Earth & Space Science

Permanent URI for this collectionhttps://hdl.handle.net/10315/27626

Browse

Now showing 1 - 20 of 216

Access status: Open Access ,
Development and Impedance Control of Lightweight Shape Memory Alloy Actuated Robotic Arms
(2026-03-10) Cormier, Auri; Bazzocchi, Michael
Shape memory alloys (SMAs) have been proposed as a lightweight actuator for robotic applications. Their low mass makes them an excellent candidate for aerial manipulation, where reducing the mass of a manipulator can increase usable payload and save energy. Existing SMA-actuated robotic arms are restricted in their size, strength, and degrees of freedom (DOFs). This thesis develops two 3-DOF, SMA-actuated robotic arms suitable for aerial manipulation. The developed arms are significantly more capable than existing SMA-actuated arms and can manipulate masses of up to 300 g. A novel compliant joint design is introduced that allows for simultaneous control of antagonistic SMA actuators. The joint provides direct measurements of the torque applied by each actuator, allowing for the development of a multi-level impedance controller. Finally, an extended Kalman filter formulation is introduced and integrated into the multi-DOF robotic arm.
Access status: Open Access ,
Advancements in modeling soil carbon using remotely sensed data
(2026-03-10) Pittman, Rory Clifford; Hu, Baoxin
This research investigated the resolving of soil carbon and interconnected properties in conjunction with vegetation and landform attributes for a boreal context within the Great Clay Belt region of northern Ontario, Canada. Objectives entailing sophisticated statistical inference of extracted soil samples, an endemic tree species classification and enhanced digital soil mapping of carbon were explored, yielding research contributions and practicable results. Statistical inference was achieved by adapting generalized estimating equations (GEEs) to estimate and assess refined mean differences of soil carbon between specific land cover types and tree species groupings. Furthermore, the feasibility of GEEs for digital soil mapping was demonstrated by predicting carbon concentrations with associated confidence intervals. Regarding boreal vegetation modeling, a pixel-based tree species classification was developed for natural settings at the remarkable spatial resolution of 2 m over an area of 100 square km. Afterwards, digital soil mapping of carbon was implemented by employing data-driven approaches via encoder-decoder (ED) neural networks formulated for relatively smaller data sets. Using EDs, modeling accuracy for soil carbon was augmented with coefficients of determination (R-squared) increasing to over 0.5. Finally, knowledge-based approaches to unravel linkages between soil carbon and interrelated properties were accommodated through structural equation modeling (SEM). A framework was devised to effectively facilitate SEM for digital soil mapping. The SEM quantified impacts on soil properties from covariates relating to soil formation factors, supporting the discernment of vegetational and environmental drivers for bulk density, carbon and carbon-to-nitrogen ratio. Uncertainty with prediction was also ascertained. A normalized entropy metric constituted from the top two contending groupings was conceived, which better encapsulated prediction uncertainty when compared to conventional entropy. Quantile mapping was also incorporated to uncover insights regarding prediction uncertainty from model ensembles. Structured query language (SQL) was harnessed to efficiently derive and generate rasterized covariates from LiDAR point cloud data. This novel computation consisted of a more detailed digital terrain model (DTM), as well as covariates pertaining to vegetational structure with canopy height model (CHM) and a gap fraction. These covariates were successfully exploited as predictors for modeling with both digital soil mapping and tree species classification.
Access status: Open Access ,
Multi-Head Encoder (MHEnc)-based Fusion Strategies for Land Cover and Land Use Classification Using Sentinel-1 and Sentinel-2
(2026-03-10) Ham, Taewoong; Hu, Baoxin
This study aims to develop a fusion strategy guided by data characteristics to improve LULC classification from multi-source data. The Multi-Head Encoder framework was designed as a backbone, integrating parallel convolutional heads with different dilation rates through learned attention mechanisms to process features across multiple spatial scales from fine local detail to broad landscape context. Building on this framework, fusion strategies were systematically investigated for heterogeneous sensor combinations and homogeneous temporal scenarios, establishing data-driven guidelines. To understand the underlying mechanisms, Grad-CAM analysis was employed, revealing that Early Fusion can lead to optical dominance over SAR information while Middle Fusion preserves modality-specific contributions. These insights guided the development of a hybrid fusion strategy that achieved 96.83% Overall Accuracy on the spatially independent test area. Patch boundary effects were investigated, and a central-area approach was evaluated, improving performance by up to 13% for attention-based SAR architectures. This integrated framework of multi-scale processing, systematic fusion evaluation, and interpretability analysis enables principled, data-driven classification design.
