Geology Reports⌕ Search

SEARCH · Geology Reports

Results for “Algorithms”

Search indexed USGS publications on groundwater, aquifers, geologic maps, mineral resources and earthquakes. Explore source records by subject and place.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,081 records · Page 60Linked to original sources
USGS2024

Using crustal-scale refraction data of joint inversions of Rayleigh-wave dispersion curves and H/V spectral ratios for Atlantic Coastal Plain velocity structure, eastern U.S.

Shallow shear‐wave velocities ( ⁠⁠ V s ) sometimes are estimated from joint inversions of horizontal‐to‐vertical (H/V) spectral ratios and surface‐wave dispersion curves derived from ambient noise or small active sources. Here, we evaluate carrying out these inversions using Rayleigh‐wave dispersion curves computed from crustal‐scale P ‐wave seismic refraction data. We use data from the 2014–2015 Eastern North American Margin (ENAM) experiment in Virginia and North Carolina, but similar seismic refraction data sets have been acquired over sedimentary basins of interest for seismic hazard studies, including in major urban areas. The ENAM project deployed a pair of ∼215 km long, northwest–southeast linear arrays with ∼300 m receiver spacing to record 11 dynamite shots, and 80 continuously recording seismometers with 5–6 km spacing along the same arrays to record offshore airguns. The arrays crossed the onland portion of the Atlantic Coastal Plain sediments, which are a seaward‐thickening wedge of Cretaceous and younger sediments deposited mostly on crystalline bedrock. We compute Rayleigh‐wave dispersion curves from 3 to 9 km long portions of the receiver arrays on each side of the dynamite shots, and we compute ambient‐noise H/V ratios from the continuously recording seismometers. We use a genetic inversion algorithm in which forward velocity models in each “generation” are evaluated for misfits compared to the observed data, with subsequent generations constructed from the models with the smallest misfits. Velocities to depths of 500 m are defined well, as shown by a narrow range of velocities in the best‐fit models, by the consistency between multiple inversion runs at a site, and by forward modeling of site responses. The resulting velocity cross‐section of the Coastal Plain strata has seaward‐dipping contours in the thinner portions of the Coastal Plain but smaller dips in the deeper portions. We interpret these results as showing that velocity contours in the ACP strata are influenced by a combination of lithology and overburden pressure. Results demonstrate that existing seismic refraction data have the potential for determining detailed shallow shear‐wave velocity profiles.

eastern United States↗
USGS2024

Site response in the Walnut Creek–Concord region of the San Francisco Bay, California: Ground motion amplification in a fault-bounded basin

Thirty‐seven portable accelerometers were deployed in the eastern San Francisco Bay communities of Walnut Creek and Concord to study site response in a fault‐bounded, urban, sedimentary basin. Local earthquakes were recorded for a period of two years from 2017 to 2019 resulting in 101 well‐recorded events. Site response is estimated by two methods: the reference site spectral ratio method and a source‐site spectral inversion method. The reference site spectral ratio method allows investigation of the variability of site amplification with source azimuth and frequency. The source‐site spectral inversion method yields the best least‐squares fit to site response for a database of ground‐motion records. Both methods show substantial amplification in the Walnut Creek–Concord basin below 2 Hz indicating strong surface‐wave development. Greater amplification is seen for sources aligned along the long axis of the basin. Inversion using close‐in sources at short distances yields lower amplification at longer periods than the entire data set due to reduced surface‐wave generation for steeper angles of incidence. Inversion of site response spectra for shallow shear‐wave velocity using a global search algorithm yields V S30 values consistent with generalized mapping results based on geology and topography but with greater variability due to local site variations. 3D finite‐element modeling shows greater amplification in the Walnut Creek–Concord basin with a basin‐edge effect likely contributing to higher ground motions. Topography is also seen to lead to increased scattering and shadowing effects.

