topic Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes in Geoprocessing Questions

Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

JessieFoest — Fri, 08 Apr 2016 15:03:12 GMT

Dear ArcGIS community,

I am trying to add attribute information of a soil type polygon layer to a biodiversity (output) polygon layer, where both polygon layers are different (and thus, there are different soil types in each biodiversity polygon). I only want to include the soil type that is most common in the biodiversity polygons (i.e. largest area) Thus, I thought, I would preform a Union, then calculate the area of each section and put this largest area and the soil type of this area in centroids. Then I wanted to perform a Spatial join of these centroids to the original biodiversity layer, but in a way that ONLY the soil types would be added if that soil type is most common based on the area I calculated before.

I have, however not found a way to do this (i.e. make the output soil type attribute dependent on if the corresponding area is larger than the other soil type areas within the biodiversity polygon).

What would be a way to approach this?

Thank you so much in advance, it really means a lot!

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

WesMiller — Fri, 08 Apr 2016 16:01:51 GMT

Could you add a field to both and calculate the areas on both then do your spatial join then do a selection area1 > area2 or area2 > area1

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

JessieFoest — Fri, 08 Apr 2016 16:22:58 GMT

Sorry, I do not understand your answer entirely? Could you please clarify it a bit further?

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

MervynLotter — Fri, 08 Apr 2016 18:17:00 GMT

Hi Jessie

I would convert my soil data to raster, and then use *Zonal statistics as

table tool* to and Majority statistics type to sect the most common soil

type in each polygon.

Regards

Mervyn

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

JessieFoest — Fri, 08 Apr 2016 20:14:41 GMT

Thank you, I think this did the trick! I will check in the morning and mark the question as answered

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

JessieFoest — Sat, 09 Apr 2016 06:56:08 GMT

Zonal statistics as table worked! Only, I get 18500 out of my 87.346 overlapping polygons? Is there a limit to the length of the Zonal Statistics as table tool, and can I work around this?

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

DanPatterson_Retired — Sat, 09 Apr 2016 07:05:09 GMT

Curtis posted a couple of links here https://community.esri.com/message/519722#comment-519722

and the blog reference is here https://blogs.esri.com/esri/arcgis/2013/11/26/new-spatial-analyst-supplemental-tools-v1-3/

designed to handle overlaps

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

MervynLotter — Sat, 09 Apr 2016 08:05:38 GMT

Hi Jessie

I have run this tool on a lot more polygons. I suspect something else my be

at play, as Dan suggests, perhaps overlapping polygons.

I gather you would have linked the zonal stats table back to your polygon

data. Select those with zero values and run zonal stats tool just on these

to see if perhaps now the balance of the units can be populated with soil

data.

Cheers

Mervyn

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

JessieFoest — Sat, 09 Apr 2016 17:27:14 GMT

Thank you all so much for the help. I've tried the new zonal statistics as table 2 tool, and run a 'Find overlapping features' analysis which did not find any overlapping polygons.

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

DanPatterson_Retired — Sat, 09 Apr 2016 17:37:50 GMT

Your earlier thread said you had overlapping zones. Is this not the case?

I hope you aren't confusing features in a featureclass to zones in a raster. You can have 1000 discrete polygons in a featureclass consisting of 10 classes... no overlaps... when converted to raster you will have 10 classes and all the apparent discrete zones have vanished since in a raster, it is the class that is important not their spatial position. Regiongroup is the tool to use when you convert vector polygons to raster to represent different spatial areas. Regiongroup assigns a unique value to each contiguous zone/area. It is equivalent to doing a multipart to singlepart operation in vector world.

So.. is there any overlap in the features that were converted to polygon?

You can alter the settings...within reason... that Arcmap uses for some options. You can access them from

Advanced ArcMap Settings utility—Help | ArcGIS for Desktop folder location if you have admin privileges on your machine.

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

JessieFoest — Sat, 09 Apr 2016 20:11:19 GMT

I think I did not confuse them? the overlapping referred to the original situation where I had two different polygon layers (e.g. the soil type polygons and the biodiversity polygons).

To clarify: Originally, I had a polygon (feature) layer for biodiversity (and the boundaries of this layer are to be the zones in the end-result), and a polygon layer for soil type. I converted the polygon layer for soil type into a raster, where each pixel contained information on the soil type number only (i.e. different classes?). Then, I used the Zonal statistics as table tool with as ' input feature zone data' the biodiversity polygons, as 'zone field' a copy of the OBJECTID (to make it more stable), as 'Input value raster' the soil raster file, and as statistics type 'Majority'.

Thank you for taking the time to look into this question. I really appreciate it.

Re: Spatial join: Joining all attributes to a polygon layer IF they correspond with the highest value of one of the attributes

DanPatterson_Retired — Sat, 09 Apr 2016 21:40:49 GMT

intersection or combination might be the term I would use for two different layers. Overlap often refers to just that within a layer, like when you buffers can 'overlap' in a layer. semantics aside..

If you have two raster layers and you want to do 'stuff' with them you need to ensure that you are aware of the extent of both (left,right,top,bottom), their cell sizes, whether the edges of the cells align (snap raster issues), and their cell sizes.

Given the above, you could have taken two rasters, each representing a different parameter and each having a different set of classes. In your case you have not one...but two rasters in integer form that represent some variable... soil class and biodiversity. In a case like this you might want to look at the 'association' (don't use correlation as the term) between those two. The tool to use in that case would be the Combine tool

Combine—Help | ArcGIS for Desktop

This will tell you the various combinations of the classes between the two variables. In effect, does X in the Biodiversity layer tend to be more prevelant in areas where Soil Y occurs. It does this by producing a raster that shows the unique combinations between the two inputs and their classes. the cell counts relative to their total presence in the landscape gives you data akin to producing a contingency table in a Chi-squared analysis. So, don't rule out this as an option...the zonal statistics stuff involving means, standard deviations only are useful when one of the rasters represent a variable measure on an interval/ratio scale ... like precipitation, temperature, income, age... relative to a raster classified on a nominal basis ... like soil type, census tracts, townships etc... Why? there is no such thing as a mean biodiversity, because biodiversity is at best an index and not a physical quantity.