r.slope.aspect

[Helena Mitasova]

>	I have a question about r.slope.aspect.  Let me first explain the 
>background to the project.  We have scanned in two 7.5 minute USGS contour
>separates (with 20 ft contour intervals), cleaned them (using LT4X), and 
>interpolated the DEM with a cell size of 5 meters.  My question is this..
>
>	In the GRASS version 4.1 user reference manual, it mentions that 
>Helena Mitasova of USACERL observed that most cells with very small slope end
>up having bias towards categories 0,45,...,360.  So in response, the new option
>min_slp=value was added.  Does anyone have any ideas as far as what that value
>should be ?  Is it a factor of the computation or would this value be different
>for each study area ?  

The bias is due to integers in DEM which creates in areas with
low slopes steps in DEM. If you had interpolated the DEM with elevations
in cm you should not have that problem. Another thing you can do
is to use s.surf.tps which interpolates the DEM and computes the
slope and aspect at the same time, so there is no problem with
integers ( and no bias in histograms for aspect).
 If you don't want to go back and rerun the interpolation,
 run r.slope.aspect with min_slp=0,
look at aspect and if you get lots of noise and various other weird
things (like strips) in the areas with low values of slope, check the slope
in these areas and use that for min_slp. Its value is a factor of both
the data, method and resolution you have used for the computation of DEM
 so it is often different for differnet applications.

I hope this helps

Helena Mitasova

helena at zorro.cecer.army.mil