No subject

[chuck at pollux.GEOG.UCSB.EDU]

>Fine so far - well, we get mostly (you guessed)
>negative numbers in the output, which means very

If you are confident that your DEM covers the entire watershed, take the
absolute value of the accumulation output (i.e., r.mapcalc new = 'abs(accum)').

>Is  there any alternative to waiting hours for this
>mumbo jumbo?  

If you decrease the resolution of the cells (reduce the size of the problem),
r.watershed will run faster. Decreasing resolution is a good way to play with
any long running program. Running r.watershed with only several thousand
cells total should take less than a minute. That way, you can dicker with 
the options until the output looks appropriate. Only then go to the 
fine resolution and let the program run over-nite.

Historical Note: This version of r.watershed is being used because it is
several times faster than the previous incarnation.

>below, I installed a mask around the watershed, I even 

If you didn't create a buffer around the watershed before masking the rest,
you are bound to make the watershed be filled with negative numbers.

>Also, the program runs ok without an input 
>overland flow map - when you give it a runoff
>layer it generates the stange output.

Check your input overland flow map. Does it contain negative values?
What are the patterns of strange output compared to the overland flow map?

Chuck Ehlschlaeger