Access status: Open Access ,
Resilient, Low-Cost Navigation in Urban Environments with GNSS Precise Point Positioning, Inertial Measurement Unit and Precise Clock Sensor Fusion
(2025-11-11) Yang, Sihan; Bisnath, Sunil B.
The Precise Point Positioning (PPP) measurement processing technique for Global Navigation Satellite Systems (GNSS) is widely applied in scientific and commercial applications that require sub-metre level accuracy, particularly in areas with few obstructions. PPP is expanding into mass-market applications at the consumer level. However, the technique suffers from the inherent disadvantages of GNSS-based technologies. Obstructed environments, such as urban canyons, downgrade the solution. The study presents a novel solution that provides stronger resilience and improved performance in challenging environments by implementing clock-aided algorithms and augmenting the GNSS receiver with both an inertial measurement unit (IMU) and an external clock as a frequency reference. The PPP algorithm was augmented with clock-specific modifications, including the receiver clock coasting (RCC) and receiver clock modelling (RCM) algorithms. The study began with building a prototype triple-sensor fusion using higher-end hardware and was successfully tested by collecting and processing real-world, kinematic GNSS data. Horizontal rms errors have been reduced from 1.49 m with the traditional dual-sensor solution to 1.14 m with the novel solution. With multiple simulated outages, the horizontal rms errors were reduced from 2.07m to 1.36m using the RCM solution and 0.55musing the RCC solution. To address the requirement of consumer-grade applications, a mass-market-friendly variant using low-cost hardware was built and tested in real urban canyons. The positioning accuracy over time was found to improve by 10–40% inmost datasets when compared to traditional PPP/IMU fusion using identical hardware. In both fusion configurations, three-satellite GNSS solutions were verified. Finally, the study delved into position-domain IntegrityMonitoring (IM) and showed an improved integrity performance of the novel triple-sensor solution. The number of unavailable epochs under the IM criteria was reduced by half. The percentage of safe operations in the cross-track and along-track components, as determined by the criteria, increased from 97.20% and 82.34% to 99.45% and 92.67%, respectively, indicating a more trustworthy solution for safety-of-life applications. The result has proven the novel triple-sensor solution as a viable option in improving the performance and integrity for liability-critical applications such as intelligent vehicles. Further research is recommended to optimise the characterisation of the external oscillators and to achieve deeper integration among the three sensors.
Access status: Open Access ,
Mixed Reality Human Robot Interaction Interface using Hand Tracking and Stereo Vision for Unknown and Hazardous Environments
(2025-11-11) Tennakoon, Damith Deshan; Jadidi, Mardkheh Amaneh
Many industries employ human workers to perform hazardous and unsafe tasks that result in injuries and fatalities. These include on-site monitoring of radioactive environments, collisions with heavy machinery on construction sites, and repetitive, physically demanding work. Robotics solutions have been proven to improve task efficiency, product quality, and workplace safety in many industries, though it has posed a challenge to be integrated into dynamic and unstructured environments. This research aims to develop a generalized immersive robot arm teleoperation interface designed to accelerate the deployment of robotics in real-world applications through an intuitive Human-Robot Interaction (HRI) system. The methodology involves mounting a ZEDM stereo camera, attached to a 2-axis servo-actuated gimbal, on the end effector of a 6-axis robot arm. Stereo images are streamed to a Meta Quest 2 virtual reality (VR) head-mounted display (HMD), providing a first-person view (FPV) with human-like perception of the end effector's surroundings. Hand tracking is used to map the operator’s hands into the FPV, enabling omnidirectional motion control and gripper operation via gesture recognition. In-situ data is spatially mapped to the stereoscopic view, creating a mixed-reality (MR) HRI interface. Results show improved depth perception and higher control stability compared to traditional monocular video systems paired with keyboard controls. Specifically, the MR interface achieves a 39% reduction in task completion time during target-tracking experiments and demonstrates five times greater stability in trajectory control tests. This MR HRI interface promotes the safe use of robotics in unknown or hazardous environments, allowing operators to perform high-risk tasks remotely using only HMDs and natural hand gestures.
Access status: Open Access ,
Aqueous Alteration in the Tarda Meteorite: Isotopic and Geochemical Analyses of Secondary Minerals
(2025-11-11) Wilson, Bennett John; Daly, Michael G.
