Dem Grades & Accuracy

Here was an image of how it was off when filling in water. I have no idea because it's 70M DEM data or the UTM is off.

Sample-of-DEM-tata-and-Visual-Earth-being-off.jpg


Of course when i do the Niagara gorge it looks much better because it's bigger, but if you look carefully you can still see the issue on the east side of the river over the map with room to move over the west side (side).

My-Trainz-Screenshot-Image.jpg


I'm just curious if others have encountered this and the best method. The only thing I can come up with is fill in the road grid and start to move the track and roads
to compensate for the discrepancy. Then make sure I attach the track to the ground once I'm satisfied everything is lined up. I take it the DEM is accurate enough on position and the UTM is the one off here.

Again all advice is appreciated.

Thanks

Sean
 
Sean,
I'm not sure I understand the situation exactly, but it the UTM is correct, but in the wrong location, you can drag those around like any other object. And if they are on their own layer, you can lock the layer, so they don't get accidently moved. (I know, I've done it...)
 
I never thought of that. Still trying to figure out what TransDem does I guess. So the object is formed to the elevation and not a flat plane like a Basemap. I raised the elevation of one and I can see it. So if the map is a bit off your SOL. Overall 40-70M off is not the end of the world. Just going to look a bit strange by the water I guess.

I could paste the roads and move them along with the rails plotted. Too bad Basemapz doesn't form that nicely to the surface.

Still not sure what the best solution is to remedy. I can see the water line is off in DEM but the image is showing not in the correct location. When I move the image it hugs the earth. I'll have to think of something.

Thanks

Sean
 
Last edited:
Sean,

Your tiles will remain in the same exact location because they are given real world coordinates. I have taken TransDEM routes generated by someone else, chopped parts of that route away and merged in a route that I created. There was little if any noticeable difference between which route ended and which began.

You might also find this information interesting regarding Canadian data. The source of your data may explain the anomalies you are experiencing. This is from the TransDEM user manual on pages 179 - 180:

Canada
The Canadian government provides DEMs and topographic maps in an even more
convenient way than the geo servers mentioned before.


All web links verified at the time of writing (12/2015).
TransDEM directly supports Canadian DEMs of type CDED and topographic maps in a
variety of formats, all available from http://geogratis.gc.ca/geogratis/search?lang=en.
DEMs and maps from these sources have the same sheet lines, so DEM and map will be congruent.


The largest scale maps are 1:50000. Matching DEMs have a resolution of 0.75 arc secs (about 15 x 23 m at 50°N).


Maps
At the time of writing topographic suitable maps for TransDEM are offered as
● Toporama, downloadable GeoTIFF images for automatic georeferencing.
● Toporama WMS, compatible with TransDEM client.
● CanMatrix: downloadable full map images with legend in GeoTIFF format for
automatic georeferencing.
● CanTopo: downloadable full map images with legend in GeoTIFF or GeoPDF
format for automatic georeferencing.
● CanVec, downloadable multilayer vector data in ESRI shape format.
Raster maps are based on NAD83 datum which is equivalent to WGS84 for TransDEM
purposes.


Toporama
Downloadable Toporama raster maps come in UTM and lat/long projection. For
TransDEM, choose UTM. Format is GeoTIFF and georeferencing in TransDEM will be
automatic. These are margin-less maps but detail is not at maximum for 1:50000 scale.
Toporama WMS
Toporama Web Mapping Service can be accessed with the TransDEM WMS client. The
URL is http://wms.ess-ws.nrcan.gc.ca/wms/toporama_en

CanMatrix
These are hi-res printable maps with a legend and margins (map collar). There is more
detail than in Toporama. They come in UTM coordinates and GeoTIFF format for
automatic georeferencing but need to be pre-processed due to their size and the margins.




CanTopo
This is a new generation of hi-res printable maps with a legend and margins (map collar).
CanTopo is available in GeoTIFF or GeoPDF format for automatic georeferencing but
needs to be pre-processed due to its size and the margins. GeoPDF comes with a “neat
line” to enclose the map frame.


CanVec
CanVec is multilayer vector data in ESRI shape format. The level of detail is equal to
Toporama. TransDEM can read and use the data but cannot distinguish the layers. The
recommended way is to use one layer at a time.


