Fishnet features have different length and area

[KyttMacManus]

Hi everyone,

I am running into a problem with the create fishnet tool.  It seems that everytime I try to create a 30 Arc Second regular fishnet there are slight differences in the length, and area measurements of the resulting features.

If I create the file and then choose to select by attribute so I can see the number of unique values in the Shape_Area field the result is several possible values.

If I select one of them there is a structure to the selection such that it results in a checkered type pattern.

I have tried several manners of running the tool playing with parameters such as the coordinate system of the output, but always see the strange results.

Has anyone else experienced this?  This is not the normal behaviour I would expect from the tool given what I have read about how it works and what it is supposed to do.

Any advice would be appreciated.

Many thanks,
Kytt

[DaleHoneycutt]

Something is amiss, but I don't know what.  Particularly disturbing is this : "If I select one of them there is a structure to the selection such that it results in a checkered type pattern."  It sounds like the Create Fishnet tool didn't construct the polygons properly.

One thing you can try is running the Feature To Polygon tool.  As input to Feature To Polygon, enter the fishnet.  You'll get a new feature class and this feature class should have properly formed polygons (no "checkered type pattern").

If this doesn't solve anything, respond to the thread with the parameter values you used to create the fishnet.  One way to easily get the parameter values is to use the Results window (10.0 and higher).  Right-click the result of your fishnet run and click "Save as Python snippet", then paste that snippet in your response.

[KyttMacManus]

Thanks for the fast reply Dale, much appreciated.

I tried running the Feature to Polygon tool, but that didn't correct the problem.

Here are the parameters I used to create my fishnet:

arcpy.CreateFishnet_management("D:/LECZ/output/ARE_output.gdb/fishnet_test","51 22","51 32","0.0083333333","0.0083333333","0","0","57 27","NO_LABELS","D:/LECZ/fldgrids/arefld","POLYGON")

It is worth mentioning that the differences appear some 9 decimal places deep, I have been thinking about whether that precision would even matter for my purposes.

However, there is certainly a structure to the differences.

I am running ArcGIS
Release Version: 10.1
Product Version: 10.1.1.3143


Thank you for your help with this,
Kytt

[DaleHoneycutt]

I'll need the dataset D:/LECZ/fldgrids/arefld  which you're using as an extent template.  (Sorry, I thought that the Python snippet converted the template extent to raw numbers).

I tried this w/o a template extent, entering 20 rows by 20 columns and setting the output spatial reference to WGS84 (since I couldn't provide a template dataset, the output spatial reference is <Unknown>.)  I did get slightly different values in length and area, probably similar to what you're seeing. 

Investigating your statement that "If I select one of them there is a structure to the selection such that it results in a checkered type pattern."  I finally understand what you were doing -- you selected all polygons whose length or area were the same.  Doing so creates a selected set that is a checkerboard pattern.  I thought there was an issue with the structure of the polygons -- that you were selecting one random polygon and a checkerboard would appear (which would be terrifying, frankly).  

Back to the coordinate values -- in order to investigate further, I need your template dataset ... or I need to know the exact XY Resolution, XY Tolerance, and Extent.  If you could create a file geodatabase that just has your template dataset, then zip the file geodatabase and send it to me (dhoneycutt@esri.com), it would go a long way towards explaining what's going on.  Just to give you an idea, I created a template dataset in WGS84 with the default tolerance (very small), used that in Create Fishnet, and my length/area numbers were all the same.  I'm thinking that the XY Tolerance of your template dataset isn't small enough.  What's its lineage?