Water appeared on some carbonaceous asteroids within the first few million years of Solar System history and modified the other accreted components in a process known as aqueous alteration. Recovered meteorites from such bodies, termed carbonaceous chondrites, are composed of diverse assemblages of secondary minerals and organic compounds that differ depending on the fluid environment the meteorite experienced. Since carbonaceous asteroids likely seeded the early planets with prebiotic organic matter, understanding the conditions that governed aqueous alteration is critical for evaluating their role in the origin of life. This dissertation investigates the aqueous alteration history of Tarda, a new and unusual C2-ungrouped carbonaceous chondrite that fell in Morocco in 2020. The secondary mineralogy of Tarda implies exposure to significant aqueous alteration, and their investigation enables the fluid environment of Tarda to be constrained. The goal of this dissertation is to constrain the temperature, timing, evolution, and chemistry of the fluid responsible for altering the Tarda meteorite. Chapter 1 first establishes a novel, non-polar sample preparation procedure capable of polishing Tarda and other clay-rich samples for sensitive microanalytical techniques. Hexane, toluene, and mineral oil were identified as effective polishing liquids for such samples, which prevent clay minerals from swelling. Chapter 2 uses this polished surface and employs in situ secondary ion mass spectrometry (SIMS) on dolomite and magnetite to analyze oxygen, carbon, and manganese-chromium isotopes to constrain the timing, temperature, and evolution. Here, dolomite and magnetite were found to have precipitated approximately ~4.563 billion years ago at ~90˚C, from a relatively evolved fluid after significant water-rock interaction. Chapter 3 employs transmission electron microscopy (TEM), energy-dispersive spectroscopy (EDS), and atom probe tomography (APT) to investigate the nanoscale chemistry of magnetite framboids in Tarda. These analyses reveal discrete boundary enrichments in elements such as Ti, Si, Na, Mg, Ca, and Mn, capturing a chemical record of the altering fluid from which the framboids precipitated. The observed element distributions and inferred surface charge conditions constrain the fluid to alkaline pH (>5.4) and support a chemically diverse, cation-rich environment. Together, these chapters present a multi-technique, multi-scale investigation into the aqueous alteration history of Tarda. Beyond providing new insights into Tarda itself, this work contributes to broader efforts to reconstruct parent body fluid histories and evaluate the potential for carbonaceous asteroids to host environments favorable to prebiotic organic synthesis.
Access status: Open Access ,
Digital Twin Platform for Drone Applications
(2025-11-11) Haridevan, Amal Dev; Shan, Jinjun
Simulation tools and digital twins are vital to UAV research, enabling efficient prototyping, testing, and deployment of algorithms in dynamics, control, computer vision, and deep learning. Despite their importance, existing simulators face limitations in modularity, scalability, and multidisciplinary integration, while the persistent sim-to-real gap remains a major challenge. Nevertheless, simulators are indispensable, particularly for deep reinforcement learning (DRL), where extensive data collection in real-world settings is infeasible due to hardware and safety constraints. This thesis presents a modular, scalable, and transferable simulation platform for quadrotor UAVs, designed to support research in control systems, DRL, and embodied AI. The platform incorporates a high-fidelity digital twin, modeled using Blade Element Theory, with parameterized aerodynamic forces tailored to the Quanser QDrone2 and adaptable to other quadrotor architectures. A dedicated Control library, built within the Gazebo ecosystem, facilitates rapid prototyping and benchmarking of UAV controllers under a unified framework. To extend applicability to autonomous navigation, computer vision, and state estimation, Gazebo plugins are developed to enable efficient same-process interfacing and integration with both ROS1 and ROS2, thereby ensuring seamless transfer of algorithms from simulation to hardware. To advance DRL research, a suite of deterministic and stochastic environments is introduced. These environments balance reproducibility and realism by modeling communication delays, timing variations, and other uncertainties, while remaining compatible with widely used DRL libraries. Their parallelizable design enables large-scale data collection within hours, significantly accelerating training. Experimental validation confirms the platform’s fidelity and effectiveness in bridging the gap between simulation and real-world UAV deployment
Access status: Open Access ,
Planetary Radar and Geomorphology in the Cryosphere: Investigating Glacial-Periglacial Landsystems in Phlegra Montes, Mars and Yukon Territory, Canada
(2025-11-11) Andres, Chimira Nicole; Smith, Isaac B.
