Removing extra vertices in TIN, or 3D visualizing in any other way

[ZiadGewely]

1. I wanted to visualize the topographic surface of Alexandria, so I got some points that was already taken using GPS devices and Total Station. I used "Topo to raster" and created a TIN using that raster, it was full of vertices which wasn't that satisfying for me. I download DEM from internet tried then raster to TIN always didn't satisfy me. Lastly I created contour lines using the raster I made using the points features, with a certain contour intervals, and I got this TIN

   This is the closest thing I got to what I wanted

   

[RPGIS]

Hi @ZiadGewely,

Do you know the relative accuracy of the DEM. Meaning do you know the overall resolution because this can significantly affect the overall outcome of a TIN. I have created several TINs in the past using lidar data sent in by our vendor and it makes a significant difference in the overall outcome of the TIN dataset. I don't know the level of detail you are looking for but I may be able to provide some modest help with this.

[ZiadGewely]

Hi @RPGIS ,

Thank you for the insight. I'm somewhat familiar with relative accuracy of a DEM but I'm still learning about it. The DEM I used was from USGS Earth Explorer (30m SRTM), so the resolution is a limiting factor here. But I suspect that the 'noise' is because of the point density. I'm aiming to create a smooth cartographic visualized surface using a DEM or even using these points I have on a local scene, to compare it to another surface I have of the same area, so the final input is going to be the two surfaces together, so I wanted this visualization to be the closest thing to reality without too many vertices.

[RPGIS]

Noise can be a myriad of factors aside from density alone. Sometimes having greater point density often results in more accurate DEMs due to either nearest neighbor clustering or standard deviation. The thing you may also want to investigate is the variation in z values since a cluster with higher z averages will skew the resulting TIN. I would recommend checking the z groupings or z value doesn't have a high deviation amongst clusters or there are too many individual z values with greater variation which could be contaminating the overall dataset.