I am studying the impact of urbanization on the discharge levels of a river during flooding events. For this, I need to determine the catchment's percent urbanized area through land classification from the years 1992 to 2019.

I've thought of using Landsat images, fmask, ArcMap, eCognition, and the following procedure:

  1. Use the Fmask algorithm for cloud cover removal.
  2. Fill the removed portions using another (but temporally close) Landsat image with ArcMap.
  3. Clip the final image with the catchment shapefile.
  4. Have eCognition determine the percent urbanized area.

My problem is how to fill the portions that were removed by Fmask (step 2).

Is this procedure a viable method?

Moreover, I get confused as to how to process the Landsat images in ArcMap what with it having multiple bands.

And although I haven't really encountered any significant problem with eCognition aside from the somewhat low resolution image I get when working with my Landsat 8 data, I seek any suggestion as to how I can effectively use the program for determining the percent urbanized area.

I am relatively new at GIS.

  • Hi Jesi Martin Maglana, welcome to GIS stackexchange! I made some small improvements to the formatting and wording of your question. If I removed anything important, you can fix it by the edit link under the question. – csk Mar 7 '19 at 18:37
  • What happens when you try your procedure? – PolyGeo Mar 7 '19 at 19:30
  • When I used eCognition to process the Landsat images, I found that the latter was highly pixelated. This made it quite difficult to assess whether or not the initial ruleset I made was suited for the task. Is there any way to have finer data which still covers 1992-2019? – Jesi Martin Maglana Mar 26 '19 at 13:02

Your Answer

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

Browse other questions tagged or ask your own question.