Note: As with all vector data in TransDEM, a DEM or another map of the area has to be
loaded first.


Documentation for the CanVec layers can be found here:
http://ftp2.cits.rncan.gc.ca/pub/canvec/doc/CanVec_feature_catalogue_en.html
The railway layer is coded *_TR_1020009_*.shp


DEMs
Canadian DEMs of type CDED have their own format, which is of file type .dem, but
different to MicroDEM .dem. TransDEM supports CDED type .dem, but file type needs to
be set either explicitly to CDED or to “all elevation data”.


These DEMs have the same naming scheme as the associated maps.
CDED DEMs are in geographic coordinates. TransDEM automatically converts them to UTM, a procedure similar to handling SRTM DEMs.
 
That information is perfect! I was on their site but got a 404 so I just went to their home page and found what I needed.

This information is very useful. I'll pass it along.

Thanks

Sean
 
Purchased TransDem just waiting for Bank draft to clear. Thanks for everyone's help. I'll be likely playing around with this middle of next week. Enough of what I've seen has sold me to support the project.

Thanks

Sean
 
I thought I would bump this one since I know I've been playing with this for over 6 weeks now, and I highly suggest it over eyeballing it. Simply put there is so much that the eyeball misses.

Despite the good tutorial by the author on merging DEMS, I eventually went another direction this weekend. I've probably generated routes now for over 6 weeks. Didn't understand the 5m ability and had to play between long/short load times vs. good/perfect UTM and DEMs.

I've gone as wide as 13 tiles to the route and 3 UTM's. Granted this is not needed in many areas, it does give the good views I like from the escarpment overlooking Lake Ontario on CP climbing it. While CN hugs along the lake until slowly making it's climb in St. Catharines towards Niagara Falls. Once above the escarpment your really not seeing anything far except in rural areas that don't have tree blocks.

As tempting as it was to do all the lines I wanted in Niagara at once I stuck to the CN Grimsby Sub and merged it manually with the CN Stamford Subdivision. This allows me to have plenty of operations as I originally intended (before DEM tempted me in doing much more such as CP and the short line). So sticking with 1 railroad and being happy with the track locations at yards and industries would be a good starting point. Since I could actually run some sessions yet because I set my yard in stone could now save and continue on the next while working on the route. It also will allow me to trim out the unwanted tiles (with the extra UTM's I don't need). My goal is NOT to generate another DEM on these subs. I've already started over from scratch one and I don't want to do it again :) I save versions by the data of creation if I want to go back.

Google Earth is a great side reference but their elevation is also limited to data similar to a DEM and usually does not include man made objects. Roads from overpasses often come down to meet the track. Wider roads sometimes does the opposite and the track comes up to meet it. But one thing that is indispensable is living in the region you are modeling. I've made several trips to every grade crossing (and yes despite trespassing in some rare instances I am finding the tops and lower ends of the grades). I take my video camera with me and most cases if there are enough grade crossings is sufficient without walking any track. I monitor with a scanner and ATCS so I know when trains are coming when I do want to venture a bit further. But the camera with it's zoom is very good at capturing where the low and high points are. It also unlike the Google Mobile's views gives you a much further depth view of what is happening.

The other tool I like is the actually driving it. Many times looking at the grade profile you can see where the natural grade is. So this works as a great tool. I don't think I'll be using the smooth spline tool because even with this class 1 it's just too wide. Yes I'm running my trains through the ground of the DEM like it's a tunnel in some places there should be a cut. Perhaps I may use it there eventually if it looks good but some of the cuts are narrow and same with the embankments.

But I appreciate the help along the way. Hopefully those who have followed me along and are new to this have learned from my mistakes (I've made plenty of them). Just be patient and have fun.

Thanks

Sean
 
Now I did hear some place warning not to use smooth spline on a DEM terrain because it cuts into everything else around it.

One user -- I've forgotten who -- suggested using a narrower spline than a typical track spline, which would reduce the width of the Smooth Spline effects. That would require replacing the track spline with a narrower spline, then using the smooth spline tool, then swapping the track spline back in afterwards. Perhaps too much work when you have to preserve the height of the spline points.

