Browsing by Author "Lary, David J."
Now showing 1 - 14 of 14
- Results Per Page
- Sort Options
Item Applying Deep Neural Networks and Ensemble Machine Learning Methods To Forecast Airborne Ambrosia Pollen(MDPI AG, 2019-06-04) Zewdie, Gebreab K.; Lary, David J.; Levetin, E.; Garuma, G. F.; Zewdie, Gebreab K.; Lary, David J.Allergies to airborne pollen are a significant issue affecting millions of Americans. Consequently, accurately predicting the daily concentration of airborne pollen is of significant public benefit in providing timely alerts. This study presents a method for the robust estimation of the concentration of airborne Ambrosia pollen using a suite of machine learning approaches including deep learning and ensemble learners. Each of these machine learning approaches utilize data from the European Centre for Medium-Range Weather Forecasts (ECMWF) atmospheric weather and land surface reanalysis. The machine learning approaches used for developing a suite of empirical models are deep neural networks, extreme gradient boosting, random forests and Bayesian ridge regression methods for developing our predictive model. The training data included twenty-four years of daily pollen concentration measurements together with ECMWF weather and land surface reanalysis data from 1987 to 2011 is used to develop the machine learning predictive models. The last six years of the dataset from 2012 to 2017 is used to independently test the performance of the machine learning models. The correlation coefficients between the estimated and actual pollen abundance for the independent validation datasets for the deep neural networks, random forest, extreme gradient boosting and Bayesian ridge were 0.82, 0.81, 0.81 and 0.75 respectively, showing that machine learning can be used to effectively forecast the concentrations of airborne pollen.Item Cloud Detection and PM2.5 Estimation Using Machine Learning(2021-12-01T06:00:00.000Z) Yu, Xiaohe; Lary, David J.; Hicks, Donald A.; Chun, Yongwan; Yuan, May; Qiu, FangEarth observation (EO) is the gathering of information about the physical, chemical, and biological systems of the planet via remote-sensing technologies, supplemented by Earthsurveying techniques, which encompasses the collection, analysis, and presentation of data. Research on exploring effective methods for earth observation data analysis has increased over the years because of the increasing amount of data generated by earth observation systems, such as remote sensing imagery and weather radars. Researchers have therefore taken an interest in machine learning, a technique that allows computer algorithms to learn from samples. In general, the more comprehensive our training samples are, the better the machine learning performance will be. This feature makes machine learning an ideal approach for analyzing earth observation data. Particulate matter of fine size, such as particulate matter 2.5 (PM2.5), poses a severe health risk to humans and is associated with many different health problems. PM2.5 concentrations are influenced by factors such as meteorological conditions, local population density, and the geographic context. As a result of the large quantity of information provided by Earth observation, they become a valuable tool for studying PM2.5. They are huge and come from different platforms, with different spatial and temporal resolutions, and in different formats, which challenge the approaches for PM2.5 studies. This dissertation shows how machine learning methods can be used to address these challenges in three subtopics connected to modeling and estimation for PM2.5. Satellite-based remote sensing products provide important variables that can be used to study regional and global PM2.5, such as the Aerosol Optical Depth (AOD). Nevertheless, AOD products in cloudy areas cannot be retrieved, and the quality of AOD data in nearby cloud areas cannot be guaranteed. Accordingly, the first study aims to detect cloud pixels based on remote sensing images. This study investigates the cloud detection with a set of machine learning models on four subsets of 88 Landsat8 images that have been carefully labelled by analysts. Four subsets of training data are used to train 16 machine learning models with different input feature selections. The performance of these models is then compared with that of the Fmask algorithm, which is widely used for cloud detection. When testing on the 88 annotated images, the best performance was observed with a model that incorporates unsupervised self-organizing map (SOM) classification results among the input features. In comparison with Fmask4.0, the model improves the correctness