USA Building Footprints Map

[jbensman]

Thanks for posting the shapefiles-I was disappointed it only covered the National Forests, but I want to use some of the data.  One concern is it is from USGS maps which I have seen pretty far off.  Have you had any expericience with its accuracy?

Is this data available by state or with a state field?  Clipping by state can be a pain.  Do they have this in points?  I've added lots of NPS building footprints.  I've found the smaller ones don't work too well due to the grid.  What I usually do is sort by size and only use the bigger shapes.  Then I use ExpertGPS to convert them to waypoints and use a POI for the smaller files. 

Where did you get the data?

Thanks again for posting it.

[yogazoo]

It doesn't cover only National Forests. Check out the map I linked to with screenshots of the data. It shows Missoula, MT and Helena MT with those buildings added. There are alot of buildings and the accuracy is excellent here in Montana. Again, you can see how they line up on my screenshots. Clipping to your area should be pretty easy depending on what software you use to do it. If you have ArcGIS it's a snap to clip the  shapefile. If you use GPSedit you can clip a box around your state and then manually delete the fragments you missed outside your states boundary. SHouldn't take more than a few minutes

This is the most comprehensive and accurate dataset regarding building footprints I've ever seen. It's also difficult to find, which is why I posted it.

The coverage is good although some areas are covered more comprehensively than others. If your in an area or state that has good coverage it's really nice to have as more visual candy/reference points for your maps

[jbensman]

What it looks like to me is they traced the buildings on every USGS 24k quad that has some NF land in it.

What I was really curious about was how accurate it was in the field-are the buildings where they are shown and are they still there?  I've been impressed with the accuracy of the NPS building footprints.  But since the source for this file is 24K USGS maps which can be way off and really old data, I am curious on the accuracy.  I might try to find some overlap with NPS differential GPS data and see how they compare. 

[yogazoo]

Your comment would explain why there seem to be straight line boundaries between areas of coverage and areas of non coverage. Good observation. It would also explain why Montana (where I live) is so well represented by data in this layer.

I'd be curious to hear the results of your comparison between NPS and USFS data. From everything I've seen, the footprints that are present are all current (havent found one that wasn't yet). I'm sure there are a few buildings that have been torn down since the compiling of this data but there would seem to be very few.

[maps4gps]

Just spend over an hour on a post and got a Windows message saying the website was too busy and closed the site.  

In summary - footprints of larger buildings occuring in 7-1/2 quads with have some Forest Service lands within them, digitized from those USGS 24k quads with some more recent data from other sources.  

----  The above from comparison of the footprints with 24k DRG, recent aerial image, and USGS 24k SDTS (digital line file of 24k topo).

[Boyd]

Am I missing something here? Where are the files you're talking about?

[jbensman]

I added some of it in IL to MY Trails, unfortunately, I am removing it all due to accuracy issues.  I hope the data quality is better in other states!

Alter importing the data for Southern IL, I first looked at the State Parks where My Trails has gps locations of many buildings.  In all the parks, there was hardly any data, and 100% of the data for the parks was wrong a couple of things were close, but most was way off.  I used GoogleEarth to also verify the data was off.

Next I looked at SIU as it had building footprints for the univeristy's buildings.  Some of the buildings covered up roads.  I used GoogleEarh to check out the locations (making waypoints of building centers and tracing some of the buildings). I also did visual comparisions.   Some of it was fairly close and other stuff was way off.  Several buildings had the correct shape but were roated wrong.  Some locations were significnatly off and others were off just enough to make them cover up a road.  Some buildings were not there.  Others had significnatly differnet sizes and shapes.

I then looked at the city of Carbondale on my GPS.  Several of the buildings covered the streets.

In the past, I have tried getting building locations from USGS 24K maps and they almost always were off-just like trail locations tend to be off.

Can give a link on where you found the data?  The metadata indicates they have additional data from the maps digitized.  I'd like to see what they have. 