It might be easier to nudge the terrain vertexes up and down to get what you need after you establish reasonable track gradients.
 
This post might be seen as somewhat off-topic, but is TransDEM related. During the Corona virus shutdown in March, to stay busy I began a new route; non-protypical but current era North America, but if North America had good passenger trains like Europe and Japan. I looked on Google Maps for interesting terrain to use for a large city and suburbs, with ocean and some mountainous terrain. I settled on the area around Hakodate, Hokkaido, Japan. Loaded TransDEM, downloaded the SRTM dem, and created the terrain only new route. Started building my route, using the best parts of my previous route (towns mainly) pasted in. Soon I decided I needed some awesome background scenery. Using Google Maps, I looked for island volcanos (island because I wanted the base level to be sea level, not too big). I found many off the coast of Japan and Alaska. Using TransDEM I made a few into their own routes, trim, trim, merge, merge, with minimal texturing and trees since its all background.

Bottom line my route now has Mt Edgecumbe and Little Sitkin in the background. Making this terrain from scratch would take forever and not be as good as the real thing.
 
Yeah I remember seeing the narrow spline post. It does sound like it's going to be too much hassle though. I'm already trying to plot in the spline points with the actual altitude of places like crossings that I can easily measure or estimate within reason.

The camera has come in handy with the zoom as from 2 locations you can see both the upgrade, downgrade, summit (or valley) in various locations. The only thing I can determine is what that altitude actually is at that location away from grade crossings. The only thing I can safely guess is if no visible difference in grade takes places between 2 grade crossings, then I can easily use the grade between those crossings as a safe assumption for the grade to the spline point.

However, issues do occur with this method if 2 points of the grade can't be measured within reason. This has made me wonder about getting an altimeter. I was looking at them the other night and though they are as detailed as 1m I've seen some mixed reviews. I assume it's people who hike or climb who rely on this over a larger period of time. Because the altimeter is adjusted by barometric pressure. But in some cases it sounds like cheap design as well (hard to tell). In most cases if I was to make an adjustment at a fixed height location nearby, I should have enough time to journey over (yes on property) to the peak or valley and get a measurement. Of course I'm not recommending anyone do the later.

I'm curious though if anyone has an altimeter and what they would recommend. I'm seriously thinking about getting one but I don't want to spend hundreds of dollars. I've seen some as low as $20. I was looking at one that was in a watch that is about $120. I'm afraid if it's off by 5m it would be useless. Even 2 to 3 meters difference after adjustment would really throw off a grade.

Thanks

Sean
 
Im in the process of building a 200Km section of the NSW Northern railway (now closed) using 5 meter resolution DEM. The difference between this and previous DEM’s Ive used based on the NASA STRM files is amazing. The area I’m modeling runs basically north along some of the highest areas of the Great Dividing Range and has the highest station on NSW at Ben Lomond (1363m altitude) and steep grades of 1:40. The 5m DEM of this area is detailed enough to show cuttings and embankments where the track runs. The general look of the scenery, mountains, valleys, rivers etc is much more realistic than Ive seen before with the STRM based routes. Ive used TransDEM to lay in the track and build UTM tiles which help lay in the final scenery in enough detail to show every road, house, tree and fence line etc. Then in Surveyor it’s a matter of fine tuning the grades and alignments which Ive found is much easier with the 5M DEM based ground. In some cases the track is very close to the documented grades for the area, so the ground files are very accurate to reality. Also using the 5 meter grid in Surveyor helps so that when smoothing track you don’t get such a wide area affected.

So my message is, try to get hold of the most highly detailed DEM file you can.
PG
 
Yeah unfortunately the DEM I have is one down from it and even Buffalo, NY doesn't have that coverage yet. I was fortunate enough to have a USGS image of my area across the border (but it was also missing a section from it's map in some parts but not all of Canada amazingly). So for the missing parts I believe I had 15m data from Canada and the US data is 1/3 arc. Fortunately other than the escarpment there is not that many grades and is relatively flat once above and below the escarpment.

Now you brought up an interesting point I didn't think of as I haven't tried it since adapting the 5m grid to the track only in my last DEM build. Perhaps the smoothing may not be as bad so I might give it a try or at least try and carve out a path and see how things look. I never thought of that to be honest.