by 10.11% and reduces the cloud omission error by 6.39%. Focusing on the other 8 independent validation images that were never sampled as part of the model training, the model trained on the second largest training subset with additional 5 input features has the best overall performance. Compared with Fmask4.0, this model improves the overall correctness by 3.26% and reduces the cloud omission error by 1.28%. In the second study, high temporal resolution PM2.5 models are developed based on data from weather radar systems and the meteorological data from the European Centre for MediumRange Weather Forecasts (ECMWF). A dataset covering the period from July 2019 to June 2021 was collected for model training, which included the Next Generation Weather Radar (NEXRAD) retrieved from a repository on Amazon Web Services (AWS), meteorological data from ECMWF, and the PM2.5 ground observations from 31 sensors deployed across Dallas county, Collin county, and Tarrant county. The models are classified in groups to demonstrate the effectiveness of NEXRAD in high temporal PM2.5 modeling. The model utilizing NEXRAD data achieves an 0.855 score of the correlation of determination (R2 ), while the model without NEXRAD has a 0.7 R2 for PM2.5. The third study establishes a nationwide PM2.5 estimation model by using high temporal resolution AOD data from the GOES-16 geostationary satellite, meteorological variables from ECMWF and a set of ancillary data from a variety of sources, which achieves 3.0µg/m3 and 5.8 µg/m3 as the value of mean absolute error (MAE) and root mean square error (RMSE). The model performances are then further evaluated by time, elevation, soil order, population density, and lithology. The historical PM2.5 estimation surfaces are then reconstructed and the PM2.5 surfaces during the period of California Santa Clara Unite (SCU) Lightning Complex fires are demonstrated.Item Coupling Physical Measurement with Machine Learning for Holistic Environmental Sensing(2021-05-01T05:00:00.000Z) Wijeratne, Lakitha Omal Harindha; Lary, David J.; Stefan, Mihaela C.; Glosser, Robert; Chen, Lunjin; Slinker, Jason D.; Heelis, Roderick A.The interest in characterizing the abundance and nature of airborne particulates has been increasing over the last decade, driven in large part by the rising awareness of the manifold health impacts of airborne particulates. Since regulatory observations of airborne particulates are usually made with expensive instruments, the number of sensors that can be deployed is naturally limited by the costs involved. This dissertation describes the substantial progress we have made in the physical sensing of airborne particulates by providing low-cost, high-quality observations of airborne particulates by utilizing advances in low-cost laser-based sensors, that can be deployed at scale, coupled with machine learning used for accurate calibration of these low-cost sensors. The abundance of airborne particulates is usually quantified by an integrated mass density in µg/m3 over the airborne aerosol size distribution (e.g. PM2.5, the integrated mass density of all airborne particulates with a diameter of up to 2.5 microns). A persistent feature of all airborne observations of particulates is the variability over small temporal and spatial scales. This persistent and ubiquitous variability underscores the value of being able to deploy a large number of low-cost sensors that can make accurate measurements every few seconds, 24/7. Taking this into account, I have built, calibrated, and deployed a large number of sensors across the Dallas-Fort Worth (DFW) Metroplex in Texas as a part of my dissertation work. Other physical measurements can also be utilized in accurate assessment of airborne particulates. Just as weather RADARs are used to examine the spatial and temporal distribution of atmospheric precipitation, we show that if we use machine learning, we can also employ the weather RADARs to examine the spatial distribution of airborne particulates. CO2 has gained a lot of attention in recent years due to global warming. It is considered the principal anthropogenic greenhouse gas driving global warming. As a result, CO2 levels must be monitored and controlled. The present study describes how machine learning can be used to calibrate a low-cost CO2 sensor which is already part of the sensor systems that I have built and deployed. This dissertation provides an overview of how low-cost physical sensing can be combined with machine learning to provide environmental sensing systems at scale, thus using physics in service of society.Item Estimating the Daily Pollen Concentration in the Atmosphere