79717 research outputs found
Sort by
Laws, Regulations and Action Plans for per- and polyfluoroalkyl substances found in Michigan drinking water supplies
Get PDFPer- and polyfluoroalkyl substances (PFASs) have emerged in the last half century as concerning global contaminants. PFASs have been found in drinking water systems causing negative health impacts for those who rely on this as their primary source of drinking water. PFASs are man-made industrial chemicals composed of carbon chains bonded to fluorine and other substances and cause detrimental impacts to the environment and human health (TOMWC, 2019c). While PFASs are not a new substance, the adverse effects are just starting to be realized. In response, Michigan is pursuing a leadership role in policy, research, training and clean-up/remediation plans for PFAS contamination with the implementation of the Michigan PFAS Action Response Team (MPART). The State of Michigan is poised to be one of the first states to enact Maximum Contamination Levels (MCLs) for some of the most impactful PFASs.
This capstone project seeks to determine the effectiveness of the current legislation and PFAS Action Plans in protecting Michigan citizens from PFAS contaminated drinking water. A review of the current and proposed federal and Michigan legislation was completed as well as the impacts associated with PFASs exposure which identified the success of current policies with regards to the environment, the ecosystem, human health, the economy and the socio-political scene. Additionally, local Northern Michigan government officials and employees of environmental organizations were surveyed to ascertain their opinions of the effectiveness of MPART, the current regulated PFASs levels and to determine what more can be done to assist local areas with current and future PFAS contamination. Key findings of the literature review and the survey illustrate a need for stricter and more detailed legislation that include nationwide MCLs for individual PFASs, further research on the impacts, with emphasis on human health. In addition, it was found that while it might be too early to determine the effectiveness of MPART, there is evidence that local officials lack the necessary training to adequately administer best practices to help mitigate PFAS contamination
Satellite data ingestion tool
Get PDFSatellite data ingestion tool is an automated database application for processing and archiving the ocean and land data that is broadcasted by a remote sensing satellite. This Ingestion system has a two-tier architecture, with data processing algorithm forming the first tier and the database server forming the second tier. The raw satellite data is an HDF (Hierarchical Data Format) file. An HDF reader has been developed that reads this satellite data file to extract the required data. A program has been written, that converts HDF files into images. A database schema has been developed in such a way that all important parameters of the satellite file can be inserted into it along with the locations of data and image files.
This report consists of a detailed description of design and implementation of this ingestion tool along with the design of the database schema
Local Climate Zone Classification Using Random Forests
Get PDFObjective: The goal of this project is to recreate aspects of the article “Comparison between convolutional neural networks and random forest for local climate zone classification in mega urban areas using Landsat images” (Yoo et al., 2019), where methods for predicting LCZ classes for four large cities throughout the world were compared. To do so, a small training dataset from the 2017 Institute of Electrical and Electronics Engineers (IEEE) Geoscience and Remote Sensing Society (GRSS) Data Fusion Contest (Tuia et al., 2017) was used as ground truth for LCZ classes. This was combined with Landsat 8 satellite input data to create a series of models, which were then compared to a larger, full LCZ layer for each city and assessed for accuracy. The primary types of models considered in the Yoo et al. (2019) work were random forests and convolutional neural networks. However, for this project the focus will be only on random forests. In addition, this investigation targets just Hong Kong. This city was chosen because each LCZ that is classified has at least 4 polygons. Finally, here, a classification scheme like the one used by the World Urban Database and Access Portal Tools (WUDAPT) project, denoted as Scheme 1 in Yoo et al. (2019) will be the focus, with comparisons between accuracy at different values of a tuning parameter.
