Aspects of Modelling Geographic Information in Geodesy and Geodynamics

Abstract

We studied aspects of modelling geographical information from the geodetic and geodynamic viewpoints. The data for our studies were acquired by a variety of methods: laser scanning, levelling and satellite positioning. The major subject of this dissertation is quality and its measures. We study various modelling approaches on geodetic data aimed at use cases in the fields of geodesy, geodynamics and geographic information science. The dissertation discusses internal and external quality aspects of the modelling and the data used. The main objectives of the research are related to data modelling aspects and fitness for use within the fields of study as expressed quantitatively in various quality measures. The novelty of the dissertation is in the application of appropriate quality measures, like precision or accuracy, in these fields of research from the viewpoint of fitness for use, which for the various models depend on input data precision as well as on the envisaged applications of the model product. This dissertation has three main topics. Firstly, the modelling of gravity based heights was studied. Three different modelling methods were used: kriging, fuzzy modelling and bilinear affine transformation. We studied the use of geostatistical and geodetic methods for the construction of digital elevation models and height transformation surfaces. Within these methods also quality measures were formulated showing their applicability in height modelling. We found that the quality of the results is dependent on the data point distribution and the availability of precise geoid heights. Secondly, precision measures for gravimetric geoid determination were derived for two test areas. For this study three error sources were investigated: the error of omission, the aliasing error and the out-of-area error. We showed that error sources were dependent on the spatial extent and accuracy of the gravimetric measurements. Thirdly, the modelling of post-glacial land uplift was investigated, using two methods: land uplift prediction by least-squares collocation, and Glacial Isostatic Adjustment (GIA) modelling using two different ice models and fitting the Earth model parameter values. In the land uplift recovery study, possibilities were investigated for projecting the land uplift forward in time. From this study a statistical model for predicting land uplift rate from point velocity rates using a relatively simple formulation was derived by the least squares collocation technique, with error propagation into the predicted land uplift. The GIA modelling study gave us experience in building land uplift models. We found that the two methods, land uplift rate prediction and physical GIA modelling, though being very different, are giving similar accuracy measures for the derived land uplift values.