University of Isfahan Geography and Environmental Planning 2008-5362 35 1 2024 03 20 Forecasting Magnitudes and Locations of Potential Earthquakes along Railway Lines in Hormozgan Province Using Artificial Neural Network (ANN) Forecasting Magnitudes and Locations of Potential Earthquakes along Railway Lines in Hormozgan Province Using Artificial Neural Network (ANN) 111 128 28047 10.22108/gep.2023.136336.1569 FA Mohsen Pourkhosravani Associate professor, Department of Geography, Shahid Bahonar University of Kerman, Kerman, Iran Ali Mehrabi Associate professor, Department of Geography, Shahid Bahonar University of Kerman, Kerman, Iran Zahra Amirjahanshahi Master's degree in Natural Hazards, Shahid Bahonar University of Kerman, Kerman, Iran Journal Article 2023 01 09 Abstract Predicting the magnitude and location of earthquakes can significantly mitigate the impact of this natural phenomenon. Anticipating potential earthquake locations can enhance infrastructure resilience and reduce vulnerability. This study aimed to forecast and analyze the magnitude and location of potential earthquakes along the railway lines in Hormozgan Province using intelligent Artificial Neural Network (ANN) algorithms. The model utilized earthquake location, magnitude, and depth data from the International Institute of Seismology and Earthquake Engineering, as well as fault lengths in the region as the input variables. The findings revealed 32 potential earthquake points in the study area with projected magnitudes ranging from 4.3 to 5.2 on the Richter scale. The earthquake prediction-based zoning of Hormozgan Province indicated that the southern and central parts (north of the Strait of Hormuz) were at a high risk. Consequently, the rail lines in this area were more susceptible. Specifically, Tunnel No. 23 was situated in a high-risk zone and Tunnels 21, 22, and 23 were in close proximity to earthquakes with magnitudes exceeding 5 on the Richter scale. Keywords: Forecast, Railway Lines, Earthquake, Artificial Neural Network (ANN), Hormozgan Province. Introduction Earthquakes represent one of the most intricate and nonlinear natural phenomena. Their complex nature and system variability make predicting their magnitudes and locations seemingly impossible. However, forecasting these aspects of earthquakes can significantly mitigate the damage caused by such events. Anticipating the locations of potential earthquakes can bolster infrastructure and facilities in these areas, reducing their vulnerability. Consequently, the quest for reliable methods to predict the timing, location, and magnitude of earthquakes has been a focal point of recent research. Artificial Neural Networks (ANNs) have emerged as powerful tools for earthquake prediction, offering several key advantages. Firstly, they excel at learning complex, nonlinear environments. Secondly, they make no assumptions about data distribution and thirdly, they exhibit flexibility in handling incomplete or missing data (Vellido et al., 1999, p. 53). Overall, ANNs have demonstrated success in various domains, including system identification, approximation and estimation, optimization, and behavior prediction (Cigizoglu & Kisi, 2006, p. 236). Hormozgan Province situated in the folded Zagros belt harbors numerous faults and has experienced destructive earthquakes in the past, indicating its high seismic potential. Therefore, this study sought to address the following questions: What is the likelihood of high-magnitude earthquakes occurring in Hormozgan Province? And where are the potential locations of these earthquakes? Materials & Methods This applied study aimed to forecast the magnitudes and locations of potential earthquakes in Hormozgan Province using the ANN algorithm. The simulation utilized earthquake location, depth, and magnitude data for events exceeding 4 on the Richter scale in the study area, along with fault length, as the model inputs. The prediction of earthquake magnitudes was carried out using the Perceptron neural network, while the Cohen's neural network was employed to forecast potential earthquake locations. Specifically, the Perceptron neural network was utilized for magnitude prediction and the Self-Organizing Feature Map (SOFM) neural network was employed for location prediction. Research Findings In general, the seismic potential of faults to generate earthquakes is influenced by seismic history, tectonic movement, and fault dimensions. Through the application of ANNs, a total of 32 potential earthquake locations were predicted with projected magnitudes ranging from 4.3 to 5.2. Discussion of Results & Conclusion The study's findings indicated the prediction of 32 potential earthquake locations in the study area with projected magnitudes ranging from 4.3 to 5.2 on the Richter scale. Consequently, zoning of Hormozgan Province based on these predictions revealed that the southern and central parts of the