19304 research outputs found
Sort by
Advanced Deep Learning Approaches for Forecasting High-Resolution Fire Weather Index (FWI) over CONUS: Integration of GNN-LSTM, GNN-TCNN, and GNN-DeepAR
No full textWildfires in the United States have increased in frequency and severity over recent decades, driven by climate change, altered weather patterns, and accumulated flammable materials. Accurately forecasting the Fire Weather Index (FWI) is crucial for mitigating wildfire risks and protecting ecosystems, human health, and infrastructure. This study analyzed FWI trends across the Continental United States (CONUS) from 2014 to 2023, using meteorological data from the gridMET dataset. Key variables, including temperature, relative humidity, wind speed, and precipitation, were utilized to calculate the FWI at a fine spatial resolution of 4 km, ensuring the precise identification of wildfire-prone areas. Based on this, our study developed a hybrid modeling framework to forecast FWI over a 14-day horizon, integrating Graph Neural Networks (GNNs) with Temporal Convolutional Neural Networks (TCNNs), Long Short-Term Memory (LSTM), and Deep Autoregressive Networks (DeepAR). The models were evaluated using the Index of Agreement (IOA) and root mean squared error (RMSE). The results revealed that the Southwest and West regions of CONUS consistently exhibited the highest mean FWI values, with the summer months demonstrating the greatest variability across all climatic zones. In terms of model performance on forecasting, Day 1 results highlighted the superior performance of the GNN-TCNN model, achieving an IOA of 0.95 and an RMSE of 1.21, compared to the GNN-LSTM (IOA: 0.93, RMSE: 1.25) and GNN-DeepAR (IOA: 0.92, RMSE: 1.30). On average, across all 14 days, the GNN-TCNN outperformed others with a mean IOA of 0.885 and an RMSE of 1.325, followed by the GNN-LSTM (IOA: 0.852, RMSE: 1.590) and GNN-DeepAR (IOA: 0.8225, RMSE: 1.755). The GNN-TCNN demonstrated robust accuracy across short-term (days 1–7) and long-term (days 8–14) forecasts. This study advances wildfire risk assessment by combining descriptive analysis with hybrid modeling, offering a scalable and robust framework for FWI forecasting and proactive wildfire management amidst a changing climate
Optimization of Salicylic Acid Biosynthesis Through a Growth-Coupled Metabolic Driving Force
No full textSalicylic acid (SA) is an industrially significant compound acting as an active ingredient in skincare, a precursor to anti-inflammatory drugs such as Aspirin, and an important plant hormone. Current production methodologies rely on chemical synthesis that utilize non-renewable feedstock, necessitating the development of sustainable alternatives. Microbial biosynthesis offers a promising solution, though achieving commercially viable yields remains a challenge due to metabolic constraints and genetic instability. To address these limitations, product synthesis can be coupled with cellular viability, which enforces an evolutionary pressure that enhances product formation. This study demonstrates the implementation of a pyruvate-driven metabolic driving force in Escherichia coli for enhanced SA biosynthesis, utilizing either glucose or glycerol as renewable carbon sources. The release of pyruvate as a byproduct of SA biosynthesis addresses the endogenous pyruvate requirements of pyruvate-deficient strains, restoring cellular homeostasis. Through the optimization of carbon flux, growth conditions, and precursor availability, we achieved SA titers of 2991.8 ± 58.1 g/L and 3088.5 ± 42.2 g/L in shake flask cultures utilizing glucose and glycerol, respectively
Microstructure Engineering of All-Solid-State Composite Electrodes
No full textDriven by the immense global demand for energy, the development of advanced energy storage systems is critical. Solid-state batteries (SSBs) have emerged as promising alternatives to traditional lithium-ion batteries for next-generation energy storage due to their enhanced safety and higher energy densities. Composite cathodes are pivotal in determining the areal capacity and specific energy of SSBs, making their design, fabrication and characterization an important research direction. This dissertation focuses on electrode microstructure engineering and the development of novel organic active materials for high-performance SSBs. It begins with a review of key challenges associated with composite cathodes and current solutions. Chapter 2 explores strategies using organic cathodes to address these challenges from microstructural, mechanical, and chemical perspectives, along with a discussion on mechanical failures in conversion cathodes and potential optimization strategies. Chapter 3 and Chapter 4 focus on constructing favorable microstructures in organic cathodes. We identified that the unfavorable organic cathode microstructure is attributed to the mechanical mismatch between the soft organic compound and the relatively hard solid electrolyte. By employing solvent-assisted processing approach or manipulating the hardness of sulfide-based solid electrolyte, we successfully transformed the microstructure from “electrolyte-in-active material” to “active material-in-electrolyte”, thereby improving both cathode fraction and electrode-level