Browsing by Author "Sun, Jiajia"
Now showing 1 - 20 of 23
- Results Per Page
- Sort Options
Item Analyzing Shear-Wave Splitting in the Permian Basin and Evaluating the Measurements with Deep Learning(2023-12) Guzman, Veronica Valentina; Li, Aibing; Mann, Paul; Sun, Jiajia; Savvaidis, AlexandrosThe Permian Basin, the largest oil-producing basin in the United States, has experienced increasing seismic activity since 2009, including a few large earthquakes with ML5.0 and above in the Delaware and Midland basins. We have conducted shear-wave splitting (SWS) analysis from local earthquakes at ten TexNet stations to understand the increasing and intensifying seismicity in the area. In the Delaware Basin, the fast orientations from individual events vary in a broad range for all stations, indicating a complex fracture system in the upper crust even though the averages are consistent with the local fault strikes or the maximum horizontal stress. Fast orientations with large angles from the local stress appeared after the 2020 ML5.0 earthquake, that I interpret as slip on less favorable fracture planes due to increased pore pressure. SWS orientations in the Midland Basin are parallel with geologic fault strikes to the southwest of Midland City and parallel with the regional SHmax direction to the northeast of Midland SWS analysis requires processing a large quantity of seismograms and inspecting the measurements for quality control, which is laborious and time-consuming. I have developed a new automated procedure to categorize SWS results using the method of Convolutional Neural Networks and Support Vector Machines. The existing SWS dataset in the Delaware Basin is used to test the reliability and scalability of this automated method. The new method significantly reduces the processing time for SWS parameters and shows a high degree of accuracy. The method is also successfully applied to new SWS data in the Delaware Basin. I expect a broad usage of this automation method in other SWS studies from local earthquakes.Item Building Static Geological Models in the Presence of Missing Values and Sparse Data(2019-12) Adelman, Melanie Milner 1988-; Sun, Jiajia; Yarus, Jeffrey; Casey, John F.Earth modeling is a critical tool available for geoscientists and engineers to assess risk and make important decisions in drilling, completing, and producing wells. It requires input of available data such as well log and seismic data, to generate the present-day distribution of rock and fluid properties of a petroleum system and its elements. Model accuracy is highly dependent on the quality and quantity of available data. Two fundamental issues confronted in earth modeling is the occurrence of missing values and sparse data. Missing values refers to gaps in well logs that remain unrecorded or intervals of unrecorded data attributed to various sources. Sparse data occurs when there is a lack of data to build a reliable model. The missing value and sparse data problems often occur in unconventional reservoirs where well logs are often not run, or poor hole conditions can often prevent continuous data collection. Such problems can also take place in conventional reservoirs. The lack of data and occurrence of missing values make it challenging to create reliable earth models and understand their uncertainty. The objective of this study is to solve the missing values issue using statistical imputation, machine learning, and geostatistical techniques to predict proxy values where data values are missing. The sparse data problem may also be obviated by the integration of a 3-D finite-volume basin model, containing estimates of key variables, collocated with reservoir petrophysical well data. A finite-volume of petrophysical, geochemical, and geomechanical properties derived from the finite-volume basin model provide “soft” data where “hard” well data does not exist. Often, the creation of an earth model from sparse data is achieved by the use of a secondary variable such as seismic data, which is not always available and can be expensive to acquire. A finite-volume basin model can be effectively used as a co-variable in place or in addition to seismic, is inexpensive compared to seismic and requires a minimum set of data input which is commonly accessible from public domain sources. The proposed methods of missing value imputation or prediction and collocation of a finite-volume basin model with local petrophysical well data can provide qualitative and quantitative information essential for the generation of a more reliable earth model. Missing values are typically handled by petrophysicists and geostatisticians as a pre-modeling procedure prior to simulation during earth modeling. The time allocated to data preparation far exceeds the amount of time focused on earth modeling. Big data exacerbates this dilemma along with problems associated with missing values and sparse data. The techniques suggested in this thesis are scalable to high performance computing, machine learning, and automated environments, providing an opportunity to reduce the amount of time spent on data quality control and increase earth model reliability.Item Effects of Rock and Fluid Properties on Seismic Dispersion and Attenuation in Sandstone(2021-05) Wei, Qianqian; Zhou, Hua-Wei; Han, De-Hua; Sun, Jiajia; Li, HuiA better understanding of the relationship between dispersion/attenuation and rock/fluid properties is of great interest in improving hydrocarbon identification and reservoir characterization. Presently it is less investigated at seismic frequency range, limited by the reliable laboratory data measured under varying physical conditions. Additionally, the roles of fluid type and distribution in enhancing wave dispersion and attenuation are still poorly understood. With this concern, I perform three groups of laboratory measurements on wave dispersion and