Dissolved Organic Carbon Dynamics in a Salt Marsh Creek

Codden, Christina ProQuest Dissertations & Theses Northeastern Unive 2020 해외박사(DDOD)

Salt marshes are blue carbon systems that sequester carbon at higher rates than many terrestrial ecosystems due to a coupled relationship between high primary production and slow decomposition in anaerobic sediments. Annually, this coupled relationship allows for over 10 Tg of organic carbon to be sequestered in global salt marsh sediments alone, or a storage equivalent of over 55,000 Blue Whales per year. In turn, this storage ability enables salt marshes to help mitigate increasing atmospheric CO2. Despite high primary production in salt marshes and their ability to help mitigate increasing atmospheric CO2, a long-standing question remains in coastal carbon cycling and ecology which asks: Is a fraction of salt marsh produced carbon, prior to sequestration or mineralization, exported (i.e., outwelled) as dissolved organic carbon (DOC) to the coastal ocean? Answering this question of salt marsh DOC outwelling is critical for quantifying the significance of salt marsh carbon outwelling in comparison to total salt marsh carbon storage, total salt marsh primary production, and broader coastal carbon cycling.Because the question of DOC outwelling first arose on the Georgia coast and because the Georgia coast houses some of the most productive salt marshes in the world, this dissertation focuses on analyzing DOC outwelling in Groves Creek, a tidally-driven salt marsh creek on the Georgia coast. Groves Creek was additionally chosen as it lacks a freshwater head and has limited freshwater input, making the analysis of marsh-only DOC fluxes through the estuarine water possible without confounding results from terrestrial DOC input. In Groves Creek and other Georgia salt marsh creeks, DOC is a master variable that controls the light field, initiates photochemical reactions, and provides sustenance to microbes. The dynamics of DOC in these systems are complex as multiple DOC sources, sinks, and patterns of mixing occur. The complexity in salt marsh DOC dynamics plus the failure of past studies to capture export trends in marsh-derived DOC at both high-temporal resolution and across seasons may explain why it remains unclear whether salt marshes generally export DOC (i.e., outwell).Thus, at a Groves Creek study station, this dissertation sought to answer the question of salt marsh DOC outwelling over three research captures. At Groves Creek study station, Chapter 1 captured hydrology (water level, velocity, flow) at 10-minute resolution over 16-months using an in situ Acoustic Doppler Profiler (ADP) deployed in the creek bed over 7 deployments. After data collection, the hydrology record indicated that the ADP instrument was not deployed in precisely the same location of the creek bed for all deployments. Thus, to make hydrology comparable over the entire study, hydrology records required alignment using a novel alignment approach in which non-tidal signals from individual ADP deployments were added to an extrapolated tidal signal based upon three already aligned deployments.Chapter 2 went on to assess DOC concentration at Groves Creek study station at the same temporal resolution and study length as Chapter 1. As no in situ instrument exists that could directly measure DOC concentration, DOC was estimated in Chapter 2 through the use of site-specific machine learning and linear algorithms coupled with optical and other low-to-zero cost predictors (e.g., water level, salinity, local rainfall) collected at high-temporal resolution. Models were trained using 306 discrete lab-based DOC measurements collected as water samples from the study station. These discrete samples served as ground truth. Work from Chapter 2 included the first-ever incorporation of non-linear machine learning to estimate DOC concentration. By combining DOC concentration (Chapter 2) with water flux (Chapter 1), plus measured salinity (Chapter 3), Chapter 3 was able to calculate DOC fluxes at Groves Creek and ultimately assess the long-standing and inconclusive topic of DOC outwelling. Chapter 3 provided the first-ever estimation of both high-temporal (10-minute) and cross-seasonally (16-month) resolved DOC fluxes.Results show Groves Creek is hydrologically complex with ebb-dominated tidal asymmetry and often more water flowing into the main channel than out (Chapter 1). Since the marsh is hydrologically balanced overall, net imported water likely drained the marsh via unsampled flow paths (e.g., smaller channels, overmarsh flow at marsh edge). Concerning DOC estimation (Chapter 2), at seasonal timescales, machine learning (mean absolute error (MAE) 3.7%) modestly improved upon the accuracy of linear methods (MAE 6.5%) but offered substantial instrumentation cost reductions (~90%) by requiring only cost-free predictors (online data) or cost-free predictors in combination with low-cost in situ predictors (temperature, salinity, depth). At intratidal timescales, linear methods proved ill-equipped (median Pearson’s correlation coefficient (R) 0.55) to predict DOC concentration compared to machine learning (median R 0.87–0.94), and again machine learning offered a substantial instrumentation cost reduction (~90%). Thus, one of the main advances set forth in this dissertation is a novel, improved accuracy, and lower-cost method to estimate DOC concentrations in complex aquatic ecosystems. The results of this portion of the dissertation, as presented in Chapter 2, are under a second round of review at Limnology and Oceanography: Methods.Chapter 3 marks the culmination of my PhD research by combining hydrologic fluxes (Chapter 1) and DOC estimates from the two top-performing machine learning algorithms (Chapter 2) to estimate net DOC fluxes through Groves Creek and test the hypothesis that salt marshes outwell DOC (Chapter 3). DOC flux results show that cumulative net DOC-flow and DOC-salt relationships were largely conservative, indicating DOC outwelling was not supported over most of the study period at the Groves Creek study station. However, during summer 2014, the conserved DOC-flow and DOC-salt relationships were disturbed with a loss of DOC from the marsh relative to salt and water fluxes. This discursion from conservative behavior marked a short-lived period of DOC outwelling from the marsh creek to the estuary in summer 2014 during which an estimated 5.7 to 42.1 tons of DOC were exported. Although this is a modest carbon flux, the outwelled DOC remains a significant net term in the marsh carbon budget (e.g., up to 12% of the annual organic carbon sequestration in Groves Creek salt marsh) and an important process to capture in mechanistic models of long-term carbon production, export, and storage for marshes and other blue carbon ecosystems. Results also indicate DOC outwelling from salt marshes may occur as a pulse during highly productive summer months. Resolving these hot moments of DOC export at high-temporal resolution across larger salt marsh ecosystems is required to assess the true extent and quantitative significance of DOC outwelling to coastal carbon cycles, coastal ecology, and the carbon budgets of salt marshes.

