We observe frequent air traffic events in continuous seismic waveforms recorded for about 30 days by 1,108 vertical geophones in a tight array on the San Jacinto fault zone. The waveforms of the air traffic events resemble tremor or collections of small earthquakes. However, time‐frequency analysis reveals clear Doppler effects that can be modeled with basic equations and fitted well with parameters corresponding to airplanes and helicopters. The flying traces can be inverted by fitting the parameters at each station across the entire array. About 31 air traffic events are detected per day in the relatively remote study area, with peak activity from about 8:00 to 18:00 and significantly fewer events between 23:00 and 4:00 local time. The average event duration is about 200 s, so they cover together >7% of the day. To estimate the total time covered by earthquakes, we derive a scaling relation logτ(M) = 0.41M + 0.89 between earthquake duration and magnitudes using data of 266 earthquakes in the magnitude range 0 ≤ M ≤ 3. The results indicate that in most places, the duration of air traffic events is likely to exceed considerably the total time covered by earthquakes.

Proper classification of nontectonic seismic signals is critical for detecting microearthquakes and developing an improved understanding of ongoing weak ground motions. We use unsupervised machine learning to label five classes of nonstationary seismic noise common in continuous waveforms. Temporal and spectral features describing the data are clustered to identify separable types of emergent and impulsive waveforms. The trained clustering model is used to classify every 1 s of continuous seismic records from a dense seismic array with 10–30 m station spacing. We show that dominate noise signals can be highly localized and vary on length scales of hundreds of meters. The methodology demonstrates the complexity of weak ground motions and improves the standard of analyzing seismic waveforms with a low signal‐to‐noise ratio. Application of this technique will improve the ability to detect genuine microseismic events in noisy environments where seismic sensors record earthquake‐like signals originating from nontectonic sources.

Correct identification and modeling of anthropogenic sources of ground motion are of considerable importance for many studies, including detection of small earthquakes and imaging seismic properties below the surface. To understand signals generated by common vehicle traffic, we use seismic data recorded by closely spaced geophones normal to roads at two sites on San Jacinto fault zone. To quantify the spatiotemporal and frequency variations of the recorded ground motions, we develop a simple analytical solution accounting for propagation and attenuation of surface waves. The model reproduces well‐observed bell‐shaped spectrograms of car signals recorded by geophones close to roads, and it can be used to estimate frequency‐dependent $Q$‐values of the subsurface materials. The data analysis indicates $Q$‐values of 3–40, for frequencies up to 150 Hz for road‐receiver paths at the two examined sites. The derived $Q$‐values are consistent with attenuation factors of surface waves previously obtained with other methods. The analytical results and analysis procedure provide a highly efficient method for deriving $Q$‐values of shallow subsurface materials.

Understanding earthquake foreshocks is essential for deciphering earthquake rupture physics and can aid seismic hazard mitigation. With regional dense seismic arrays, we identify immediate foreshocks of 527 0.9 M 5.4 events of the 2019 Ridgecrest earthquake sequence, including 48 earthquakes with series of immediate foreshocks. These immediate foreshocks are adjacent to the mainshocks occurring within 100 s of the mainshocks, and their P waves share high resemblances with the mainshock P waves. However, attributes of the immediate‐foreshock P waves, including the amplitudes and preceding times, do not clearly scale with the mainshock magnitudes. Our observations suggest that earthquake rupture may initiate in a universal fashion but evolves stochastically. This indicates that earthquake rupture development is likely controlled by fine‐scale fault heterogeneities in the Ridgecrest fault system, and the final magnitude is the only difference between small and large earthquakes.

We develop a semiautomated method for estimating with second seismic moments the directivity, rupture area, duration, and centroid velocity of earthquakes. The method is applied to 41 southern California earthquakes with magnitude in the range 3.5–5.2 and provides stable results for 28 events. Apparent source time functions (ASTFs) of P and S phases are derived using deconvolution with three stacked empirical Green's functions (seGf). The use of seGf suppresses nongeneric source effects, improves the focal mechanism correspondence to the analyzed earthquakes, and typically allows inclusion of 5 to 15 more ASTFs compared with analysis using a single eGf. Most analyzed earthquakes in the Trifurcation area of the San Jacinto Fault have directivities toward the northwest, while events around Cajon Pass and San Gabriel Mountain tend to propagate toward the southeast. These results are generally consistent with predictions for dynamic rupture on bimaterial interfaces associated with the imaged velocity contrasts in the area. The second moment inversions also provide constraints on the upper and lower bounds of rupture areas in our data set. Stress drops and uncertainties are estimated for elliptical ruptures using the derived characteristic rupture length and width. The semiautomated second moment method with seGfs can be used for routine application to moderate earthquakes in locations with good station coverage.