All code and higher resolution images for this project can be found on GitHub at https://github.com/erickabsmith/masters-project-lcz-classification
Benefits and Trends of Sustainable Building Initiatives in Parks and Protected Areas
Get PDFEnsuring sustainable development is a growing societal concern, providing rationale to examine the current role and efforts towards implementing sustainable building initiatives within the United States. Sustainable building initiatives refer to the design, operation, and maintenance of a building while considering its conservation efforts and pollution reduction performance. These initiatives are of importance as buildings are significant consumers of resources and sources of emission. Parks and protected areas are tasked with finding balance between management of the natural and built environment. As such, parks and protected areas provide an applicable focus area regarding sustainable building initiatives. The overarching goal of this capstone report is to review the application of sustainable building initiatives within parks and protected areas as means of aligning with managerial objectives while simultaneously providing ecological, social, and economic advantages. Various case study examples examined throughout the project display avenues of effective implementation of sustainable building initiatives. Through this capstone project, it is recommended that policy be directed at maintaining and enhancing protection of intact ecosystems while considering externalities of necessary development. In addition to implementing sustainable building features, it would benefit global sustainability goals for agencies to incorporate interpretive displays to communicate the advantages of their efforts and provide calls to action for visitors. Finally, it is also of high importance to focus future efforts on renovations and increasing a site’s resilience as new construction is far less prevalent than previously constructed sites that are in need of upgrades
A benchmark suite for parallel processors : part I
Get PDFAdvanced computer architectures are centered around the parallel computer systems. This project is focused on the experiment on two parallel computer architectures : Sequent Balance 21000 shared memory multiprocessor and Cogent XTM distributed system.
A set of benchmark programs are implemented using "C" language and Dynix parallel programming library on the Sequent system and "Linda" parallel programming primitives on the Cogent XTM.
In this paper, the following seven programs are discussed and the benchmark results are presented.
1. Parallel Matrix Multiplication
2. Disk File I/0
3. Saturating
4. Parallel Enumeration Sort
5. Memory Transfer
6. System Math Functions
7. Linpack routine
Spruce-Fir Moss Spider (Microhexura Montivaga) Monitoring Plan
Get PDFOn June 20, 2019, the U.S. Fish and Wildlife Service (USFWS) announced it would conduct a 5-year status review of the Spruce-Fir Moss Spider as part of the process mandated by the Endangered Species Act. Although the Spruce-Fir Moss Spider has been listed as endangered since 1994, minimal research has been done on its basic ecology, and there is presently no long-term monitoring plan in place. This inhibits proper management of the species (USFWS 2019).
It is also worth noting that few field surveys have been done on the Spruce-Fir Moss Spider because there are only a handful of people with sufficient experience and skill to find and identify this cryptic species , and because frequent surveys would disturb the spider’s limited and fragile habitat. Furthermore, given the Spruce-Fir Moss Spider’s endangered status, observers must employ non-lethal survey methods. Traditional methods that kill the target species would undermine conservation efforts (Lecq et al. 2015). This further complicates research on the species.
In light of these challenges, as well as limited availability of human and material resources, this plan focuses on occupancy rather than abundance. Compared to abundance studies, occupancy studies tend to be more cost effective and also more appropriate for cryptic species (Dibner et al. 2017).