attenuation for typical porous sandstones at both seismic and ultrasonic frequencies and under vacuum-dry and fluid-saturated conditions. More specifically, sandstone samples are fully or partially saturated by a series of fluids: methane, butane, water, and glycerin, aiming to investigate effects of fluid viscosity and distribution on wave dispersion and attenuation. The experimental data suggests that distinct dispersion and attenuation can be found even at vacuum-dry conditions, especially for sandstones with relatively high clay contents. This finding might contradict the previous knowledge of no dispersion and attenuation in dry rocks, but has been extensively certified through a series of laboratory data in this study. Nevertheless, the comparison with fluid saturated data indicates that pore fluid related mechanisms are still the dominant cause for the dispersion and attenuation in sandstones. Significant dispersion and attenuation occur in the presence of relatively small amounts of gas both for partial glycerin and partial water saturation, yet varying in their magnitudes and characteristic frequencies. Generally, the overall characteristic frequencies shift to a relatively lower frequency range with the decrease of rock permeability or the increase of fluid viscosity. A complete attenuation curve is firstly observed in glycerin-saturated conditions at measured seismic frequency. Based on porous modeling analysis, the mesoscopic fluid flow in response to a heterogeneous fluid distribution in the pore space, might be the dominant mechanism accounting for the observed dispersion and attenuation in partially fluid-saturated rocks. The associations among velocity dispersion and wave attenuation, rock permeability, and fluid properties in the laboratory provide a potential indicator for the presence of fizz gas or high-permeability zones in fields during seismic surveys.Item Forecasting Petrophysical Trends Based on Well Logs and Seismic Attributes Using Supervised Machine Learning Algorithms in the Teapot Dome Field, Wyoming(2018-12) Chen, Po-Hsu 1993-; Zhou, Hua-Wei; Sun, Jiajia; Li, Xin-GongThe objective of this research is to forecast petrophysical trends at the Teapot Dome field, Wyoming using supervised machine learning algorithms based on a combined use of well logs and seismic attributes. Thirteen instantaneous attributes and two reservoir properties, porosity and water saturation, were selected to set up the training structure after depth conversion and data re-sampling. Three algorithms, including kernel ridge regression, decision tree, and artificial neural network were proposed to build machine learning models. The optimization was done by minimizing the errors between the model-predicted results and the values of the targets. The performances of the models were quantified by the mean absolute errors, and each individual trained model was compared by a cross-validation approach. Further evaluation was done by three testing wells that were not used in the training. The tests indicate that, for my data and test parameters, decision tree is more useful than kernel ridge regression and artificial neural network in terms of the testing accuracy and effectiveness. Feature selection in decision tree was tested as a useful tool to reduce the data dimensionality, so that the redundant features and information can be removed without damaging the training accuracy. Based on the trained model, the distribution of porosity, and water saturation were simulated effectively and efficiently in the Teapot Dome field.Item Geoid Inversion For Mantle Viscosity With Convolutional Neural Networks(2022-04-28) Kerl, Jacob; Colli, Lorenzo; Sun, Jiajia; Ghelichkhan, SiavashThe Earth’s radial viscosity profile affects the extent to which density contrasts in the mantle manifest themselves in the long-wavelength shape of the non-hydrostatic geoid. Consequently, geodynamicists have used the observed geoid along with an estimate of the Earth’s density structure to invert for the viscous structure of the mantle. The inversion procedure, however, is one that has no direct solution, relies upon density estimates that have a high degree of uncertainty, and is computationally costly. In this study, we provide a first attempt at using machine learning as a low-cost solution to the geoid inverse problem. We carry out two separate solutions with convolutional neural networks (CNNs) and compare the results. The first solution uses two CNNs, where the first predicts viscous-response kernels from data characterizing the geoid and density structure of the Earth. The second network then predicts a radial viscosity profile from the viscous-response kernels. The second solution, instead, predicts a radial viscosity profile directly from the geoid and density data. We find that in the two-network solution, both CNNs make accurate predictions on the test data, but when predictions from the first network are fed into the second as input, the results are poor. We find that the single network solution allows us to obtain a smooth, long-wavelength estimate of the Earth’s radial viscosity profile.Item Geomorphological Analysis of Longhorn Cavern in Burnet County, Texas via Integrated Ground and Airborne LIDAR Data(2021-05) Sims, Drew A.; Wang, Guoquan; Li, Hong-Yi; Sun, JiajiaLonghorn Cavern is located within Burnet County, Texas in the Llano Uplift region. Longhorn Cavern encounters flooding via water percolation through the cavern ceiling from water discharging through sinkholes on the surface. This paper focuses on the four key topics which are: (1) generate an accurate and high-resolution 3-dimensional model of the cavern from a TLS (Terrestrial Laser Scanning) survey to provide a surface expression of it, (2) produce meshes from the collected data of the cavern for volumetric calculations of the cavern itself and the volume of the strata between the surface and cavern ceiling (3) characterize sinkholes within the area underneath the forest canopy, and (4) to delineate the water percolation/cover collapse zones from these sinkholes through a cavern to surface expression. To build a 3-dimensional model of the cavern, a LIDAR survey was performed in this study. TLS data was collected over the span of 3 days that was composed of 39 scans throughout Longhorn Cavern from the Sam Bass Entrance through the Hall of Gems room. The output of this LIDAR data was a high-resolution 3-D model of Longhorn Cavern containing a point density 457 points per m2 and a length of 460 m. The average standard deviation error is 2.21 between registered scans. Meshes were generated from this point cloud and the volume of the cavern was 16,035.47 m3. The volume of the strata above the cavern was 287,058.38 m3. 