I have just tested v.surf.icw in some detail. In general, it is a very nice module that produces far more realistic results than plain euclidean v.surf.idw in the right circumstances.
I made some observations and have some questions/comments I'd like to share: Line 150: This produces an error message from the shell if any of the env vars does not exist (cosmetics, really): if [ $GIS_FLAG_V -eq 1 ] || [ "$GRASS_VERBOSE" -gt 1 ] ; then Line 234: r.cost $VERBOSE -k in=tmp_icw_area_$$ out=cost_site.$NUM coordinate=$EASTING,$NORTHING = r.cost $VERBOSE -k in=tmp_icw_area_$$ output=cost_site.$NUM coordinate=$EASTING,$NORTHING Otherwise, it won't play any longer with newer versions of r.cost that have the "outdir=" option. Line 239: This changes the original input data without any warning or documentation about it: # so the divisor exists and the weighting is huge at the exact sample spots # more efficient to reclass to 1? r.mapcalc "cost_site.$NUM = if(cost_site.$NUM == 0, 0.1, cost_site.$NUM)" Apparently, this is done to avoid a divison by zero in the standard IDW formula (line 246): EXPRESSION="1.0 / pow(cost_site.$NUM $DIVISOR, $FRICTION )" If so, it needs to be documented. I actually used costs normalized to [0,1] and ran into trouble here. I realize now that wasn't a good idea, because a movement costof "0" between two spatially distinct locations is physically implausible. However, this behaviour still needs to be document, perhaps advising users to use a minimum cost of "1"? Line 246: The variable $DIVISOR is never initialized (empty) but still used in the expression: EXPRESSION="1.0 / pow(cost_site.$NUM $DIVISOR, $FRICTION )" In addition, I would like to know where the second IDW formula comes from (-r flag). Any literature references? When would this be the preferable formula? Thanks, Ben ------ Files attached to this email may be in ISO 26300 format (OASIS Open Document Format). If you have difficulty opening them, please visit http://iso26300.info for more information. _______________________________________________ grass-dev mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/grass-dev