I hope that the information presented here will contribute in some small way to the development of a long-term monitoring plan for this difficult to monitor, endangered species
Discrete Morse Theory and Persistent Homology
Get PDFIn this paper we explain at the theoretical level how discrete Morse theory can provide us a more efficient approach to compute persistent homologies. In achieving so we also provide a framework for discrete Morse theory to be applied to persistent homology for other purposes
A Multivariate Analysis of Pacific Arctic Surface Waters
Get PDFThe Pacific Arctic is undergoing rapid biogeochemical changes in response to warming air temperatures caused by climate forcing. This is manifesting as changes in seasonal sea ice thickness and sea ice extent, as well as changes in primary production within surface waters. The data and samples analyzed here were collected on the Healy 1203 October cruise in 2012. Utilizing a Surface Underway System, water was manually sampled to acquire nutrient data. A Semi Automated Filtration System was used for particulate organic carbon (POC) sample acquisition, and an Equilibrated Inlet Mass Spectrometer was used to collect O2/Ar ratios. Surface Underway System allowed for high resolution data acquisition over a large physical and temporal area, and we are thus able to pinpoint several acute instances of physical and biological processes over the Bering, Chukchi and Beaufort Seas by using a combination of physical, nutrient, dissolved oxygen, and particulate data. The variables analyzed here are temperature (°C), salinity, nitrate (NO3 - ), ammonium (NH4 + ), silicate (SiO4 4- ), particulate organic carbon (POC), and O2/Ar ratios. The results show the Bering Sea had the highest average temperatures (4.76°C), followed by the Chukchi (1.85°C) and lastly the Beaufort (1.78°C). Nitrate, ammonium, silicate, and particulate organic carbon (POC) values were significantly different across basins (p>0.05) and O2/Ar ratios as well, with the highest values located in the Bering and the lowest values located in the Beaufort in everything but silicate (Beaufort max=26.2 μM). When compared temporally, nitrate and SiO4 4- concentrations within the Bering Sea were not significantly different (p<0.05). Ammonium concentrations were significantly different between transits in the Bering, as well as POC concentrations and O2/Ar ratios. Nutrient data within the Chukchi revealed no significant differences between transits, however O2/Ar ratios were significantly different. Nitrate concentrations within the Beaufort were significantly different when compared temporally; however, all other variables did not have significant differences in concentrations. This analysis identified five instances of vertical and horizontal mixing features (two in Bering Sea, two in Chukchi Sea, one in Beaufort Sea), two biological responses (one in the Bering Sea, one in the Chukchi Sea), and two instances of river input (one in the Bering Sea, one in Beaufort Sea). We also identified two notable variations (Chukchi Sea) that do not closely follow expected responses to these events, implying the occurrence of process beyond the scope of this analysis. Limitations to this analysis include a lack of data deeper than surface waters, which limits our ability to further understand and classify the variations shown; therefore, future work should include data within different depths of the water column
Exploring Outlier Exposure Methods for Deep Margin-Based Anomaly Detection
Get PDFIn standard training regimes, one assumes that the classes presented to a model constitute all of the classes that the model will encounter when it is deployed. In real deployment scenarios, however, a model can sometimes encounter situations or objects that it has never seen. When these scenarios are safety-critical, a model’s response to an out-of-distribution (OOD) input can be the difference between success and catastrophic failure. In some cases, one has access to some OOD examples and can exploit these during training to make the model more robust at deployment time. This technique is known as outlier exposure (OE), and it has been shown in the literature to improve novelty detection performance.
In this project, two OE strategies, one established and one speculative, are explored in two different deployment scenarios: one where OE data are representative of the OOD examples that will be seen during test time and one where they are not. Lastly, an attempt is made to better understand these OE techniques when used in tandem with deep margin-based classifiers, an approach that has not yet appeared in the literature
Self-adaptive Subsets for Stretched and Rotated Textures in Digital Image Correlation
Get PDFDigital image correlation (DIC) uses pairs of images, one in a reference state and one after in situ manipulation of the object, to measure full-field displacements and strains by tracking the motion of small image subsets. Subset selection is always a crucial part of DIC technique as it has a large impact on the resolution and accuracy of overall results. In traditional DIC with controlled and optimized isotropic speckle applied to the object, subsets of the same fixed size are appropriate. However, for applications where naturally occurring material texture is used for correlation, ideal subsets would adjust to local inhomogeneity and anisotropy. In this paper, three methods are introduced to adapt the subset to its local material texture or speckle pattern under three different scenarios: 1) unevenly distributed speckles, 2) stretched speckles, and 3) stretched and rotated speckles. The adaptive subset will automatically adjust its size based on the quality of local speckles and the stretch ratio of reference image. For rotation situation, the estimated rotation angle of each subset will be provided with high accuracy