1-meter aerial LIDAR data of the surface around Longhorn Cavern was downloaded from the USGS 3-DEP project database, and a DEM was generated from it to delineate sinkholes around Longhorn Cavern State Park. 3 high-risk sinkholes water percolation/cover-collapse zones were discovered which include Sinkholes 28, 29, and 39. The volume above the cavern for sinkhole 28 is 13,254.21 m3, sinkhole 29 is 7,411.16 m3, and sinkhole 39 is 3,450.99 m3.Item Geophysical characterization of the Elk Creek Carbonatite based on 3D joint inversion and geology differentiation(2021-05) Li, Kenneth H.; Sun, Jiajia; Kass, Mason A.; Sager, William W.The Elk Creek Carbonatite located in southeastern Nebraska hosts the largest known niobium deposit in the United States. Most of the known niobium is hosted within magnetite-dolomite carbonatite, a dense and highly magnetized unit within the carbonatite. The shallower lithology of the carbonatite has been well explored by boreholes, but the deeper lithology remains poorly understood. Three-dimensional joint inversion of airborne gravity gradiometry and magnetic measurements was performed, producing a structurally coupled density and susceptibility model. Geology differentiation, a process of classifying the recovered subsurface models into distinct units, was then carried out to develop a 3D quasi-geology model. Physical property measurements based on drill core samples and analysis of inverted physical property values in the spatial domain were used for geology differentiation. The resulting quasi-geology model, an approximation of the subsurface geology, shows the spatial distribution of various geological units in 3D, and includes units at greater depths than previous studies on the region. I identified 11 geological units with each characterized by a distinct combination of density and susceptibility values. These units include the country rock surrounding Elk Creek, various carbonatites, mafic rocks, the niobium target zone, and additional distinct geological units which have not been previously classified. Geology differentiation also identifies the known niobium ore zone and indicates the existence of a significant volume of dense and strongly magnetized rocks below the deepest boreholes. These rocks are likely to be associated with unexplored niobium mineralization. This thesis work is the first attempt at constructing a 3D quasi-geology model in the study area based on airborne geophysical measurements, and demonstrates the added value of 3D geophysical inversions and geology differentiation when it comes to mineral exploration under thick sedimentary overburdenItem GPS AND INSAR INTEGRATED FAULTING, SUBSIDENCE, AND SEASONAL GROUND DEFORMATION MONITORING IN HOUSTON, TEXAS, USA(2022-12-08) Liu, Yuhao; Wang, Guoquan; Robinson, Alexander C.; Sun, Jiajia; Lee, HyongkiFor approximately 100 years, the Houston region has been adversely impacted by ground movements associated with active faulting, subsidence, and seasonal ground deformation, which have caused costly recurring damages to private and public infrastructure, including buildings, roadways, parking lots, and utility lines. The rapidly growing population in the Houston region means that the ongoing ground movements must be vigilantly monitored. In this study, we have introduced detailed methods using GPS, InSAR, and GPS-enhanced InSAR (GInSAR) to monitor faulting, subsidence, and seasonal ground deformation. The Houston region has numerous gravitationally induced “down-to-the-coast” growth faults that represent slow sliding of the land mass towards the Gulf of Mexico. The Long Point Fault is one of those active urban faults belonging to the complex normal fault system. We use a GPS array with 12 permanent stations installed along the two sides of the 16-km-long fault scarp to assess the activity of the Long Point Fault. The six-year continuous GPS observations (2013-2018) indicate that the Long Point Fault is currently inactive, with the rates of dip-slip and strike-slip being below 1 mm/year. GInSAR-derived subsidence rates (2015-2019) also suggest no considerable differences between the hanging wall and footwall sides along the Long Point Fault. Current surficial damages in the Long Point Fault area are more likely caused by ongoing uneven subsidence (~ 1 cm/year) and its induced horizontal strains, as well as the significant seasonal subsidence and heave, rather than deep-seated or tectonic-controlled fault movements. For mapping both the long-term (multiple years) and short-term (inter-annual. Seasonal) subsidence in the Houston region, we use the GInSAR method, integrating GPS and Sentinel-1A InSAR datasets covering the entire Houston region from 2015 to 2019. The root-mean-square (RMS) of the detrended InSAR-displacement time series is able to achieve a level of sub-centimeter accuracy, and the uncertainty (95% confidence interval) of the InSAR-derived subsidence rates is able to achieve a couple of millimeters accuracy per year for 5-year or longer datasets. The GInSAR