Structural Damage Characterization with Special Attention to Under-Constrained Problems

Memarzadeh, Esmaeil ProQuest Dissertations & Theses Northeastern Unive 2022 해외박사(DDOD)

Structures can get damaged by severe events such as earthquakes and hurricanes or deteriorate over time. Therefore the need to find cost-effective and reliable inspection and monitoring solutions for structures such as bridges, wind turbines, and buildings is important. Structural Health Monitoring (SHM) is the process of using damage detection and characterization techniques to determine whether a structure is in a healthy state or a damaged state.Damage localization and quantification, collectively referred to as damage characterization, can be addressed as a parameter estimation problem. In this setting, the location and extent of damage are inferred from the model parameters that are estimated from features extracted from the measurements. The measurements are collected from the sensors. For success, the features from the measurements must be sensitive to the parameters and have low variability to non-damage-related changes. Eigenvalues can be measured more precisely than eigenvectors and, for this reason, are widely used as features for damage characterization.An issue in using eigenvalues only for parameter estimation is that the number of eigenvalues extracted from the measurements can often be less than the number of model parameters that are candidates for updating, making the problem under-determined. While the number of candidate parameters is large, one expects that only a few will change due to damage. A solution that has only a few non-zeros is called sparse. Sparsity has been added as a constraint to the under-determined parameter estimation problem to obtain a solution that will likely be aligned with what happens in reality. Sparsity has been exploited in the last ten years or so using a linearized approximation. In this dissertation, the error resulting from the linear approximation is examined, and approaches that consider the nonlinearity are presented.Feature selection for parameter estimation is another item that is treated in this dissertation. Output feedback control has been used to increase the sensitivity of the eigenvalues to parameters in the last twenty years. Gains have been designed to obtain closed-loop systems with eigenvalues that have more sensitivity to damage. Applying output feedback to a structure requires that the structure is equipped with controllers, which can be a limitation. Virtual output feedback, however, only requires measuring the open-loop input-output data. The closed-loop matrix is formed offline after system identification. Virtual output feedback can be used for feature selection. It is shown that replacing the open-loop eigenvalues with more sensitive closed-loop eigenvalues will also increase their variability. Notwithstanding, Virtual output feedback can still be used to use multiple sets of closed-loop eigenvalues instead of open-loop eigenvectors. This is shown to provide better conditioning in the parameter estimation problem and more robustness to noise.This dissertation presents a damage detection method based on nonlinear output feedback as the last item. Unlike the virtual output feedback, hardware and controllers are required to perform the tests in this approach. The objective is to announce whether the structure is damaged or non-damaged by observing a feature from the nonlinear system. Using nonlinear output feedback in a linear system will generate a nonlinear closed-loop system. Nonlinear systems have features that do not exist in linear systems. We used the period of a Limit Cycle (LC) for damage detection. The limit cycle is obtained by applying the nonlinear feedback law at the point of actuation in the structure. The sensitivity of the period of the limit cycle is orders of magnitude larger than the change of period in the open-loop setting while showing robustness to non-damaged related variabilities such as noise, environmental changes, and model error.

