Creating the Terrain Model

After manipulating or editing the Point Cloud (as needed), the next step is creation of the terrain model. The entire Point Cloud can be used, or the Point Cloud Presentation can be used to limit the import to certain classifications, or a fence may be placed in the area of interest. The fence should be placed prior to starting the tool.

Two tools are supported in the upper left corner of the dialog:

Settings	Description
Rename Terrain Model	renames the specified Terrain Model.
Remove Terrain Model	removes the specified Terrain Model from the list.

The following options are supported in the right pane of the dialog:

Terrain Models -options if there is already a terrain model in the active *.dgn file and how to combine with the data to be imported.
Filter - Filtering may be a useful option for large datasets.
Feature Definition - feature definition to be used for imported terrain. If none is selected, the terrain model is drawn using the active symbology.
Geographical Coordinate Systems (GCS) - If the software can determine the geographical coordinate system of the file to be imported, the fields are populated. If set to Unknown, the software uses the MicroStation design units. This section is used if the resultant terrain model needs to be in a different GCS.

Filter by Tile TIN

This tool supports two filtering algorithms. One is Tile based and the other Tin or triangulation based. From empirical studies, the tiling algorithm is faster and typically produces a 30% to 50% reduction in file size. The TIN algorithm typically produces a 70% to 90% reduction.

The tiling algorithm is a recursive divide and conquer algorithm that divides the LIDAR data set into tiles. A best fit plane is calculated for each tile and LIDAR points are removed if they fall within the user set Z tolerance to the plane.

The TIN algorithm filters LIDAR points if they fall within the user set Z tolerance of the triangle planes. The TIN algorithm first tiles the LIDAR points into tiles with a maximum of 2 million points and then repetitively triangulates each tile filtering out points. This algorithm running on a 1.67 GHz machine with 1 GB of memory will typically take about 12 minutes to filter 30 million LIDAR points.

Settings	Description
File Option	Tile or TIN filter
Z Tolerance	Z Tolerance is common to both algorithms and is basically the variation in the Z coordinate that the surface is allowed to move during the filtering process. Typically for the first invocation of the filtering function, the Z tolerance should be set from 0.5 to 1.0 for imperial data sets and from 0.25 to 0.5 for metric data sets. Depending on the outcome and desired result, the Z tolerance can be varied up or down.
Min. Tile Points (Tile)	A tile will not be subdivided if it has less than this number of points. Typically set this to five.
Max Tile Divisions (Tile)	Allowable level of recursion allowed and is the number of times the initial tiling set can be subdivided. Typically set this to five.
Start Tile Length (Tile)	The LIDAR data set is initially divided into tiles of this size, prior to recursion to the minimum tile points. The setting of this parameter requires some knowledge of the distance between the LIDAR points, which requires an inspection of the LIDAR points in MicroStation to determine. Typically set this to 10 times the distance between the LIDAR points.
Coarse Filters (TIN)	Filters more points with some blurring of ridges and valleys.
Fine Filter (TIN)	Filters fewer points with less blurring of ridges and valleys.
Reinsert Points (TIN)	Reinsert points adds points back if the iterative filtering on the terrain removes points which are within the tolerance. For example, this process could remove points at the top of a hill and the second iteration could remove points around the high point making the point out of the thinning tolerance. The iterative process can happen a maximum of five times.

Workflow

Open or reference the DGN file containing point cloud.
Set the view attributes of the cloud to display all points which are desired in the terrain model
Optionally make a fence around points to extract.
Start the Terr ain from Point Cloud command.
Filter, if desir ed.
Set the feature Definition and Triangulation Option.
Datapoint to accept and build the terrain model, named Point Cloud Surface_1 by default.
Rename the terrain model, if desired.