[GRASSLIST:9003] Re: r.terraflow: wrong data type

[Florian Kindl]

> On Nov 10, 2005, at 4:52 PM, Andrew Danner wrote:
>
>> Helena,
>>
>>   I have attached a patch against the 6.1 cvs tree that should fix at
>> least the filled file having the wrong type. Internally, the code was
>> using floating point elevations, but when writing the data out to a
>> grass grid, it converted to integer type. I tested it on a sample  
>> data
>> set and r.terraflow now should produce a filled grid that is of FCELL
>> type.
>>
>>   The flow accumulation is done on an internal representation that  
>> does
>> have floating point elevations, so if the flow accumulation  
>> routine is
>> generating unwanted results, it is not related to this output  
>> issue. The
>> problem with integer grids was simply an error when converting from
>> terraflow internal format to the grass raster format.
>>
>> -Andy

On 11.11.2005, at 00:38, Helena Mitasova wrote:

> Andy, thank you very much, that explains it. In fact I never looked  
> at the filled DEM
> only at the resulting flowaccumulation. Florian, I will submit the  
> fix and then can you please
> test it again - the problems that you may see in the  
> flowaccumulation result may not be
> due to integers but due to many nested sinks that were filled.   
> What kind of data and resolutions are you using?
>
> Helena
>
>

Helena, Andy,
thank you very much for your quick help and effort.

The patch fixed the problem. Now the filled DEM looks as it should.

Helena, regarding your question on the kind of data:
It is very high resolution data (1 meter or less) from airborne laser  
scanning.
So I suspected that maybe filled sinks lead to unwanted results -  
that's why I wanted to to take a look at the filled DEM in the first  
place.

Thank you that this is possible now :)
- Flo.