I decided to only use the 5m grids on the boards with the route as I have a 13 baseboard width across the region for long views where applicable. Trying to get 5m on everything was killing my load times :) But I'll give it a try tonight. Glad you mentioned that!

Thanks

Sean
 
Sean, that narrow spline thing was mine. I use both at the same time - track on one layer, and spline on a working layer. I lay the spline over the track, do the smoothing, then delete and do little touchups. Good luck with it!
Ron
 
One thing that shows up that is very easily fixed is culverts and creeks. Fortunately I have a good track map that shows where they are (because they aren't always easily visible). But a good clue is when you track dips sharply and comes back up. Just follow the track in Surveyor up to the point the dip starts and ends and add spline points. Level out the area to match both sides then replace the track.

The video I took recently of local grade crossings and zooming is a huge benefit as well, because it shows the grade when you zoom in and can see miles in front of you at an exaggerated profile.

The smooth features is good but unfortunately if your to do that with ditches it's going to make an average. I actually tried that technique first and didn't like the results. Of course when you don't have the visuals or actual data some times that may be the best or only option.

Thanks

Sean
 
Yeah I remember seeing the narrow spline post. It does sound like it's going to be too much hassle though. I'm already trying to plot in the spline points with the actual altitude of places like crossings that I can easily measure or estimate within reason.
...

Thanks

Sean
I try to grade the whole route fixing the track spline height and then terraform the ground to match the track. I use survey maps or track charts of the route that have grade data. Or engineer it myself if little or no info is available. For the EBT RR I model I work to the survey maps produced in 1917. These give elevation data and grade and indicate distance along the track as station distance. There's also an indication of the ground profile below the track and the vertical transitions used at changes in grade. Data on all curves and locations of all junction, bridges, trestles, tunnels and crossings are also indicated.

The 10m DEM data I'm using for the current version has most of the man made cuts and fills and also natural stream and river beds. In previous versions I had a mix of 10 and 30m DEM data and of course the definition wasn't as good but still usable. I decided to hold to the track line I laid out from the survey data and terraform the ground to suit. I use track with embankment/fill below it so I try to avoid the smooth spline tool and never use it where the fill is visible. Where the track is "submerged" in wide cuts and needs to be uncovered it is sometimes useful. For narrow cuts not aligned with the baseboard grid I just avoid it. When I do use it I mostly end up terraforming to reduce the width of the affected areas.

From my experience using a "narrow or wide" spline to change the width of ground the smooth tool affects doesn't seem to work. The width affected is ~40m regardless. But using 5m vs 10m baseboards does have an affect reducing it down to ~30m wide. I have splines I use for invisible track that are only 0.1m in width. I also have a 100m wide track spline I use as a track tool in mountainous areas to help engineer track grades where I have to do the design myself. Fixing grade/elevations of the tool shows the intersection of the spline and the terrain within 50m each side of the centerline. It helps to see if reasonable cuts and fills are possible or I need to move the track a bit while keeping the grade set into the tool. The smooth tool has the same affect on both these splines as seen in the pic below. As noted the 10m baseboard in the foreground shows ~40m width in the total width affected while the 5m baseboard in the back shows ~30m total width affected.

My-Trainz-Screenshot-Image.jpg


Bob Pearson
 
Last edited:
Actually that is very useful. I came to that same conclusion last night. I'm going to have to figure out how to use embankment fill. Is that an option in surveyor or the actual track you use? For now I did it manually and filled in a culvert last night with no issues. I was planning to paint in ballast around it but for now I'm not at that stage as I have full UTM's in place while I put everything down.

Thanks

Sean
 
Actually that is very useful. I came to that same conclusion last night. I'm going to have to figure out how to use embankment fill. Is that an option in surveyor or the actual track you use? For now I did it manually and filled in a culvert last night with no issues. I was planning to paint in ballast around it but for now I'm not at that stage as I have full UTM's in place while I put everything down.

Thanks

Sean

Jointed Rail has track that has a deep ballast that will fill in underneath so the track isn't floating above the surface. User jim_spb made some embankment splines of various widths and textures. They're on the DLS as j_embankment. I recommend downloading them all. The textures are easily changed and you and replace his with those of your own choice. For other embankments, I recommend looking at those that begin with FMA there are various under-track embankments of various widths that may work for you.

