Hello i;m performing a land-use land-cover change detection with landsat data. i have downloaded ETM image with scan line error, so i wanna use landsat gap-fill add on.
Dear Jibrin,
The answer to your question depends (1) on the ultimate goal(s) of the analysis, (2) on the method(s) you are planning to use to replace the missing values, and (3) on the expected quality of the result.
If land use (and/or land cover) is the only application you are considering, and if you are only discriminating between a few well-defined and well-separated land cover categories and/or if you are using a simple classification approach, you can probably generate your land cover map first and then fill the missing points using nearest neighbors: each 'pixel' in your map must belong to one of a small number of classes, and it should not be too difficult to implement that assignment.
However, if you anticipate that the same Landsat scene might be used later for another purpose or by another user, then it would be wiser to correct the original data first, and generate the land cover map once the scene is complete. This way, you will have a better basis for any and all applications. However, you should perhaps also adopt a more careful approach to specify the missing values, which are physical measurements.
Whichever way you decide to go, you should assess the validity and performance of the outcome, or at least determine what might be the 'cost' of an error in your map, and select the approach that minimizes that cost.
Good luck in your investigation, Michel.
- Badges
- Science topic
- Similar topics
- Correction