Environmental monitoring refers to a host of activities involving the sampling or sensing of diverse properties from an environment in an effort to monitor, study and overall better understand it. While potentially rich and scientifically valuable, these data often create challenging interpretation tasks because of their volume and complexity. This thesis explores the efficiency of Computer Vision-based frameworks towards the processing of large amounts of visual environmental monitoring data. While considering every potential type of visual environmental monitoring measurement is not possible, this thesis elects three data streams as representatives of diverse monitoring layouts: visual out-of-water stream, visual underwater stream and active acoustic underwater stream. Detailed structure, objectives, challenges, solutions and insights from each of them are presented and used to assess the feasibility of Computer Vision within the environmental monitoring context. This thesis starts by providing an in-depth analysis of the definition and goals of environmental monitoring, as well as the Computer Vision systems typically used in conjunction with it. The document continues by studying the visual out-of-water stream via the design of a novel system employing a contrast-guided approach towards the enhancement of low-light underwater images. This enhancement system outperforms multiple state-of-the-art methods, as supported by a group of commonly-employed metrics. A pair of detection frameworks capable of identifying schools of herring, salmon, hake and swarms of krill are also presented in this document. The inputs used in their development, echograms, are visual representations of acoustic backscatter data from echosounder instruments, thus contemplating the active acoustic underwater stream. These detectors use different Deep Learning paradigms to account for the unique challenges presented by each pelagic species. Specifically, the detection of krill and finfish is accomplish with a novel semantic segmentation network (U-MSAA-Net) capable of leveraging local and contextual information from feature maps of multiple scales. In order to explore the out-of-water visual data stream, we examine a large dataset composed by years-worth of images from a coastal region with strong marine vessels traffic, which has been associated with significant anthropogenic footprints upon marine environments. A novel system that involves ``traditional'' Computer Vision and Deep Learning is proposed for the identification of such vessels under diverse visual appearances on this monitoring imagery. Thorough experimentation shows that this system is able to efficiently detect vessels of diverse sizes, shapes, colors and levels of visibility. The results and reflections presented in this thesis reinforce the hypothesis that Computer Vision offers an extremely powerful set of methods for the automatic, accurate, time- and space-efficient interpretation of large amounts of visual environmental monitoring data, as detailed in the remainder of this work. / Graduate

Detecting change in stream channel features over time is important in understanding channel morphology and the effects of both natural and anthropogenic influences. Channel features historically, and now currently, are being measured using a variety of ground survey techniques. These surveys require substantial time commitments and funding to complete. Light Detection and Ranging (LiDAR) is an airborne laser mapping technology that holds promise to provide an alternative to ground-based survey methods. For this study, ground surveys were used to verify the accuracy of data collected using airborne LiDAR. Fifty nine cross-sectional profiles were surveyed in the Little Creek watershed at Cal Poly’s Swanton Pacific Ranch and compared to LiDAR-generated profiles of the same location. LiDAR data were collected in two flights during April and May of 2002. The vertical accuracy of LiDAR elevations was determined to be 0.610 m RSME based on a point-to-point comparison of the elevation of ground survey points in each cross-sectional profile to the corresponding LiDAR elevation. The average ground spacing of the LiDAR survey within the study area was one point every 5.2 square meters. In comparison to ground surveys it was found that with this level of vertical precision and horizontal resolution it would be difficult to detect change in bankfull channel characteristics of a relatively small channel, such as Little Creek. These difficulties are largely attributed to poor point coverage in forested, steep, and mountainous terrain, along with technological limitations of LiDAR that have since improved.

Multimedia environmental monitoring involves capturing valuable visual and audio information from the field station. This will permit the environmentalists and researchers to analyze the habitat and vegetation of a region with respect to other environmental specifics like temperature, soil moisture, etc. This thesis deals with the development of a test bed for multimedia monitoring by capturing image information and making it available for the public. A USB camera and a Single board computer are used to capture images at a specified frequency. A web-client is designed to display the image data and establish a secured remote access to reconfigure the field station. The development includes two modes of image acquisition including a basic activity recognition algorithm. Good quality images are captured with the cost for development of the system being less than 2 hundred dollars.

