Trees¶

Generating the elevation grids (rasters)
Generating the trees with FINT
Converting FINT to points
Last requirements
Custom trees / other

This section describe the steps needed to generate the optional tree stems for them to be considered in the simulations. Preparation of the digital gridded terrain and surface models (DTM and DSM rasters) necessary for the automatic identification of stems with the FINT tool of the ecorisQ group is first discussed. The conversion of raster grids in GeoTIFF format to ASCII format is also discussed. These steps are followed by using FINT to identify the stems and then using the stnFINT2points tool to convert them to a point cloud. Then, the prerequisites for the simulations are quickly detailed. Finally, some additional information concerning the validation of the results generated with FINT is presented together with the instructions for manually modifying the stems for example to assign certain species or diameters of trunks to given sectors.

Generating the elevation grids (rasters)¶

The FINT tool of the ecorisQ group makes it possible to automatically estimate the position and the diameter of the tree stems based on the extrema (maxima) of the gridded elevation model (raster) of the difference between the surface model (DSM) and the terrain model (DTM) ( DEM of Difference [DoD] ). It is therefore necessary to generate the elevation grids (rasters) from the elevation data used for the project. This is detailed in the following subsections.

If the project is initially carried out from DTM and DSM, it is still necessary to ensure, for an optimal functioning of FINT, that these were generated by considering the maximum elevation of the points in each cell, rather than the average elevation commonly used. This limits the number of false positives of FINT at the cliffs and improves the estimation of the height and diameter of the trunks of Trees, reducing manual post-processing. Also, we must ensure that the DSM used only represents the area linked to the trees, without the infrastructure. Otherwise, additional post-processing will be necessary in order to manually remove the trunks identified by mistake around the infrastructures (e.g. power lines, viaducts, buildings, etc.).

LiDAR to DTM ¶

This step consists in generating a gridded digital terrain model (DTM) optimized for the identification of trunks with FINT from classified point cloud elevation data. The extent of the points used for vegetation and terrain must be identical before generating the rasters, for the proper functioning of FINT which requires DTM and DSM grids aligned and of the same extent. It may be necessary to pre-cut the raw data to the desired extent (e.g. with the Interactive Segmentation Tool and a vertical view without perspective or Cross Section) before continuing with the other instructions.

LiDAR to DSM ¶

This short step quickly repeats the instructions from the previous step to generate the gridded digital surface model (DSM) optimized for the stem identification with FINT.

GeoTIFF to ASCII ¶

This step consists of converting the DTM and DSM elevation grids to ASCII raster (.asc) format. If the footprint of the rasters is not identical, a clipping may be necessary for the proper functioning of FINT which requires DTM and DSM grids aligned and of the same size. It may be necessary to cut the rasters to the desired extent (e.g. with the QGIS Clip raster by extent or Clip raster by mask layer tools, or the ArcGIS clip [Data Management] tool) before continuing with the instructions.

Generating the trees with FINT ¶

This short step shows how to identify tree stems with FINT from DTM and DSM elevation grids. The next step is to use the estimated trunk diameters at breast height (DBH) to determine the height of the trees. It is therefore recommended to use the default relation: DBH = H^1.25. Either way, the DBH and FE Ratio can be modified or replaced by custom values later in CloudCompare with the Scalar fields/Arithmetic tool (see section "Customs trees / other" ).

It may also be necessary to filter the results generated with FINT in order to remove any identification errors due to artefacts, infrastructures present in the DSM or significant elevation differences at the cliffs. This manual post-processing is easier to perform once the trunks have been converted to a point cloud for viewing in CloudCompare. This is why the following sections focus on converting to a point cloud. See the last section "Customs trees / other" for manual post processing.

Converting FINT to points ¶

This step shows how to convert the trunks identified by FINT in the previous step into a point cloud with the stnFint2points tool. The latter uses the relation DBH = H^1.25 to fix the height to which the position of the trunks must be extrapolated by equally spaced points from the elevation of the DTM.

These tree stem center points are then used by stnParabel to generate conical trunks with diameters at breast height (1.3m) equivalent to the DBH value. Large trunks are therefore more likely to intercept rocks, and the same is true for the probability that a large rock will intercept a trunk.

It is possible at this stage to define up to three different FE ratios for the tree stems identified according to the altitude at which they are found, to simulate different distributions of species. See the section "Customs trees / other" to manually set the position of certain species.

Last requirements ¶

Finally, you have to make sure that you only have the position of the trunks, and the 3 scalar fields of DBH, FE Ratio and height from the ground: [X, Y, Z, DBH, FE Ratio, ΔZ].

Any [R, G, B] colors and surface orientation [Nx, Ny, Nz] must be removed before exporting the points in ASCII format.

The ASCII file must be named "tree_pts.txt" in order to be correctly taken into account when importing the terrain for the simulations.

Custom trees / other ¶

As mentioned in the section "Generating the trees with FINT", it may be good to validate the tree stems identified with FINT in order to remove any "false positives" by segmenting them out with the Interactive Segmentation Tool. It is also possible to remove tree stems incorrectly identified at the cliffs if they are excessively high. This can be done with the Scalar fields/Filter by Value tool using the excessive DBH as the maximum limit, visually validating the range to use in the properties ( SF display params ) beforehand.

It is also possible to apply a manual distribution of certain species by cutting out the trunks of a given sector and modifying their FE Ratio values. A similar approach can be used to modify DBHs. For example, the first species trunks can be extracted with the Scalar fields / Filter by Value tool and the "Split" option by isolating them from their FE Ratio. Then, their DBH values can be modified with the Scalar fields/Arithmetic tool before these trunks are reintegrated with those of other species.

The segmentation of the tree stems can also be used for example to combine two different outputs of stnFINT2points in order to obtain a random distribution of different species for a given sector. For example, an output A) with 20% of species 1 and 80% of species 2 can Locally be replaced by an output B) of 50-50% species distribution.