An investigation on the use of point cloud data in evaluation of local geoid models

The aim of this thesis study is to investigate the feasibility of Digital Terrain Model and local geoid model from point cloud data obtained by airborne LIDAR (Light Detection and Ranging) and unmanned aerial vehicle (UAV). Various accuracy and resolution of Digital Terrain Model data obtained from different techniques are used for several purposes in a variety of applications today. In addition to this, geoid models are practically needed for transforming the ellipsoidal heights obtained by the GNSS technique into the regional height datum. Our geodetic infrastructure in our country does not currently providing a geoid model with a few centimeters of accuracy to satisfy the needs of applications. Large Scale Map and Map Information Production Regulation (BÖHHBÜY) forms the basis for local geoid model production in relatively small regions. In this thesis research, the contribution of the technology of obtaining the point cloud is also investigated and examined in the production of the local geoid model with the accuracy of centimeters. İzmir Bergama region is chosen as the test area for the thesis study. The existence of rural and urban areas, presence of human production structures within the scope of this field is important in terms of contributing to the aim of this study. The fact that the study area has a heterogeneous character in terms of construction makes it possible to obtain different conclusions and gain experience within the scope of Digital Terrain Model production. In the thesis study, different software algorithms are used to process point cloud data obtained by LIDAR technology and UAV photogrammetry. These data and the land height data obtained by the algorithms are used for local geoid modelling and testing studies. The thesis studies are started with the results of the literature search on the data used, the methods for the application and the basic principles of the examined phenomenon. The literature section of the thesis is begun with the definition of coordinate systems. The reasons for the use of coordinate systems and their importance are indicated. Criteria evaluated in the classification of coordinate systems and definitions of the classified coordinate systems are made. The types of transformations made between coordinate systems are defined. The mathematical structure of Bursa-Wolf and Moledensky models used for similarity transformations in the study are explained. In the continuation of the literature study, the definition of the geoid, ellipsoid of revolution concepts and their importance in geodetic studies are explained. Some ellipsoids used in the world and the parameters of these ellipsoids are given. After the explanations of the basic definitions on the geoid and ellipsoid of revolution, related with these definitions, the geodetic datum is defined as well. The parameters of various datums and the regions where these datums are used are given. The definitions of the datums of the dataset used in the thesis study have are provided. The definition of geoid determination, the historical development of the methods used to determine the geoid and the geoid determination methods including the one that is used in the thesis study are also explained. The literature studies are continuing with the differences between Digital Terrain Model (DTM), Digital Surface Model (DSM) and general definition of Digital Elevation Model (DEM) concepts used in thesis application. In the literature study, methods used to produce with Digital Elevation Model are mentioned and explained in detail. It is stated within the thesis in which method the point cloud data is used for the Digital Terrain Model production. The theoretical and technical specifications of the airborne LIDAR system and UAV photogrammetry are explained to give a better understanding for the dataset used in the thesis study. The components of both airborne LIDAR and UAV photogrammetry systems are given and the interoperability of the components are described. Coordinate transformations that are applied while producing an airborne LIDAR data are defined and explained. The theory of structure from motion which is used to produce point clouds from images are also studied for the point cloud produced by UAV photogrammetry. The airborne LIDAR data used in the study is produced by the General Command of Mapping within the scope of the project of testing the LIDAR data. Airborne LIDAR point cloud data; is received in ITRF datum and ".las" format, covering the Bergama region and divided into grids. Since the airborne LIDAR point cloud is produced in ITRF datum, it has ellipsoidal altitude information. The point cloud data obtained by the UAV photogrammetry is produced within the scope of General Directorate of Geographical Information Systems and has orthometric height information. The height information in both ellipsoidal and orthometric systems are used for the determination of the geoid by the GNSS / Levelling method and the geoid heights are determined by a surface modelling approach. In order to generate Digital Terrain Models from point cloud data, several filters must be applied. With the help of the filters non-ground points in the point cloud is removed and plain surface is obtained. There are several approaches to filtering point cloud data, and filtering methods are classified according to these approaches. The filtering methods used to generate Digital Terrain Models from point cloud data are explained in the thesis work. In the thesis application section, the information about the approach of the filters applied to the data is given in details and explained with the Digital Terrain Model images. The computer hardware that are used in the study and the experienced constraints in the processing of the point cloud data are given as well. In order to overcome the faced limitations, various optimization approaches are applied in processes of the point clouds with large data sets, and strategies are evaluated to avoid possible problems in data quality. In order to avoid the limitations caused by the computer used in the thesis study, the UAV point cloud data is divided and processed. The application of the thesis begins with the removal of the blundered points from the point cloud data obtained by airborne LIDAR and UAV photogrammetry. In order to remove coarse faulty points, first the "statistical outlier removal" (SOR) filter is used and then the noise filter is applied. Digital Terrain Models are produced with different methods and with various software for the purpose of being used in geoid modelling from the data generated after filtering. The usability of the Digital Terrain Models produced are validated in the geoid modelling study. The sources of errors in the results are investigated and discussed in conclusions and recommendations for minimizing the errors are provided. In the numerical application of the thesis, the Digital Terrain Model productions which will cover the whole of the study area are produced by using different methods on divided and filtered point clouds. Firstly, it is tried to produce a surface model covering the working area by combining the Digital Terrain Models which are produced from the divided point cloud data. As a result of this experiment, problems related to data loss and topography display are observed at Digital Terrain Model common points. Interpolation at common points is tried to remove these observed problems and to obtain a much precise Digital Terrain Model. Data loss is eliminated by interpolation, but the problems of topography display are not resolved. As a second method to implement the Digital Terrain Model production that covers the study area, point density of UAV and airborne LIDAR point cloud data are reduced. The clouds with the reduced point density are combined in itself and two point clouds are generated for UAV and airborne LIDAR. From those point clouds covering the study area, Digital Terrain Models are produced. The models produced by this method have created problems in the representation of topography in areas having steep slopes in the study area. Finally, within the scope of the study, the density of the UAV and airborne LIDAR point clouds are left unchanged. The filtering processes performed at the beginning of the thesis is showed that the computer-based constraints were overcome using certain optimization strategies in computations. Digital Terrain Model production is performed successfully from the point clouds covering the study area for the UAV and airborne LIDAR point clouds. Digital Terrain Models are produced with two software using two different approaches. The visuals and statistics of the models produced by both software are given in the application section of the thesis. From the Digital Terrain Models that produced from point clouds, new surface models are produced. New surface models are produced by subtracting the orthometric heights from the ellipsoidal heights obtained through the generated Digital Terrain Models. This subtraction process is made by using several software. The calculation errors and their reasons are discussed. Hence these errors in generation of the statistics in generation of differences among the grids are studied and tried to be avoided in the execution of the differentiations. In the results; the produced geoid surfaces are compared with national regional and global geoid models. Some drawn conclusions are found applicable that the point clouds by the different platforms for geoid generation with GNSS/Levelling method.