Using Machine Learning and NEXRAD Weather Radar Data(Springer International Publishing, 2019-06-07) Zewdie, Gebreab K.; Lary, David J.; Liu, Xun; Wu, Daji; Levetin, E.; Zewdie, Gebreab K.; Lary, David J.; Liu, Xun; Wu, DajiMillions of people have an allergic reaction to pollen. The impact of pollen allergies is on the rise due to increased pollen levels caused by global warming and the spread of highly invasive weeds. The production, release, and dispersal of pollen depend on the ambient weather conditions. The temperature, rainfall, humidity, cloud cover, and wind are known to affect the amount of pollen in the atmosphere. In the past, various regression techniques have been applied to estimate and forecast the daily pollen concentration in the atmosphere based on the weather conditions. In this research, machine learning methods were applied to the Next Generation Weather Radar (NEXRAD) data to estimate the daily Ambrosia pollen over a 300 km × 300 km region centered on a NEXRAD weather radar. The Neural Network and Random Forest machine learning methods have been employed to develop separate models to estimate Ambrosia pollen over the region. A feasible way of estimating the daily pollen concentration using only the NEXRAD radar data and machine learning methods would lay the foundation to forecast daily pollen at a fine spatial resolution nationally. © 2019, Springer Nature Switzerland AG.Item Holistics 3.0 for Health(MDPI AG, 2014-07-24) Lary, David J.; Woolf, Steven; Faruque, Fazlay; LePage, James P.; Lary, David J.Human health is part of an interdependent multifaceted system. More than ever, we have increasingly large amounts of data on the body, both spatial and non-spatial, its systems, disease and our social and physical environment. These data have a geospatial component. An exciting new era is dawning where we are simultaneously collecting multiple datasets to describe many aspects of health, wellness, human activity, environment and disease. Valuable insights from these datasets can be extracted using massively multivariate computational techniques, such as machine learning, coupled with geospatial techniques. These computational tools help us to understand the topology of the data and provide insights for scientific discovery, decision support and policy formulation. This paper outlines a holistic paradigm called Holistics 3.0 for analyzing health data with a set of examples. Holistics 3.0 combines multiple big datasets set in their geospatial context describing as many areas of a problem as possible with machine learning and causality, to both learn from the data and to construct tools for data-driven decisions.Item Improving Electrofacies Prediction by Combining Supervised and Unsupervised Learning Methods(December 2023) Ippolito, Marco Michele 1979-; Lumley, David; Ferguson, John F.; Delk, Nikki; Brikowski, Thomas H.; Lary, David J.Electrofacies classification from well logs is an indispensable part of seismic interpretation and is important in the determination of sequence stratigraphy, and ultimately reservoir characterization. Although there have been improvements in the tools used to perform this task, it remains laborious, subjective, and error-prone. Achieving a proper classification is complicated by increasing dataset sizes as well as the need for correlated multidisciplinary models. Recent developments in machine learning provide an opportunity to assist interpreters in accomplishing this task while also improving the accuracy of classification results. Applications of machine learning methods for automating facies classification from well logs have previously been explored, however these have mostly focused on evaluations or comparisons of individual algorithms or of ensembles of homogeneous agents. The proposed methods combine heterogeneous agents to enhance prediction accuracy and expedite the assessment of large datasets. This approach seamlessly integrates supervised and unsupervised learning techniques, effectively capitalizing on their individual strengths and mitigating their inherent limitations. The overarching objective is to offer valuable support to geoscientists by not only improving prediction accuracy beyond the constraints of current methodologies, but also by significantly accelerating the evaluation process for progressively expanding datasets. To accomplish this, supervised learning, which establishes a direct mapping from the data domain to the solution domain while introducing some bias to generalize the mapping, is coupled with unsupervised learning, which operates without reliance on similar generalization bias or predefined training data but