Register | Login
Attackpoint - performance and training tools for orienteering athletes

Discussion: Seeking advice from the Mapping Geeks

in: Orienteering; General

Mar 10, 2008 8:27 PM # 
Hello--I am trying to improve on the topographic resolution of the base maps for our club, which at this point are essentially USGS topo maps. At long last I have found a relatively cheap source for digital terrain model data in our area that has 10-ft vertical resolution.

Can someone recommend software and/or a procedure to read the data in, interpolate the data to create 5-m elevation contours, and then get those contours into OCAD with the correct positional coordinates? And, of course, cheap is better :-)

The source data are "available in Arc/Info Generate, AutoCAD DWG & Microstation In-Roads formats." I'm clueless as to what these formats are...

Mar 10, 2008 9:07 PM # 
I'm not much help here right now. I used to be, but Brian Coleman with RMOC would definitely be a guy to contact.
Mar 10, 2008 9:15 PM # 
Where I have seen this done it has generally not met the "cheaper is better" option. It would have been with the Arc package.
Mar 10, 2008 9:35 PM # 
I agree with Ratlaf: RMOC's Brian Coleman would be a good contact for you. I think his focus is more on mapping in general (rather than O-specific maps). But, he's working on a GIS degree, and one of his projects was to automate the USGS-to-OCAD process.

Mar 10, 2008 9:40 PM # 
p.s. Are you also aware of this O-Map discussion group?
Mar 11, 2008 7:26 PM # 
Thank you--I will follow up with your suggestions
Mar 13, 2008 2:41 AM # 

This is what we use in our club to get things prepared for OCAD. Fairly easy to use and only costs about $400. Cheap for GIS software.
Mar 13, 2008 2:49 AM # 
....and you can download a free evaluation copy to make sure it works on the data format you need.
Mar 19, 2008 5:28 PM # 
CactusMagnet: Please post anything that you learn here. Automated transfer of USGS data to OCAD would be very nice.
Mar 20, 2008 2:22 AM # 
Not wanting to come off condescending, because I realize that resources drive these decisions, but I'm not sure why you would want to do this. If you're going to interpolate USGS contours, you're accepting the idea that your contours won't be particularly good. In that case, why is the interval important? Simply using the contours as is and calling it 3m would seem much simpler.
Mar 20, 2008 2:47 AM # 
I've run on a map made with interpolated contours like this, at a reasonably high-profile event. The consensus was that it... really was not very good at all. To put it politely.
Mar 20, 2008 3:34 AM # 
My first step would be to visit the county government GIS department, which varies from the office of city planning to the assessor's office to the engineer's office.

They often have 2foot contour line engineering maps for the whole county. The quality of the 2ft contours varies widely. It's often the product of a digital elevation model that has way too much averaging. Sometimes it's quite good. A typical automated product (and I am no expert here, so take this with a grain of salt) takes a low-res digital elevation model (DEM), makes a "Triangular Irregular Network" (TIN), which is a mathematical surface of triangles, which is no better than the DEM averaging, and then uses that TIN to develop contours of any requested interval. When you look at them on a screen or printed out, you can see triangular artifacts in the contour shapes. Again, the problem is that the DEM is often low res. I've heard of DEMs that return one elevation result for an area of 100x100 feet. 30x30 feet is also a typical value.

Here's a TIN I found with google. You can see the terrain shapes aren't realistic---you can see obvious triangular artifacts.

That GIS office almost always has the ability to print maps with contours and an orthographic aerial photo. The ones I have visited are used to printing engineering maps at 1 inch = 200 feet (1:2400). They often can print to whatever scale you want, typically 1:2500 or 1:5000. Some offices can't or won't print to an arbitrary scale. They want to give you a 1:2496 map (1 inch = 208 feet), which is so close not to matter (usually).

You should also ask if the state or county has had lidar data taken from aircraft. This is considered the next big thing, but only a few people know how to do it. Eddie has done some of this, and the results appear to be amazing. I can't find any of the stuff I remember seeing a few years ago. The website below is one of the few about lidar for orienteering:

LIDAR Basemaps for Orienteering

I know I could use some hand-holding to figure out how to make basemaps from lidar data, and I suspect there are others.
Mar 20, 2008 8:52 PM # 
LiDAR based O-mapping! Talk about expensive though! As a proud mapping geek who works with LiDAR datasets on a daily basis I can profess to its power ...we recently mapped a 900+ sqkm area of ground elevation in remote northern Ontario, completely tree covered I might add, using LiDAR and achieving sub 10cm vertical accuracy throughout.
Mar 21, 2008 2:34 AM # 
Actually, the LiDAR data is free for much of the costal US. Inland areas are getting it a bit slower; but where the govt. has done it, the data is public domain. Of course, there's still a lot to do after acquiring the raw data, but it's definitely the future of base mapping.
Mar 21, 2008 3:31 AM # 
All of Ohio was lidar mapped in the winter of 2006 for Northern Ohio and 2007 for Southern. And as ebuckley said, it's all public domain. I'm pretty sure they used 2m post sizes unless the county specifically paid extra for 1m posts.

My understanding is that the lidar last return data is essentially a high-res 2mx2m or 1mx1m Digital Elevation Model, rather than the 100x100 foot ones I mentioned earlier. This makes the TIN and the eventual contours very detailed, even noisy. If I recall Eddie's work correctly, there are some noise reduction steps to improve the data.
Mar 21, 2008 5:51 AM # 
Thats about right. The "last return" lidar data still has quite a bit of leftover vegetation/buildings etc that has to be filtered spatially to get the bare earth you need for contouring. Usually the agency that has acquired the data also post-processes it to produce a clean bare earth DEM or TIN and provides that as one of the products (in addition to the raw "point cloud" data). You also get the return intensity, which is effectively an IR photo of the area at the same posting as the elevation data. Agencies often make contours from the final DEM (usually 2-foot) and provide these in vector formats.