energy density. Chapter 5 and Chapter 6 introduce novel conductive organic materials for solid-state batteries. First, we demonstrate that high malleable organic materials can accommodate mechanical stress from active material volume change during cycling. The intimate interfacial contact between organic material and solid electrolyte enables operation under low stack pressure. Second, we examine the physical properties of lithium-containing organic materials, revealing that the reduced form of organic material exhibits improved chemical stability when combined with sulfide electrolyte. This highly reversible interface within the cathode ensures long-term cycling stability. Finally, I address mechanical failure in sulfur-based conversion cathodes in Chapter 7, where delamination at the cathode-solid electrolyte interface leads to significant polarization during cycling at low stacking pressure. To mitigate this issue, a strategy using FeS2 particles coated with organic materials is proposed to enhance cycling stability
Development of Multi-modal Optical Coherence Tomography Imaging Systems and Probes
No full textOptical Coherence Tomography (OCT) is a powerful imaging modality widely used in medical diagnostics and biological research for its non-invasive, high-resolution, and depth-resolved properties. Additionally, OCT has demonstrated remarkable versatility in various applications. In this thesis, we develop multi-modal OCT systems. We have designed and built a new class of endoscopic OCT that enables imaging inside the human body and a new type of functional OCT that provides information beyond structural characteristics. We combine the chemical specificity of mid-infrared spectroscopy with the morphological capabilities of OCT systems, potentially opening a new class of biomedical applications. First, we develop a swept-source OCT (SS-OCT) system that serves as a benchtop tool and a scalable platform for further functional extension. A conventional galvanometer-scanner-based fiber optics SS-OCT system has been developed with an axial resolution of 7.24 μm at 1310 nm, achieving max imaging depth of 5.98 mm in the air. The A-scan rate is 100 kHz defined by the swept-source and lateral resolution of 24.8 μm defined by the scanning objective. The system has demonstrated a high sensitivity of 96.88 dB and low sensitivity roll-off of 0.6 dB over 5.5 mm imaging depth. We demonstrate the utility of SS-OCT in endomicroscopic applications by developing forward and side-viewing probes. Both classes of probes were designed with cost-efficient scanners while maintaining high performance. The side-viewing probes use a 6 mm diameter micro stepper motor and a custom Gradient Index (GRIN) achieving 26 μm lateral resolution and real-time imaging with a frame rate of 100 fps. The forward-viewing probes are built using a piezoelectric (PTZ) cantilever scanner and a micro lens, enabling a 4 mm diameter probe size while achieving a large field of view (2.25 mm2). Finally, we have developed a novel photothermal mid-infrared spectroscopic imaging (MIRSI) OCT technology. Unlike prior approaches, this new form of functional OCT can provide label-free chemical contrast. The system utilizes a pulsed mid-infrared laser to introduce a modulated photothermal signal into the OCT. This technology combines the three-dimensional (3D) morphological imaging capabilities of OCT with the molecular sensitivity of MIRSI through endogenous contrast without adding nanoparticles of contrast agents. Our approach can obtain co-registered OCT and photothermal MIRSI images with a lateral resolution of 21 μm and 15.6 μm, respectively. The 25 and 45 μm polystyrene (PS) beads and biological samples, including mouse kidneys and mouse brains, are used to demonstrate spatial and spectral imaging fidelity
Foundational Concepts in Cardiology, an Introduction to Pressure Volume Loops and the Cardiovascular System
No full textThis video introduces fundamental concepts in cardiology to medical and other health science students.Clinical Sciences and Administratio
Skyflower: A Reusable Tethered Lunar Landing System
No full textAs part of the Artemis program’s broader objective to establish a long-term human and robotic presence on the Moon, NASA and its partners must overcome the significant challenge of delivering substantial payload mass to the lunar surface. While commercial partnerships under programs like CLPS (Commercial Lunar Payload Services) and HLS (Human Landing System) have made progress, these efforts primarily address small-scale payload delivery or crewed missions. The current state of the art lacks dedicated solutions for high-mass, uncrewed cargo delivery—a critical gap for enabling infrastructure deployment, in-situ resource utilization, and sustained lunar operations. This thesis investigates Skyflower, a conceptual reusable lunar landing system specifically designed to address this need. Skyflower reimagines the architecture of planetary cargo delivery by adapting the tethered offloading approach used in the Martian Skycrane. By deploying cargo from a hovering lander via tether, the system minimizes plume-surface interaction (PSI) and improves landing precision in complex terrain. The lander is designed for reusability and