does not offer a direct mapping between the data and solution domains. This fusion is achieved through the utilization of a joint probability density function (PDF) derived from the supervised classification. The PDF serves to guide the identification of clusters outlined by unsupervised learning. This multi-agent approach can effectively detect bias introduced during training for as many as one in five samples present in well log data, and forms the basis for generating a probability distribution for individual samples, rather than simply assigning a discrete classification for each sample. Consequently, this distribution proves valuable in more accurately representing the continuous nature of well log signals, and captures the intrinsic continuity present within lithological regimes. A modified version of this approach can be applied to rapidly identify regions of interest within complex depositional environments which may involve hundreds or even thousands of boreholes equipped with extensive suites of well log data. This innovative adaptation streamlines the evaluation process, reducing an interpretation task that could have consumed months to just hours. As a result, geoscientists can dramatically scale up their interpretation efforts, increasing their output substantially. The research concludes by establishing a foundation for comparing the predictive accuracy of the proposed approach with that of traditional petrophysical analysis and of core interpretation; where these proven methods serve as benchmark references for the evaluation.Item Magnetic Storm Effects on the Occurrence and Characteristics of Plasma Bubbles(May 2023) Adhya, Purbi 1993-; McKeown, Stephen; Anderson, Phillip C.; Heelis, Roderick A.; Valladares, Cesar; Lary, David J.; Glosser, RobertDuring geomagnetic storms actions of prompt penetration electric fields (PPEF) during the main phase disturbs the equatorial ionosphere. In addition, disturbance dynamo electric fields (DDEF) can follow during the recovery phase to further modify the plasma dynamics of the low-latitude ionosphere. The eastward PPEF and westward DDEF cause sudden or ongoing upward and downward plasma drifts that can cause changes in seasonal-longitudinal occurrence patterns of plasma bubbles. Ionospheric irregularities like plasma bubbles occur all around the year within equatorial latitudes, but their occurrence varies seasonally and with changes in solar activity. The short-term changes in the plasma vertical drift during storms cause enhancement or suppression in the occurrence and intensity of plasma bubbles. Our study investigates the changes in the plasma bubble occurrence pattern and characteristics during different phases of storms. The Communications/Navigation Outages Forecast System (C/NOFS) satellite mission was designed to investigate the ionospheric conditions that lead to the formation of plasma irregularities. We have studied the effects of magnetic storms on the formation and evolution of plasma bubbles during the satellite’s lifetime (2008-2014). During this period encompassing solar minimum and maximum conditions, many magnetic storms of varying intensity developed. Each storm was isolated and divided into initial, main, and recovery phases based on the SYM-H index data observed from geomagnetic observatories. Interplanetary Magnetic Field (IMF) data measured by the Advanced Composition Explorer (ACE) satellite was used to observe fluctuations in magnetic fields during storms. Measurements of plasma density from the Plasma Langmuir Probe (PLP) were used to identify plasma bubble occurrences, and determine their local times, depths, widths, etc. A bubble detection algorithm was developed to detect bubbles from the plasma density data. Measurements of plasma vertical velocities from the Ion Velocity Meter (IVM) were used to determine evening PRE peak velocities and bubble internal vertical velocities. Analysis of 109 storms of varying intensities with available bubble and PRE data between May 2008 and August 2014 has revealed that the most intense plasma bubbles occur during a storm’s main phase when Bz turns southwards as PRE velocities tend to increase during those times. New bubbles develop with large PRE values and the bubble lifetime extends into the recovery phase. Comparisons of bubble depths and internal vertical velocities between the storm’s main phases and quiet periods before have shown significant improvement during storms. The augmentation of the plasma bubbles’ depth and internal velocity become more prominent when the bubble intensities were low during the quiet period before the storms. Furthermore, bubble intensities decrease