California↗
USGS2025

Site response and wave propagation effects in the eastern United States

Fourier amplitude spectra from regional earthquakes in the eastern United States are used in a parametric inversion for source, path, and site effects. Five earthquakes are selected for analysis during the installation of the United States National Seismic Network (US), Earthscope’s USArray Transportable Array (TA), and other temporary arrays to maximize station coverage. A global search algorithm is used to solve for site response from 0.1 to 15 Hz, corner frequency, geometrical spreading ( r - γ ), and frequency dependent anelastic attenuation in the form Q(f) = Q o f α . Tradeoff between moment and geometric spreading is handled by fixing the moment. The tradeoff between corner frequency and Q(f) is solved by selecting the value of corner frequency that minimizes an objective function defined over all stations. Values of site response and attenuation parameters show a strong spatial correlation with the physiographic provinces of the eastern United States. Site response for the Atlantic Coastal Plain is consistent with previous work using spectral ratios relative to a reference site, defined by strong resonance peaks correlated with the thickness of sediments. Site response for the other physiographic provinces is markedly different from the coastal plain, with a lack of distinct resonance peaks and a broad moderate high at frequences from 0.1 to 0.5 Hz consistent with the hard-rock geology of the regions. Like site response, Q(f) has a strong correlation with physiographic province, showing lower values on the coastal plain and higher values inland. Geometric spreading exponent, γ, decreases with increasing hypocenter distance from just above 1 at a few tens of kilometers to 0.9 at 500 km. The limited range in geometric spreading values is attributed to starting the Fourier transform window at the S ‐wave arrival for all distances and averaging over multiple wave types.

eastern United States↗
USGS2015

The Red Atrapa Sismos (Quake Catcher Network in Mexico): assessing performance during large and damaging earthquakes.

The Quake‐Catcher Network (QCN) is an expanding seismic array made possible by thousands of participants who volunteered time and resources from their computers to record seismic data using low‐cost accelerometers (http://qcn.stanford.edu/; last accessed December 2014). Sensors based on Micro‐Electromechanical Systems (MEMS) technology have rapidly improved over the last few years due to the demand of the private sector (e.g., automobiles, cell phones, and laptops). For strong‐motion applications, low‐cost MEMS accelerometers have promising features due to an increasing resolution and near‐linear phase and amplitude response ( Cochran, Lawrence, Christensen, and Jakka, 2009 ; Clayton et al. , 2011 ; Evans et al. , 2014 ). Each volunteer computer monitors ground motion and communicates using the Berkeley Open Infrastructure for Network Computing (BOINC, Anderson, 2004 ). Using a standard short‐term average, long‐term average (STLA) algorithm ( Earle and Shearer, 1994 ; Cochran, Lawrence, Christensen, Chung, 2009 ; Cochran, Lawrence, Christensen, and Jakka, 2009 ), volunteer computer and sensor systems detect abrupt changes in the acceleration recordings. Each time a possible trigger signal is declared, a small package of information containing sensor and ground‐motion information is streamed to one of the QCN servers ( Chung et al. , 2011 ). Trigger signals, correlated in space and time, are then processed by the QCN server to look for potential earthquakes.

Seismological Research Letters↗
USGS2016

Reply to “Comment on ‘Ground motions from the 2015 Mw 7.8 Gorkha, Nepal, earthquake constrained by a detailed assessment of macroseismic data’ by Stacey S. Martin, Susan E. Hough, and Charleen Hung” by Andrea Tertulliani, Laura Graziani, Corrado Castellano, Alessandra Maramai, and Antonio Rossi