Mars has abundant water-ice across its surface and in the subsurface, particularly in the form of glacier and permafrost ice. These cryosphere elements drive the development of numerous landforms that are also observed on Earth, enabling comparative studies between both planets. In particular, the mid-latitude region of Mars (30-50°N) is where thousands of viscous flow features (VFFs) also called debris-covered glaciers, are situated. In this PhD thesis, I use a combination of orbital radar sounding, high-resolution imagery, topographic data, and geomorphological techniques to analyze glacial landforms in the Phlegra Montes region of Mars. This study is supported by geophysical field investigations in the Canadian Arctic, a well-established analogue for Martian permafrost and glacial landscapes. Together, these approaches provide new insights into the structure, dynamics, and detection of shallow ice systems on Mars and Earth. Three major findings are presented. First, I report the discovery of a debris-covered glacier with terraced topography as well as a hanging glacier in Phlegra Montes, Mars detected using SHARAD (radar) data, representing the first known observation of such features on Mars. Secondly, I provide evidence supporting past interpretations that VFFs and the latitude- dependent mantle (LDM) are separate, yet overlapping systems. Lastly, through field-based radar and sedimentological surveys in Tombstone Territorial Park, I identify the most suitable radar frequency range for detecting shallow subsurface ice (1–5 m), offering direct guidance for future Mars missions using radar instrumentation. These findings enhance our understanding of Mars' cryosphere and its recent glacial history. They also support mission planning efforts for future exploration and subsurface ice mapping, for example and/or including the International Mars Ice Mapper (I-MIM) and Mars Life Explorer (MLE) missions.
Access status: Open Access ,
Adapting a Cryogen-Free Measurement System for Geological Materials: Method Development for Thermal Conductivity Measurements Below 300 K with Applications to Planetary Science
(2025-11-11) Gilmour, Cosette Marie; Daly, Michael G.
The thermal properties of airless planetary bodies like asteroids are essential to understanding their thermal evolution. While remote sensing missions collect thermal data, this is limited to surface properties. As such, little is known about internal heat flow. Thermal models are used to predict thermal evolution, but the limited availability of thermal property data constrains their accuracy. To improve these models, meteorites provide a way to investigate thermal properties under controlled laboratory conditions, as they are preserved fragments of planetary bodies. This study reports thermal conductivity measurements of meteorites between 5 and 300 K under vacuum (< 10^(−4) mbar), acquired using a Cryogen-Free Measurement System (CFMS) that was adapted for geological applications. Measurements were first collected for single-crystal minerals and obsidian to refine sample preparation and measurement procedures. This included developing techniques for handling friable samples, establishing thermal equilibration protocols, and implementing a data quality assessment method based on empirical observations of instrument performance. Above 100 K, radiative heat loss contributes to the measured thermal conductivity, resulting in values that exceed the true conductive behaviour. A correction procedure was applied using low-temperature model fits specific to each material. The single-crystal minerals exhibit anisotropic thermal conductivity and show trends consistent with phonon-dominated transport, where thermal conductivity is expected to increase as T^3 at low temperatures, peak, and then decrease as 1/T at higher temperatures; however, this full behaviour was not always observed within the measured range. In contrast, obsidian shows a plateau followed by a gradual increase in thermal conductivity with temperature, consistent with the behaviour of amorphous solids. Meteorites exhibit more complex behaviour. Despite being composed mainly of crystalline materials, phonon transport is suppressed by porosity and grain boundaries, resulting in thermal conductivity trends similar to disordered materials. These factors also drive anisotropy due to structural heterogeneity. The results of this study agree well with literature values, validating the use of the CFMS for geological analysis. The data collected in this study will support improved modelling of heat flow in small planetary bodies.
Access status: Open Access ,
Fully Distributed Event-Triggered Robust Cooperative Control for Multi-Agent Systems
(2025-11-11) Wang, Hao; Shan, Jinjun
Cooperative control of multi-agent systems (MASs) is essential in applications such as surveillance, formation flying, and load transportation, offering improved functionality and robustness compared to single-agent configurations. However, many existing control protocols rely on global network information, limiting their applicability to varying communication topologies. This thesis addresses the challenge of achieving fully distributed cooperative control under limited communication, computation, and energy resources. A systematic design methodology for event-triggered control schemes is proposed, enabling protocols to depend solely on local information. First, an adaptive sliding-mode-based event-triggered formation control framework is developed for leader-follower MASs with disturbances, ensuring finite-time sliding-surface reachability and Zeno-freeness. Second, an adaptive dynamic event-triggered approach with integral sliding surfaces and variable triggering intervals is designed to enhance resource efficiency. Third, for networked Euler–Lagrange systems with uncertainties, a nested adaptive sliding-mode estimator and robust event-based control strategy are introduced to compensate for nonlinearities and disturbances. Fourth, a fully distributed adaptive dynamic event-based control scheme addresses time-varying formations under switching topologies, input saturation, and unreliable communication. All proposed strategies are theoretically validated using Lyapunov methods, ensuring stability and convergence, and experimentally verified with multiple quadrotors, demonstrating effective consensus, formation maintenance, and communication efficiency. The results highlight the theoretical significance and practical applicability of fully distributed event-triggered cooperative control for MASs in dynamic and resource-constrained environments.