mapping results suggest moderate ongoing subsidence (~ 1 cm/year) in northern Waller County, western Liberty County, and the city of Mont Belvieu in Champers County, which were not recognized in previous GPS-based investigations; the GInSAR mapping results also suggest that previous investigations overestimated the ongoing subsidence in western Montgomery County. This study indicates that hydraulic-head changes in the Evangeline aquifer are the primary cause of ongoing long-term and seasonal subsidence in the Houston region. The former is dominated by inelastic deformation, and the latter is dominated by elastic deformation. Both could cause infrastructure damage. This study demonstrated the potential of employing the GInSAR methods for near-real-time subsidence monitoring in the greater Houston region. The near-real-time monitoring would also provide timely information for understanding the dynamic of groundwater storage and improving both long-term and short-term groundwater resource management.Item GPS Monitoring and Land Subsidence in the Houston Metropolitan Area(2018-05) Kearns, Timothy Joseph 1982-; Wang, Guoquan; Murphy, Michael A.; Lee, Hyongki; Sun, JiajiaThis study aims to (1) establish a stable local reference frame (Houston16) for integrating all available GPS observations within the Houston metropolitan area and conducting precise subsidence and faulting monitoring over time and space, (2) evaluate the effects of groundwater withdrawal on land subsidence, and (3) assess the possible effect of petroleum production on ongoing minor subsidence within Galveston County. In order to realize Houston16, 15 long-term (> 5 years) stable GPS stations outside the greater Houston area were selected as reference stations. The precision (stability) of the local reference frame is less than 1 mm/year. Long-term observations from approximately 200 GPS and 13 extensometer stations indicate that the southeast Houston area has ceased subsiding at historic rates (< 3 mm/year) as a result of groundwater withdrawal regulations, but to the north and in western portions of the Houston metropolitan area subsidence bowls have developed and are expanding. Moderate subsidence (2 to 3 cm/year) is currently occurring in The Woodlands, Jersey Village, and Katy areas. Slight land uplift at the level of 2 to 3 mm/year has also been observed along the Houston Ship Channel and within the Houston downtown area. Groundwater level measurements from 170 wells screened in the Chicot aquifer and 320 wells screened in the Evangeline aquifer were investigated in order to evaluate the interaction between land subsidence and groundwater level change. The results further verify that groundwater withdrawal is the primary driver of land subsidence within the greater Houston area. Subsidence of 5 to 9 mm/year was observed within the southeastern region of Galveston County near the city of La Marque from 2005 to 2012, despite that groundwater levels within the Chicot and Evangeline aquifers have been restored to the preconsolidation head and have remained stable for over two decades. In order to evaluate the contribution of petroleum withdrawal to the local subsidence, 3,570 petroleum wells across 457 leases and 217 fields within Galveston County were investigated. Production from these petroleum wells includes the gas, condensate and oil fractions. Within the localized subsiding area, 2.7 times more petroleum was produced per square mile compared to the surrounding area that had not experienced subsidence during the period from 2005 to 2012. After peak production in 2004, petroleum withdrawal was gradually reduced, but subsidence continued until 2013, and then localized subsidence decreased to its current rate of 3 mm/year. The results suggest that petroleum withdrawal could be the primary cause of the localized subsidence observed from 2005 to 2012.Item Improving the Interpretation of Magnetic Tensor Data Using Deep Learning(2023-05-05) Barker, Keenan; Sun, Jiajia; Zhu, Jennifer; Wang, GuoquanThe accurate interpretation of magnetic tensor data can be difficult to perform without a strong knowledge of local geology and experience in reading magnetic data. I examined ways in which machine learning techniques can be applied to magnetic tensor data to automatically locate possible kimberlite targets and a method to sharpen smoothness based inversion models to provide a clearer image of the subsurface. While machine learning networks like the U-Net have shown success in other fields for image processing, these methods have not been used extensively in geophysics for magnetic interpretation. I trained the U-Net to predict kimberlite pipe locations by forward modelling magnetic susceptibility models to corresponding magnetic tensor data. I examined the use of different neural network architectures and methods for calculating loss to determine the effect on prediction accuracy. The U-Net was adapted in order to sharpen inversion models by changing from a two dimensional layer architecture to three dimensions. To train this second U-Net I first created a smoothing function which closely matches the effects of a smoothness based inversion and then applied this smoothing function to three dimensional magnetic susceptibility models. I also compared the effectiveness of using a general smoothing function against a smoothing function specifically designed to match a smoothness inversion. This study shows that the use of the U-Net architecture in the field of magnetics shows great promise in the areas of automatically detecting targets and sharpening inverted models.Item Machine Learning for Reservoir Characterization and Time-Lapse Seismic Analyses(2020-08) Hussein, Marwa; Stewart, Robert R.; Johnston, David H.; Zheng, Yingcai; Sun, JiajiaSuccessful reservoir management requires a clear understanding of reservoir distribution, quality, heterogeneity, and