We present a data‐driven approach to clustering or grouping Global Navigation Satellite System (GNSS) stations according to observed velocities, displacements or other selected characteristics. Clustering GNSS stations provides useful scientific information, and is a necessary initial step in other analysis, such as detecting aseismic transient signals (Granat et al., 2013, https://doi.org/10.1785/0220130039 ). Desired features of the data can be selected for clustering, including some subset of displacement or velocity components, uncertainty estimates, station location, and other relevant information. Based on those selections, the clustering procedure autonomously groups the GNSS stations according to a selected clustering method. We have implemented this approach as a Python application, allowing us to draw upon the full range of open source clustering methods available in Python's scikit‐learn package (Pedregosa et al., 2011, https://doi.org/10.5555/1953048.2078195 ). The application returns the stations labeled by group as a table and color coded KML file and is designed to work with the GNSS information available from GeoGateway (Donnellan et al., 2021, https://doi.org/10.1007/s12145-020-00561-7 ; Heflin et al., 2020, https://doi.org/10.1029/2019ea000644 ) but is easily extensible. We demonstrate the methodology on California and western Nevada. The results show partitions that follow faults or geologic boundaries, including for recent large earthquakes and post‐seismic motion. The San Andreas fault system is most prominent, reflecting Pacific‐North American plate boundary motion. Deformation reflected as class boundaries is distributed north and south of the central California creeping section. For most models a cluster boundary connects the southernmost San Andreas fault with the Eastern California Shear Zone (ECSZ) rather than continuing through the San Gorgonio Pass.
Flood hazards and their disastrous consequences disrupt economic activity and threaten human lives globally. From a remote sensing perspective, since floods are often triggered by extreme climatic events, such as heavy rainstorms or tropical cyclones, the efficacy of using optical remote sensing data for disaster and damage mapping is significantly compromised. In many flood events, obtaining cloud-free images covering the affected area remains challenging. Nonetheless, considering that floods are the most frequent type of natural disaster on Earth, optical remote sensing data should be fully exploited. In this article, firstly, we will present a critical review of remote sensing data and machine learning methods for global flood-induced damage detection and mapping. We will primarily consider two types of remote sensing data: moderate-resolution multi-spectral data and high-resolution true-color or panchromatic data. Big and semantic databases available for advanced machine learning to date will be introduced. We will develop a set of best-use case scenarios for using these two data types to conduct water-body and built-up area mapping with no to moderate cloud coverage. We will cross-verify traditional machine learning and current deep learning methods and provide both benchmark databases and algorithms for the research community. Last, with this suite of data and algorithms, we will demonstrate the development of a cloud-computing-supported computing gateway, which houses the services of both our remote-sensing-based machine learning engine and a web-based user interface. Under this gateway, optical satellite data will be retrieved based on a global flood alerting system. Near-real-time pre- and post-event flood analytics are then showcased for end-user decision-making, providing insights such as the extent of severely flooded areas, an estimated number of affected buildings, and spatial trends of damage. In summary, this paper’s novel contributions include (1) a critical synthesis of operational readiness in flood mapping, (2) a multi-sensor-aware review of optical limitations, (3) the deployment of a lightweight ML pipeline for near-real-time mapping, and (4) a proposal of the GloFIM platform for field-level disaster support.
Uninhabited aerial vehicle synthetic aperture radar (UAVSAR) observations 2009–2017 of the Yuha Desert area and Global Positioning System (GPS) time series encompassing the region reveal a northward migrating pattern of deformation following the 4 April 2010 Mw7.2 El Mayor‐Cucapah (EMC) earthquake. The north end of the EMC rupture exhibits an asymmetric pattern of deformation that is substantial and smooth northeast of the rupture and limited but with surface fracturing slip northwest. The earthquake triggered ~1 cm of surface coseismic slip at the Yuha fault, which continued to slip postseismically. 2.5 cm of Yuha fault slip occurred by the time of the 15 June 2010 Mw5.7 Ocotillo aftershock and 5 cm of slip occurred by 2017 following a logarithmic afterslip decay 16‐day timescale. The Ocotillo aftershock triggered 1.4 cm of slip on a northwest trend extending to the Elsinore fault and by 7 years after the EMC earthquake 2.4 cm of slip had accumulated with a distribution following an afterslip function with a 16‐day timescale consistent with other earthquakes and a rate strengthening upper crustal sedimentary layer. GPS data show broad coseismic uplift of the Salton Trough and delayed postseismic motion that may be indicative of fluid migration there and subsidence west of the rupture extension, which continues following the earthquake. The data indicate that the Elsinore, Laguna Salada, and EMC ruptures are part of the same fault system. The results also suggest that north‐south shortening and east‐west extension across the region drove fracture advancing step tectonics north of the EMC earthquake rupture.
