Geography Reference
In-Depth Information
beam which is reflected from the certain height level of an object (i.e. tree) to
identify the bottoms and/or tops of the vegetation.
Vegetation analyses and operations based on the Digital Canopy Height Models
(DHCM) are the major part of this article. The main aim of the authors is to use
LIDAR data to reconstruct large scale areas of vegetation in the 3D by raster driven
distribution of surface objects. The output is presented in the photo-realistic model of
the surface, which is based on the real-world values derived from the LIDAR data.
To reach the given aim, first the quadrilateral and polygonal model of the terrain is
created. Next, filter for the tree extraction from the raw LIDAR point cloud is created
using Python. Filtered data representing real trees are then transformed into the 8 or
16 bit black-white raster. This structure is used to place the individual objects to the
correct locations in the terrain. The same approach is utilized for the cloud distribution.
A very small abstraction of factors in the recreation of the faint vegetation details
in the large sized areas provides the possibility to perform photorealistic spatial, or
view-shed based, analyses. Proposed approach of this text works with the idea to
derive the most important attributes of small objects into the black-white raster
structure. Pixel values are suitable to lead the distribution of 3D polygonal models
in the area in the form of proxy objects. Every tree has the appropriate height and
also a proximate position.
Authors utilize slope based polygonal terrain model system, which was developed
by them to match the resolution of the quadrilateral grid with the terrain adaptively.
The most hardware demanding task is to recreate polygonal models for the
vegetation and the environment globally. However, this task is the key issue for
preserving the faint details of surface objects. Linking particular digital models
together (e.g. terrain, buildings, vegetation, clouds, etc.) and visualizing them in the
form of a photorealistic and scientific output (real properties of objects are pre-
served) is a multidisciplinary approach which involves different scientific fields.
Used Data, Software and Hardware
There were two data inputs available. The
first one
is the detailed surface scan
downloaded from the OpenTopography.org, which contains the point clouds of the
Pennsylvania forests. Scanned values are in the both cases stored as XYZ coordi-
nates. This data set was used during proposal of the whole procedure of creation of
vegetation model (vegetation reconstruction).
The
second input
is the LIDAR scan for the area size roughly 10
20 km
situated in the Czech Republic, Europe. Data set is scanned by aircraft Turbolet
L-410 FG. Two types of LIDAR models are created as an output. The first type is
the Digital Surface Model 1st Generation (DSM 1G), which includes all objects on
the surface (terrain + vegetation, buildings
). The second type is the Digital
Model of the Relief 5th Generation (DMR 5G) (Belka
2012
). This data set is
used for the final 3D photorealistic visualization of a larger area of interest.
The infra-red NOAA imagery from its archive is used to illustrate utilization of
the proposed procedure in the case of clouds.
...