In recent years, people have raised their alertness to the hazard of air pollution. Sulfur dioxide is one of the most dangerous chemical compounds among those air pollutants. A study on removing sulfur dioxide from an air stream by adsorption using wood chips as the adsorbent is presented in this thesis. The reason for using wood as an adsorbent is that wood is a porous material and possesses a large surface of cell cavities which can hold a great amount of moisture. As sulfur dioxide gas is passed through the wood bed, it would be either condensed in the cell space of the wood by intermolecular attraction, adsorption or dissolved in the moisture held in the wood. This work was started with a review of literature. Then related references were collected and a proposal written. Douglas fir was chosen for the experiment because it is the most common kind of wood in the Pacific Northwest. After the process and proper equipment was set up, woodchips were screened and dried to prepare for further experiments. It was decided to use three different concentrations of sulfur dioxide. For each of the concentrations of sulfur dioxide, five levels of moisture (0%', 11%, 20%, 50% and saturated) were assigned to the selected woodchips. Fifteen combinations or experiments were done for the research. The results of the experiments show that dry wood (0% moisture content) had comparatively low characteristics in the adsorption of sulfur dioxide. For instance, at an influent so2 concentration of 1.12 ppm., about· 6 grams of dry woodchips adsorbed 29.37 µg. of sulfur dioxide in comparison to 2Q90.5 µg. of SO2 adsorbed in the same weight of woodchips but saturated with moisture. At an influent SO2 concentration of 1.83 ppm., the adsorption of sulfur dioxide increased from 7.73 µg. for the dry wood to 745.15µg. in the water saturated wood. For an influent SO2 concentration of 4.60ppm., dry wood adsorbed 15.26 µg. of SO2 while the moisture saturated wood adsorbed 1446.2 µg. The amount of dry woodchips used in above mentioned experiments were all about 6 grams. These data show that the moisture saturated wood adsorbed about 90 times the amount of sulfur dioxide that the dry wood adsorbed. It is clear that the wood adsorptivity increased with increasing moisture content. It was also found that wood adsorptivity and retention time were affected by the different flow rate of carrier gas. The figures show that most of the data fit a Freundlich equation. Other equations were developed to calculate the adsorptivity and retention time by obtaining the influent and effluent concentration of sulfur dioxide through the adsorbent bed.

The western gray squirrel (Sciurus griseus griseus Ord) occur only in the Pacific states and in a small corner of extreme western Nevada. Field studies of aspects of the ecology of the most widely distributed subspecies of western gray squirrel (S. griseus) were conducted at three sites in the ponderosa pine-Oregon white oak zone on the eastern slopes of Mt. Hood, Wasco County, Oregon, between 1981 and 1989. From 1981 through 1987, a Hunter Cooperation Program provided remains of squirrels shot during the annual late summer-early autumn hunting season. Other data were gathered through capture and release studies, radiotelemetry, and habitat analysis. The age structure of these populations, as revealed by examination of squirrels killed by hunters, is predominantly mature animals. Also, the population as estimated by hunter success during the Hunter Cooperation Program and by field observations appears to be declining. In the sites studied, western gray squirrels have two seasons of reproductive activity annually. Some individuals mate from January through March; their young emerge from the nest in May and June. Other individuals mate during May and June; their young emerge in August and September. The latter matings are usually the most productive. Analysis of nest trees and their surroundings revealed that nests are most likely to be located in mature trees that have well-developed crowns and occur in stands with a high degree of canopy closure. Nest trees were usually located within approximately 180 meters of permanent water and on sites with a south-southeasterly exposure. Home ranges of squirrels examined in this study were usually large compared to home ranges reported for this species elsewhere in its range. Given the cost of having large home ranges, it seems possible that the squirrels at these study sites exist in less than optimal ecological circumstances. Human factors that may have contributed to the decline of the populations in this study may include the nature of logging activities and the timing of the annual hunting season. Limiting factors other than human activities may include competition with other mast-consuming animals (mule deer, elk, wild turkeys, and three other species of squirrels) for limited and variable mast crops.