province (north of the Strait of Hormuz) were situated in high-risk zones. This heightened risk could make the rail lines in this area more susceptible to potential seismic events. Notably, Tunnel No. 23 was located in a high-risk area and Tunnels 21, 22, and 23 were in close proximity to earthquakes with magnitudes exceeding 5 on the Richter scale. Abstract Predicting the magnitude and location of earthquakes can significantly mitigate the impact of this natural phenomenon. Anticipating potential earthquake locations can enhance infrastructure resilience and reduce vulnerability. This study aimed to forecast and analyze the magnitude and location of potential earthquakes along the railway lines in Hormozgan Province using intelligent Artificial Neural Network (ANN) algorithms. The model utilized earthquake location, magnitude, and depth data from the International Institute of Seismology and Earthquake Engineering, as well as fault lengths in the region as the input variables. The findings revealed 32 potential earthquake points in the study area with projected magnitudes ranging from 4.3 to 5.2 on the Richter scale. The earthquake prediction-based zoning of Hormozgan Province indicated that the southern and central parts (north of the Strait of Hormuz) were at a high risk. Consequently, the rail lines in this area were more susceptible. Specifically, Tunnel No. 23 was situated in a high-risk zone and Tunnels 21, 22, and 23 were in close proximity to earthquakes with magnitudes exceeding 5 on the Richter scale. Keywords: Forecast, Railway Lines, Earthquake, Artificial Neural Network (ANN), Hormozgan Province. Introduction Earthquakes represent one of the most intricate and nonlinear natural phenomena. Their complex nature and system variability make predicting their magnitudes and locations seemingly impossible. However, forecasting these aspects of earthquakes can significantly mitigate the damage caused by such events. Anticipating the locations of potential earthquakes can bolster infrastructure and facilities in these areas, reducing their vulnerability. Consequently, the quest for reliable methods to predict the timing, location, and magnitude of earthquakes has been a focal point of recent research. Artificial Neural Networks (ANNs) have emerged as powerful tools for earthquake prediction, offering several key advantages. Firstly, they excel at learning complex, nonlinear environments. Secondly, they make no assumptions about data distribution and thirdly, they exhibit flexibility in handling incomplete or missing data (Vellido et al., 1999, p. 53). Overall, ANNs have demonstrated success in various domains, including system identification, approximation and estimation, optimization, and behavior prediction (Cigizoglu & Kisi, 2006, p. 236). Hormozgan Province situated in the folded Zagros belt harbors numerous faults and has experienced destructive earthquakes in the past, indicating its high seismic potential. Therefore, this study sought to address the following questions: What is the likelihood of high-magnitude earthquakes occurring in Hormozgan Province? And where are the potential locations of these earthquakes? Materials & Methods This applied study aimed to forecast the magnitudes and locations of potential earthquakes in Hormozgan Province using the ANN algorithm. The simulation utilized earthquake location, depth, and magnitude data for events exceeding 4 on the Richter scale in the study area, along with fault length, as the model inputs. The prediction of earthquake magnitudes was carried out using the Perceptron neural network, while the Cohen's neural network was employed to forecast potential earthquake locations. Specifically, the Perceptron neural network was utilized for magnitude prediction and the Self-Organizing Feature Map (SOFM) neural network was employed for location prediction. Research Findings In general, the seismic potential of faults to generate earthquakes is influenced by seismic history, tectonic movement, and fault dimensions. Through the application of ANNs, a total of 32 potential earthquake locations were predicted with projected magnitudes ranging from 4.3 to 5.2. Discussion of Results & Conclusion The study's findings indicated the prediction of 32 potential earthquake locations in the study area with projected magnitudes ranging from 4.3 to 5.2 on the Richter scale. Consequently, zoning of Hormozgan Province based on these predictions revealed that the southern and central parts of the province (north of the Strait of Hormuz) were situated in high-risk zones. This heightened risk could make the rail lines in this area more susceptible to potential seismic events. Notably, Tunnel No. 23 was located in a high-risk area and Tunnels 21, 22, and 23 were in close proximity to earthquakes with magnitudes exceeding 5 on the Richter scale.

https://gep.ui.ac.ir/article_28047_7187dc529fb1ba340c19bd3d8b4ca801.pdf