is integrated into a broader operational framework centered on the Lunar Gateway. This orbital hub supports payload handling, refueling, and maintenance, enabling Skyflower to function as part of a sustainable logistics network in cis-lunar space. Unlike conventional systems that prioritize surface-based infrastructure and one-way missions, Skyflower emphasizes orbital coordination, system longevity, and modularity. The research employs an iterative systems engineering (SE) approach modeled on NASA’s lifecycle processes, progressing from stakeholder needs to system architecture and subsystem-level definitions. Through multiple design iterations, this thesis advances the technical detail of the lander’s key subsystems, including propulsion, guidance and navigation, tether deployment, and payload handling. While the current design remains at a conceptual level, future work must focus on detailed computational simulations, environmental modeling, and physical testing to increase the Technology Readiness Level (TRL) of the system. Ultimately, Skyflower represents a forward-looking approach to scalable lunar logistics, offering a path toward more efficient and reusable cargo transport solutions on the Moon
Simulation of Weak Polyelectrolyte Brushes
No full textUsing molecular simulations, the stimuli-responsive behavior of weak polyelectrolyte brushes (PEBs) and weak polyampholyte brushes (PABs) was examined with respect to pH, salt concentration, monomer sequence and grafting density. While PEBs have only acidic or basic monomers, PABs contain both types of monomers, resulting in more complex interactions that require advanced simulation methods to properly investigate. As expected, the simulation results of PEBs are qualitatively consistent with previous experimental and theoretical studies, showing a crossover from the "osmotic" brush regime into the "salted" brush regime as salt concentration is increased. In contrast, the ionization state, brush height, lateral structure, and chain conformations of PABs vary with pH and salt concentration in ways that are qualitatively different from PEBs. While grafting density has a moderate impact on the response of PABs, monomer sequence and salt concentration play significant but different roles depending on pH conditions. At extreme pH conditions, salt concentration has a more prominent effect on PABs, because only one type of monomer is charged, and electrostatic repulsion is the dominant factor. Within the neutral pH range, monomer sequence becomes more important in regulating brush behavior due to electrostatic attractions prevailing when all acidic and basic monomers are charged. The findings in this work indicate that monomer sequence and salt can be leveraged to adjust the pH response of weak PABs to achieve the desired outcome when designing stimuli-responsive materials
Unequal Access: How Race, Sex, and Class Shape Career and Technical Education (CTE) Participation
No full textBackground: Despite the growing importance of Career and Technical Education (CTE) in preparing students for successful careers, disparities in participation and completion rates persist across different demographic groups, particularly in diverse districts like Fort Bend ISD (FBISD). Purpose: This quantitative case study examines educational equity within FBISD CTE programs by investigating the relationships between student demographics (race and sex) and CTE program completion rates among the class of 2024. The study addresses the following research questions: (1) How do race and sex associate with CTE completer status in FBISD's class of 2024? (2) How do these associations vary across the eleven FBISD high school campuses? Methods: A quantitative approach using descriptive statistics and logistic regression analysis was employed to analyze student data provided by FBISD for the 6,554 seniors in the class of 2024. Results: The study revealed significant disparities in CTE program completion within FBISD, with only 30% of the class of 2024 achieving completer status. Asian students demonstrated the highest completion rates. Intersectional analysis showed that Black and Asian females had higher completion rates than their male peers. Logistic regression analysis, accounting for race/ethnicity, socioeconomic status, and sex, proved that Asian students were 1.5 times more likely to complete a CTE program than White students. Also, the logistic regression highlighted that GPA was a strong predictor, with each one-point increase in GPA significantly increasing the odds of completion. Conversely, students receiving special education services faced lower completion rates. Conclusion: The findings necessitate targeted federal, state, and local policy changes. At the federal level, the reauthorization of Perkins V should prioritize standardized definitions, mandated data reporting, and targeted funding for equity initiatives. The Texas Legislature should align policies with federal mandates and provide supplemental funding at the state level. Locally, FBISD should focus on equitable access to early career exploration, targeted support for underrepresented groups, and data-informed decision-making. By utilizing these suggestions, FBISD can close opportunity gaps and ensure equitable CTE outcomes for all students by implementing these measures