by the end of the recovery phase along with the decline in the PRE velocities. The growth and decline of the bubble occurrence and characteristics signify the important roles of PPEF and DDEF during storms on the low-latitude ionosphere.Item Phytophthora Megakarya and Phytophthora Palmivora, Closely Related Causal Agents of Cacao Black Pod Rot, Underwent Increases in Genome Sizes and Gene Numbers by Different Mechanisms(2017-03-01) Ali, Shahin S.; Shao, Jonathan; Lary, David J.; Kronmiller, Brent A.; Shen, Danyu; Strem, Mary D.; Amoako-Attah, Ishmael; Akrofi, Andrew Yaw; Begoude, B. A. Didier; ten Hoopen, G. Martijn; Coulibaly, Klotioloma; Kebe, Boubacar Ismael; Melnick, Rachel L.; Guiltinan, Mark J.; Tyler, Brett M.; Meinhardt, Lyndel W.; Bailey, Bryan A.; Lary, David J.Phytophthora megakarya (Pmeg) and Phytophthora palmivora (Ppal) are closely related species causing cacao black pod rot. Although Ppal is a cosmopolitan pathogen, cacao is the only known host of economic importance for Pmeg. Pmeg is more virulent on cacao than Ppal. We sequenced and compared the Pmeg and Ppal genomes and identified virulence-related putative gene models (PGeneM) that may be responsible for their differences in host specificities and virulence. Pmeg and Ppal have estimated genome sizes of 126.88 and 151.23Mb and PGeneM numbers of 42,036 and 44,327, respectively. The evolutionary histories of Pmeg and Ppal appear quite different. Postspeciation, Ppal underwent whole-genome duplication whereas Pmeg has undergone selective increases in PGeneM numbers, likely through accelerated transposable element-driven duplications. Many PGeneMs in both species failed to match transcripts and may represent pseudogenes or cryptic genetic reservoirs. Pmeg appears to have amplified specific gene families, some of which are virulence-related. Analysis of mycelium, zoospore, and in planta transcriptome expression profiles using neural network self-organizing map analysis generated 24 multivariate and nonlinear self-organizing map classes. Many members of the RxLR, necrosis-inducing phytophthora protein, and pectinase genes families were specifically induced in planta. Pmeg displays a diverse virulence-related gene complement similar in size to and potentially of greater diversity than Ppal but it remains likely that the specific functions of the genes determine each species' unique characteristics as pathogens.Item Providing Physical Insights into the Morphology of Spatial and Temporal Distributions of Atmospheric Aerosols Using Machine Learning(2018-05) Wu, Daji; 0000-0002-1791-1009 (Daji, W); Lary, David J.The concentration of airborne particulate matter (PM$_{2.5}$) is a significant environmental and health issue. Many tools have been used to examine the relationship between PM$_{2.5}$ abundance and meteorological variables. Some of these relationships are non-linear, non-Gaussian, and even unknown. Machine Learning provides a broad range of practical solutions to help examine and provide physical insights into these relationships. In this thesis we have used a variety of machine learning approaches. Unsupervised machine learning was used to classify the morphology of PM$_{2.5}$ seasonal cycles in East Asia. Machine learning is able to objectively classify the seasonal cycles, and without apriori assumptions, is able to clearly distinguish between urban and rural areas. We show an example of this in the Sichuan Basin of China. Further, a supervised machine learning approach, random forest is able to identify the key factors associated with each distinct shape of the seasonal cycle, such as the key role placed by the surface type and the built environment. While random forests can be improved by using an optimized ensemble of machine learning approaches (boosting \& bagging), which explores a variety of ensemble methods to choose the algorithm with the best performance with tuned hyperparameters. This optimized approach automatically provides the most important meteorological and surface variables associated with PM$_{2.5}$ concentration. The variables highlighted by optimized machine learning were then examined together with five traditional meteorological features via multiple linear regression (MLR) models, which provide comprehensive physical mechanistic insights into the effect of these variables on the variation of the PM$_{2.5}$ annual cycles, e.g., how these environment variables interact with PM$_{2.5}$ in specific areas. Lastly, the SHapley Additive exPlanation (SHAP) values, which is a consistent measurement of individualized feature attributions in ensemble tree models, were employed to get more information about the