I'm currently working on a small visibility project for which I would like to use a combination of the following two datasets:
1) 5m grid based on digitized maps/fieldwork (most detailed; doesn't cover the whole region) 2) 30m elevation data from ASTER (less detailed; covers the missing areas)
I would like to leave the 5m data 'as is', as it covers the most important part of the study area and is of better quality/more detailed than the ASTER DEM. My workplan was consequently to resample the ASTER data to a cellsize of 5 and merge both datasets, resulting in a (partly interpolated) 5m DEM of the total study area.
Things aren't going as smoothly as I would like, though. Decreasing the cell size of the ASTER raster (:)) results in various kinds of artefacts when calculating the viewshed, no matter the new cell size or the resampling technique that is used. To the eye, there is nothing wrong with the resampled DEM, problems only arise when basing calculations on the new raster.
All resampling was done using the 'Resample' tool, adjusting the output cell size and resampling technique as needed.
My question is: how can I decrease the cell size of the ASTER DEM without introducting these artifacts? Also, feel free to point out if I'm going about this problem the wrong way.
When you say 'merge' did you use the Merge tool or how did you combine the two? Also I am not sure what you mean, 'partially interpolated' because you are using a resampling technique, not necessarily an interpolation. My recommendation would be to combine the two rasters with the CON statement and then choose the finer pixels over the resampled Aster values. This technique will avoid resampling and maintain the original values of the finer resolution data. You must be sure that your pixels line up with the Snap Raster environment setting. That way you can have the best of both rasters. Be advised if the values at the edge are way off, you will still see a 'stair step action' unless you choose to blend/merge. If you are getting artifacts, it may be due to the resampling/merging that is occurring when the pixels do not line up and the values are very different. I am including a blog post on adding two rasters together that discusses this technique.
When you say 'merge' did you use the Merge tool or how did you combine the two? Also I am not sure what you mean, 'partially interpolated' because you are using a resampling technique, not necessarily an interpolation. My recommendation would be to combine the two rasters with the CON statement and then choose the finer pixels over the resampled Aster values. This technique will avoid resampling and maintain the original values of the finer resolution data. You must be sure that your pixels line up with the Snap Raster environment setting. That way you can have the best of both rasters. Be advised if the values at the edge are way off, you will still see a 'stair step action' unless you choose to blend/merge. If you are getting artifacts, it may be due to the resampling/merging that is occurring when the pixels do not line up and the values are very different. I am including a blog post on adding two rasters together that discusses this technique.
Many thanks for your help. I'd already tried the CON function the way you suggest (raster snapping enabled), but alas, the same glitches arose. The good news is that I found a work-around earlier today. I let Global Mapper do the cell size calculations instead of Arc, and the resulting Aster 5m grid was, to my surprise, completely error-free. The new Aster file was then combined with the 'better' raster using CON, and the final result is flawless.
It still eludes me what Arc's problem is though. It didn't matter which method I'd use to change the cell size of the Aster DEM. Export data / resample / mosaic to new raster / raster calculator (con), they all led to the aforementioned artifacts.
Whatever the reason, I can now continue. Once again, my thanks :).
