فیلترکردن داده‌های خرد لرزه‌ای با تکنیک کاهش مصنوعی فرکانس نمونه‌برداری برای تخمین بسامد آبرفت‌های بنیادی و عمق سنگ بستر مهندسی

کیفیت اندازه‌گیری و روش پردازش خردلرزه‌ها تاثیر قابل‌توجهی بر دقت تخمین پارامترهای دینامیکی ساختگاه دارد. در این مقاله، به‌منظور افزایش دقت تخمین فرکانس غالب آبرفت، از تکنیک کاهش مصنوعی فرکانس نمونه‌برداری و روش واریانس پیش‌رونده استفاده شده است. این روش‌ها به‌ویژه در مناطق شهری که تحت‌تاثیر اختلالات محلی قرار دارند، می‌توانند به‌طور موثری نویزهای اضافی را حذف کرده و دقت تخمین را افزایش دهند. با استفاده از پنجره لغزان، واریانس پیش‌رونده سیگنال محاسبه و از آن به‌عنوان پایه‌ای برای حذف محدوده داده‌هایی که تحت‌تاثیر اختلالات گذرا قرار گرفته‌اند، استفاده می‌شود. نتایج حاصل از اعمال این روش‌ها بر روی داده‌های خردلرزه‌ای ثبت شده در شهر میبد، نشان‌دهنده عملکرد موثر این تکنیک‌ها در حذف اثر اختلالات محلی و بهبود دقت تخمین فرکانس غالب آبرفت است. این مطالعه نشان می‌دهد که با تغییراتی در روش پردازش داده‌ها، می‌توان به‌طور قابل‌توجهی از اثرات مخرب اختلالات محلی کاست و فرکانس غالب زمین را با دقت بیشتری تعیین کرد. با توجه به اقتصادی بودن این روش، تحلیل خردلرزه‌ها می‌تواند به‌عنوان یک روش پایه‌ای در مطالعات اولیه ژئوتکنیکی در مناطق مختلف مورد استفاده قرار گیرد.

filtering of microtremor data with the technique of artificial reduction of sampling frequency to estimate the frequency of fundamental alluvium and the depth of engineering bedrock

the phenomenon of resonance caused by the behavior of alluvium during an earthquake affects the occurrence of damage. this phenomenon occurs when the dominant period of alluvion is equal to the dominant period of structures (mukhopadhyay and bormann, 2004; kanli et al., 2006; kvasnička et al., 2011). therefore, determining the dominant frequency of the soil with high certainty is particularly important and is the primary goal of this paper. the method of using microtremors to determine the site’s response to earthquakes and determine the dominant frequency of the soil is of interest, and by measuring them at the ground level, information about the vibration characteristics of the soil can be obtained. (kerh chu, 2002; kianoush et al., 2023a,b; jamshidi et al., 2024; khoshmagham et al., 2024).adib et al. (2015) conducted the site classification of ardakan city in yazd province based on the earth’s natural frequency by the microtremor data. while identifying the dynamic characteristics of the earth in this area, the level of compatibility of the land classification is suitable for the studies of the site effect based on the geotechnical, geophysical, and microtremor data with the regulations of dynamic design of buildings. similarly, bagheri et al. (2017) emphasized the importance of denoising seismic data to improve the quality of records, which is essential for accurate interpretation. they proposed a novel method combining frequency-offset deconvolution (fxd) and decision-based median (dbm) filtering to enhance the signal-to-noise ratio (s/n) and effectively suppress random noise.adib and others (adib et al., 2015) classified the site effects using a fractal model based on the analysis of microtremor data, frequency amplification index, and vulnerability (k-g)) in the city of meybod. nogoshi and igarashi (1970, 1971) improved, this classification and found four types of soil, including (1) hard soil and weak rock with a frequency of 6.2 to 8 hz, (2) hard soil with a frequency of about 4.9 to 6.2 hz, (3) relatively soft soil with a frequency of 2.4 to 4.9 hz and (4) soft soil soft with a frequency of less than 2.4 hz that were separated in this city (wahba et al., 2024; labuta et al., 2025). bagheri and riahi (2016, 2018) also highlighted the effectiveness of dbm filtering in seismic data processing, demonstrating its superiority in suppressing random noise and improving s/n. this classification aligns with findings by khodagholi and bagheri (2020), who introduced local least squares polynomial (llsp) smoothing to diminish seismic random noise, thereby enhancing data quality and interpretability.the quality of microtremor data and processing methods considerably affects the accuracy of estimating dynamic soil parameters. in this paper, an artificial sampling frequency reduction technique has been proposed for removing high-frequency perturbations from microtremor data, and a running variance method has been used to improve the automatic detection of data sections infected by local perturbations. in this method, the running variance of the signal was calculated using a sliding window. then, the resulting variance signal was used to remove the portions of data affected by transient perturbations. the proposed methods have been applied to the data recorded for the meybod city in the north of yazd province. by comparing the results of the proposed techniques and the standard methods on microtremor data in urban areas, it is clear that the proposed methods have successfully removed the effects of local transients and their redundant fluctuations such that there is no sharp amplitude variation in the residual signal. the results of the simulation confirm this claim.additionally, the ease of parameter setting in the running variance approach makes it superior to the ratio method. discarding the fewer amounts of contaminated data is an effective method, especially in regions with extensive noise. this paper shows that some variations in microtremor data processes could reduce the destructive effects of local transients on them and consequently improve fundamental site frequency estimation based on the spectral ratio of microtremors. therefore, considering the low cost of this approach, microtremor analysis can be used as a primary method for initial geotechnical studies in various regions. as a result, it increases the accuracy and degree of confidence in estimating the dominant frequency based on the spectral ratio based on the microtremors data.