Access status: Open Access ,
Attitude Estimation Algorithms and Comprehensive Error Analysis in the Generic Multi-sensor Integration Strategy
(2025-11-11) Brunson, Benjamin James Loker; Wang, Jianguo
Modern multi-sensor integrated kinematic position/attitude systems serve a diverse range of industries, including precision agriculture, construction, mobile mapping, and autonomous driving. These industries have diverse needs that must be met by their position/attitude systems, which emphasizes the need for kinematic position/attitude systems that are highly customizable and robust. Lower-cost sensors have become increasingly prevalent in modern position/attitude systems, and accurately modeling the systematic errors in these sensors is of paramount importance for high-accuracy applications. This dissertation focuses on extending and refining the tools for comprehensive error analysis in the Generic Multi-sensor Integration Strategy (GMIS), with a particular focus on statistical analysis in pre-and post-processing environments. This research leverages the strengths of the GMIS to rigorously characterize the sensor performance and systematic errors of tactical grade MEMS IMU sensors, even in situations where a reference solution is unavailable. The principal research contributions of this dissertation are as follows: 1. A generalized framework for imposing system state equality constraints in a Kalman Filter, which allows for the direct estimation of the constrained observation, predicted system state, and process noise residuals, along with their associated covariance matrices. This framework opens opportunities for comprehensive error analysis in system state constrained Kalman Filters, including: residual analysis, reliability analysis, and Variance Component Estimation (VCE). 2. The formulation of three unique attitude models under the GMIS framework: the Roll-Pitch-Heading (RPH) model, the Direction Cosine Matrix (DCM) model, and the Quaternion model. This research defines each attitude model’s corresponding system model, system state constraints, and IMU observation models within the GMIS. This work uniquely uses the process noise residual vector to conduct VCE to directly compare the performance of each different attitude model. Each of the three attitude models used in this research performed similarly over a road test dataset, and the DCM model in particular exhibited resistance to a sudden trajectory variation that was captured by the IMU. 3. The formulation of Multi-IMU (MIMU) array modeling in the GMIS, allowing for the separate estimation of IMU systematic errors and stochastic properties for every sensor in the MIMU array. This work allows for a detailed, time-varying analysis of constituent IMU sensors’ stochastic properties via VCE alongside rigorous Fault Detection and Exclusion (FDE) for each sensor in a MIMU array. None of this is possible under other data fusion methods for MIMU arrays. The a posteriori position/attitude standard deviations were compared between a MIMU and Single-IMU (SIMU) solution for the same road test dataset, and the MIMU solution provided an average accuracy improvement of ca. 14-16% in the estimated position, 30% in the estimated roll and pitch, and 40% in the estimated heading. 4. A generalized framework for the pre-analysis of online sensor calibration procedures under the GMIS. The proposed framework uses a candidate trajectory with a defined sensor configuration to estimate the following quantities: minimum significant values for individual systematic error parameters; minimum detectable values for observation biases from each sensor; and observability analysis for each sensor systematic error being modeled in the Kalman Filter. This framework allows for the establishment of best field procedures and sensor configurations without requiring the construction and testing of physical systems. This framework was used to evaluate three different case studies to determine the observability of IMU systematic error states, alongside their minimum estimable values and minimum detectable errors under different calibration maneuvers. Across all three case studies, the accelerometer biases generally had minimum estimable values within 1 cm/s and the gyroscope biases generally had minimum estimable values within 3.5 ‘/s. Accelerometer/gyroscope scale factor errors and the reference-IMU lever arm vector had their observability much more dependent upon the specific calibration maneuver being used. This research takes great strides towards establishing the GMIS as a fully matured strategy for sensor integration that has much more modeling flexibility than traditional approaches. The only remaining challenge is addressing the increased computational load of the GMIS relative to the TMIS, and this is a high priority for future work. This research’s contributions lay a strong foundation for future research in adaptive filtering techniques, rigorous fault detection and exclusion, and comprehensive stochastic modeling of the sensors that are being used in a position/attitude system.
