DEM Earthworks

[PortLineParker]

Hi all,

I'm interested in hearing from those of you who have modelled a real life area using DEM and hearing about how you have tackled the surrounding topography. I've built on a few DEMs now and, as I'm sure anyone who has tried can attest, they seldom line up perfectly with the track and gradient profiles - this being especially so for railways that are no longer in existence, as embankments and cuttings have been obliterated to make way for housing/new roads/new railway/industrial units/trailways etc.

I'll give an example; on my current project this station is sat at the foot of the Mendip hills and should be relatively smooth sloped in the surrounding areas. Not flat, but no real embankments or cuttings. You can see below the dilemma I face as this will involve a lot of 'earthworks' to level and smooth (oh how I wish we had a dedicated smooth ground tool) the ground around the area to represent the station as it was in the mid to late 1950s. I know I'll do it and I know that it will be worth it, but I also know it will most likely be a royal pain in the backside and will take potentially several days to get right.

My question, then, is how do you all address this on your routes? Is there a specific method you use? Do you try and tackle it meter by meter or try to do as much as you can with a single click? I'm really interested to hear about your methods and I might just pick up a trick or two!

Cheers,

PLP

 

[martinvk]

As I understand it, and I could be wrong, DEM elevation is the height of the ground or whatever is there on top of it. I noticed this in urban areas that I know relatively flat and yet show a roller-coaster like DEM. Even more obvious are airports where runways and other such areas show hills and valleys.
As for smoothing it, I use a very wide brush and a minimal sensitivity setting to gently bring the terrain in line with what I think it should look like. If it's really flat like a parking lot, then it's easy but if there is a natural slope, it takes more work.
I've heard that the newer LIDAR data penetrates the vegetation and shows the actual ground elevation. Once this is more widely available, it should make DEM created terrain easier to use.

 

[dangavel]

It doesn't always work well, as it depends on the design of the asset, , you need to find something that sticks close to the terrain and some splines do not., but I often use road/rail splines to give smooth gradients or cuttings for that matter . I know what you mean about DEMS inaccuracy, the one I am using at the moment is incredibly inaccurate , roads are all out of whack, heights seem arbitrary, often up to 10 metres out from the real thing, they are better then working from scratch, but they often do leave a lot of be desired.

I am currently having to lay out a complex road system for a suburb, all built on a long incline that slopes upwards south and downwards to the west.I found using the terrain tools were not a lot of use in thei situation , the solution was fencing, by putting these splines along every 50 ft or so for the boundaries of the houses and running another across the centre of each row of houses , smoothing out the splines saved me many hours of work as it more or less gave the gradual incline that I needed . Of course , as I've never attemepted anythign like this before on this scale , it took about 3 attempts to find the right road system that would work on these sorts of slopes , dont bother to attempt using crossroads or T junctions , they just will not comply to steep slopes as they cant be tilted , I used a road system thet joined seamlessly and ran pavement splines alongside the roads, it took ages, but it was the only thing that worked in this situation, hope this helps.

 

[PortLineParker]

Hi both, thanks for your replies. @dangavel I have used a similar technique with splines in the past, though with the new HD terrain you have to be careful as it is very fine and can go very wrong very quickly! The DEM I have is a very good one and 90% of the route, to my surprise, did fit perfectly with very little effort but as I was saying to the creator of it the other day, the gradient profile on one particular section doesn't match perfectly - the track ends up about 6m underground so there are some quirks, but that is with any DEM and as you suggested @martinvk, is likely to be some sort of development since the railway closed.

Are you using HD terrain or the classic 5m/10m grid? I have found HD to be incredibly fine and an absolute pleasure to work with, albeit once I got past the not inconsiderate learning curve, but it does have it's limitations. One of those is that when you smooth the ground under two separate splines, you get an odd ridge appear - this is evident if you look at the terrain above the buffer stop and between the water column; you can see a lighter section which is in fact a dent. Using 5m grid this would have automatically married to the neighbouring section so I am making use of embankment splines to hide that but even so it's not a major deal breaker. The main obstacle for me at the moment is creating a smooth, flowing hillside that merges with the surrounding terrain. No doubt I'll post some screenshots once I make a start but it's an interesting discussion to have with others who have experienced it.