We thank Andrea Tertulliani and his colleagues for their interest in our article on the 2015 Gorkha earthquake (Martin, Hough, et al., 2015), and for their comments pertaining to our study (Tertulliani et al., 2016). Indeed, as they note, a comprehensive assessment of macroseismic effects for an earthquake with far‐reaching effects as that of Gorkha is not only critically important but is also an extremely difficult undertaking. In the absence of a widely known web‐based system, employing a well‐calibrated algorithm with which to collect and systematically assess macroseismic information (e.g., Wald et al., 1999; Coppola et al., 2010; Bossu et al., 2015) in the Indian subcontinent, one is left with two approaches to characterize effects of an event such as the Gorkha earthquake: a comprehensive ground‐based survey such as the one undertaken in India following the 2001 Bhuj earthquake (Pande and Kayal, 2003), or an assessment such as Martin, Hough, et al. (2015) akin to other contemporary studies (e.g., Nuttli, 1973; Sieh, 1978; Meltzner and Wald, 1998; Martin and Szeliga, 2010; Ambraseys and Bilham, 2012; Mahajan et al., 2012; Gupta et al., 2013; Singh et al., 2013; Hough and Martin, 2015; Martin and Hough, 2015; Martin, Bradley, et al., 2015; Ribeiro et al., 2015), based primarily upon media reports and other available documentary accounts.

Seismological Research Letters↗
USGS2017

Earthquake Early Warning ShakeAlert System: Testing and certification platform

Earthquake early warning systems provide warnings to end users of incoming moderate to strong ground shaking from earthquakes. An earthquake early warning system, ShakeAlert, is providing alerts to beta end users in the western United States, specifically California, Oregon, and Washington. An essential aspect of the earthquake early warning system is the development of a framework to test modifications to code to ensure functionality and assess performance. In 2016, a Testing and Certification Platform (TCP) was included in the development of the Production Prototype version of ShakeAlert. The purpose of the TCP is to evaluate the robustness of candidate code that is proposed for deployment on ShakeAlert Production Prototype servers. TCP consists of two main components: a real‐time in situ test that replicates the real‐time production system and an offline playback system to replay test suites. The real‐time tests of system performance assess code optimization and stability. The offline tests comprise a stress test of candidate code to assess if the code is production ready. The test suite includes over 120 events including local, regional, and teleseismic historic earthquakes, recentering and calibration events, and other anomalous and potentially problematic signals. Two assessments of alert performance are conducted. First, point‐source assessments are undertaken to compare magnitude, epicentral location, and origin time with the Advanced National Seismic System Comprehensive Catalog, as well as to evaluate alert latency. Second, we describe assessment of the quality of ground‐motion predictions at end‐user sites by comparing predicted shaking intensities to ShakeMaps for historic events and implement a threshold‐based approach that assesses how often end users initiate the appropriate action, based on their ground‐shaking threshold. TCP has been developed to be a convenient streamlined procedure for objectively testing algorithms, and it has been designed with flexibility to accommodate significant changes in development of new or modified system code. It is expected that the TCP will continue to evolve along with the ShakeAlert system, and the framework we describe here provides one example of how earthquake early warning systems can be evaluated.

California, Oregon, Washington↗
USGS2020

Leveraging deep learning in global 24/7 real-time earthquake monitoring at the National Earthquake Information Center

Machine‐learning algorithms continue to show promise in their application to seismic processing. The U.S. Geological Survey National Earthquake Information Center (NEIC) is exploring the adoption of these tools to aid in simultaneous local, regional, and global real‐time earthquake monitoring. As a first step, we describe a simple framework to incorporate deep‐learning tools into NEIC operations. Automatic seismic arrival detections made from standard picking methods (e.g., short‐term average/long‐term average [STA/LTA]) are fed to trained neural network models to improve automatic seismic‐arrival (pick) timing and estimate seismic‐arrival phase type and source‐station distances. These additional data are used to improve the capabilities of the NEIC associator. We compile a dataset of 1.3 million seismic‐phase arrivals that represent a globally distributed set of source‐station paths covering a range of phase types, magnitudes, and source distances. We train three separate convolutional neural network models to predict arrival time onset, phase type, and distance. We validate the performance of the trained networks on a subset of our existing dataset and further extend validation by exploring the model performance when applied to NEIC automatic pick data feeds. We show that the information provided by these models can be useful in downstream event processing, specifically in seismic‐phase association, resulting in reduced false associations and improved location estimates.

