topic Re: Strange Raster Pattern in ArcGIS GeoStatistical Analyst Questions

Strange Raster Pattern

BrookeHodge — Thu, 09 Sep 2010 19:25:06 GMT

Hi,

I have been using Ordinary Kriging in Geostatistical Analyst to derive species distributions based on sightings per unit effort of 5 minute squares summarized to points. I have 12 different species I'm doing this for and I have been getting strange outputs after I export to raster. I have attached a word document that has examples of what I'm talking about. About half of them look great, very smooth and visually appealing (A), but the other half have this weird star like pattern (B). I use the same methods and parameters (and have tried many others to see if I get different results) but I keep getting the same output. There is not a great deal of variation between the input data of those that look good and those that don't look good and can not figure out why the outputs looks so different. Has anyone had similar issues, know what may be causing this, or know of anyway to correct this problem?

Thanks!

Re: Strange Raster Pattern

EricKrause — Thu, 09 Sep 2010 19:54:43 GMT

At the Search Neighborhood screen inside the Wizard, try changing "Neighborhood type" to Smooth. If that doesn't work (or if your version does not have the Smooth option), try increasing the Major and Minor semiaxes. If that still doesn't fix it, let me know.

Do the geostatistical layers look smooth before you convert them to raster?

Re: Strange Raster Pattern

BrookeHodge — Mon, 13 Sep 2010 17:56:41 GMT

Thanks for your reply Eric. I tried the smooth neighborhood like you suggested. While this did help the star like pattern, the output was still weird. It looks pretty much the same, but now just more circular (B), no matter how large or small the major and minor semiaxes. Also, if the I make the semiaxes too large, I don't get the detail I want (C).

And in answer to your question, the geostatistical layer looks good before exported to raster when it is symbolized in filled contours (D), but, if I sybolize it as a grid, you can pick up on the weird pattern (E).

Any other thoughts? Thanks so much for your help!

Re: Strange Raster Pattern

BrookeHodge — Mon, 13 Sep 2010 18:13:15 GMT

Disrard my last thread. I just tried increasing the smoothing factor and it worked! It's not perfect, but it does look much better now. I think part of it is just the data I have.

Thanks again!

Re: Strange Raster Pattern

EricKrause — Wed, 15 Sep 2010 17:37:25 GMT

Glad I could help.