Cheers,

PLP

 

[dangavel]

Hi both, thanks for your replies. @dangavel I have used a similar technique with splines in the past, though with the new HD terrain you have to be careful as it is very fine and can go very wrong very quickly! The DEM I have is a very good one and 90% of the route, to my surprise, did fit perfectly with very little effort but as I was saying to the creator of it the other day, the gradient profile on one particular section doesn't match perfectly - the track ends up about 6m underground so there are some quirks, but that is with any DEM and as you suggested @martinvk, is likely to be some sort of development since the railway closed.

Are you using HD terrain or the classic 5m/10m grid? I have found HD to be incredibly fine and an absolute pleasure to work with, albeit once I got past the not inconsiderate learning curve, but it does have it's limitations. One of those is that when you smooth the ground under two separate splines, you get an odd ridge appear - this is evident if you look at the terrain above the buffer stop and between the water column; you can see a lighter section which is in fact a dent. Using 5m grid this would have automatically married to the neighbouring section so I am making use of embankment splines to hide that but even so it's not a major deal breaker. The main obstacle for me at the moment is creating a smooth, flowing hillside that merges with the surrounding terrain. No doubt I'll post some screenshots once I make a start but it's an interesting discussion to have with others who have experienced it.

Cheers,

PLP

I am constructing two routes in 2019 and although I have 2022 Platinum , I never use it, I do not subscribe to any Trainz version and never will, so HD terrain isn't available to me until it arrives at the standard version of Trainz, which will probably be around 2028 at the rate we are going. .

 

[martinvk]

.. dont bother to attempt using crossroads or T junctions , they just will not comply to steep slopes as they cant be tilted , ...

One advantage of S2.0 is the ability to tilt just about any object. Nice for track - road crossings at grade. Haven't tried it for road intersections, there are too many and I don't have the patience to fiddle with each one.

 

[dangavel]

One advantage of S2.0 is the ability to tilt just about any object. Nice for track - road crossings at grade. Haven't tried it for road intersections, there are too many and I don't have the patience to fiddle with each one.

I don't see how crossroads would work , unless you could tilt each of the 4 road entries in different directions and even then, the centre section would probably stay flat. This was my problem , crossroads every 200 feet or so on sloping ground which ended up with a series of flat sections at each crossroad followed by fairly steep slopes, it looked awful and it was really difficult to get the whole road to be straight as well even when using rulers.

 

[martinvk]

Here we see some road intersections tilted at various angles and joined by road splines. The splines themselves don't twist so the joint between them and the crossings are not as smooth as I would like it to be. The crossing in the foreground has been tilted in two directions at once, 4 degrees on one axis and -2 degrees on the other axis with the result that each of the spline attachment points are at a different elevation. Still using a 5 m grid so the ground isn't a smooth as it could be.

 

[JCitron]

I don't see how crossroads would work , unless you could tilt each of the 4 road entries in different directions and even then, the centre section would probably stay flat. This was my problem , crossroads every 200 feet or so on sloping ground which ended up with a series of flat sections at each crossroad followed by fairly steep slopes, it looked awful and it was really difficult to get the whole road to be straight as well even when using rulers.

I run roads across each other and do not fix the splines. I smooth the roads though underneath. This prevents the roads from squishing down the terrain and in the process losing the hill they are on. Where there are intersections, it takes a bit of practice to get the roadway smooth on the terrain without burying parts of it under the nearby slope but it can be done.

DEM files end up being inaccurate due to many reasons. Among them is the satellite picking up nearby buildings and hills causing the terrain to smooth out between them. This is mitigated a bit using higher resolution DEM data but that's not always available. Overall, 1/3-arc second seems to work well if you can get that and it's a compromise. 1/9-arc second resolution is the best but it's not available in all areas.

In my area, the railroad line comes in at 4 meters above the ground after it crosses over the Merrimack River and continues north towards Portland Maine. From where the station is located to past the downtown, is a steep hill on the left of the line that runs right down to the track edge, or even below in some places due to the line being on an elevated grade. On the right of the railroad grade are former shoe and textile mills. The end result is the railroad grade is now a roughly smoothed edge up against Washington Hill and smooths out into the former mill buildings on the right. Where the line crossed the Merrimack, then over Washington Street just before the station, the grade obliterated Water Street and Railroad square where the station is, causing yet another big blob due to Washington Hill being so close to the tracks and Water Street coming in lower and intersects Washington.

With the topographic map textures on the terrain and Google Earth on another monitor, I measured the heights at various locations and then winged it. I knew where buildings were supposed to go based on the topo map plus my familiarity of the area since I live here. I had to do a lot of smoothing and quite a bit of digging out to get the terrain.

