Accuracy of terrain vs Accuracy of Tin

2846
2
07-08-2010 07:44 AM
YaroslavZaitsev
New Contributor III
Good afternoon all. I am trying to figure out what is the difference between terrains and tins. I have tons of lidar data that I initially processed 2 years ago. Then SWFWMD issued a letter saying that some of their lidar data was not correct and then we discovered shifts in data by as much as 50-100ft.
To this day biggest issue with tins came from the fact that lidar data comes in huge blocks of data and tin has 90% crash rate making generation of lidar next to impossible. Terrain on the other hand does exactly what it supposed to do and is capable of handing lidar data with no issues.

My workflow with tins was las to multipoint conversion, then generation of tins (over and over until each 40 acre chunk would actually not crash on me) then tin to raster then focal statistics on raster and then raster to contour.

My workflow with terrains was las to multipoint. shapefile to geodatabase. create terrain/build terrain. terrain to raster. focal statistics on raster and then raster to contour.

When I compare tin vs terrain derrived contours I see very little difference and yet it is important to find out if that small change is substantial enough to continue having a tin crash fest or to start relying on terrains to generate contours.

We do volume calculations and I know I know tins are supposed to be used for lidar.. but when we run volumes we take derrived contours and convert them to tin and then to raster for volume analysis. Due to server space concerns it just worked out that way for us to store contours and clip them out to project boundaries and reverse engineer the raster from them.

Can I with some level of confidence figure out whether contours derrived from tin are significantly different from contours derrived from terrain. Just visually they look very very similar minus little pockets here and there.

Has anyone had any luck relying on terrains?
Tags (1)
0 Kudos
2 Replies
RustyRex
Occasional Contributor
Note that terrains also use TINs.  There could be several variables that are causing the differences that you are seeing including the terrain pyramid level or interpolation method you are using when going terrain to raster.

There are no major flaws in a normal TIN or a Terrain so you should probably evaluate the two final datasets instead of the tools or methods.  You might want to get benchmarks and find out which dataset has the higher absolute accuracy and then make up your mind.  You could compare the datasets at difference steps in your process to find out when the differences are introduced.  Finally, you might also do some sample volume calculations to see if the differences are acceptable for your application.

And yes we have had luck relying on terrain, but in the workflow you described there are several processing steps with several variables so you never know when error might get introduced.
0 Kudos
MichaelRoss
New Contributor II
I wonder if the same interpolation method is used for Terrain -> Raster and TIN-> Raster.
0 Kudos