Access status: Open Access ,
Spatial Quantum Computation in Graph Optimization Problems in Transportation Applications
(2025-11-11) Nourbakhshrezaei, Amirhossein; Jadidi, Mojgan
Efficient transportation management is an important aspect of urban sustainability, impacting economic growth, environmental sustainability, and quality of life. This research explores the potential of Quantum Computing (QC) to address spatial optimization problems in transportation systems. By leveraging the principles of quantum mechanics, this research aims to enhance the efficiency and effectiveness of transportation networks through QC-based solutions to challenges such as reducing the spread of viruses on road networks, dynamic rebalancing of \nomenclature{BSS}{Bike Sharing Systems}Bike Sharing Systems (BSS), clustering of BSS stations, and Feature Selection (FS) for predictive models. The study begins by examining the potential of QC in solving combinatorial optimization problems, specifically focusing on minimizing exposure to COVID-19 during city journeys. A novel QC-based approach is developed for the dynamic rebalancing of BSS, which is a critical component of BSS management. The research further explores the clustering of BSS stations using \nomenclature{QML}{Quantum Machine Learning}Quantum Machine Learning (QML) techniques to enhance system management and improve user satisfaction. Additionally, this research introduces a QC-based FS method to improve the accuracy of predictive models, utilizing spatial data to optimize station placement and service availability. The proposed methodologies are validated through different experiments and real-world data, demonstrating significant improvements in computational efficiency and solution quality compared to traditional methods. Overall, this research advances the application of QC in transportation systems, providing a QC-based framework for future studies and practical implementations in urban transportation management. It highlights the transformative potential of QC in addressing pressing urban mobility challenges, paving the way for more sustainable and efficient transportation networks.
Access status: Open Access ,
Determining the temporal and spatial variations of pollutants in Toronto
(2025-07-23) Singh, Samir Harpal; Gordon, Mark D.
This thesis aims to assess the spatial and temporal variations of pollutants in Toronto. Annual, seasonal, and monthly pollutant trends were computed from as early as 1974 to as late as the end of 2022. Decreasing NO, NO2, NOx and CO trends were found, perhaps associated with the elimination of coal-fired power plants, and vehicular technology improvements, leading to emissions reductions. Applying two different techniques to trace pollutant measurements from air quality stations upwind, NO, NO2, NOx , and CO concentrations were found to be relatively high around Downsview and Pearson Airport, as well as the York University power plant. Additionally, relatively high NO2 and NOx levels were denoted around Highway 401, suggesting vehicular and aircraft emissions sources. The greatest concentrations of PM2.5, PM10, O3, and SO2 were concluded to originate from outside the city. Income information was obtained by neighborhood and related to the spatial distribution of pollutants in Toronto. A negative relationship was found between NO and NOx levels and income. This is because the lowest and highest income neighborhoods (generally located in north and south Toronto, respectively) mostly coincided with relatively high and low NO, NOx concentrations, respectively. TROPOMI satellite measurements over a gridded plane in Toronto also display the highest and lowest NO2 and HCHO levels over Pearson Airport and Lake Ontario.
