[GRASSLIST:940] Re: Surfer Binary Grids and r.in.bin

Glynn Clements glynn.clements at virgin.net
Mon Aug 11 12:55:48 EDT 2003


David Orme wrote:

> I'm trying to import binary Surfer (6.x) files and I've got a problem 
> with the import through r.in.bin
> 
> The grid values in the binary file are byte reversed, signed, 32 bit 
> floats following IEEE-754. If I import the data section of the file 
> (ignoring the header) using:
> 
> r.in.bin -sbf bytes=4 input=file output=test ....
> 
> the file imports without problem and r.info for the resulting map shows 
> the expected min and max for the file (from cellmisc/test/r_range). 
> r.report however shows '*' for all cells. The range of values is from 
> approximately -5 to 90, although the 'no data' value is a very large 
> float (1.7014100091878280e+38).

r.in.bin has the "anull=" option to specify the value which is used
for nulls. However, I don't know how reliable this is for floats, as:

a) the check is done using ==, so any rounding error will cause
problems.

b) if the value is considered to be NaN (as is the case for GRASS'
internal null value), NaN isn't equal to itself, so the test will
always fail.

-- 
Glynn Clements <glynn.clements at virgin.net>




More information about the grass-user mailing list