Interoperability through Realizability: Expressing High-Level Abstractions Using Low-Level Code

Patterson, Daniel Baker ProQuest Dissertations & Theses Northeastern Unive 2022 해외박사(DDOD)

Large software systems are, inevitably, multi-lingual. This arises for complex socio-historical reasons, as large systems persist for years or decades, while the people working on them and the languages, libraries, and tools available to them change. Looking to these systems, I identify the interoperability challenge: that it is more difficult for programmers to reason about multi-lingual systems than about single-language programs. A corollary is that many of the key theorems about languages are proven in the absence of interoperability, reality notwithstanding.In this dissertation, I identify realizability models as a key tool for addressing the interoperability challenge. Realizability models, which use target-level behavior to inhabit source types, allow the behavior of disparate source languages to be brought together. In doing so, we can recover the type of formal language-based reasoning critical to proving universal properties upon which programmers rely. In this dissertation, the property on which we focus is type soundness, which we explore through a variety of case studies and via two different interoperability mechanisms. The first mechanism, which models how typical foreign-function interfaces work, allows foreign values to be imported at existing types. Realizability models are used to demonstrate the soundness of the conversions that happen at the boundaries. The second mechanism, which better models how programmers wish interoperation worked, allows foreign code to be imported at novel types, thus allowing new behavior to be brought in. Even as the source-level mechanism is quite different between these two approaches, the underlying realizability models are similar, underscoring the central thesis: that such realizability models are an effective way of reasoning about cross-language interoperation.

Computationally Efficient PMU-based L1 Estimators for Large Power Systems

Xu, Chenxi ProQuest Dissertations & Theses Northeastern Unive 2018 해외박사(DDOD)

Phasor Measurement Units (PMUs) are increasingly deployed in power systems because of their nice characteristics like fast data acquisition rate and GPS clock synchronization. With the explicit usage of PMU measurements, Least Absolute Value (LA. The first part of this dissertation presents two centralized LAV SEs incorporating Zero Injection (ZI) measurements into the LAV state estimation formulation using direct enforcement and Kron reduction, respectively. Based on the current circumstance that VLSI power grids are usually divided into several independent and non-overlapping zones, the second part of this dissertation presents several multi-area distributed LAV SEs. The first algorithm combines a well-known LP decomposition method: Dantzig-Wolfe (DW) decomposition with the LAV SE considering the motivation that LAV can be formulated as an LP problem and multi-area state estimation measurement matrix has the. The second algorithm uses a two-stage set-up to assure adequate robustness around zone boundaries. All zones run their own SEs and the estimated boundary bus states, together with measurements between zones, are both used as measurements for the. The third algorithm generates one or several additional zones covering all boundary buses and their direct neighbors. This new zone and all existing zones run their SEs simultaneously in parallel. Results are collected and reconciled to provide. The fourth algorithm creates one or several "copies" of the system. Each copy contains one way of system zone partitioning. All buses appear at least once as an internal bus in these copies. All zones in all copies run independent SEs. An algori. The above multi-copy algorithm is implemented and further tested on a high-performance multi-core computer using parallel processing. Above algorithms are implemented on different test systems with sizes ranging from 30-bus to 16216-bus and the corresponding simulation results are presented in this dissertation.

