The ability to rapidly assess the spatial distribution and severity of building damage is essential to post-event emergency response and recovery. Visually identifying and classifying individual building damage requires significant time and personnel resources and can last for months after the event. This article evaluates the feasibility of using machine learning techniques such as discriminant analysis, k-nearest neighbors, decision trees, and random forests, to rapidly predict earthquake-induced building damage. Data from the 2014 South Napa earthquake are used for the study where building damage is classified based on the assigned Applied Technology Council (ATC)-20 tag (red, yellow, and green). Spectral acceleration at a period of 0.3 s, fault distance, and several building specific characteristics (e.g. age, floor area, presence of plan irregularity) are used as features or predictor variables for the machine learning models. A portion of the damage data from the Napa earthquake is used to obtain the forecast model, and the performance of each machine learning technique is evaluated using the remaining (test) data. It is noted that the random forest algorithm can accurately predict the assigned tags for 66% of the buildings in the test dataset.

Ergodic site amplification models for active tectonic regions are conditioned on the time-averaged shear wave velocity in the upper 30 m ( VS30) and the depth to a shear wave velocity isosurface ( zx). The depth components of such models are derived using data from sites within many geomorphic domains. We provide a site amplification model utilizing VS30 and depth, with the depth component conditioned on type of geomorphic province: basins, valleys, and mountain/hills. As with current models, the depth component of our model is centered with respect to the VS30-scaling model using differential depth δzx, taken as the difference between a site-specific depth and a VS30 -conditioned average depth. Using data from southern California, we find that long-period site response for all sites combined exhibits relative de-amplification and amplification for negative and positive differential depths, respectively. Individual provinces exhibit broadly similar trends with depth, but amplification levels are on average stronger in basins such that little relative de-amplification occurs at negative differential depths. Valley and mountain/hill sites have, on average, weaker amplification levels but stronger scaling with δzx. Site-to-site standard deviations vary appreciably across geomorphic provinces, with basins having lower dispersions than mountain/hill sites and the reference ergodic model.

The 2019 Ridgecrest earthquake sequence produced a 4 July M 6.5 foreshock and a 5 July M 7.1 mainshock, along with 23 events with magnitudes greater than 4.5 in the 24 hr period following the mainshock. The epicenters of the two principal events were located in the Indian Wells Valley, northwest of Searles Valley near the towns of Ridgecrest, Trona, and Argus. We describe observed liquefaction manifestations including sand boils, fissures, and lateral spreading features, as well as proximate non-ground failure zones that resulted from the sequence. Expanding upon results initially presented in a report of the Geotechnical Extreme Events Reconnaissance Association, we synthesize results of field mapping, aerial imagery, and inferences of ground deformations from Synthetic Aperture Radar-based damage proxy maps (DPMs). We document incidents of liquefaction, settlement, and lateral spreading in the Naval Air Weapons Station China Lake US military base and compare locations of these observations to pre- and postevent mapping of liquefaction hazards. We describe liquefaction and ground-failure features in Trona and Argus, which produced lateral deformations and impacts on several single-story masonry and wood frame buildings. Detailed maps showing zones with and without ground failure are provided for these towns, along with mapped ground deformations along transects. Finally, we describe incidents of massive liquefaction with related ground failures and proximate areas of similar geologic origin without ground failure in the Searles Lakebed. Observations in this region are consistent with surface change predicted by the DPM. In the same region, geospatial liquefaction hazard maps are effective at identifying broad percentages of land with liquefaction-related damage. We anticipate that data presented in this article will be useful for future liquefaction susceptibility, triggering, and consequence studies being undertaken as part of the Next Generation Liquefaction project.

We introduce procedures to validate site response in sedimentary basins as predicted using ground motion simulations. These procedures aim to isolate contributions of site response to computed intensity measures relative to those from seismic source and path effects. In one of the validation procedures, simulated motions are analyzed in the same manner as earthquake recordings to derive non-ergodic site terms. This procedure compares the scaling with sediment isosurface depth of simulated versus empirical site terms (the latter having been derived in a separate study). A second validation procedure utilizes two sets of simulations, one that considers three-dimensional (3D) basin structure and a second that utilizes a one-dimensional (1D) representation of the crustal structure. Identical sources are used in both procedures, and after correcting for variable path effects, differences in ground motions are used to estimate site amplification in 3D basins. Such site responses are compared to those derived empirically to validate both the absolute levels and the depth scaling of site response from 3D simulations. We apply both procedures to southern California in a manner that is consistent between the simulated and empirical data (i.e. by using similar event locations and magnitudes). The results show that the 3D simulations overpredict the depth-scaling and absolute levels of site amplification in basins. However, overall patterns of site amplification with depth are similar, suggesting that future calibration may be able to remove observed biases.

Following the Ridgecrest earthquake sequence, consisting of an M 6.4 foreshock and M 7.1 mainshock along with many other events, the Geotechnical Extreme Events Reconnaissance association deployed a team to gather perishable data. The team focused their efforts on documenting ground deformations including surface fault rupture south of the Naval Air Weapons Station China Lake, and liquefaction features in Trona and Argus. The team published a report within two weeks of the M 7.1 mainshock. This article presents data products gathered by the team, which are now published and publicly accessible. The data products presented herein include ground-based observations using Global Positioning System trackers, digital cameras, and hand-measuring devices, as well as unmanned aerial vehicle-based imaging products using Structure from Motion to create point clouds and digital surface models. The article describes the data products, as well as tools available for interacting with the products.