ERROR 010005 - ERROR 010098 - ERROR 010067

Question

There's only one forum entry for this error so I thought I would share my experience.

I ran a Kernel Density on about 750 points and outputted a 10000X16000 grid. The input points came from a point layer with a definition query applied to them and the cell-size of the output was set to 2m. I went to re-run the kernel density with a slight change to the query definition and I got Errors 010005 and 010067. The second time I changed the cell size to 5m.

After a reboot - the same errors
After trying the original definition query and cell size - the same errors
I even tried to run the points through a useless IDW and I got:
ERROR 010098: Insufficient number of points and 010067.

I exported the query definition results (250 points) to a new feature class and both tools work great. I have a kernel density raster of 250 points and an IDW of their latitudes.

ArcMap 10 SP2

JimW1 · Answer

Pretty is fine, Jim, but don't let it complicate your analysis.  Do the analytical work however it should be done and then prettify the result.  This is usually much more efficient.  A good strategy in such cases is to do the work on a coarse grid using a smoother interpolator than IDW (such as kriging or splines).  You can then (quickly and easily) resample the coarse grid to make it look nice and smooth for presentation.  That will save loads of time and anguish.

Another philosophy is that when good science or statistics are not used, or the data are rough and uncertain, or you have to use approximate methods for whatever reason, then create maps and graphics that make this crudeness apparent: leave them blocky and tessellated and low-res so the viewer is not deceived.

Bill you must be a teacher. Each input takes less than than 5 seconds to run so your workflow is academic at best.

Please contribute to the topic of thread - I need my definition queries to work not a lecture on SA.

JimW1 · Answer

This is just a comment, Jim: the approximations inherent in (any) kernel density or IDW calculation indicate that this grid is likely to be far, far too detailed for the information inherent in just 750 points.  You could probably do an excellent job on a 20 m grid and then, if you really need results on a 2 m grid, resample the 20 m grid down to 2 m.  This will save you about 98% of your computing time and likely avoid the errors you have been getting.

Thanks Bill but this is for hot-spot reference. The results are for images that are 90% marketing and 10% statistics and easily digested. I am going for pretty over science(y). I have a main features class of 1.4 million records so having to make new data for each subset I am trying to display is a bit annoying.