Coupling Methods for Wireless Intra-Body Communication and Sensing

Banou, Stella ProQuest Dissertations & Theses Northeastern Unive 2022 해외박사(DDOD)

Advances in miniaturized bio-compatible Internet of Things (IoT) device design and wireless connectivity have resulted in rapid strides towards realizing the vision of connected health and ubiquitous monitoring of physiological conditions. Core enablers of this capability are wearable and implanted IoT devices, albeit with limitations arising from their low energy storage and computational power. This thesis goes beyond the RF-only communication standards by exploring alternate communication modalities that are more amenable for inter- and intra-body communication. In summary, this thesis explores the conductive and radiating nature of the human body as a channel for three non-RF coupling communication methods - Galvanic, Magnetic and Capacitive coupling.An implementation of Galvanic Coupling-based beamforming is presented for implant to wearable communication. The key idea here is to exploit the conductivity of human tissue and transmit weak electrical signals by coupling them via electrodes to muscle tissue in a way that concentrates energy at the receiver location. Following that, we focus on realizing a relay network of IoT devices for both implant-implant and implant to on-skin sensor communication using Magnetic Resonance Coupling. The advantage of this method over Galvanic Coupling is that the former reduces attenuation when signals pass through human tissue. This thesis enhances the scope of the connected health paradigm to now include sensing for proximity and for automated encouraging of healthy habits that mitigate the spread of communicable diseases using Capacitive Coupling.Finally, a novel human antenna field to sense and communicate with wearable and implantable IoT devices in the near field is designed. This thesis proposes the utilization of a capacitively induced human body antenna field to connect multiple wearable and implantable devices to each other via Intra-body Communication. Such a paradigm not only supports multi-cast data transfer but also helps in sensing the environment to detect the presence of other patients and caregivers, unlike current RF-only approaches. The human body antenna approach is validated via simulation and experiments on a phantom to study the frequency response, sensitivity and antenna design parameters of the resulting human body antenna system. Results reveal a communication and sensing range of 2.5 meters while operating at 75-95 MHz with a single wearable device with capacitive electrodes of size 30x50 mm. To ensure safety in such capacitively induced human body antenna for Wireless Body Area Network (WBAN) communications, we perform a power budget analysis to prove that the proposed approach is indeed a low-power and robust communication and sensing system for health monitoring purposes. The final part of this dissertation completes the full cycle of data flow, from implanted to wearable devices and finally connects the body network to the computational cloud for the next generation of IoT-enabled healthcare.

Bridging the Gap Between Theory and Practice Through Data- and Task-Based Visualization Recommendation Systems

Pandey, Aditeya ProQuest Dissertations & Theses Northeastern Unive 2022 해외박사(DDOD)

Data visualization plays a critical role in data analysis and information dissemination. An essential step in the data visualization pipeline is mapping data and task requirements to appropriate visualization techniques. However, this step is prone to errors because it requires visualization creators to be familiar with a substantial body of visualization design guidelines and best practices. This dissertation aims to mitigate the challenges associated with the visualization design process by building systems and curating design guidelines that can support visualization practitioners in choosing appropriate visualization techniques based on their data and task requirements.This dissertation is composed of four parts. The first part of the dissertation presents two visualization design studies: CerebroVis and Portola. These studies led to the creation of tree and network visualizations that solve critical domain problems in medical diagnosis and cybersecurity. Through these studies, the dissertation motivates the need for resources to help visualization creators in mapping data and task requirements to visual encodings. The second part of the dissertation contributes a task abstraction framework for tree visualizations which supports visualization designers to more specifically abstract the goals of their users and better understand their data analysis needs. The framework also enables visualization researchers to systematically and exhaustively curate task-based design guidelines for tree visualizations. The third part of the dissertation contributes visualization design guidelines for glyph visualizations, timelines, and tree visualization. In addition to the guidelines, this part also discusses the challenges of curating design guidelines from existing empirical research and potential ways to overcome the challenges. Finally, the last part of the dissertation presents three data- and task-based visualization recommendation systems which put the theory into practice: GenoREC, NESTED, and MEDLEY. GenoREC is a domain-specific visualization recommendation system designed to support genomics analysts. GenoREC uses a knowledge-based approach to recommend domain-specific genomics visualization based on the common data formats and analysis tasks in genomics. NESTED supports visualization creators by recommending the appropriate tree visualization technique, corresponding interaction, and supporting widgets to analyze hierarchical data. Finally, MEDLEY presents a mixed-initiative interface that assists in dashboard composition by recommending dashboard collections (i.e., a logically grouped set of views and filtering widgets) that map to specific analytical intents. Contributions of this thesis pave the way for future work that can extend information visualization theory and bridge the gap between theory and practice by providing users with visualization recommendation tools and systems.

