Ticket #388 (closed Defect: fixed)
Sample Rasters tool fails with RuntimeError: The feature class ... contains a point (...) which does not appear in the table output by the ArcGIS Spatial Analyst Sample tool
|Reported by:||jjr8||Owned by:||jjr8|
|Component:||Tools - Spatial Analysis||Version:|
The ArcGIS geoprocessing output looks similar to this:
Executing (Sample Rasters): ArcGISRasterSamplerSampleRasters E:\NaGISA_Workshop\Laura\PISCO\Dati_replicati\Pisco_dati_replicati.mdb\analysis_points2 E:\NaGISA_Workshop\Laura\PISCO\Dati_replicati\season3_sst SST # NEAREST # # # E:\NaGISA_Workshop\Laura\PISCO\Dati_replicati\Pisco_dati_replicati.mdb\analysis_points2 Start Time: Sun Oct 11 22:03:58 2009 Running script ArcGISRasterSamplerSampleRasters... Sampling rasters: points=E:\NaGISA_Workshop\Laura\PISCO\Dati_replicati\Pisco_dati_replicati.mdb\analysis_points2, rasters=[u'E:\\NaGISA_Workshop\\Laura\\PISCO\\Dati_replicati\\season3_sst'], fields=[u'SST']... RuntimeError: The feature class E:\NaGISA_Workshop\Laura\PISCO\Dati_replicati\Pisco_dati_replicati.mdb\analysis_points2 contains a point (-122.84889, 38.0122222) which does not appear in the table output by the ArcGIS Spatial Analyst Sample tool. The coordinates of this point differ from all of those in the output table by an amount greater than the XY tolerance of the feature class (8.52706238952e-015). This should never happen. Please contact the author of this tool for assistance. The following consequences resulted from the original error: Sampling failed.
This failure can also occur in the Sample Rasters Listed in Fields tool.
Internally, the tool calls the ArcGIS Spatial Analyst's Sample tool to perform the sampling. The problem relates to how the MGET tool compares the coordinates output by the ArcGIS tool to the coordinates of the input points. Currently, the MGET tool compares coordinates using the smallest XY tolerance possible for the input features. I discovered a situation in which the coordinates differed by exactly twice that tolerance. I'm not sure why that would occur but I suspect it is valid but just not documented in ArcGIS. I will increase the MGET tool's tolerance by a factor of 2. This should have virtually no consequence on the accuracy of the tool, as the increased tolerance will still be something like 14 digits of precision.