Evaluation of Performance and Ranking of Local, Regional, and Global Attenuation Relations Based on Strong-Motion Datasets Recorded in Seismotectonic Provinces of Iran

Abstract

Selecting the most appropriate attenuation relations is one of the key stages in conducting seismic hazard analysis. Neglecting this factor can significantly affect the analysis results and make them unreliable. In this study, the Iranian dataset was first extracted from the Iranian Strong-Motion Network (ISMN). This dataset includes records from accelerometer stations with known average shear-wave velocity in each seismotectonic province, for earthquakes with moment magnitudes of 4.5 and above, and distances ranging from zero to a maximum of 200 km, up to 5 Jan 2025. Then, the performance of 10 attenuation relations developed globally, regionally for the Middle-East, and locally in Iran, was evaluated. Given the vastness of Iran and the differences in seismic regimes in various parts, analyses were conducted separately for each of the seismotectonic provinces of Iran. For this purpose, the residuals distribution and log-likelihood (LLH) methods were used to detect biases and quantitatively evaluate goodness-of-fit, respectively. These well-established methods are commonly used for ranking and weighting. This study also introduces the local standard deviation, which provides a better assessment of the conformity between predicted and observed values. The performance of attenuation relations across all seismotectonic provinces of Iran has been thoroughly evaluated. The results of the residuals distribution and the LLH test show strong agreement in all cases. This study led to the identification of efficient relations and showed that in all five seismotectonic provinces, including Zagros, Alborz-Azerbaijan, Central-East Iran, Koppeh Dagh, and Makoran, the relations proposed by Farajpour et al. (Bull Seismol Soc Am 109:732–744, 2019), Sedaghati and Pezeshk (Bull Seismol Soc Am 107:934–948, 2017), and Kale et al. (Bull Seismol Soc Am 105:963–980, 2015) consistently ranked first to third and had the best performance in estimating experienced data. The findings of this study can significantly assist in the appropriate selection of attenuation relations for both deterministic and probabilistic seismic hazard analysis.