More recently, I've been involved in a project where a railroad hasn't been in place since 1926 when it was obliterated by a dam. I was lucky that the topographic map from a bit later in the 1930s still showed the line in place, albeit abandoned, and the heights even where the water is located. Using a bit of artistic license, I literally, not figuratively, dug out the river valley using the topo map heights and measuring those locations on Google Earth. It turned out that the terrain generated by TransDEM is quite accurate and it was only a matter of digging until I came up with a river valley again.

In another location, a hill has been carved out by a contemporary sand and gravel company, Tilcon. Federal Hill was an egg-shaped glacial drumlin and on the ca 1933 topo map, the hill is intact, but when placed on a modern DEM, there are bite marks taken out of it. Using splines, I matched up the topo lines at the significant heights marked on the map, i.e. 400 ft, 500ft, etc., and set that as a height. I then placed shorter splines that I locked between two different significant heights and smoothed the terrain, similar to placing plaster-dipped newspaper pieces over a chicken wire frame. I repeated until the hill was rebuilt.

 

[PortLineParker]

Hi John,

That's interesting to hear how you rebuilt the hill. I have a couple of reverse situations where new embankments have been created since, so I will need to merge those back into the ground at points. The spline point method for creating sloping terrain seems to be a favourite and I have used it in the past so it's good to see others do too.

A new problem for you all; here we have a viaduct crossing a quarry and then heading onto an embankment. As you can see the ground drops sharply down to the quarry floor, but whether this is prototypical or not I can't see as the imagery on Google Earth/Maps is blocked by trees. How would you all address the quarry issue - leave it as is and cover with ground foliage or would you try to smooth it somehow to give more of a slope? To me it just looks odd, especially with the embankment disappearing into a wall of ground but if that's prototypical then it is what it is - I've never modelled a quarry before so I'm willing to be led on the matter!

Cheers,

PLP

 

[mcjaco]

Quarries can leave some pretty dramatic slopes. I'd leave it as is, and cover with foliage.

 

[PortLineParker]

Brilliant, thank you - if you think it looks okay as is and not completely daft then I'm happy with that as it means less work for me!

Cheers,

PLP

 

[JCitron]

Brilliant, thank you - if you think it looks okay as is and not completely daft then I'm happy with that as it means less work for me!

Cheers,

PLP

I was going to say the same thing. Leave the quarry as-is and fill in around the bridge and embankment with the trees, and bushes. Quarries have that distinctive straight edge that sets them apart from a natural valley or watercourse.

We had lots of quarries in my area, especially those in Rockport and Westford, which were places here famous for their granite quarries. The very first railroad in Massachusetts was built to hall granite from a quarry in Quincy, MA and used locomotives and rail imported from England on granite sleepers quarried locally.

It was George Fisher (gfisher) that showed us how to use splines to smooth and raise terrain probably 20 years ago now. George has sadly left the Trainz community ages ago, which is too bad because his artistic touch and experience was what inspired many of us to model and create the routes we have today.

 

[PortLineParker]

That's great, thanks John - I'll leave it as is and just plant scenery around it as usual.

Cheers,

PLP

 

[schweitzerdude]

As I understand it, and I could be wrong, DEM elevation is the height of the ground or whatever is there on top of it. I noticed this in urban areas that I know relatively flat and yet show a roller-coaster like DEM. Even more obvious are airports where runways and other such areas show hills and valleys.
As for smoothing it, I use a very wide brush and a minimal sensitivity setting to gently bring the terrain in line with what I think it should look like. If it's really flat like a parking lot, then it's easy but if there is a natural slope, it takes more work.
I've heard that the newer LIDAR data penetrates the vegetation and shows the actual ground elevation. Once this is more widely available, it should make DEM created terrain easier to use.

I agree. Using TrainzDEM, I created a DEM of the Vancouver BC area (SRTM data) and right away noticed some anomalies.
1. What are these bumps in downtown Vancouver, which is flat? High-rise buildings.
2. Why are parts of a otherwise flat area higher than other parts? Trees.
3. What are these bumps along the Fraser River? Grain elevators.

 

[dangavel]

I run roads across each other and do not fix the splines. I smooth the roads though underneath. This prevents the roads from squishing down the terrain and in the process losing the hill they are on. Where there are intersections, it takes a bit of practice to get the roadway smooth on the terrain without burying parts of it under the nearby slope but it can be done.

