Here is example logging output:
Executing (Predict GAM From Rasters): GAMPredictFromArcGISRasters C:\HabModExample2\Model\GAM.rdata C:\HabModExample2\OceanographyRasters\Bathymetry\etopo2v2;C:\HabModExample2\OceanographyRasters\Chl\Monthly\chl199908;C:\HabModExample2\OceanographyRasters\SST\Monthly\sst\sst199908 Bathymetry;Chlorophyll;SST C:\HabModExample2\HabitatPredictions\habitat08 C:\HabModExample2\HabitatPredictions\template # true C:\HabModExample2\HabitatPredictions\stderrors08 C:\HabModExample2\HabitatPredictions\binary08 0.275160261084 true
Start Time: Wed Jul 22 16:01:14 2009
Running script GAMPredictFromArcGISRasters...
Loaded GAM from C:\HabModExample2\Model\GAM.rdata. The model was fitted with the R mgcv package.
MODEL SUMMARY:
==============
Family: binomial
Link function: logit
Formula:
Presence ~ log10(abs(Bathymetry)) + s(log10(Chlorophyll)) + SST
Parametric coefficients:
Estimate Std. Error z value Pr(>|z|)
(Intercept) -18.1695 3.3549 -5.416 6.10e-08 ***
log10(abs(Bathymetry)) -1.9870 0.4842 -4.103 4.07e-05 ***
SST 0.6532 0.1075 6.075 1.24e-09 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Approximate significance of smooth terms:
edf Ref.df Chi.sq p-value
s(log10(Chlorophyll)) 4.147 4.147 14.91 0.00555 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
R-sq.(adj) = 0.334 Deviance explained = 40.4%
UBRE score = -0.56808 Scale est. = 1 n = 303
Checking coordinate systems, extents, and cell sizes of predictor rasters and reprojecting and clipping as needed to make them conform to the template raster...
Projecting and clipping C:\HabModExample2\OceanographyRasters\SST\Monthly\sst\sst199908. Its cell size (0.0439453) does not match the cell size of the template raster (0.10000000000000001).
Projecting and clipping C:\HabModExample2\OceanographyRasters\Chl\Monthly\chl199908. Its cell size (0.083333299999999985) does not match the cell size of the template raster (0.10000000000000001).
Projecting and clipping C:\HabModExample2\OceanographyRasters\Bathymetry\etopo2v2. Its cell size (0.033333300000000003) does not match the cell size of the template raster (0.10000000000000001).
Predicting...
RPy_RException: Error in .local(.Object, ...) :
GDAL Error 3: Attempt to read past EOF in C:\Temp\GeoEcoTemp_Jason\tmphplbkf\projected0a/../info/arc.dir.
Obtaining additional debugging information. This may cause this tool to hang. If you do not see any additional messages after one minute, halt its execution.
<class 'rpy.RPy_RException'>: Error in .local(.Object, ...) :
GDAL Error 3: Attempt to read past EOF in C:\Temp\GeoEcoTemp_Jason\tmphplbkf\projected0a/../info/arc.dir.
Failed to execute (Predict GAM From Rasters).
End Time: Wed Jul 22 16:01:42 2009 (Elapsed Time: 28.00 seconds)
The root cause of this problem is a bug in the GDAL library, as described here under the email threads titled "Reading ArcInfo binary grids with the R rgdal package fails with "GDAL Error 3: Attempt to read past EOF in..../arc.dir".
The problem occurs with versions of R for which the latest version of the rgdal package (0.6-10) includes a copy of GDAL 1.6.1. Basically, GDAL 1.6.1 fails to read ArcInfo binary grids under certain circumstances (I do not understand what these are), and this failure cascades up through the rgdal R package to MGET.
The ultimate solution is for the GDAL dev team to fix the problem, release another version of GDAL (presumably 1.6.2), and for the rgdal dev team to pick up that release and issue a new version of rgdal. Until that happens, the following workarounds may allow you to bypass this problem:
- Use an older version of R, for which the latest version of rgdal only includes GDAL 1.6.0 or earlier. I was advised by Roger Bivand, an rgdal developer, that R 2.8.* and earlier include rgdal 0.6-8 and earlier, which include GDAL 1.6.0 and earlier.
- When using R 2.9.*, use the install.packages command to explicitly install rgdal 0.6-8 or earlier. Roger mentioned that this version of rgdal may not even exist for R 2.9.*, so this suggestion may not work.
