Take the 2-minute tour ×
Geographic Information Systems Stack Exchange is a question and answer site for cartographers, geographers and GIS professionals. It's 100% free, no registration required.

I'm new to terrain analysis. I have a 30m NED DEM (.tif) for the conterminous US. I would like to generate a nice looking hillshade for the US, but it looks bad - as in pixelated and rough (See image). I have tried different classification schemes and number of classes, which helps some, but not enough. I have tried majority filter, but the output looks exactly the same. Is there some sort of smoothing function that should be applied for this scale? Any suggestions much appreciated.

Edit: Using ArcMap 10.1, projected w/ NAD83 Albers Edit 2: The hillshade could use more "depth", but looks good for now (image 2).

Raw hillshade - looks rough on screen at this scale

The finished map, I'm happy with so far. Looks much better printed out as high res image, like .tiff. I may smooth out the coastal vignette area.

share|improve this question
2  
What GIS/Raster Software are you using? –  Mapperz May 7 '13 at 14:57
1  
It looks like you haven't projected the coordinates. Use a reasonable projection for the conterminous US and try again. –  whuber May 7 '13 at 15:25
    
I'm using ArcMap 10.1, but have SAGA too. The DEM is projected with NAD83. –  SoilSciGuy May 7 '13 at 16:41
1  
That's clearly not an Albers projection, despite the claim in the question. The map is using geographic (spherical) coordinates as if they were Cartesian. Regardless of the distortion problems, it's likely the hillshade is having to equate elevations in meters to plan distances in degrees, thereby making everything look about 100,000 times steeper than it really is. All you get is black, white, & gray, with almost nothing in between. Apply an honest projection to the raster and redo the hillshade. –  whuber May 7 '13 at 16:57
2  
So I exported the hillshade as pdf w/ 1:1 scale and the image looks good to my surprise. It just looks bad while viewing on screen (like this screen shot), but prints well . Whuber, the raster was projected to albers, but was being drawn improperly as you pointed out - good eye. Thanks again. –  SoilSciGuy May 9 '13 at 15:33

2 Answers 2

up vote 1 down vote accepted

The best is to resample the raster and have hillshade models of different scales. 90m resolution of the whole US will shade correctly but when you view the whole image the resampling algorithm for viewing in the screen will/might cause pixellation. If you view at about 1:1 screen scale you should see a good hillshade. Also hillshading does not work really well in very flat areas with "random" height differences between pixels, that might also cause problems in the "aesthetics" of viewing the hillshade model.

share|improve this answer
    
I noticed the random effects in flat areas you mention. I tried filtering but the output was exactly the same..maybe some issue with a .tif? Thanks for the help :) –  SoilSciGuy May 7 '13 at 16:55
    
The only way in flat areas is to mask them and give them a single value depending on the scale you are using. A filter will not work well because there is not enough difference and filter parameters would have to be very strong, oversmoothing the rest. The only way to solve for random effects in flat areas is to just remove them by assigning a single mean value to the flat area. This is of course more of a aesthetic technique than a metric technique!! –  Francesco May 16 '13 at 6:10

To smooth the hillshade, I used the focal statistics tool (in spatial analyst) with a 5x5 moving window. This can be done more than once until you're happy with the results. But as @Francesco mentioned, the hillshade will look sort of grainy on the monitor at this scale in ArcGIS. A good work around is to export map as a high res image to see the details better.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.