Access status: Open Access ,
Understanding and Modeling Marine Fog in Areas Offshore from Atlantic Canada
(2025-07-23) Teeloku, Piyush; Taylor, Peter
In recent years, machine learning (ML) has gained popularity in the field of weather forecasting, particularly in areas where numerical weather prediction (NWP) models face challenges. One such area is fog prediction. A key factor in improving fog forecasts is the accurate representation of microphysics processes, which play a crucial role in fog formation and dissipation. In this study, we propose a post-processing approach that combines the Weather Research and Forecast (WRF) model with a machine learning classifier to distinguish between fog and no-fog conditions. This research builds on previous work conducted during the Fog and Turbulence Interactions in the Marine Atmosphere (FATIMA) project, where a 3-day marine fog forecast over the Yellow Sea was provided using WRF. The project highlighted the importance of microphysics parameterization schemes in accurately forecasting marine fog. Our study spans eleven years, from 2012 to 2023, focusing on the months from April to August each year. We utilize the WRF Preprocessing System (WPS) with initial and boundary conditions provided by ERA5 reanalysis data. The features of our ML model include 2-meter temperature, U and V wind components, 2-meter relative humidity, surface pressure, and the month, day, and hour. The target variable is hourly reported visibility data from Navigation Canada on the Environment and Climate Change Canada (ECCC) website for the locations of St. John’s, Newfoundland and Labrador, and Yarmouth, Nova Scotia in Canada. When tested with data in 2024, our ML model demonstrates better performance compared to predictions based solely on the liquid water content (LWC) from the WRF model. We used various metrics to evaluate the classification, with the F1 score being the most important metric. By using the ExtraTreesClassifier model, we were able to obtain a 11% increase in the F1-score (0.69 vs. 0.62)and a 2% increase in accuracy (0.90 vs. 0.88) compared to the WRF model at St John’s. A similar performance was noted in Yarmouth. The accuracy there increased by 4% (0.86 vs. 0.82) and F1-score by 11% (0.59 vs. 0.53). This approach shows promise in enhancing the accuracy of fog prediction, offering valuable insights for aviation, marine operations, and transportation safety.
Access status: Open Access ,
Satellite-Derived Bathymetry Using Machine Learning Methods
(2025-07-23) Ngalande, Chifuniro; Armenakis, Costas
This study evaluates the performance of machine and deep learning (ML/DL) models for satellite-derived bathymetry (SDB) compared to traditional empirical and physics-based methods in high-latitude freshwater and saltwater environments. The models were trained on data from several Canadian sites, and their transferability was tested on unseen sites. DL models, U-Net, SegNet and DeepLabv3 +, achieved about two times higher F1 scores than traditional methods. From the class-wise scores, ML/DL models can reliably predict up to a depth of 8m in freshwater, due to greater water transparency, and up to 3m in saltwater, due to sediments that prevent light penetration. Traditional methods struggled in deeper and turbid waters, but performed similarly to the machine learning Random Forest model in both environments. Overall, the ML/DL models have difficulties generalizing to unseen data or data from geographic locations not included in the training process, particularly in saltwater due to poor light penetration. The depth results met the Zones of Confidence (ZOC) categories, CATZOC C and CATZOC D, of the International Hydrographic Organization (IHO) depth accuracy standards. Overall, this study highlights the advantages of machine learning over traditional methods while identifying challenges in model generalization and data diversity.
Access status: Open Access ,
Marine And Coastal Fog: Forecasts and Evaluations
(2025-07-23) Chen, Zheqi; Taylor, Peter
Accurate marine fog forecast is of importance for human activities in the coastal regions and over the ocean. However, it is still a challenge due to lack of observation and unsatisfactory model performance. This dissertation includes three studies on forecasting marine fog. The first study has summarized the performances of three different models during the FATIMA (Fog And Turbulence Interactions in the Marine Atmosphere) field campaign on Sable Island, Nova Scotia, including two WRF (Weather Research and Forecasting) models and a COAMPS (Coupled Ocean/Atmosphere Mesoscale Prediction System) model. It is found that the models can perform differently on fog prediction with similar errors in temperature, dew point, relative humidity, wind speed and wind direction. Additional tests show that the WRF model can be improved by adjusting the horizontal and vertical domains, and a spin-up time is necessary for forecasting fog. The second study has compared the liquid water content and droplet concentration observations from the field campaign to WRF variables in two periods. It is found that the WRF model produces liquid water contents up to 0.6 g m-3 while the observation has up to 0.3 g m-3, partly due to the updraft speed in the microphysics scheme being too high and causing a high activation rate towards droplets. The surface temperature of the island, not included in the GFS (Global Forecast System) data but assigned by WPS (WRF Preprocessing System) is also too high, causing the fog to incorrectly dissipate in daytime. The last study has used XGBoost, a machine learning model to post-process WRF output. XGBoost is trained with the ERA5 (ECMWF Reanalysis v5) data of St. John’s, Newfoundland and Labrador, and Yarmouth, Nova Scotia. XGBoost can predict fog or clear with given 2 m temperature and relative humidity, 10 m U and V winds, land surface pressure, their corresponding values one hour ago, hour, day and month. Tests using forecasts of the features from WRF in 2024 found that XGBoost improved the recall by up to 0.07 without decreasing the precision, compared to using WRF only. This shows the potential to combine the strength of both models.
