v.surf.rst crashes on Mac OS X

[cmbarton]

OK. v.surf.rst also crashes before it even starts. So that only interpolation routines that works for me is v.surf.idw and one routine for r.resamp.interp.

FWIW, I'm running this on a version of GRASS from the SVN trunk compiled on 1 March 2008.

Michael

[hamish]

how big is your region?

since all these things fail maybe it is some common problem?

this is all on OSX 10.5?

exact command lines and error messages, etc? g.gisenv set="DEBUG=5" ?

Hamish

[cmbarton]

This is happening with Spearfish and with the new NC dataset. I've simply set my region to match the elevation DEM. In the NC dataset, I think this is a little over 2million cells. This is not extraordinarily large and it IS the demo set for GRASS.

It is with 10.4 AND with 10.5 for everything except r.random. This is only with 10.5.

I'm also finding that r.resamp.interp says it makes maps, but it does not. With the neighbors algorithm, it makes a map that is exactly the raster points that are input. That is, there is no interpolation. For bilinear and bicubic, it makes NAN maps. I've checked this with r.info. I'll give examples below. Note that the output of the nearest routine produces only 500 points (see r.univar results)--the same number I used as points for interpolation. Also note that I just updated from the SVN trunk, did a make distclean, and compiled fresh this afternoon.

Michael

---

GRASS 6.3.svn (nc_spm_07):~ > r.info elev500bilinear

+----------------------------------------------------------------------------+ | Layer: elev500bilinear Date: Tue Mar 18 21:12:23 2008 | | Mapset: spatialtech2008 Login of Creator: cmbarton | | Location: nc_spm_07 | | DataBase: /Users/Shared/grassdata | | Title: Resample by bilinear interpolation ( elev500bilinear ) | | Timestamp: none |

|

| | | Type of Map: raster Number of Categories: 255 | | Data Type: DCELL | | Rows: 1350 | | Columns: 1500 | | Total Cells: 2025000 | | Projection: Lambert Conformal Conic | | N: 228500 S: 215000 Res: 10 | | E: 645000 W: 630000 Res: 10 | | Range of data: min = nan max = nan | | | | Data Source: | | elev_pts500@spatialtech2008 | | Source map NS res: 10 EW res: 10 | | | | Data Description: | | generated by r.resamp.interp | | | | Comments: | | r.resamp.interp input="elev_pts500@spatialtech2008" output="elev500b\ | | ilinear" method="bilinear" | | | +----------------------------------------------------------------------------+

GRASS 6.3.svn (nc_spm_07):~ > r.info elev500nearest

+----------------------------------------------------------------------------+ | Layer: elev500nearest Date: Tue Mar 18 21:12:59 2008 | | Mapset: spatialtech2008 Login of Creator: cmbarton | | Location: nc_spm_07 | | DataBase: /Users/Shared/grassdata | | Title: Resample by nearest interpolation ( elev500nearest ) | | Timestamp: none |

|

| | | Type of Map: raster Number of Categories: 255 | | Data Type: DCELL | | Rows: 1350 | | Columns: 1500 | | Total Cells: 2025000 | | Projection: Lambert Conformal Conic | | N: 228500 S: 215000 Res: 10 | | E: 645000 W: 630000 Res: 10 | | Range of data: min = 57.820839 max = 154.312943 | | | | Data Source: | | elev_pts500@spatialtech2008 | | Source map NS res: 10 EW res: 10 | | | | Data Description: | | generated by r.resamp.interp | | | | Comments: | | r.resamp.interp input="elev_pts500@spatialtech2008" output="elev500n\ | | earest" method="nearest" | | | +----------------------------------------------------------------------------+

GRASS 6.3.svn (nc_spm_07):~ > r.univar elev500nearest

100%

total null and non-null cells: 2025000 total null cells: 2024500

Of the non-null cells:

---

n: 500 minimum: 57.8208 maximum: 154.313 range: 96.4921 mean: 111.107 mean of absolute values: 111.107 standard deviation: 21.0507 variance: 443.131 variation coefficient: 18.9463 % sum: 55553.4484596252

GRASS 6.3.svn (nc_spm_07):~ >

[glynn]

Replying to cmbarton:

I'm also finding that r.resamp.interp says it makes maps, but it does not. With the neighbors algorithm, it makes a map that is exactly the raster points that are input. That is, there is no interpolation. For bilinear and bicubic, it makes NAN maps.

You are misunderstanding what r.resamp.interp does. It does not "fill" null areas (like e.g. r.surf.idw or r.fillnulls). It resamples a map, calculating cell values by interpolating between the neighbouring cells, all of which must be non-null in order to get a non-null result.

[hamish]

Hi,

I have just tested v.surf.rst and r.random from William's 6.3.0rc4 binaries on OSX 10.4, and both worked fine.

#spearfish
g.region -d
v.surf.rst in=archsites elev=arch_surf zcol=cat
r.random in=elevation.dem rast=rand_elev.1000 n=1000

(r.random was expected to work here, it was only failing with OSX 10.5)

is any one else see a problem with v.surf.rst on OSX?

worksforme, Hamish

[neteler]

priority downgraded since it works fro Hamish.

[hamish]

Michael reported on grass-dev that it now works for him. no one else has reported problems.

shrug. closing bug.

Hamish