Creating terrain reasonably accurately
-
Obviously terrain can be imported from Google Earth to some extent. I have this issue where the land I am looking at was obviously scanned / photo graphed in the height of summer and so the tree canopies covered most of the area, thus rendering the terrain I can extract from there useless. I was wondering what peoples experiences were of surveying / creating there own terrain and good practice for doing this? Can photo-match be used to do this in some way with sandbox?
Any thoughts / ideas welcomed.
Cheers.
-
Good question!
As for Google Earth, I wouldn't lean much on it if you need an accurate terrain. I know an example of the river going upstream - something screwed the algorithm. Let's stay a bit more on rivers. One would expect a usual, steep riverbanks, right? Not with Google Earth, just a slight approximation. We mustn't forget Google Earth models the entire planet and surely the issue of complexity had been taken into account. Imagine what a wonderful thing that would be: a high-poly accurate model of the Earth.
One day, maybe, but not now.Modeling terrain using photo-match? I've tried it.
The one who successfully makes an accurate model of any terrain by using photo-match during a reasonable period of time should be awarded the prize as a genius. Unfortunately, perspective has its own rules and our eyes are easily tricked.
When you watch at the building you can swear that the angle you're looking at has 90 degrees. But measurement tools show 87 or 92 or something like that. This is for instance very often case with old, simple stone medieval churches that had to be adjusted to its terrain or adjoining buildings.Yet buildings are normally still 'obedient' in most cases with photo-match.
But what to do when you have a photo of some meadow, with grass, few trees and a grazing cow? Nothing to refer to, nothing to start with...
-
Ben what do you need it for? Google Earth Terrain is decent for many things like conceptual site work, or even beyond. Is it just the fact that you can't see what is below the trees that is making it hard to use the Google Earth terrain?
Typically if you're working for someone, they will have a survey made of their site and give you that CAD survey data. You can then bring those existing conditions into your SU model. Or often you can get GIS data of an area and have pretty valid existing conditions that way. There are quite a few ways to work around poor aerial images.
So more context might be needed here. What is it for, modeling existing conditions? or proposed design? Professional work" School project? Just for fun? How important is accuracy, honestly. Don't kid yourself if its not important, don't sweat it. If it is important, you'll end up paying for it by paying a survey crew or something.
-
Thanks for the replies, I've had problems logging in, so it's taken me a while to respond. I'm looking to model some terrain with differing levels and then look at options for positioning a house and I want to see the effect it will have on two houses that are at a lower level a way down the plot. I want to design as sympathetically as possible before I submit an outline planning proposal and I really want to minimise the impact on the existing properties with respect to blocking views and light, etc. I guess it doesn't have to be super accurate, but it'd be nice to be within a 300mm or so of the average slope or level. Then I can see maybe how much I might need to consider excavating to lower the plot if there is a considerable benefit to the existing properties vs. the extra cost.
Thanks,
Ben
-
I've used Google terrain for some initial visualization but it just gives the basic lay of the land. It would really be best to make your judgements by field observation, perhaps using the model to demonstrate what you've found, not for factual data, as you might use a survey.
If this is the US. you might see if (reasonably current) USGS data is available.
-
What you are looking for are DEMs. Digital elevation models. Most cities have ones created be synthetic aperture radar, accuracy about 1 meter, or lidar, accuracy about 1 foot. Rural areas are a lot more variable. The shuttle flew a mission that did SAR from orbit, but the grid spacing is about 30 meters, and the accuracy is awful.
If USGS topo data is good enough, I vaguely recall a page on converting topo lines to terrain. Warning: topo data is not very good at the house lot size level. There is typically a semi-systematic error in the horizontal placement of contour lines equal to several times the contour interval. The shape is right, but slope inflections tend to be moved away from the centre of the photographic track.
If its critical to get it right, rent a DGPS, and take a waypoint every 10 meters. If the terrain isn't too bad, you can do a hectare an hour.
Advertisement