Seismological Research Letters↗
USGS2020

Latency of waveform data delivery from the Southern California Seismic Network during the 2019 Ridgecrest earthquake sequence and its effect on ShakeAlert

The occurrence of the 4–6 July 2019 M w 6.4 and M w 7.1 Ridgecrest earthquake sequence provided the first full‐scale test of the network and telemetry readiness of the Southern California Seismic Network (SCSN), to support the ShakeAlert earthquake early warning (EEW) system in California. ShakeAlert is a U.S. Geological Survey (USGS)‐led collaboration to detect earthquakes and, when possible, to alert the public before the arrival of the strongest shaking. The SCSN performed well in its regional monitoring role for both the 4 July M w 6.4 and the 6 July M w 7.1 earthquakes. In the EEW role, it provided timely delivery of 5 s of P ‐wave data to ShakeAlert, which issued its first alert 6.9 s after origin time. Data delivery at peak data volumes for many stations exhibited some latency, and, as a consequence, some data arrived too late for analysis by one of the EEW algorithms. We find that the average link bandwidth for each station was sufficient, because all waveform data were delivered automatically to the archive, but link capacity for many stations was insufficient for peak demand. We describe the performance of the data telemetry for the sequence, including cellular, radio, hybrid, and backhaul systems. Cellular‐based telemetry systems maintained low latency throughout strong shaking and after, but some stations, even at great distances, experienced subsequent brief increases in latency. Performance of radio links depended mostly on the signal strength of the link, with short‐distance direct shots to high‐bandwidth backhaul systems showing no latency impact, whereas stations on some long distance or marginal quality links suffered latencies of tens or hundreds of seconds. Improvements are being implemented to move telemetry links onto USGS and partner high‐bandwidth microwave systems, and to reduce dependency on less robust long‐distance radio shots.

California↗
USGS2022

Ready for real time: Performance of Global Navigation Satellite System in 2019 Mw 7.1 Ridgecrest, California, rapid response products

Global Navigation Satellite Systems (GNSSs) have undergone notable advancement in the last few decades, leading to the availability of a dataset with capabilities well beyond its original intended purpose. The proliferation of high‐rate (1 Hz or greater) GNSS receivers in areas of seismological interest now allows for routine consideration of dynamic earthquake ground motions, with centimeter‐level displacement accuracy via precise point positioning methods. Real‐time (RT) GNSS observations, from stations that are both telemetered and processed to displacement with minimal latency, have lower accuracy compared to post‐processed (PP) GNSS displacements due to imprecise knowledge of atmospheric conditions, satellite clocks, and satellite orbits in RT. Whether the quality of RT high‐rate GNSS is sufficient for use in rapid response products remains to be thoroughly examined. Here, we highlight RT GNSS displacement time series processed during the 2019 M w "> M w Mw 7.1 Ridgecrest, California, earthquake in the context of common rapid‐response products, magnitude estimation, and kinematic fault‐slip models. We discuss how these data can be used to supplement RT seismic data for rapid characterization of significant earthquakes. We find that kinematic fault‐slip models using RT GNSS data retain the general spatiotemporal characteristics of those with PP data, with subtle differences in size and amplitude of modeled slip asperities. We demonstrate the effect of these rapid seismic source models using RT GNSS data on the U.S. Geological Survey product ShakeMap—a downstream ground‐motion prediction algorithm informed by the rupture dimensions estimated in the slip model. Discrepancies in the ShakeMap estimate are minor, within ±12% change, with the most severe variation at the fault edges. Our analysis suggests that, when used in conjunction with available seismic data sources, RT GNSS is sufficient and valuable for rapid earthquake characterization.

