Tabulate Area does not sum up

268
7
12-20-2013 03:25 AM
AndreasHeugenhauser
New Contributor
Hello,
I am having a couple of Buffers and want to calculate the share of different lancover classes within them. Its ten meters Buffers, so every Buffer has a size of about 314m². Everything is functioning great so far, but when I Export a dbf into xl and sum up all the classes some classes are 30-40m² smaller. I calculated overlapping buffers and all the buffers are completely within my land cover subset. I already converted the Buffer and Landcover files to raster files with the same results.
Any suggestions?
regards, Andi
0 Kudos
7 Replies
DanPatterson_Retired
MVP Esteemed Contributor
calculate the area of your buffers, I suspect that they don't match exactly to your number since buffers are represented by n-gons (36 sided polygons often) and the buffer area will be slightly less than the theoretical value.  See if this accounts for your difference.
0 Kudos
AndreasHeugenhauser
New Contributor
Thanks for your answer. Unfortunately most of my Buffers actually have the right size which is 314m². But there are a lot which only have e.g. 280 or 290. I would understand if there was a slight variance but this is a quite significant difference. I attach a small .xlxs file. If you are intersted have a look at it.
Thanks, Andi
0 Kudos
DanPatterson_Retired
MVP Esteemed Contributor
How were those areas calculated?  Do any of the buffers overlap?  Is the source a raster data set?  I am less interested in the xls file but in how the numbers were derived prior to export to ensure that the error is not associated with the input features and their areas
0 Kudos
AndreasHeugenhauser
New Contributor
The buffer Areas/Zonedata were calculated around centroids/Points of building Polygons. Before putting them through the Tab Area tool every Buffer had an exact area of 314,159.... The land use/classfeatures I developed in eCognition and exported them. So both files are vectors. I already converted them to raster and put them through tabulate area aswell with the same result. The Buffers are overlapping, put this should not be a Problem. I allready checked, there are also Buffers which are standing alone and having wrong area values.
0 Kudos
DanPatterson_Retired
MVP Esteemed Contributor
I suspect that the problem areas are where buffers overlap, there is a thread on the forum somewhere about tabulate area with overlapping polygons which addresses this issue since buffers can't overlap in raster world for sure.  I can't put my hands on the thread right now but the solution to this issue has been addressed before

EDIT
try this thread http://forums.arcgis.com/threads/60035-zonal-statistics-with-overlapping-polygons
0 Kudos
AndreasHeugenhauser
New Contributor
Thanks anyway.
But I still dont think thats the problem. Like I said, I used raster and vector data and both times I got the exact same but wrong values.
0 Kudos
DanPatterson_Retired
MVP Esteemed Contributor
No clue then...send your data to tech support, report it as a bug is my suggestion.  If they can duplicate it then a bug fix or workaround can be implemented
0 Kudos