baffles. During field development, static reservoir models are built by incorporating well logs, core, and seismic data. However, seismic amplitude data often cannot image small faults, resolve thin reservoirs, or discriminate subtle changes in reservoir properties. Seismic attributes aid in illuminating subtle faults and stratigraphic features but, analyzing numerous individual attributes can be tedious and may have limitations for revealing small petrophysical changes within a reservoir. Using the Maui 3D dataset acquired in the offshore Taranaki Basin of New Zealand, I generate seismic attributes that are sensitive to faults as well as attributes that are sensitive to reservoir properties. I use principal component analysis (PCA) and self-organizing maps (SOM) to integrate geological information from six geometric attributes into one classification volume, which shows small faults that could affect field compartmentalization. I also develop a machine learning workflow to combine the reservoir information from eight spectral instantaneous attributes into one clustered volume. This SOM classification shows the reservoir distribution and helps to understand reservoir quality and to illuminate thin baffles. During reservoir production, time-lapse (4D) seismic data can be acquired to monitor fluid movement and constrain simulation models. However, 4D seismic data suffer from seismic interference and tuning effects. Thus, it can be challenging to monitor fluid behavior within thin reservoirs and/or to illuminate baffles. As a result, the reservoir simulation model may not capture sufficient reservoir heterogeneity detail, which can lead to a mismatch between synthetic models and observed 4D seismic responses. Using the 4D seismic data from the Maui field, I develop a systematic workflow to carefully select the attributes that best show 4D changes and integrate the preproduction and postproduction reservoir information from multiple 4D spectral instantaneous attributes into 4D classification volumes. Changes in SOM cluster patterns between baseline and monitor surveys suggest production-related changes within good to poor quality reservoirs that are primarily caused by water saturation changes as the reservoir is being water swept. Geobodies derived from the SOM clusters are used to calculate recoverable gas reserves that are compared to production data.Item Mantle Heterogeneity Through Mid-Ocean Ridge Basalts(2023-08) Ling, Xiang; Fu, Qi; Snow, Jonathan E.; Chin, Wynne W.; Sun, Jiajia; Wu, JonnyThe chemical heterogeneity of Earth's upper mantle is often examined through mid-ocean ridge basalts (MORB), which contest the prevalent assumption of a well-mixed convecting mantle. This dissertation delves into studying mantle heterogeneity from a local to regional scale. Initially, petrological models were utilized to scrutinize the source and melting conditions of potassium-enriched MORB (K-MORB) from the Central Lena Trough in the Arctic Ocean. The findings suggest a complex mantle source with phlogopite acting as the dominant H2O-K bearing mineral and indicate a binary mixing of 40% K-MORB and 60% N-MORB-like endmember, thereby providing insights into the area's tectonic evolution. Further into the study, advanced statistical methods like principal component analysis and partial least squares regression were used to decipher the isotopic compositions of Mid-Atlantic Ridge (MAR) basalts. These analyses reveal five distinct "zoo" signatures (akin to EMI, EMII, DM, HIMU, FOZO) in the mantle components and surprisingly, establish isotopic signatures as reliable predictors of geographic location along the ridge, suggesting a more potent connection between mantle geochemistry and location than previously believed. Finally, the study considers the implications of these findings on our current understanding of mantle convection. Discrepancies in the well-stirred mantle hypothesis are underscored, and recent geodynamic models proposing limited mixing and stirring between the actively upwelling mantle and sub-ridge mantle are examined with the k-means clustering and artificial Neural Network models. The examination of MAR basalts indicates that mantle parcels may have unique and prolonged geologic histories, implying a mantle convection system potentially composed of interconnected subsystems contributing to overall mantle heterogeneity. The research underscores a perspective of Earth's upper mantle that is significantly more heterogeneous and regionally influenced than traditionally understood, suggesting a strong linkage between mantle geochemistry, chemical geography, and geological history, thereby illuminating chemical geodynamics in Earth's mantle.Item Natural Hazards Monitoring Using Multiple Remote Sensing Techniques and Machine Learning Algorithms(2022-12-02) Yu, Xiao; Wang, Guoquan; Khan, Shuhab D.; Sun, Jiajia; Li, Hong-YiNatural hazards often fit into two categories according to the speed and extent of a hazard: rapid onset hazards and slow onset hazards. Both could pose considerable risks to society development, community security, and ecosystems, especially in urban areas with dense populations. Satellite-based remote sensing techniques provide essential information for preventing and mitigating natural hazards. This dissertation has demonstrated both long-term and short-term monitoring methods using multiple remote sensing techniques together with advanced machine learning algorithms. The 2021 Texas winter storm is taken as an example to exemplify the methods of monitoring rapid onset hazards (disasters), and the subsidence in Tianjin, China is used to exemplify the methods of monitoring slow onset natural hazards. In February 2021, an unprecedented winter storm dumped the snow record in Texas. It claimed hundreds of lives and evolved into a major disaster nationalwide. This study uses the differential