California↗
USGS2022

A global catalog of calibrated earthquake locations

We produced a globally distributed catalog of earthquakes and nuclear explosions with calibrated hypocenters, referred to as the Global Catalog of Calibrated Earthquake Locations (GCCEL). This dataset currently contains 18,782 events in 289 clusters with >3.2 million arrival times observed at 19,258 stations. The term “calibrated” refers to the property that the hypocenters are minimally biased by unknown Earth structure. In addition, we calculate uncertainties using empirically determined variability of the arrival‐time data itself, specific to each calibrated cluster of hypocenters. Outliers in the arrival‐time dataset are removed based on measured variability of the data. In each cluster, we estimate the empirically determined uncertainty for each set of station‐phase arrival times. We use a version of the hypocentroidal decomposition multiple event relocation algorithm specifically adapted for calibrated relocations of clusters of seismic events. Most clusters are calibrated by fitting the subset of direct crustal first arrivals ( Pg and Sg ) with a locally appropriate travel‐time model to estimate the cluster hypocentroid. A few clusters are calibrated by aligning the pattern of relative locations in space and time with one or more events for which a ground‐truth hypocenter is available from an independent source with known uncertainty, such as a nuclear explosion. Epicentral uncertainties in GCCEL typically range from 1 to 5 km with a 90% confidence interval. Most events have depth constraint from one or more sources, usually with an uncertainty of ≤5 km. GCCEL is a significant resource for research at local, regional, and global scales because it provides minimally biased absolute hypocenters, meaningful associated error estimates, and curated arrival times as a reference dataset that can be used as prior constraints in the development of new regional, national, and global earthquake catalogs; validation of new location techniques; and the generation of advanced Earth models.

Seismological Research Letters↗
USGS2023

Earthquake detection with tinyML

Earthquake detection is the critical first step in earthquake early warning (EEW) systems. For robust EEW systems, detection accuracy, detection latency, and sensor density are critical to providing real‐time earthquake alerts. Traditional EEW systems use fixed sensor networks or, more recently, networks of mobile phones equipped with microelectromechanical systems (MEMS) accelerometers. Internet of things edge devices, with built‐in tiny machine learning (tinyML) capable microcontrollers, and always‐on, internet‐connected, stationary MEMS accelerometers provide the opportunity to deploy ML‐based earthquake detection and warning using a single‐station approach at a global scale. Here, I test and evaluate tinyML deep learning algorithms for earthquake detection on a microcontroller. I show that the tinyML earthquake detection models can generalize to earthquakes outside the training set.

Seismological Research Letters↗
USGS2024

Distinguishing natural sources from anthropogenic events in seismic data

As seismic data are increasingly used to investigate a diverse range of subsurface phenomena beyond regular fast-rupturing earthquakes (Peng and Gomberg, 2010; Beroza and Ide, 2011), it is important to acknowledge that human-generated ground vibrations may be mistaken for naturally generated subsurface processes (Larose et al., 2015; Li et al., 2018). Correct discrimination of natural processes from anthropogenic noise is especially pressing given the trend in seismic detection research toward automated algorithms and machine learning methods (Yoon et al., 2015; Kong et al., 2019;Mousavi and Beroza, 2022) and the growth in seismic data collection in new environments such as urban and industry settings (e.g., Díaz et al.,2017).

Seismological Research Letters↗
USGS2025

SURF: An automated method for building nonplanar 3D fault models from earthquake hypocenters

Accurately characterizing 3D fault geometry is vital for improving our understanding of earthquake behavior and informing the development of seismic hazard models. Despite their importance, subsurface fault structures tend to be poorly constrained because of limitations in observational data. Improvements to the seismic networks and earthquake detection algorithms have increased the precision and volume of earthquake catalogs, which help illuminate detailed subsurface fault structure and provide the most direct information available about fault geometries at depth. We present a Python package to automate generating 3D fault geometries directly from hypocentral seismicity patterns. This method begins with clustering events based on their spatial density, identifying coherent patterns. Nearby clusters are then merged based on the similarity of their orientations. We fit nonplanar surfaces using support vector regression to balance surface accuracy with minimal deviations from planarity. The fault models are output as quadrilateral meshes at user‐defined resolution. In the process of generating the 3D fault surfaces, we compute the spatial density of seismicity around the surface and the planarity as quantitative metrics of the model outputs. As a proof of concept, we apply this approach to the San Andreas–Calaveras fault junction region and the 2019 Ridgecrest earthquake sequence, both in California, which contain complex subparallel faults well defined at the Earth’s surface and abundant microseismicity. These case studies demonstrate the method’s ability to model complex fault structures, including long continuous fault surfaces, crossing faults, variably dipping segments, and subparallel faults. We test the method on both standard network catalogs and double‐difference relocated catalogs. We find that our seismicity‐based fault model results align with published 3D models that incorporate additional constraints and interpretations ( Plesch et al. , 2020 ; Aagaard and Hirakawa, 2021 ). This workflow provides a low‐user‐input solution for estimating fault geometries at depth from earthquake catalogs.

