CAPÍTULO II SITUACIÓN ACTUAL
SITUACIÓN CON PROYECTO
3.2.2 Red ferroviaria y servicio
The LiDAR data provided for this study is already pre-classified into ground and non- ground classes, and was divided up into smaller tile grids for ease of handling and processing, as stated earlier. Two types of DEMs are required for use in this study, namely, the nDSM and the DSM. The nDSM is required as an input for the extraction of building roofs, while the DSM is the main input required for estimating the global solar radiation.
The processes involved in generating the two elevation models differ slightly and are therefore treated separately. The nDSM could be generated by subtracting the DTM from the DSM. It could also be created directly by interpolating a height normalized
91
LiDAR cloud point. These two methods are assessed during the stage of choosing the most appropriate methodology (see Chapter 3, sub-section 3.6.1) and are found to yield similar results. However, the process of generating the nDSM by interpolating the height normalized points is much faster and required less storage space, and is therefore chosen as the preferred method. The classification of the LiDAR data and the generation of the DEMs is discussed in the following sub-sections.
4.1.1 LiDAR Data Processing
The raw LiDAR points were processed to achieve the height normalized point cloud, using the lasheight tool from “LAStools” toolset for LiDAR processing. The lasheight tool requires the raw LiDAR data “las” file as input. The ground classified points were selected as a reference for constructing the ground TIN, with which the height of all other points was then computed, in respect to these selected ground points. The last step was to select the option of replacing the Z coordinate of points with the calculated height. This process outputs a “las” file containing height normalized LiDAR points, which was then utilised in creating the nDSM. The process was included as a sub- process in the Python algorithm to generate the nDSM.
4.1.2 DEM Generation Analysis
The elevation models are created using various processes contained within the “arcpy” library in Python. The algorithm is designed to read the LiDAR data as the sole input, while the “lasheight” tool is added as a sub-process. The LiDAR data goes into the lasheight model (“arcpy.gp.Model12”) and outputs a height normalized as described
in Section 4.1.1. The processed LiDAR then goes into
“arcpy.CreateLasDataset_management” process to create a las dataset. The las dataset is then served into the “arcpy.LasDatasetToRaster_conversion” process, this converts the las dataset into a raster using binning method. The value field, which specifies the LiDAR information to be interpolated, is set to “ELEVATION”. The next setting is the interpolation type, which is set to “Binning”; the “MAXIMUM” option for cell assignment type is chosen to assign an elevation value to each cell of the raster. Cells that do not have points within them (void cells) are assigned values using the NN interpolation method. The final output raster settings include setting the data type to floating point and setting resolution (pixel size) to 8cm to match those of the aerial
92
imagery of the study area; moreover, the NoData value is set to zero. The nDSM raster output now serve as input to calculate the slope and the aspect raster using “arcpy.gp.Slope_sa” and “arcpy.gp.Aspect_sa” process respectively. A copy of the Python algorithm is provided in Appendix 1.
4.1.3 Results
Figures 4.1(a), (b) and (c) show the sample results in respect of the LiDAR-derived products created, using the “lasheight” and “arcpy” in Python algorithm. All the procedures discussed in the analysis stage above are repeated for each of the LiDAR tiles, which contained buildings and within the scope of the study.
Figure 4.1(a): Normalized digital surface model (nDSM), (b): Slope raster, (c); Aspect raster.
4.1.4 Discussion of LiDAR Processing and DEM Generation
The elevation model generated, as shown in Figure 4.1(a), forms the fundamental input data for the subsequent analysis. This goes a long way in affirming the importance and usefulness of LiDAR data in spatial analysis and surface information generation. The LiDAR data offers a massive collection of 3D points, which can be utilised by various applications in deriving spatial information about a particular area or region.
For instance, based on the sole input of LiDAR data, which has been used to create an elevation model, various surface analysis can be carried out to yield information, such as aspect (Figure 4.1(c)), which determines the orientation or direction a particular surface faces in terms of geographic direction. Other surface information that can be generated from LiDAR-derived DEMs include slope (Figure 4.1(b)),
(a (b (c
Meter
Degree
93
contour, volume calculation, viewshed, and solar radiation, amongst others. It has also been found to be effective in object detection and extraction, as will be discussed in detail in the subsequent section (see Section 4.2). There are many other areas of application where LiDAR data is very useful, but these are beyond the scope of this study. However, it must be emphasized that investment in LiDAR data acquisition by government establishments or corporate bodies involved in spatial planning is indeed a great asset.
Having discussed LiDAR data processing and generation of DEMs, these outputs are now put to use in the analysis of building roof extraction as presented below.