Sensing and Imaging Using Multi-dimensionality Coded Compressive Material

Molaei, Ali ProQuest Dissertations & Theses Northeastern Unive 2019 해외박사(DDOD)

Reducing the cost of electromagnetic sensing and imaging systems is a necessity before they can be far and widely established as a part of an extensive network of radars. Conventional imaging systems seek to reconstruct a target in the imaging domain by employing many transmitting and receiving antenna elements. These systems are suboptimal, due to the often large mutual information existing between successive measurements. This thesis describes a new sensing technique, which is based on the use of a novel compressive reflector antenna (CRA), that is capable of providing high sensing capacity in different sensing applications. The high sensing capacity provided by the CRA enhances the information transfer efficiency from the sensing system to the imaging domain and vice versa. Thus, complexity and cost of the hardware architecture can be drastically reduced. The CRA generates spatial codes in the imaging domain, which are dynamically changed through the excitation of Multiple-Input-Multiple-Output (MIMO) feeding arrays. In order to increase the sensing capacity of the CRA even further, frequency dispersive metamaterials can be designed to coat the surface of the CRA, which ultimately produce spectral codes in near- and far- field of the reflector. This thesis describes different concepts of operation, in which a multi-dimensionality coded compressive system can be used to perform sensing and imaging. The proposed sensing technique, which may be used for imaging applications, is based on norm-1 regularized iterative Compressive Sensing (CS) algorithms. In this thesis, we also present the mathematical formulation that describes the properties of the spatial and spectral codes produced by the CRA that will be used to perform quasi-real-time imaging. The content outlined in this thesis leverages advances from multi-scale wave propagation, sparse data signal processing, information coding and distributed computing. The result will enhance the efficiency and reliability of the current beamforming systems by using novel Compressive Sensors made of traditional metallic and dielectric structures, as well as novel metamaterials and metasurfaces.

Predictability and Inequality in Complex Social Systems

Wang, Xindi ProQuest Dissertations & Theses Northeastern Unive 2021 해외박사(DDOD)

Centralization in Blockchains: Causes and Mitigations

Kiffer, Lucianna Carvalhaes ProQuest Dissertations & Theses Northeastern Unive 2022 해외박사(DDOD)

The celebrated Nakamoto consensus protocol [3], introduced in 2008, has ushered in several new consensus applications, most popularly cryptocurrencies like Bitcoin and Ethereum. There has since been a spark in this new area of study in both academia and industry, including many new systems which are now part of a multi-billion dollar industry [4]. At their heart, these protocols implement a public and immutable record of transactions known as the blockchain. The main promise and appeal of such blockchain systems is decentralized governance in an open and distributed network. I present this thesis as a comprehensive study of whether or not these systems are living up to that promise.In the first part of this thesis, I focus on one of these systems: Ethereum, the second largest cryptocurrency by market capitalization [4]. I present several measurement studies, focusing on several aspects of the network and its history. My work covers how protocol changes have historically impacted the network, how users are utilizing the blockchain in their transactions, and how the peer-to-peer network responsible for disseminating the messages in the network is operating. Across all of these studies, I observe centralizing behaviors in the network linked to peer behavior, barriers of entry into the system, and miner incentives.Based on these observations, in the second part of this thesis I ask the following questions: (1) What are the conditions under which blockchain consistency can be maintained? (2) Are there protocols that will enable clients to join the system in a lightweight manner without giving up trust? And (3) What role does the mining reward function play in incentivizing more decentralized miner participation? I address these foundational questions through rigorous theoretical analyses of existing protocols, and by developing new protocols with provable guarantees. My results include a new Markov framework for analyzing consistency guarantees for several blockchain models, a lightweight transaction verification protocol, and a class of mining reward functions that promote more decentralized miner participation.In summary, the main contributions of this thesis are extensive measurement studies of blockchain systems that expose certain security vulnerabilities and centralizing factors, and the development of analytical tools and protocols to mitigate them.

Can We Trust AI? Towards Practical Implementation and Theoretical Analysis in Trustworthy Machine Learning

Xu, Kaidi ProQuest Dissertations & Theses Northeastern Unive 2021 해외박사(DDOD)