coherence from Sentinel-1 Synthetic Aperture Radar imagery to characterize the surface disturbance due to this winter storm. Furthermore, machine learning algorithms are applied to quantify Texas statewide snow depth using surface disturbance map, SAR amplitude, precipitation, temperature, surface topography, land cover, and population. The result shows that approximately 90% of Texas accumulated over 30-mm snow depth. A consequent model accuracy of 99% demonstrates that our approach can provide an independent snow depth estimation. Tianjin is one of the large urban regions in China that has suffered from severe land subsidence induced by excessive groundwater withdrawal for approximately half a century. However, since the launch of the South-to-North Water Diversion (SNWD) project, the groundwater withdrawal has significantly been released, thus greatly easing the land subsidence worries here since about 2019. This study quantitively delineated the impacts of the SNWD on land subsidence in Tianjin using Sentinel-1A/B Interferometric Synthetic Aperture Radar (InSAR) (2014-2021), GPS (2010-2021), and groundwater data. The results show that as of 2021, the subsidence area (> 5 mm/year) has reduced by 15% of the subsiding area before SNWD; the areas of rapid subsidence (> 3 cm/year) and extremely rapid subsidence (> 5 cm/year) have reduced to about 70% and 60% of the areas before SNWD, respectively. Besides, the Principal Component Analysis (PCA), an unsupervised machine learning method, is employed to highlight primary factors controlling the recent land subsidence from the large dimension InSAR displacement results. Integrating frontier technologies, i.e., remote sensing and machine-learning algorithms, into existing natural hazards monitoring systems will significantly improve our ability to understand, communicate, and forecast natural hazards and assess risks, thus improving hazard mitigation and disaster management. Two case studies in this study demonstrate the effectiveness of the SAR and machine-learning integrated methods in both slow onset and rapid onset natural hazards monitoring.Item Predicting Magnetization Directions Using Convolutional Neural Networks(2020-05) Nurindrawati, Felicia Disa; Sun, Jiajia; Sager, William W.; Di, HaibinProper interpretation of magnetic data requires an accurate knowledge of total magnetization directions of the source bodies in an area of study. I examined the use of machine learning, specifically Convolutional Neural Network (CNN), to automatically predict the magnetization direction of a magnetic source body, given a magnetic map. CNN has achieved great success in other applications such as computer vision, but has not been attempted in the realm of magnetics. I simulated magnetic data maps with varying magnetization directions from a cubic source body, all subject to the same inducing field. Two CNNs were trained separately, one for predicting magnetization inclinations and the other for predicting magnetization declinations. I also investigated various CNN architectures and determined the optimal architectures for predicting inclinations and declinations. The method works by generating many magnetic data maps with different magnetization directions as the training data for the CNNs. In order to generate these data maps, the user needs to interpret the source body parameters from the data. In this study, I also investigated how different source body parameters can affect the prediction of magnetization directions. My study shows that machine learning holds great promise for automatically predicting magnetization directions based on magnetic data maps. Two methods to increase the accuracy of the predictions are also explored in this study. The first method is to diversify the training data set. The second method is to use U-net, another type of CNN, to interpret the shape and lateral position needed to generate the training set. Both methods have proven well in improving the accuracy of the predictions and automating the process.Item Processing Case Study with AVO and Pore Pressure Analyses: Eugene Island, Gulf of Mexico(2019-12) Kuo, Christine 1987-; Zhou, Hua-Wei; Hilterman, Fred J.; Sun, Jiajia; Verm, RichardSeismic exploration uses numerous tools and techniques for detecting hydrocarbon reservoirs in conventional plays as found on the shelf of the northern Gulf of Mexico (GOM). The tools include seismic data-processing attributes, well-log calibration, and AVO (Amplitude Versus Offset) and pore pressure analyses. While the supporting theory for each method has been well established, an integration of the methods to a real exploration project can be confusing for the uninitiated. Given a seismic dataset and well-log curves, how exactly would someone process and analyze the data using the techniques listed? This research proposes a practical workflow for creating an end product for exploration applications. Using a seismic dataset in the GOM and several well log datasets, an integration of the processed seismic image with pore pressure analysis, petrophysical analysis and calibrated seismic attributes derived from the petrophysical analysis indicates the area would be classified as a Class III AVO environment which is supported by the geologic background. The dataset is further processed to completion and then used to generate AVO and pore pressure fields. Available well logs in the area are then compared and tied to the seismic, in order to validate the presence of any potential hydrocarbon attributes. An abnormal overpressure zone is determined by seismic velocity analysis and confirmed by the structural continuity of reflections, which was bound within a fault block. The seismic attributes indicated a potential gas reservoir on the downthrown side of the fault block, which would correlate with the proximity of the abnormal pressure. The exploration tools of