Other choices, of course, include various retaining walls. This is where FMA comes in and so samplaire and others.

Remember, as much as we try to be realistic with this we can only do just so much because we are still stuck with a 5m or 10m grid. Maybe in the future when our computers can handle 8 x the resolution, or a 2.5m grid we can do stuff at a higher detail, but at this current point in time and technology even today some computers barely handle 5m terrain even in small amounts.
 
Embankments can either be built into the track spline which makes it easier to work with or you can find separate embankment/fill splines that you run under the track spline and you can usually find them under scenery splines tab. I think it complicates grading a bit but makes it easier to work with 2 splines in close proximity.

I prefer them built into the track and I have one that goes with the track splines I use for the ng track on the route. But most of that is single track territory. Separate splines allow for more variety in shape and textures and probably make sense when working with several parallel track splines.

The 4 track PRR main thru the northern part of my route is wide enough where the smooth tool actually provides about the right width. The cuts and fills associated with it also show up clearly in the 10m DEMs I'm using but typically need some cleaning up. Zapping a whole section of track with the smooth tool to adjust the ground height here is tempting.

Bob Pearson
 
I'll have take a look and find something appropriate. Most of my route is single track, and double track.

Thanks

Sean
 
So today I went back to do some walking and visual surveying. Of course the old handheld GPS I have at time showed me climbing the grade but as coverage changes it can changer very quickly (not sure how much clouds can change this as I had 30% cloud cover today approximately). But anything that is not detailed to 1M is garbage and those GPS I believe are accurate to about 30M.

I've actually been looking at some altimeters and they can get quite pricey. Again I'm not sure if I want to go that much into this hobby but when I know something is not right I want to solve it, just in my nature.

So if I haven't already this is the DEM I'm using and it's on a 5M grid on route baseboards only. The DEM was derived from USGS 1/3arc which I'm told could be anywhere from about 7 to 10m DEM matrix. The double track main where I'm surveying right now west of Grimsby, ON is about 10M wide on WNW to ESE straight line.

I do not have grade charts for CN (and wish I knew how to get them). I was fortunate to get plenty of manuals of a retired engineer with CN. I also have signal charts which show grade crossing mileage, creeks, culverts, even bridge lengths, but not grades. The DEM traces some smaller areas such as 50 mile creek well (this was my target today). This bridge at MP31.4 of the CN Grimsby Sub is 83 feet in length according to the chart.

Normally I have found that when it comes to overpasses creeks be it the road to rail, rail to creek, or rail to road the DEM will always default to the lowest position. Embankments though likely not natural and built up on a route that has been in place for a very very long time just doesn't show up. It actually shows flat. Now granted this is not a wide embankment (about 10m wide). I was able to locate the lower apex of the grade at Kelson Ave MP30.90 when I walked half a mile west. You could tell there was a grade going up and indeed we were getting higher.

By the time I got to the bridge I climbed down to what was a private road that followed the creek. I estimate that the height of the bridge was about 15 feet though I'm going to look at the video and take some images again with about a 12 foot clearance. At the very least the trek got me a side shot and I found some piers I didn't know existed and wasn't visible in Google Earth (so all is not lost).

I believe I can get a good idea of the lower apex within a meter or two since it's not that much higher than the road height on approaches. However it gets more difficult to guess without proper measuring devices or charts what the actual apex top of the grade is that I can plainly see in both regular and zoomed out optics.

Is there anyone here who can recommend a good altimeter that works at 1m accuracy? I know there are plenty out there that say it. Complaints of having to adjust all the time etc. because the weather changes. I think people there perhaps just have too high of expectations. They are also trekkers who likely climb higher altitudes and take less altimeter readings to adjust properly in changing conditions. For me I'm likely only traveling no more than a mile at a time by foot so the weather shouldn't change that much and I can take another altimeter reading to confirm this.

I guess I'm looking for some experience on this matter in the field so I can make sense of what my eyes are seeing. Thanks for all the help guys, please keep it coming. Hopefully I'm helping others in other topics as well.

Thanks

Sean
 
Back
Top