Access status: Open Access ,
Robust, Multi-Constellation, Multi-Frequency Precise Point Positioning for Instantaneous cm-level Positioning
(2025-07-23) Adu, Benjamin; Bisnath, Sunil B.
Achieving instantaneous centimetre-level 3D accuracy for Global Navigation Satellite System (GNSS) multi-constellation, multi-frequency Precise Point Positioning (PPP) with ambiguity resolution without local augmentation poses significant challenges due to multipath, atmospheric refraction, and hardware noise. This research reduces York-PPP engine limitations by separating the BeiDou-2 and BeiDou-3 GNSS constellations clock terms in the design matrix and expanding the uncombined Decoupled Clock Model from triple to quadruple frequency measurements for BeiDou-3 constellation. Processing of geodetic measurements show overall horizontal positioning errors reduced by 77% (from 20.8 cm to 4.7 cm) and 88% (from 22.8 cm to 2.8 cm) for float and fixed solutions, respectively. Furthermore, this research investigates outliers in epoch-by-epoch PPP solutions, delving into potential causes such as signal-to-noise ratio (SNR), pseudorange multipath and noise etc. Additionally, principal component analysis with Hotelling’s T-squared method is employed to detect major outlier sources. Results indicate that satellite elevation angle, signal strength, and the code-minus-carrier observable, which includes ionospheric refraction and pseudorange multipath, significantly impact positioning solution.
Access status: Open Access ,
Athermal Optical Phased Array
(2025-04-10) Papakonstantinou, Constantine Athanasios; Lee, Regina S. K.
Optical phased arrays (OPA) are of growing interest in free space communication systems due to the need for high speed communication. Generally, design of OPAs using the Silicon On Insulator (SOI) platform will rely on the thermo-optic (TO) effect to tune the refractive index, thereby controlling the phase in each channel. However this makes the design sensitive to parasitic heat, and thermal cross-talk at scale. Instead of using thermal control, thermal tolerance is introduced by reducing the TO effect, and leveraging the electro-optic (EO) effect in lieu. Metamaterial techniques for the athermal design of optical waveguides and subwavelength structures are presented for the application of designing an OPA feed network. The design of compact and athermal grating antennas suitable for OPA configurations is also explored.
Access status: Open Access ,
Building Footprint Reconstruction from Satellite Imagery Using a Deep Learning Framework with Geometric Regularization
(2025-04-10) Kamran, Muhammad; Sohn, Gunho
Since the launch of Landsat-1 in 1972, Earth observation satellites have significantly evolved, now capturing vast amounts of high-resolution imagery. These satellites continuously transmit data that helps us monitor urban expansion, plan infrastructure, and respond to natural disasters. Satellite imaging plays a crucial role in creating accurate spatial maps, offering detailed insights into the built environment. However, reconstructing maps from satellite images is a complex challenge. The field of computer vision has made impressive progress in object recognition and representation, but accurately modeling buildings as geometric primitives remains difficult. Unlike traditional pixel-based approaches, primitive-based object representation requires understanding the spatial structure and relationships that existing methods often struggle with. In this thesis, we explore four different network architectures aimed at improving our baseline model, R-PolyGCN, by integrating novel modules that enhance building footprint reconstruction. The final chapter presents Decoupled-PolyGCN, our most advanced deep learning model, that leverages Graph Convolution Network (GCN) to enhance building footprint reconstruction. By incorporating geometric regularity, multi-scale feature fusion, and Attraction Field Maps (AFM), the model generates more structured and precise building outlines from a single satellite image. Evaluations on the Wuhan University (WHU) and SpaceNet-2 datasets show that Decoupled-PolyGCN outperforms existing approaches, improving Average Precision (AP) by Average Recall (AR)% and AR by 10%. These improvements enable more accurate and reliable mapping, benefiting applications in urban planning, disaster management, and large-scale spatial analysis.
Access status: Open Access ,
Development of an Orbital LIDAR System for Mars Atmospheric Research
(2025-04-10) Baal, Davin John; Whiteway, James A.
A Monte Carlo model was developed to simulate multiple scattering and depolarization of LIDAR signals in water ice clouds and mineral dust in the atmosphere of Mars. This model was used to test the viability of an orbital LIDAR system for the study of the Martian atmosphere. It was found that a system with the same laser as the Earth-based CALIPSO orbital LIDAR system but with a receiver telescope with half the diameter of CALIPSO's is capable of producing reliable measurements of typical Martian cloud and dust conditions.

Browse

Recent Submissions