The aim of the paper is presentation of some results obtained for the 3D gravity inversion in high performance parallel computer systems (HPC). Calculations were based on CLEAN algorithm parallelized with OpenMP and MPI using HPC capacities offered by FP7 project HP-SEE. Scalability of runtime in terms of model sizes and number of computing cores was analyzed. The quality of inversion was evaluated for single and two body models, and concrete tests with field data were carried out. Runtime analysis results matched theoretical expectations for the order of magnitude of calculation time O(N^8) where N represents the linear size of the geosection, and permitted an evaluation of effective runtime for small and medium models. Two modified least squares schemes were applied for the inversion of multi-modal anomalies and respective errors compared. The link between symmetries of geosections and respective anomaly approximation errors was observed. Obtained inverted geosections are characterized by distinguished contrast of mass density between anomalous bodies and surrounding rocks. But more work is required to deal with multi-body geosections (multi-modal anomalies) to reduce the lack of discrimination capacity of the algorithm for in-depth bodies.


Article metrics loading...

Loading full text...

Full text loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error