string is not a geographic or projected coordinate system 010067

Question

I am trying to overlay two raster datasets. But I get this when trying to run the tool - zonal statistics.

("esri.SpatRefEnvironment") the input string is not a geographic or projected coordinate system
ERROR 010067_Error in executing grid expression.Zonal statistics program failed

I have been trying a bunch of different setting (extents, snap to etc) and also to convert one of the rasters to a polygon layer instead. But nothing works.

Both rasters are single band. Both have the same geo (wgs 1984) and projection (albers conical equal area). They have also the same cellsize. Both are imagine format.

However one is continious 32bit and the other is discrete 8 bits. Is that a problem?

Hm....what can i do?

StacieWolny · Answer

This is happening to me as well, using Arc 9.3.1.  Hollie is actually using our tools, so this problem might end up effecting a lot of our users, so any insights would be appreciated.

Here's a bit more info.  It's happening within a Python script (actually, a couple of them) where I've done a bunch of raster calculations (very different in each script) and in the end aggregate the results using Zonal Statistics, where the zone file is a shapefile of watersheds.  One odd thing I'm seeing in one of the rasters that fails with this error is that when I look at its Statistics, the Mean value is 1.#QNAN,  and Calculate Statistics will not work on it (CS tells me that it's invalid when I try to add it to the dialog box.)  If I turn the watershed file into a raster, initial tests are showing that Zonal Stats can work, but Zonal Stats as Table does not (using either shapefile or raster zones), but returns a different error: 
ERROR 010067: Error in executing grid expression. Zonal statistics program failed
Failed to execute (ZonalStatisticsAsTable).

Now, the QNANs first appear when I do a Lookup_sa on a raster with a bunch of fields, created by doing a join between a watershed raster and a table, then a CopyRaster to a new grid.  This new grid looks fine and has proper Statistics.  The resulting Lookup values and maps look good, but the Statistics say Min 89, Max 1701, Mean 1.#QNAN, Std dev 0, Classes 0.

Zonal Stats as Table seems to work fine on rasters that have proper statistics, but not on those with a Mean of QNAN.  Could be coincidence, but it's the only thing that looks obviously amiss.

~ Stacie

HollieHall · Answer

Hi,
I am receiving a similar error.  All of my raster files are in grid format.  Any other suggestions as how to resolve the following error?

Error running Yield calculations: ERROR 999999: Error executing function.  ("esri.SpatRefEnvironment") the input string is not a geographic or projected corrdinate systesm
Error 010092: Invalid output extent.
Failed to execute (ZonalStatistics)

Thank you in advance for any suggestions.
Hollie

curtvprice · Answer

Both rasters are single band. Both have the same geo (wgs 1984) and projection (albers conical equal area). They have also the same cellsize. Both are imagine format.

However one is continious 32bit and the other is discrete 8 bits. Is that a problem?

Hm....what can i do?

At 9.3.1, raster analysis happens only with grid data sets. Behind the scenes the raster data sets are converted to grids (in your scratch workspace or a folder, in temp, if your scratch workspace is a gdb).

My experience is you can sometimes have better luck with analysis if you convert the inputs to ESRI grid format. (In your case I'd define the raster coordinate system after the conversion just to be sure they are defined the same.)

In 10.0 the tools work natively so you are less depending on a behind-the-scenes conversion to work correctly. If you are using 10.0 I'd still try the grid format as grid and raster gdb are the "native" formats for raster tools and are likely to work the best and fastest.