| Abstract | Jedan od ciljeva ovoga rada bio je simulirati potresnu trešnju na širem dubrovačkom području. U tu svrhu, najprije je sastavljen približan 3D radni lokalni model građe kore, DF_3D. Model DF_3D sastoji se od tri glavne cjeline: gornje kore, donje kore i plašta; uključuje topografiju i batimetriju te u svakom elementu sadrži informacije o brzinama P- i S-valova te gustoćama sredstva. Novodefinirani model korišten je i uspoređen s lokalnim 1D modelom te regionalnim 3D modelom u deterministički izračunatim simulacijama niskofrekventnih zapisa gibanja tla. Vizualnom provjerom te analizom mjera kvalitete prilagodbe za vršne brzine gibanja, međukorelacije simuliranih i zabilježenih zapisa gibanja tla te vremena trajanja snažne trešnje, ustanovljeno je da je lokalni 3D model građe kore trenutno najprikladniji model za računanje niskofrekventnih simulacija na širem dubrovačkom području. Potom su pomoću hibridne metode, izračunati širokopojasni zapisi gibanja tla za deset umjerenih do jakih potresa. Zapisi su dobiveni sjedinjavanjem odvojeno izračunatih niskofrekventnih i visokofrekventnih doprinosa pomoću, determinističkog i stohastičkog pristupa. Prilikom računanja visokofrekventnih doprinosa definirana je i korištena baza Greenovih funkcija izračunatih za 1D model (DF_1D) koji je izveden iz modela DF_3D. Vizualnom provjerom, analizom mjera kvalitete prilagodbe za vršne brzine gibanja i vremena trajanja snažne trešnje, te analizom vremensko-frekventnih funkcija odstupanja ovojnica i faza simuliranih i zabilježenih zapisa, ostaje nejasno u kolikoj je mjeri hibridni pristup simuliranja širokopojasnih zapisa gibanja tla uspješan na istraživanom području. Za potpuno utvrđivanje njegove primjenjivosti nužni su detaljnija validacija i parametrizacija. Neovisno o tom rezultatu, u svrhu poboljšanja razumijevanja gibanja tla prilikom snažnih potresa na ovom području, provedene su niskofrekventne simulacije Velikog dubrovačkog potresa iz 1667. godine. Za ovaj potres simulirano je šest scenarija potresne trešnje korištenjem šest različitih kinematičkih modela konačnih izvora. Za scenarije Velike dubrovačke trešnje definirana je baza simuliranih niskofrekventnih seizmograma i video zapisa simulacija te su analizirane njihove glavne značajke. Validacija scenarija potresne trešnje provedena je na temelju usporedbe simuliranih mjera intenziteta gibanja tla horizontalne vršne brzine gibanja tla, spektralnih akceleracija za periode T = 1.0 s i Ariasovog intenziteta s opaženim makroseizmičkim intenzitetima. Utvrđeno je da kombinacija nejednolike razdiobe pomaka duž rasjedne plohe i bilateralne vremenske simetrije rasjedanja daje najveći korelacijski koeficijent, odnosno, potencijalno najbolje opisuje seizmički izvor Velikog dubrovačkog potresa. Daljnje modeliranje seizmičkog izvora, korištenog modela građe kore, metoda simulacije i validacije, nužni su da bi scenariji potresne trešnje odražavali očekivana svojstva gibanja tla i time doprinijeli boljoj procjeni determinističkog seizmičkog hazarda na širem dubrovačkom području. |
| Abstract (english) | One of the goals of this study was to simulate earthquake shaking in the wider Dubrovnik area. To achieve this, an approximate 3D local crustal structure model, DF_3D, was first created. The DF_3D model comprises three main layers: the upper crust, lower crust, and mantle, and includes topography and bathymetry. Each element within the model contains information on P- and S-wave velocities and material densities. This newly defined model was then used and compared with a local 1D model and a regional 3D model in deterministic simulations of low-frequency ground motion recordings. Through visual inspection and analysis of the quality metrics—such as peak ground velocities, correlation between simulated and recorded ground motions, and the duration of strong shaking—the local 3D crustal model was determined to be the most suitable model currently available for calculating low-frequency simulations in the wider Dubrovnik area. Using a hybrid method, broadband ground motion recordings were subsequently calculated for ten moderate to strong earthquakes. These recordings were generated by merging separately computed low- and high-frequency contributions using deterministic and stochastic approaches. In calculating the high-frequency contributions, a base of Green’s functions was defined and calculated for a 1D model (DF_1D), derived from the DF_3D model. Through visual inspection, quality metric analysis of peak ground velocities and shaking duration, and time-frequency analysis of envelope and phase deviations in simulated and recorded recordings, the effectiveness of the hybrid approach for broadband ground motion simulation in the studied area remains unclear. Further detailed validation and parameterization are required to fully assess its applicability. Regardless of this outcome, low-frequency simulations of the Great Dubrovnik Earthquake of 1667 were conducted to enhance understanding of ground motion during strong earthquakes in this area. Six earthquake shaking scenarios were simulated for this earthquake, using six different kinematic finite-source models. For these scenarios, a database of simulated low-frequency seismograms and simulation videos was created, and their main characteristics were analyzed. Scenario validation was based on a comparison of simulated ground motion intensity metrics, including horizontal peak ground velocity, spectral acceleration for periods of T = 1.0 s, and Arias intensity, with observed macroseismic intensities. It was found that a combination of non-uniform slip distribution along the fault plane and bilateral temporal rupture symmetry yields the highest correlation coefficient, potentially providing the best description of the seismic source for the Great Dubrovnik Earthquake. Further modeling of the seismic source, crustal structure model, simulation methods, and validation is necessary to ensure that earthquake shaking scenarios accurately reflect expected ground motion characteristics, thereby contributing to an improved deterministic seismic hazard assessment in the wider Dubrovnik area. |