impacts of those environmental variables in ensemble tree models. SHAP provided individualized attributions of predictors on the final output. SHAP values were calculated based on ensemble tree models and it didn't assume any linear relationships between predictors and PM$_{2.5}$ concentration like MLR. Results of these impacts given by SHAP were consistent with MLR, but more generally applicable.Item Providing Wavelength Resolved Irradiance Measurements by Using Machine Learning(2021-12-01T06:00:00.000Z) Zhang, Yichao; Biewer, Michael C.; Lary, David J.; Stoneback, Russell; Lou, Xinchou; Anderson, Phillip C.; Da Silveira Rodrigues, FabianoSunlight incident on the Earth’s atmosphere is essential for life and is the driving force for atmospheric photo-chemistry. Atmospheric photo-chemistry is central to understanding urban air quality and the host of associated human health impacts. In this dissertation, two solutions were proposed to address the current lack of real-time wavelength-resolved solar irradiance data across cities. Our first solution is based on the machine learning calibration of low-cost light sensors. These calibrated sensors have a strong performance and can be readily deployed at scale across dense urban environments to measure the wavelength resolved irradiance on a neighborhood scale. This work has been published in MDPI (Zhang et al., 2021). Our second solution is based on the comprehensive dataset from public environmental sensors. We developed another machine learning model to estimate the wavelength resolved solar irradiance from solar zenith angle, earth distance, and multiple environmental dataset, such as relative humidity, total column ozone, earth surface reflectance, and radar reflectivities in the sky. All these factors can be accessed from the public datasets of weather stations and remote sensing systems. Using this solution, wavelength resolved solar irradiance can be estimated in a neighborhood scale, without implementing any additional sensors.Item Survey on the Estimation of Mutual Information Methods as a Measure of Dependency Versus Correlation Analysis(2015-02) Gencaga, Deniz; Malakar, Nabin K.; Lary, David J.; Gencaga, Deniz; Malakar, Nabin K.; Lary, David J.In this survey, we present and compare different approaches to estimate Mutual Information (MI) from data to analyse general dependencies between variables of interest in a system. We demonstrate the performance difference of MI versus correlation analysis, which is only optimal in case of linear dependencies. First, we use a piece-wise constant Bayesian methodology using a general Dirichlet prior. In this estimation method, we use a two-stage approach where we approximate the probability distribution first and then calculate the marginal and joint entropies. Here, we demonstrate the performance of this Bayesian approach versus the others for computing the dependency between different variables. We also compare these with linear correlation analysis. Finally, we apply MI and correlation analysis to the identification of the bias in the determination of the aerosol optical depth (AOD) by the satellite based Moderate Resolution Imaging Spectroradiometer (MODIS) and the ground based AErosol RObotic NETwork (AERONET). Here, we observe that the AOD measurements by these two instruments might be different for the same location. The reason of this bias is explored by quantifying the dependencies between the bias and 15 other variables including cloud cover, surface reflectivity and others.Item The Physical Characterization of Human Autonomic Responses and Health in a Variety of Environmental and Social Contexts(December 2023) Fernando, Bharana Ashen 1992-; Lary, David J.; DeJong, Jeff L.; Glosser, Robert; Lumata, Lloyd; King, Lindsay J.; Lumley, DavidThe human body responds to environmental stimuli in a variety of ways. Combining the physical measurements autonomic responses such as eye movement, heart rate and sweat response may provide a robust characterization of millisecond human interactions with the surrounding environment. Furthermore, the environment directs health outcomes through long-term exposures, which is explored in conjunction with hospitalization data. This thesis also explores a method to detect blinks that would aid in quantifying concepts such as cognitive load, and finally, a software suite for comprehensively analyzing brain, eye, and audio data to dynamically explore the coupled nature of metrics and an application of these techniques to a variety of everyday and specialized activities. Various machine learning techniques are used to analyze this high-dimensional feature-rich data space