# Help, I have 10,000+ .shp files covering 3 UTM zones with unknown projections

I was handed a ton of property files with no projections attached but was told they are all in UTM zone 21S. Well once I merged them all together and projected them I realized that the data covers 3 UTM zones 20S, 21S, and 22S. So basically each property file could be one of 3 UTM zones. Is there a way to figure out what UTM zone each property file belongs too without checking each one at a time? I can python code if need be.

I'm using ArcMap 10.2

I tried that. Here is a more detailed explanation of my situation. These files were original downloaded by UTM zone. So all properties in UTM z 20S were downloaded together with a projection of UTM z 20S, all properties in UTM z 21S were downloaded together with a projection of UTM z 21S ect. The shape files I received didn't come with the .prj files and I have no way to obtain them. So when I project them all with the same coordinate system say UTM z 21S, they overlap, all of the property files from UTM z 20S and 22S are overlapping the correctly referenced properties of z 21S. The misplaced properties are located in the correct area within the UTM zone, just the wrong zone. Does that make sense?

• What software did you use to merge them all together? Mar 27, 2014 at 22:24
• It is not uncommon for a shape to fall into a neighboring UTM zone, so it's possible your data may be correct. Mar 27, 2014 at 23:00
• If the coordinates of 20 and 22 fall outside of the normal extents of zone 21 - then it is likely they are all projected to 21, even though technically they don't belong there! Mar 27, 2014 at 23:29
• That is correct dklassen, that is what is happening to me Mar 27, 2014 at 23:37
• You basically have three overlapping zigsaw puzzles. If all the properties within each UTM zone are adjacent, and form a connected block, then a programmatic solution is possible: it would leave you with three blocks and you would have to determine which one is in which zone. In the absence of such adjacency information, even a programmatic solution will fail: the best you could hope for is to identify clusters of adjacent properties and you would have to manually examine each cluster. Such programming would be complex in any case: maybe it's easiest to re-download the data? Mar 28, 2014 at 19:22

I have edited this answer to reflect the new information in the question. I am leaving the old solutions for anyone who might find the question in future searches but actually have those issues.

As dklassen pointed out they may have been created in 21S even though they don't 'belong' there, in which case bringing them in/defining them as what they should be would be incorrect. You'd want to bring them all in as 21S and then project those necessary from there to the other zones - or leave them as-is, depending on your purposes. Essentially since the projections aren't included you have no way of knowing beyond what you were told, regardless of what they should have been created in.

As far as identifying which ones should be where, you could do it by using shapefiles for the zones and then doing a select by location - either programatically with the individual files (beyond my knowledge) or as two selections with them all merged to a single database.

However in your case this will not work, as you have clarified (contrary to the comments on the question) the data was created in the different zones, hence if you define them to the same they will come in overlapping as you stated. Unfortunately without projection information there are only two other possible ways I can think of to determine which zone they should be in. Either you would have to be familiar with the shapes (and even if you were with 10k to sort that's a great deal of work), or there would have to be something in the attribute data that would distinguish which zone they belong to. Not necessarily the zone itself coded to an attribute, but perhaps other reference information such as a 'county' or 'state' or 'province' or city or whatever other geopolitical boundary the property is described as being in that you could locate. Otherwise you just have a bunch of random shapes and how, even checking individually, would you know which of the overlapping ones was in the correct zone and which wasn't?

Another possibility occurs to me, and it's a remote one, is using the timestamps on the files to determine which were downloaded with which, assuming each group/zone was downloaded at the same time and modifications haven't been done to alter all of the different timestamps (modified/created/etc).

• +1 I was going to suggest intersecting the shapefiles with a UTM feature class.
– Fezter
Mar 28, 2014 at 0:31
• I added more to my post Mar 28, 2014 at 15:26
• Thanks for the detailed answers, me and my work group literally tried everything suggested but to no avail as there is not identifying attribute. I found out the person who originally downloaded the data and it turns out he is only a programmer who created a bot to grab there files from a website and he has no GIS knowledge so we are hopping that he just missed grabbing the .prj files. I'll keep everyone updated. Mar 30, 2014 at 17:05

There are a few Reference systems shapefiles data that ESRI provide them when we install ESRI ArcGIS from DVD. You could find it in the "C:\Program Files (x86)\ArcGIS\Desktop10.x\Reference Systems" and in it has a shapefile that has UTM zones, but the coordinate system is in Geographic Coordinate System WGS 1984. So perhaps when you load them your UTM with unknown in ArcMap and then bring utm shapefile use Fezter's suggest with intersect to know which utm it is in .

• Since the shapefiles have no projection or coordinate information, all you know is that each feature lies in one of three UTM zones. Intersecting it with any of those zones will not provide any additional information. Mar 29, 2014 at 21:31