Seçil Turan Karaoğlan

Jeofizik verilerinin ters çözümünde günümüzde türev tabanlı optimizasyon yöntemlerinin yerini metasezgisel yöntemler almaya başlamıştır. Arama uzayını kapsamlı taraması, iyi bir başlangıç modeline ihtiyaç duymaması ve her optimizasyon problemine kolay uyarlanabilmesi gibi avantajları nedeniyle doğadan esinlenerek oluşturulan birçok metasezgisel yöntem, jeofizik ters çözüm çalışmalarında kullanılmaktadır. Bu çalışmada, popülasyon tabanlı bir metasezgisel yöntem olan Yarasa Algoritmasının doğal potansiyel verilerinin ters çözümünde kullanımına yönelik incelemeler sunulmuştur. Öncelikle Yarasa Algoritmasının kontrol parametrelerini belirleme çalışmaları yapılmış, elde edilen kontrol parametreleri kullanılarak da gürültüsüz ve gürültü eklenmiş kuramsal doğal potansiyel verilerinin ve iki adet arazi verisinin ters çözümü çalışmaları gerçekleştirilmiştir. Ayrıca, kuramsal ve arazi verilerine ait model parametrelerinin güvenilirliğinin belirlenmesi amacıyla Metropolis-Hasting algoritması kullanılarak, kestirim parametrelerinin olasılık yoğunluğu histogramları, istatistiksel olarak da test edilmiştir. Kuramsal verilerin ters çözüm sonuçları ilgili verilere ait doğru değerlerle, arazi verilerine ait ters çözüm sonuçlarıysa literatürdeki diğer çalışmalarda bulunan sonuçlarla kıyaslanarak Yarasa Algoritmasının doğal potansiyel verilerinin ters çözümünde efektif bir yöntem olarak uygulanabilir olduğu görülmüştür

In this work, we introduce application of a hybrid algorithm (DE/PSO) to estimate the model parameters from residual gravity anomalies due to some simple geometrical bodies. This algorithm combines differential evolution (DE) and particle swarm optimization (PSO). To investigate the performance of the hybrid algorithm, test studies were carried out using synthetic and field data sets. The synthetic data sets include noise-free and noisy synthetic anomalies. Two published gravity anomalies from Cuba and Canada were used as the field data. In the hybrid algorithm, DE and PSO yield [premature] solutions separately and share their best solutions during an iterative process. An openly accessible metaheuristics package (NMOF) in R programming environment was used to implement the hybrid algorithm. Through simulations using synthetic anomalies, DE/PSO algorithm was successful to provide improved results. In comparison to the solutions from the single algorithms (DE and PSO), the DE/PSO algorithm shows more effectiveness in terms of accuracy and convergence. The true model parameters of noise-free and noisy synthetic gravity anomalies were recovered well by the hybrid algorithm. The results of inversion for the field examples are characterized by low residual values between the observed gravity anomalies and the calculated ones.

Local and regional networks are needed to better examine the regions with complex tectonism, such as western Anatolia. Local seismic networks may provide critical data to both scientists and local authorities. They are also required as a base for future early warning systems which Izmir, the largest city in the western Anatolia, does not have one. With the support of Dokuz Eylul University, Engineering Faculty and the Department of Geophysical Engineering the installation of a new local network DEUNET established on February 15, 2019, to obtain new and detailed information about Izmir and its surroundings by observing microearthquake activity as well as major and damaging earthquakes in the region. The obtained maps indicate that the network in this period has the capability to detect and locate local, regional and teleseismic earthquakes occurred. According to the results, an intense seismic activity was observed in the northeast and southwest of the station network from its establishment to the present. These recent active areas affected by two large earthquakes which one is called Aegean Sea – Samos earthquake and the other is Manisa earthquake occurred and affected Izmir City and its surroundings. We also examined the focal mechanism solutions of some recorded earthquakes in this paper. In addition, the ground characteristics of the station locations were examined using the recorded earthquakes. Since Dokuz Eylul University is the only university in Western Anatolia that offers seismology education, this network will also provide significant benefits for seismology education.