and to automate the extraction of interesting events.Item Using a Comprehensive Characterization of the Physical Environment and Machine Learning to Forecast the Abundance of Airborne Pollen(2019-05) Zewdie, Gebreab K.; 0000-0002-1797-8967 (Zewdie, GK); Lary, David J.It is known that approximately 50 million Americans have allergic diseases. Airborne pollen is a significant trigger for several of these allergic diseases. Among all sources, Ambrosia (ragweed) is known for its abundant production of pollen and its potent allergic effect. It is prevalent in North America and the Northern temperate regions in general. Hence, estimating and predicting the daily atmospheric concentration of pollen (ragweed pollen in particular) is useful for both people with allergies and for the health professionals who care for them. In this study we show that a suite of meteorological and land surface parameters along with atmospheric trajectory analysis can be used together with machine learning to successfully estimate (forecast) the daily pollen concentration. Our main data sources are from the MERRA (Modern-Era Retrospective analysis for Research Applications), ECMWF (European Centre for Medium Medium-Range Weather Forecast) as well as the NEXt-generation weather RADar (NEXRAD). We used supervised machine learning methods ranging from linear models such as the Bayesian ridge, the random forest and gradient boosting (ensemble tree based learners), simple and deep neural networks and support vector machines. The performance of the different machine learning methods are independently validated using a test data partitioned based on the holdout method from the total dataset. Additionally, in order to estimate pollen over a large spatial scale, we developed neural network and random forest machine learning models to estimate pollen over a 300 km × 300 km region centered at a NEXRAD radar site at a resolution of 0.5 km × 0.5 km . In this case the models are developed over a 10 km × 10 km area solely on the basis of NEXRAD parameters and applied to all pixels having enough NEXRAD measurements. The feasibility of estimating the daily pollen concentration using only the NEXRAD radar data and machine learning methods would lay the foundation to forecast daily pollen at a fine spatial resolution over the contiguous United States. The results show that despite very different approaches used by the neural network and random forest, the two machine learning methods highlighted high levels of pollen in approximately the same regions. Various independent variable importance estimation techniques were used to calculate and rank the relative importance of the available weather and land surface parameter for estimating pollen abundance. Surface albedo, soil temperature, vegetation greenness fraction, wind speed were among most influential parameters for forecasting allergic pollen. Among the NEXRAD measurements the reflectivity and direction of the wind were the top predictors. The physical interpretation of each predictor variable and its influence on the prediction of allergic pollen are presented towards the end of the dissertation.Item Using Remote Control Aerial Vehicles to Study Variability of Airborne Particulates(Libertas Academica Ltd, 2015-08-04) Harrison, William A.; Lary, David J.; Nathan, Brian J.; Moore, Alec G.; Harrison, William A.; Lary, David J.; Nathan, Brian J.; Moore, Alec G.Airborne particulates play a significant role in the atmospheric radiative balance and impact human health. To characterize this impact, global-scale observations and data products are needed. Satellite products allow for this global coverage but require in situ validations. This study used a remote-controlled aerial vehicle to look at the horizontal, vertical, and temporal variability of airborne particulates within the first 150 m of the atmosphere. Four flights were conducted on December 4, 2014, between 12:00 pm and 5:00 pm local time. The first three flights flew a pattern of increasing altitude up to 140 m. The fourth flight was conducted at a near-constant altitude of 60 m. The mean PM_{2.5} concentration for the three flights with varying altitude was 36.3 μg/m³, with the highest concentration occurring below 10 m altitude. The overall vertical variation was very small with a standard deviation of only 3.6 μg/m³. PM_{2.5} concentration also did not change much throughout the day with mean concentrations for the altitude-varying flights of 35.1, 37.2, and 36.8 μg/m³. The fourth flight, flown at a near-constant altitude, had a lower concentration of 23.5 μg/m3. © 2015, the authors, publisher and licensee Libertas Academica Limited.