Boyd, look at most recent maps and download building footprints, it is a shapefile, not a map

[Boyd]

Cool, just downloaded. Unfortunately, no data for the mid-atlantic area where my mapmaking efforts are concentrated. Are you from Southern Illinois by any chance? That's where my mother's family came from.

[jbensman]

I live in Alton IL-Its right out of St Louis

I forgot to mention, some good came out of testing the data.   My POIs had the SIU buildings and most of them were way off.  In checking the data, I have the buildings in the right places now.

[Boyd]

Nice - I was born and raised in St. Louis myself. 

[Seldom]

If you overlay a USGS Quad file on the building footprints file, it appears that the footprints are laid out in some quads but not in immediately adjacent ones where municipalities continue.  And some municipalities are missed altogether.  In my neighborhood, Parowan, UT and Brian Head, UT (Populations 3000 and 1000?) are included, but Cedar City, UT (Population 20,000+?) is missing entirely.  The northern half of Bryce Canyon,NP is included, but not the south, and only the northern half of the town of Tropic, UT (Pop 1000?) is drawn.

It looks like the outlines are only drawn in quads with National Forest land.  Dixie NF in the examples above.  All subtypes appear to be drawn in the quad, whether included in the NF or not, but there are some quads with NF, but no buildings.  Assume no subtypes?

[maps4gps]

The source data is in geodatabase format.  Yogazoo converted the polygons (footprints); not the points which represent smaller structures. 

In comparing, any given pair of FS building footprint, USGS DRG, USGS SDTS vertor file, or DA aerial image; a building may conincide, be slightly different (position and/or outline), be very different, or not exist.  Even some buildings in SDTS (digitized from the 24k topos) were very different from the DRG image of that topo.  My comparisons were done for small areas west, nw and sw of Denver.

In visits with a GIS person at the local NF office over the past few years,  they seam to process/include data for an entire 7-1/2' quad if any part of a National Forest is within that quad.  This may or may not be true for other offices and probably depends on the resources (funds) available to any one office.

[yogazoo]

Sorry to hear about the accuracy issues in Ill. In Montana the data presents as very accurate. In fact I have one of those buildings right accross the street from my house and I can walk around the building, virtually outlining the polygon on my maps. I don't blame you for removing them from your trails map if the building locations are that bad. You can look at the screenshots on my Montana Topo map page and see for yourself http://www.gpsfiledepot.com/maps/view/87that the accuracy is right inline with the high accuracy of our states transportation layer.

I added the layer to this site based on the accuracy in Montana and I really didn't check out other states. As far as sheer numbers of buildings I can't find a larger dataset out there. Living in Montana, at least the west half, your almost never out 7 1/2 from Forest Service lands. Also, Montana is one of, if not the leader in geospatial data available to the public in America. One of the most amazing datasets we have is a layer I loaded to GPS with POI loader (and provided the .gpi file on this site ) that gives you the exact location of every house structure in the state. You can search by address or NAME OF THE OWNER! If someone calls me and asks me to come over to their house all I need is their name! We also have every private land owner parcel mapped with the name of the owner. It's called Cadastral and I've made a map for Garmins with it. There's nothing like being able to pointer over any lot or parcel you walk by and see the name of the person who owns it.

I guess I'll count my blessing here and keep making maps with our plethora of stellar datasets.

[Seldom]

You can search by address or NAME OF THE OWNER! If someone calls me and asks me to come over to their house all I need is their name!

Guess that's OK if you're invited.  Kinda scary privacy wise if you aren't.

[Boyd]

Also, Montana is one of, if not the leader in geospatial data available to the public in America.

Actually New Jersey has really extensive data available through a number of portals, including 1 foot per pixel orthophotography for the whole state, and very detailed vector landcover data at 12k resolution. I've tried to find good landcover data for adjoining states (PA, NY, MD, etc) and it just isn't out there. Most of the NJ counties have their own GIS sites with a wealth of data also, including tax maps which provide property owner data.

Of course, that is not as much of an accomplishment as Montana, since NJ is less than 10,000 square miles.