DEM files end up being inaccurate due to many reasons. Among them is the satellite picking up nearby buildings and hills causing the terrain to smooth out between them. This is mitigated a bit using higher resolution DEM data but that's not always available. Overall, 1/3-arc second seems to work well if you can get that and it's a compromise. 1/9-arc second resolution is the best but it's not available in all areas.

In my area, the railroad line comes in at 4 meters above the ground after it crosses over the Merrimack River and continues north towards Portland Maine. From where the station is located to past the downtown, is a steep hill on the left of the line that runs right down to the track edge, or even below in some places due to the line being on an elevated grade. On the right of the railroad grade are former shoe and textile mills. The end result is the railroad grade is now a roughly smoothed edge up against Washington Hill and smooths out into the former mill buildings on the right. Where the line crossed the Merrimack, then over Washington Street just before the station, the grade obliterated Water Street and Railroad square where the station is, causing yet another big blob due to Washington Hill being so close to the tracks and Water Street coming in lower and intersects Washington.

With the topographic map textures on the terrain and Google Earth on another monitor, I measured the heights at various locations and then winged it. I knew where buildings were supposed to go based on the topo map plus my familiarity of the area since I live here. I had to do a lot of smoothing and quite a bit of digging out to get the terrain.

More recently, I've been involved in a project where a railroad hasn't been in place since 1926 when it was obliterated by a dam. I was lucky that the topographic map from a bit later in the 1930s still showed the line in place, albeit abandoned, and the heights even where the water is located. Using a bit of artistic license, I literally, not figuratively, dug out the river valley using the topo map heights and measuring those locations on Google Earth. It turned out that the terrain generated by TransDEM is quite accurate and it was only a matter of digging until I came up with a river valley again.

In another location, a hill has been carved out by a contemporary sand and gravel company, Tilcon. Federal Hill was an egg-shaped glacial drumlin and on the ca 1933 topo map, the hill is intact, but when placed on a modern DEM, there are bite marks taken out of it. Using splines, I matched up the topo lines at the significant heights marked on the map, i.e. 400 ft, 500ft, etc., and set that as a height. I then placed shorter splines that I locked between two different significant heights and smoothed the terrain, similar to placing plaster-dipped newspaper pieces over a chicken wire frame. I repeated until the hill was rebuilt.

"I run roads across each other and do not fix the splines. I smooth the roads though underneath. This prevents the roads from squishing down the terrain and in the process losing the hill they are on. Where there are intersections, it takes a bit of practice to get the roadway smooth on the terrain without burying parts of it under the nearby slope but it can be done."
That's what i do as well, the issue is with the non spline crossroads that remain flat , they are really only designed to work on the slightest of inclines. in general, roads are one of the weakest asset types that we have and some of the most time consuming items there are to get to look acceptable , i've never come across a road set which has lower curbs every now and again for driveway sections, it can be done by lowering the pavements one by one but its extremely time consuming .

 

[afe_uy]

In my country, the terrain is very soft and detailed SRTM data is not available, plus I find TransDem or its similar complicated.
As some mentioned before, I simply lay a long section of the main line, adjusting Spline Points to the corresponding terrain height.
Then, for the surroundings I lay other rails approximately following the height lines of the topographic map of the area.
After applying Smooth Spline, I remove this unnecessary rails.

 

[JCitron]

"I run roads across each other and do not fix the splines. I smooth the roads though underneath. This prevents the roads from squishing down the terrain and in the process losing the hill they are on. Where there are intersections, it takes a bit of practice to get the roadway smooth on the terrain without burying parts of it under the nearby slope but it can be done."
That's what i do as well, the issue is with the non spline crossroads that remain flat , they are really only designed to work on the slightest of inclines. in general, roads are one of the weakest asset types that we have and some of the most time consuming items there are to get to look acceptable , i've never come across a road set which has lower curbs every now and again for driveway sections, it can be done by lowering the pavements one by one but its extremely time consuming .

There are some road splines that have a curb on one side. When I need a road cut-in, I use those but only where it's visible because it's a real pain raising and lowering roads and a real pain constantly switching splines.

The non-spline assets are like plastic models we stick into plaster and matte-medium paste. Unless they are made with twistable edges, there's no way to get them to work on slopes.

 

[rwk]

How does the real world deal with intersections on slopes? If one road is slanted and the other road flat, when they join is the part where they join flat or sloped?

 

[blueodessey]

Brilliant, thank you - if you think it looks okay as is and not completely daft then I'm happy with that as it means less work for me!

Cheers,

PLP

Beauty is in the eye of the beholder.