Treatment of parameter and model uncertainties through a sensitivity analysis of the seismic hazard for Johannesburg
Date
2021
Authors
Manzunzu, Brassnavy
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Seismic hazard analysis is a dynamic process, which requires regular updating with emerging new information. In this regard, quest for new information is always there. However, in order to reduce costs, minimise labour while producing reliable results, efforts must be focused on areas that are influential to the final calculated hazard. In this study, we analyzed the uncertainties associated with various models and input parameters in the computation of the seismic hazard for Johannesburg. Several uncertainties were assessed and evaluated at different steps in the seismic hazard process. The evaluated uncertainties include uncertainty in the compilation of the seismic catalogue, the selection of ground motion prediction equations (GMPEs), the development of seismic source models as well as the computation of recurrence parameters. A comprehensive homogeneous catalogue was compiled consisting of both historical and instrumental earthquakes. Different declustering techniques were tested on our catalogue on their suitability to remove dependent events in order to come out with a Poissonian distributed catalogue. Both location and magnitude uncertainties were characterized in the catalogue processing. Location uncertainty reduction helps in identification of more refined seismic source zones while magnitude uncertainties influence the computation of recurrence parameters. The catalogue was used in the development of a seismotectonic model. The area under study is characterized by diffuse low magnitude earthquakes, which made it very difficult to identify linear sources of earthquakes. Consequently, areal source zones were developed based on the available and recently determined information on seismicity, focal mechanisms, and active faults. The zones identified in this study were compared to previously identified tectonic source zones in the region. It was noted that although the more information on focal mechanisms and possible active faults were gathered, only a few refinements could be made to the source model. Although there were a few modifications to the previous tectonic studies, sensitivity analysis results showed large variability emphasizing the importance of accurate seismic source zoning in the region. Careful selection of Ground Motion Prediction Equations (GMPEs) was done considering local available data. All the uncertainties were utilized in the sensitivity analysis. Sensitivity analyses were performed for 2% probability of exceedance in 50 years for the city of Johannesburg. Sensitivity analysis results indicates that the five most important inputs that have a profound impact in controlling the final hazard levels are (1) GMPEs, (2) seismic source model, (3) temporal variation of the seismic catalogue, (4) catalogue homogenisation process, (5) method of computing recurrence parameters. Coefficient of variation values in excess of 0.2 were obtain for these parameters. It shows that the largest contributor to uncertainty in the assessment of Peak Ground Acceleration (PGA) and acceleration at spectral periods 0.15 and 2.0 seconds, which are important for the response of buildings, is the selection of GMPEs. Next are the uncertainties in the processing of the seismic catalogue and the recurrence parameters. Seismic hazard is not very sensitive to the variability in bedrock velocity. The research findings from this study may serve as a useful guide to facilitate further research studies on influential parameters and models for future seismic hazard assessments. Even so, it should be noted that the degree of sensitivity of the parameters depends on the study site. A meticulous study of these parameters will lead to the reduction of uncertainties in future assessments
Description
A dissertation submitted to the Faculty of Science, University of Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Doctor of Philosophy, 2021