Quantitative Methods for the Analysis of Neuroglia in Micrographs
No full textAstrocytes are an important subclass of glial cells with an extremely intricate structure that play critical roles in maintaining or regulating a variety of functions within the Central Nervous System(CNS). These cells participate in a variety of CNS activities, from the maintenance of the blood-brain barrier and the regulation of blood flow to the modulation of synaptic activity and responsive changes after brain injury. However, despite their great importance, the mechanisms through which astrocytes work within the CNS remain not fully understood. One of the reasons lies in the complex morphological interactions that astrocytes have with other CNS cells, in particular with neurons and other glial cells. One major challenge to further understand the role of astrocytes comes from the limitations of available quantitative image analysis tools that are not very efficient at capturing their elaborated shapes, high heterogeneity and complex networks. This limits the possibility of studying in detail the alterations of astrocytes, their responses to various inputs and their interactions within the CNS. To help addressing these challenges, in this dissertation we developed a new deep learning framework that enables the automatic identification and analysis of GFAP-immunolabeled astrocytes directly from brightfield and fluorescent micrographs based on a state-of-the-art deep learning platform in object detection -You Only Look Once(YOLOv8). For this task, we have customized YOLOv8 to develop optimized models for astrocyte detection, improving its ability to accurately identify these cells even under a highly complex and challenging environment and introduced a novel automated counting method of those astrocytes. Our framework has been tested through a large collection of numerical experiments applied on different imaging modalities and conditions. These experiments have demonstrated that our approach is very competitive compared to traditional and state-of-the-art techniques. In particular, in cases of relatively dense packing of astrocytes, our approach seems quite effective. It is expected that our framework will facilitate studies on these critical CNS cells with a more accurate and efficient way to detect and analyze astrocytes
Large Eddy Simulation and Reduced Order Modeling to Contain the Computational Cost of Ocean Simulations
No full textThe two-layer quasi-geostrophic equations (2QGE) provide a simplified yet effective model for stratified, wind-driven ocean dynamics. Its direct numerical simulation (DNS) is challenging due to the size of the typical computational domain and the need for high-resolution meshes to resolve the full spectrum of turbulent scales. This dissertation presents three approaches to overcome these computational challenges by using filtering stabilization techniques, data-driven reduced order modeling (ROM), and randomized numerical linear algebra methods. First, we introduce a filtering stabilization technique that enhances the accuracy of low-resolution simulations. This method employs linear and nonlinear differential low-pass filters to adaptively apply artificial viscosity, enabling the use of significantly coarser meshes while maintaining accuracy. The linear filter introduces the same amount of artificial viscosity over the entire domain, while the nonlinear filter determines where and how much artificial viscosity is needed through an indicator function. Numerical experiments for a double-gyre wind forcing benchmark demonstrate that this approach achieves speedups ranging from 30 to 300 times compared to traditional high-resolution simulations. Second, we develop a data-driven ROM for the 2QGE that integrates proper orthogonal decomposition (POD) with long short-term memory (LSTM) networks. This model captures the dominant features of the 2QGE and enables drastic reduction in the computational cost, which is especially beneficial for parametric studies. By training LSTM networks on modal coefficients extracted from high-fidelity snapshots, our approach achieves predictive accuracy even when retaining only 10–20% of the singular value energy of the system. The resulting ROM attains computational speedups of up to 1E+07 compared to a DNS. Finally, we enhance the efficiency of the ROM by incorporating randomized numerical linear algebra techniques. Specifically, we replace deterministic POD with randomized POD to reduce the cost of extracting the dominant modes of the system, allowing for an efficient computation of the reduced basis, even as the dimensionality of the parameter space increases. Through this approach, we achieve approximately 700 times speedup in computing the reduced basis. By combining rPOD with the filtering stabilization technique in the offline stage, the computation cost of snapshot collection is reduced by 24 times compared when using DNS. In the online stage, the final rPOD-LSTM has been significantly extended the applicability of ROMs to high-dimensional parameter space successfully capturing large-scale ocean dynamics under varying physical conditions while achieving speedup factors exceeding 4.72E+05