seismic data analysis, well log analysis, AVO analysis, and pore pressure analysis are found to be strongly correlated by this study.Item Regional Scale Mineral Exploration Through Joint Inversion And Geology Differentiation Based On Multi-physics Geoscientific Data(2020-05) Kim, Jae Deok; Sun, Jiajia; Melo, Aline Tavares; Khan, Shuhab D.Modern mineral exploration focuses in underexplored regions where the terrain and geological environments make the discovery of mineral ore deposits increasingly difficult. Airborne geophysics is widely used in regional scale mineral exploration because it provides rapid collection of multiple types of geoscientific data over large areas. The availability of multi-physics data is potentially useful because the complementary information contained in the multiple data sets can be integrated into a common Earth model consistent with all available data and prior information. However, quantitative integration of multi-physics and regional scale airborne geophysical data is rarely reported in literature. The goal of this research is to develop a workflow for quantitative integration of airborne gravity and magnetic data for mineral exploration. I focus on two important components of the workflow: joint inversion and geology differentiation. Joint inversion allows density and susceptibility models to constrain each other at the inversion stage, resulting in structurally similar physical property models and enhanced correlations between inverted density and susceptibility values. Geology differentiation makes use of the jointly inverted physical property values and builds a 3D quasi-geology model that shows the spatial distribution of various geological units. Prior geological information from various sources are also used when performing geology differentiation. The proposed workflow is first tested on a synthetic data set before being applied to a set of airborne gravity and magnetic data in central British Columbia. I have successfully identified multiple geological units that are consistent with airborne geophysical data and prior geological information. I have also identified potential targets for future detailed geophysical surveys in an area that lies beneath a thick glacial sedimentary cover. My work provides guidance for follow-up detailed geophysical surveys in the study area and highlights the benefits of integrated interpretation of multi-physics geoscientific data. I am confident that the proposed workflow can be easily extended to the integration of other types of geoscientific data.Item Surface Deformation Analysis of the Houston Area: Investigating Contributions of Faults, Salt Domes, and Major Storms(2019-08) Crupa, Wanda Elena 1994-; Khan, Shuhab D.; Suppe, John; Sun, Jiajia; Cao, NingThe Houston area has undergone significant ground deformation in the last century, with the main factor being attributed to groundwater/natural gas withdrawal. However, subsidence can be due to groundwater withdrawal or excess loading brought about by heavy precipitation. Houston has recently been subjected to multiple flooding events which appear to be increasing in frequency. The Houston area is also home to faults and salt domes that contribute to surface deformation. The effect that these factors have had on ground deformation has not previously been studied; certain components of ground motion have been misinterpreted, or largely ignored in scientific studies and when making policies. In this study we investigate the contributions of surface and groundwater to subsidence using data collected over the past 30 years to model/predict groundwater fluctuations and look at the correlation with faults/salt domes and GPS data to see how surface deformation patterns have changed in recent years. The high rate of salt motion coupled with CO2 injection has resulted in uplift in southern Harris County, which acts to alleviate groundwater/gas withdrawal induced subsidence. Observed fault motion along the Long Point-Eureka Heights system is correlated with groundwater trends from 2006-20107. The northern Houston area shows strong subsidence of up to 16 mm/yr and an elongated subsidence bowl. The weakened aquifer systems in the north and southwest are more susceptible to intense subsidence and major flooding. These trends may be matched in the Woodlands and southwest Harris County in the future.Item Tectonic Evolution of the Late Cretaceous-Cenozoic Lesser Antilles Subduction System and the Mesozoic-Recent Rifted-Passive Margin of Northern Morocco Using an Integration of Geologic, Seismic Reflection, and Gravity Data(2021-12) Miller, Benjamin Donald; Mann, Paul; Rotzien, Jonathan R.; Sun, JiajiaThe largely submarine, 850-km-long, late Cretaceous to recent Lesser Antilles subduction system consists of the Aves Ridge remnant arc, the Grenada back-arc basin, the Lesser Antilles volcanic arc (LAVA), the Tobago forearc basin, and the Barbados accretionary prism. Differing tectonic models have been proposed to explain the origin of these five main components of the Lesser Antilles magmatic terranes and their intervening sedimentary basins. Chapter 2 of this thesis tests three of these differing tectonic models using three dip-direction and one strike-direction gravity transect. Based on the integration of the crustal thickness information from the gravity transects and a compilation of radiometric dates, the tectonic model that best fits the gravity profiles and dates include: 1) an eastward shift about 190 km of the volcanic arc from the late Cretaceous Aves Ridge to its present-day location of the LAVA during the Oligocene; this eastward shift was likely a response to slab rollback; and 2) a westward shift of 180 km from the inactive Outer Arc to the presently active Inner Arc (13-5 Ma); this westward shift was likely a response to the subduction of elevated areas of fracture zones in the Atlantic