Commented Mar 18, 2023 at 17:26


Your Answer

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