California↗
USGS2025

Searching for seismic precursors - The Barry Landslide hazard clean up

The Barry Landslide, located in Barry Arm of Prince William Sound, Alaska, poses a major hazard due to its steep, unstable slopes and the potential for a massive landslide-generated tsunami. With an estimated volume of 500–700 million cubic meters, the Barry Landslide could trigger highly destructive waves. In this study, we focus on seismic signals from the Barry Landslide, which are critical for providing timely tsunami warnings. Since the summer of 2020, the region has been instrumented to monitor the landslide, but the seismic record is complicated by the presence of nearby glaciers and frequent regional earthquakes. Among these signals, we analyze a specific class of short-duration, high-frequency seismic events that exhibit strong seasonal variability, increasing in rate from late summer to mid-winter before ceasing abruptly in late winter or early spring. Our analysis suggests that the source of these signals is likely near or beneath Cascade Glacier, adjacent to the landslide, rather than within the landslide mass itself. We apply detection algorithms to construct a time history for this signal type, which we then compare with environmental factors like precipitation, temperature, and slope displacement data from ground-based radar and remote sensing. Correlations indicate that these seismic events may be driven by seasonal hydrological changes, particularly the freeze-up of subglacial water pathways. While these events are not directly linked to landslide motion, they serve as indirect markers of subsurface hydrological conditions that influence slope stability. Our findings highlight the complex interplay between glaciers, groundwater, and landslide dynamics, emphasizing the need for multi-parameter monitoring to assess evolving geohazards in the region.

Alaska↗
USGS2026

Digitizer Suite: The Albuquerque Seismological Laboratory Digitizer Testing Suite

Laboratory testing of digitizers and seismometers helps ensure that prior to deployment the instrumentation can produce high quality data and is operating within specifications. In this work we detail the software package called: the Albuquerque Seismological Laboratory (ASL) Digitizer Test Suite. This Java software package provides several algorithms to verify various performance parameters of digitizers commonly used for recording analog seismic instruments. The goal of these tests is not to be exhaustive, but to identify common failures that could compromise the integrity of seismic data being recorded on the digitizer. For example, Sandia National Laboratories (e.g., Slad and Merchant, 2018) routinely do comprehensive testing of digitizers for various monitoring missions. While these tests reports are valuable for comprehensively characterizing a recording system, it would be resource intensive to conduct such tests on every seismic recorder used in a network. We focus on tests that include ways to estimate the sensitivity, timing, self-noise, and clip-level of the digitizer, as well as the fidelity of the signal being recorded. The software is publicly available and provides a way for the community to verify the integrity of a digitizer using a minimum amount of outside equipment.