More and more statewide and countywide data is available for free or for a nominal fee nowadays. County GIS departments, some state ones, and major city GIS depts are good places to look. Maryland has statewide coverage that is distributed by county, and each county does it a little differently. Baltimore county for example sells the data (currently $90 for "4 tiles," which covers several km^2). They provide the raw data, intensity, and a bare-earth dem. Cecil and Howard county 2m posting data is available for free from a NOAA coastal erosion website. Pennsylvania is in the process of collecting statewide data that will all be free. The western 1/3 is already done, and the middle 1/3 is due this summer. I did a test with a piece near Pittsburgh for Ed Hicks to compare a recent photo-derive d base to a base from lidar using this data and the contour match was excellent. Connecticut has statewide data at 4m posting, which is very coarse. Its about the same res as you'd get from USGS data, but its more accurate (that is, it matches reality even though the res is low). Louisiana has statewide coverage via FEMA and the Army corps of Engineers for flood mapping. Its all available for free online. Most of North Carolina is available through CLICK, as is some of the LA data, and there is alot of coverage of Puget Sound free online.

From what I've seen, data that is 2m posting or better will get you
the same quality you get from a standard photo-derived basemap. If your average point spacing is much larger than 2m then you start to miss detail in the contours that is important for O. As you go to lower and lower spacing approaching 1m you get incredible ammounts of extra detail - beyond what you need for just the contours.

The Balto county data is just over 1m posting and I can pick out rootstocks under canopy from the hole behind the rootball, large boulders and most trail treads. Other info you can add from the lidar are roads, trails, buildings, clearings, ditches, streams, pits, depressions, knolls and in some cases you can judge vegetation thickness from the quality or lack thereof of the vegetation filtering. You can also tell the difference between evergreen and deciduous, and you can see things under most pine canopy. In one case we were able to map a ruined barbed-wire fence in the middle of the forest using average vegetation height. The fenceline followed an old farm field edge and the average tree height in the old field was 5m shorter than the forest on the other side. You couldn't tell just standing under the trees.

Francis Hogle pointed out to me that one of the tougher problems with the lidar data while drafting is being careful not to put too much detail from the lidar base on the final map, as you can often see things in the lidar that are just too subtle for an orienteer to notice. I can make out 100 year old farm field boundaries under canopy that are just 15cm above the average terrain that are impossible to find when walking through the woods.

Anyways, a number of O maps have already been produced from Lidar - notably QOC's 2007 US Champs map last fall, and a map north of Baltimore that Ted Good and others have fieldchecked for an A-meet next spring. I know Mark Dominie is working with lidar bases in the NE these days. As soon as central and eastern PA are covered by the state, SVO and DVOA will be in basemap heaven. This will exacerbate the current shortage of good fieldcheckers in the US. We're going to see alot more lidar-derived maps from now on.

Regarding cactusmagnet's original question, I looked into the three file formats he listed. Two of them at least seem to be simple ascii text, so might be easy to manipulate. However that globalmapper demo might be a good place for him to start. It claims to be able to read at least one of his 3 formats. Looks like a pretty capable package, but I haven't used it myself. As jj and others pointed out though, most of the USGS DEM data is pretty poor no matter how you slice it. The 30m DEMs are way too low resolution. The 10m DEMs provide about the same resolution (same information content) as the printed USGS topos, so even though its easy to re-contour those at 5m intervals, you haven't gained any new info and they are pretty bad to begin with. I've converted 2-foot vector data into grid, made TINs and then re-contoured at 5m (with smoothing to overcome coarse jaggies). That works pretty well as long as you always keep the limitations of the original data in mind. You never get something for nothing. Changing the contour interval or making the paper USGS data digital doesn't improve the original content - just changes its form. Avoid the USGS data if you can. If its all you've got, go ahead but be very careful with it. Some might suggest a white-paper base would be better than using mis-information to start, but that does require some genuine fieldchecking skill (which I certainly don't have).

There are a number of free GIS tools out there. Notably the GDAL libraries in C or Python, but these do require some operator skill - even just to get an executable program the average user can run. I've been working on a set of simple lidar/gis tools for orienteers which Greg Lennon and James Scarborough have been helping with. These are written in a proprietary language called IDL but its possible to export compiled applications without requiring a license (i.e. its free). We've been testing the first basic tools distributed on mac OSX, windows, and linux. The idea here is not to produce a huge general-purpose GIS library, but to provide the basic tools an average club user would need to make an OCAD basemap from lidar data (and maybe read a few other useful vector formats like shapefiles - although OCAD9 can read shapefiles directly now I think). I'm easliy distracted by shiny objects though, and racing season is looming. Plus my apt is now surrounded by tens of km^2 of lidar basemap, which is very distracting. Greg has some other free contouring and display tools linked at, some of which can write DXF for import into OCAD.
Mar 25, 2008 12:34 AM # 
Wow! I fully expected that there would be a lot of expertise out there on this subject, and I have not been disappointed. Thanks for the comments.

I'm rather envious of states that will be putting Lidar imagery into the public domain. Arizona appears to be years behind in that thinking. Besides the 100 ft posting/ 10 foot elevation data that I referred to earlier (which you have now convinced me would be next to useless), the only other data that I have found is 2 foot or 4 foot elevation data at $90/km^2--but only available in limited areas. This will bring the investment in data + software to well over $1k to get one base map--but that may be the best I'll find for the next few years. Of course, being the paranoid that I am, I'll be worried that any public land that the county has surveyed at finer resolution probably means that the land is about to be swapped for the next new housing development. But that's life in the sun belt, I guess...

This discussion thread is closed.