oceanic. Chapter 3 of this thesis describes the 950-km-long, Jurassic Agadir-Essaouira-Rharb marginal rift underlies the Mesozoic-Recent, rifted-passive margin of northwestern Morocco. Seismic data integrated with gravity models show that the 8-15 km-thick necked zone of rifted continental crust is overlain by a single marginal rift that concentrates the thickest (1-5 km) area of Jurassic salt. The four 170-280-km-long, dip 2D gravity models created along the coast of northern Morocco were constrained by several previous refraction studies and delineate three crustal provinces of the rifted-passive margin: 1) early Jurassic oceanic crust of the deepwater area of the Atlantic Ocean; 2) thinned continental crust that includes the marginal rift on the slope and outer shelf; and 3) full-thickness and unstretched continental crust of the shelf and land areas of Morocco. The integration of the gravity models with the seismic reflection data helped constrain the width and depth of the marginal rift, which was not well imaged from seismic reflection data.Item The Challenges and Solutions of Green Seismic Survey: Imaging Deep Seafloor Structures with Cruise TN-373 Reflection Data(2023-08) Li, Zhehao; Zhou, Hua-Wei; Sager, William W.; Sun, Jiajia; Gonzalez, AlfonsoSeismic surveys in the past can be viewed as a trade-off between data quality and survey cost. As technology advances, the gap between these two demands has lessened. Nowadays, a new factor is emerging as more regulations are aimed at curbing the environmental impact of seismic surveys. I define “green” seismic surveys as those that employ natural or low-energy seismic sources to minimize the environmental impact on the surrounding area. However, lower energy can drastically reduce the quality of the collected seismic data. This dissertation aims to develop strategies that optimize the efficiency of future acquisition by properly balancing three factors: quality, cost, and environmental impact. Chapter 2 introduces the seismic dataset from Cruise TN-373, an NSF project that utilized small airgun sources to image deep seafloor structures. The research vessel sailed across the South Atlantic Ocean from November through December of 2019 to collect more than 3000 km of 2D reflection seismic lines. The dataset modernizes existing seismic data over the area to expand our understanding of the Walvis Ridge formation, as well as to assist in the selection of the ocean floor drilling sites for the IODP Expedition 391. Chapter 3 reviews the challenges and opportunities of future acquisition using small sources by using TN-373 as a case study. I recommend ways to optimize survey parameters for future opportunities of integrating green seismic surveys into conventional surveys. I also summarize key challenges to data fidelity including image resolution, image artifacts, and uncertainty in image position. Chapter 4 documents the processing flows of TN-373, focusing on pre-processing, deconvolution, velocity model building, and migration algorithms for imaging deep marine sediments. Common marine noises are attenuated by classifying them based on origin, coherency, and frequency to increase the signal-to-noise ratio (SNR). To improve resolution by deconvolution, I compared the far-field source wavelet derived using three common methods and discovered that accurately estimating a far-field source wavelet relies on the directivity and SNR of the data. I further assess the uncertainties in our velocity models and identify commonly observed artifacts. In Chapter 5, I study the state-of-the-art methods in the separation of simultaneous source data. The lack of amplitude in low-energy sources can be increased by raising source density and therefore increasing data quality. I demonstrate a neural network approach to effectively reduce unwanted crosstalk noise and show potential for improved efficiency and better separation than existing algorithms. This dissertation offers insights into optimizing the efficiency and quality of future acquisitions while minimizing their environmental impact. The findings and recommendations can guide prospective seismic surveys to becoming more sustainable and responsible through the ongoing energy transition.Item The Correlation between Current Land Subsidence and Groundwater Levels in Montgomery County, Texas(2020-08) Zhou, Fanbo; Wang, Guoquan; Sun, Jiajia; Mayerich, DavidHouston, a city in Texas, has been affected by land subsidence for nearly a century. Houston and its surrounding areas have been suffering damage associated with faulting and land subsidence severely. In recent years, due to the implemented groundwater regulations, the overall land subsidence rate in the southeastern regions of the Houston metro area has been decreased. However, moderate subsidence (1 to 2 cm /year) is still ongoing in the northern and northwestern regions. Montgomery County is one of the northern areas suffering from infrastructure damages affected by land subsidence. Land subsidence in this region is caused by the overuse of groundwater in Jasper and Evangeline aquifers. This study investigated the vertical ground deformations that are recorded by the Global Positioning System (GPS) and groundwater levels data within Montgomery County. The U.S. Geological Survey (USGS), Lone Star Groundwater Conservation District (LS- GCD), Harris Galveston Subsidence District (HGSD), The University of Houston (UH), and other local agents have been continuously monitoring groundwater and land subsidence for over ten years. By processing USGS groundwater monitoring data and permanent GPS data, this study researched on the correlation between water levels of groundwater and land subsidence. GPS stations recorded land subsidence rates and provided basic information for predicting future land subsidence, which is crucial to better adjust groundwater regulations in the future.