18 April 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Flooding is a frequent extreme weather event that causes significant financial and societal losses. According to the International Disaster Database (EM-DAT), during January through July 2023, 87 flooding events caused about 2,000 deaths and $13 billion in damages, globally. Among the impacted, lowand medium-income countries with resource scarcity tend to experience high mortality, displacement of people, unmitigated damages, and long-term recovery. Currently, several hydrologic models and Earth Observation (EO) datasets are used to forecast flood severity and impacts. However, not all of these models are globally operational or publicly available. The variability in outputs in terms of accuracy, scale, and content also limits their usage for emergency response activities. The Model of Models (MoM), an ensemble approach, integrates hydrologic models and EO datasets (i) to forecast flood risk (probability of occurrence) globally every 24-hours at a sub-watershed level, which is used (ii) to disseminate alert messages and potential impact information to at-risk communities using the Pacific Disaster Center's DisasterAWARE platform. MoM is operational and designed to assist countries with flood risk management and mitigation by providing early warning and situational awareness information. An accuracy assessment of MoM from user-perspective across nine different flood types revealed that (i) the model reliably generated early warning for 100% of the flooded sub-watersheds in seven events, and (ii) during 2022 flooding, 61% and 89% of the flooded sub-watersheds that were identified to be in Watch and Warning categories in Pakistan and Chad, respectively, were detected to be flooding by the Copernicus Global Flood Monitoring system.
(1) Background: For realizing an effective flood detection of Synthetic Aperture Radar (SAR) images in water regions, a post-classification comparison algorithm with Multi-depth Flood Detection Convolutional Neural Network (MDFD-CNN) is proposed in this paper. The MDFD-CNN is used to classify and extract the water regions in the SAR image, aiming at solving the flood detection after high-resolution SAR water regions extraction in complex terrain. (2) Methods: What the MDFD-CNN owning a two-branches network with different depth is used for extracting respectively the water regions in the bi-temporal SAR images using the saliency detection. In the process of recognizing the water regions, to improve the reliability of the training samples and reduce the number of training samples, MDFD-CNN uses the SAR backscatter coefficient significance detection to obtain the salient regions of the extracted training samples. Moreover, a piecewise back-propagation is adopted to optimize the network. With the detected water regions of the bi-temporal images, a post-classification comparison is implemented for detecting the changes in water regions. (3) Results: Compared with the manually extracted features used by conventional methods, what the MDFD-CNN extracted is more robust. Meanwhile, experimental results illustrates the effectiveness, especially strong noise immunity, of the proposed method. (4) Conclusions: The proposed method offers a valid way for flood detection in water regions of SAR images in realistic environments.
In this study, we present a method for the joint inversion of receiver function and ambient noise based on Bayesian inverse theory (Tarantola, 1987, 2005). The nonlinear inversion method of the complex spectrum ratio of receiver functions (Liu et al., 1996) has been extended to perform the joint inversion of the receiver function and ambient noise with a global scanning of the crustal Poisson's ratio. The forward problem of the Rayleigh‐wave phase dispersion is solved in terms of a modified version of the fast generalized R/T method proposed by Pei et al. (2008, 2009). Our numerical tests show that (1) the dependency of inversion results on initial models is removed and the model's parameter is estimated reliably even in the case of using a vertically homogeneous model as the initial guess for the crust structure; (2) because the frequency band consistency of the receiver function with the phase dispersion obtained from ambient noise is much better than that with seismic surface waves, the S‐wave velocity structure in depth of 0~80 km can be well estimated by the joint inversion of receiver function and noise‐derived phase velocity dispersion in the period range of 2~40 s, and the space resolution of the shallow structure near to the surface can reach 1 km; (3) global scanning of the Poisson's ratio is not only in favor of data interpretation of the receiver function and ambient noise, but also provides a reliable estimation of the crustal Poisson's ratio. The joint inversion of receiver function and ambient noise recorded at station KWC05 of the western Sichuan seismic array shows that the crustal thickness beneath the station reaches 44 km and the crustal S‐wave velocity structure manifests a high‐speed upper crust and low‐speed middle‐lower crust in depth of 24~42 km. The Poisson's ratio averaged over the crust is 0.262 and that over the low‐velocity zone is 0.27.