Seismological Research Letters↗
USGS1989

A satellite-based digital data system for low-frequency geophysical data

A reliable method for collection, display, and analysis of low-frequency geophysical data from isolated sites, which can be throughout North and South America and the Pacific Rim, has been developed for use with the Geostationary Operational Environmental Satellite (GOES) system. Geophysical data primarily intended for earthquake hazard and crustal deformation monitoring are digitized with either 12-bit or 16-bit resolution and transmitted every 10 min through a satellite link to a bank of UNIX-based computers in Menlo Park, California. There the data are available for analysis and display within a few seconds of their transmit time. This system provides real-time monitoring of crustal deformation parameters such as tilt, strain, fault displacement, local magnetic field, crustal geochemistry, and water levels, as well as meteorological and other parameters, along faults in California and Alaska, and in volcanic regions in the western United States, Rabaul, and other locations in the New Britain region of the South Pacific. Various mathematical, statistical, and graphical algorithms process the incoming data to detect changes in crustal deformation and fault slip that may indicate the first stages of catastrophic fault failure. Alert trigger levels based on physical models, signal resolution, and previous history have been defined for particular instrument types. Computer-driven remote paging and mail systems are used to notify appropriate personnel when alarm status is reached. The system supports continuous historical records of low-frequency geophysical data, software for extensive analysis of these data, and programs for modeling fault rupture with and without seismic radiation, as well as providing an environment for real-time attempts at earthquake prediction.

Bulletin of the Seismological Society of America↗
USGS1995

Source parameters and crustal Q for four earthquakes in South Carolina

Two three-component seismometers (one surface and one borehole) were re-installed on the Savannah River Site (SRS), South Carolina in July 1992 to determine attenuation in the Coastal Plain sediment wedge and source parameters of local earthquakes. Four earthquakes M ∼ 1.8 to 3.6 were recorded during the next 6 months. The largest event was located near Summerville within the meizoseismal area for the 1886 Charleston earthquake. Two shocks were located 50 km to the east near Neeses, and one was located 20 km north of the SRS near Aiken. Although source parameters have been determined from strong motion data and short-period regional networks for east coast earthquakes, such as the Saguenay, Nahanni, and Mt. Laurier earthquakes (e.g., Atkinson, 1993 ), these are some of the first source parameters determined from broad-band digital recorders. Seismograms for the Summerville event are also available from Chapel Hill, North Carolina and Blacksburg, Virginia, providing estimates of t * beyond 200 km. Here we determine source parameters such as moment, stress drop, and the attenuation parameter t * using a non-linear least-squares algorithm. We do not correct for site response because the deepest borehole is not below the Coastal Plain sediments (about 300m thick at this site) and because only one station is available for most of the data. Values of t* are marginally higher from seismograms recorded at the surface when compared to records from the 91m depth borehole seismograph. A value of 170-200 bars was determined for the Brune stress drop of the Summerville event using the borehole data, which is high compared to a value of 50 bars usually specified for modeling strong motion in western North America, but similar to other estimates for eastern North America. A higher stress drop leads to a higher seismic risk because peak acceleration is approximately proportional to stress drop. Moreover, mid- to upper-crustal Qs are in the range of 2,000 to 3,000, which would permit the propagation of high frequency seismic waves. A comparison of the surface records from the SRS with a record from the USGS dense array at Parkfield, CA for an event at about the same distance range and moment as the Summerville event-SRS case shows that the peak acceleration of the Summerville event is 16 times higher than that for the event from California (stress drop of 21 bars).

South Carolina↗
USGS2007

Random forests for classification in ecology

Classification procedures are some of the most widely used statistical methods in ecology. Random forests (RF) is a new and powerful statistical classifier that is well established in other disciplines but is relatively unknown in ecology. Advantages of RF compared to other statistical classifiers include (1) very high classification accuracy; (2) a novel method of determining variable importance; (3) ability to model complex interactions among predictor variables; (4) flexibility to perform several types of statistical data analysis, including regression, classification, survival analysis, and unsupervised learning; and (5) an algorithm for imputing missing values. We compared the accuracies of RF and four other commonly used statistical classifiers using data on invasive plant species presence in Lava Beds National Monument, California, USA, rare lichen species presence in the Pacific Northwest, USA, and nest sites for cavity nesting birds in the Uinta Mountains, Utah, USA. We observed high classification accuracy in all applications as measured by cross-validation and, in the case of the lichen data, by independent test data, when comparing RF to other common classification methods. We also observed that the variables that RF identified as most important for classifying invasive plant species coincided with expectations based on the literature. ?? 2007 by the Ecological Society of America.

Ecology↗