The Application of Crustal Models in Regional Modelling of the Earth's Gravity Field

 One of the ultimate goals in geodesy, a 1 cm geoid model, is still unreachable for most of the areas worldwide. Several theoretical, methodological, numerical and data problems will have to be resolved in order to achieve it. The main motivation of this research is in making methodological and empirical contribution towards resolving some of the open problems in the regional gravity field and geoid modeling. Topographic and density effects which affect short and very-short wavelengths of the gravity field have been traditionally modelled using the constant parameters of the Earth’s crust. As such parameters are only an approximation, this has been a limitation in more accurate filtering and reduction of the gravity data. Therefore, a methodology was developed which allows inclusion of surface and three dimensional crustal models in all steps of geoid determination. Prior to this, surface crustal density models were developed based on the inversion methods according to Pratt-Hayford, Airy-Heiskanen, and Parasnis-Nettleton. Additionally, three-dimensional crustal models EPcrust and CRUST1.0 were included in the computations. As a result of including crustal density models, the accuracy of developed gravimetric geoid models was improved from 1 to 3 cm. The second major focus of research was related to the problem of the diversity of possible geoid computation methods and dozens of ways to perform reduction of the gravity field. The comparison of two widely used geoid modelling approaches was performed: Royal Institute of Technology (KTH) and Remove-Compute-Restore (RCR). Furthermore, compute step in RCR approach may be performed using several spectral and spatial methods. Therefore, different geoid computation methods were compared, including analytic Stokes integration using different deterministic modifications of the Stokes’ kernels, planar and spherical Fast Fourier Technique (FFT), flat-Earth and 3D least squares collocation (LSC). KTH approach, being a relatively straightforward geoid modelling approach compared to the RCR, was used for the analysis of the influence of all input models and parameters on the accuracy computed geoid models. From the large number of computed geoid solutions, two final gravimetric and hybrid geoid models for Croatia were selected HRG2018-RCR and HRG2018-KTH having standard deviation of ±3.0 cm and ±3.5 cm. The accuracy of geoid models was validated on GNSS/levelling points with seven parametric models using a unique cross-validation fitting methodology. Few other aspects of regional gravity field modeling were researched: i) investigation of the influence of input models and parameters in obtaining residual gravity field used in the RCR approach, ii) validation of the accuracy of global geopotential models, and iii) validation of gridding methods for several types of gravity anomalies.