A thesis submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the requirements for the degree of Masters in Science Centre for African Ecology, School of Plant Animal and Environmental Sciences, 2011. / Habitat selection occurs across a range of different spatial scales and is influenced by a variety of factors, ultimately determining how animals distribute themselves across the landscape. Studying the decisions that an individual animal makes across different levels of selection, from its choice in dietary item to predator avoidance strategies, is a fundamental link in understanding the response of groups of animals and ultimately entire populations that may provide insight into population performance. The study formed part of a broader study focused on the decline of rare antelope species. Specifically this study was aimed at establishing the home range and resource use of sable antelope in a region where they were initially expected to be thriving. The levels of selection covered in this study are: the location of home ranges of an individual or social group within the landscape; the use of various habitat components within the home range; and the procurement of food items within those habitats. At the highest level, the broad objectives were (1) to determine where sable occupied home ranges within the landscape, indicating the suitability of various landscape units to sustain sable populations and (2) to determine the relative use of habitat types within those home ranges that may enable sable to avoid predation and acquire resources required to survive and reproduce. At the lowest level of selection the characteristics of forage selection and how the grass quality in the different vegetation types during different seasons affects the success of sable herds was explored. The broad objectives were (1) to determine the effect of seasonal flooding and rainfall on grass greenness in the floodplains and upland vegetation types and the consequent use of those vegetation types by sable antelope and (2) to determine how exploitation of resources in the floodplains and in the uplands contributed to the nutritional status of sable. I additionally quantified the time spent browsing and determined the composition of the browse component of the diet of sable. Adult female sable from each of three adjacent sable herds were fitted with GPS collars providing hourly GPS co-ordinates. Adaptive LoCoH was used to determine home range location and annual, seasonal and core home range extents. A vegetation map was created and the number of GPS locations within each vegetation type was counted to determine their relative use in relation to availability within the home ranges. GPS collars were used to locate herds daily so that foraging observations of browsing and characteristics of the grasses grazed could be attained. Acceptability and dietary contributions of grass species and browse were determined for each sable herd during different seasons. The availability of grass species on the floodplain grasslands and in the upland grasslands and woodlands was estimated. Water and the availability of key resources posed a constraint on where sable home ranges were established. Sable simply did not occupy the region in the north of the study area further than 7 km from permanent water and floodplain grasslands. Herds generally avoided open savanna, mopane woodlands and Kalahari apple-leaf woodlands characterised by sparse grass cover, particularly during the dry season. Home ranges were relatively small compared 4 to the range estimates from herds in Kruger National Park. There was no obvious seasonal difference in home range extent nor were there large areas of overlap between home ranges of adjacent herds. Observations during the study indicated that competitor species, including zebra and wildebeest, concentrated on the floodplain grasslands. Throughout the year H. dissoluta was the most strongly favoured grass species and contributed most to the diet of the sable herds in both the wet and dry season. During the dry season sable herds expanded their diet to include Aristida stipitata and Aristida meridionalis which are generally considered poor forage value species for cattle but that retained some greenness. Additionally, the contribution of browse, especially the leaves of Croton megalobotrys, Philenoptera nelsii and Combretum mossambicense and the flowers of Kigelia africana, constituted an important bridging resource during the extended dry season. Crude faecal protein levels remained above the suggested maintenance levels throughout the annual cycle. Crude faecal protein levels were elevated prior to calving when sable spent more time foraging on the floodplain grasslands where high value forage species such as Paspalum scrobiculatum, Panicum repens and Urochloa mossambicense and sedges were eaten. Indications are that the constraint posed by the distribution of water within the landscape, rather than resource limitations within occupied home ranges, are the primary limitation to population performance in the Kwedi concession. / Wilderness Safari, the Wilderness Wildlife Trust, the Conservation Foundation, Classic Africa and National Research Foundation (NRF) of South Africa.

Effective risk communication is based on appropriate risk characterization. A reevaluation of the 1987 US EPA radon risk categorization of Unicoi County Tennessee was conducted using in-home radon concentrations, determined in a long-term monitoring study. Radon concentrations were measured in 69 homes using Electret Passive Environmental Radon Monitors (E-PERM’s), following standard methods. Radon concentrations determined in this study (avg. 4.03 ± 3.04) were significantly higher than those measured in the USEPA study (avg. 1.96 ± 1.08). Using this study’s data, the risk categorization was recalculated with the US EPA Radon Index Matrix Model. The model re-categorized Unicoi County from a moderate to a high risk zone classification. These results suggest that the health risks associated with in-home radon concentrations are inaccurately categorized and communicated to the citizens of Unicoi County, Tennessee.

The purpose of this study was to perform an indoor air quality (IAQ) investigation at the Center for Applied Reproductive Science (CARS) to assess whether VOCs exist at levels dangerous to embryo. Formaldehyde, n-hexane, benzene, and styrene concentrations were measured at six locations. Formaldehyde concentrations were comparable to office and residential indoor air. N-hexane, benzene, and styrene were not detected. In addition, acetaldehyde, ethanol, and isopropyl alcohol were detected. IAQ parameters (carbon dioxide, temperature, humidity, pressure, and particulates) were measured at 22 sites monthly for one year. Temperature and humidity readings were within Environmental Protection Agency recommendations. Particulate concentrations were below Occupational Safety and Health Administration standards. Pressure readings indicated the facility was under a negative pressure. Carbon dioxide concentrations exceeded recommendations established by American Society for Heating, Refrigeration, and Air Conditioning Engineers (ASHRAE). Recommendations include assessing air intakes to assure dampers are adjusted to allow 15 ft3/min/person of fresh air established by ASHRAE.

Fluctuating asymmetry (FA) is commonly thought to be a predictor of environmental stress. However, the relationship between FA and established indicators for environmental stress has received little attention. In this study, 10-38 specimens of the freshwater fish Rhinichthys atratulus were collected from 15 natural populations under varying amounts of environmental stress. Asymmetry measurements in three bilateral characters of the specimens were used to investigate the relationship between FA and established indicators of environmental stress. Significant differences in the magnitude of FA were observed between sampling locations. However, the relationship between estimates of FA and established indicators produced varying results. The present study concludes with a discussion on the usefulness of FA as a bioindicator for environmental stress and implications for future studies.

The aim of this report is to contribute to the base of knowledge on environmental monitoring by increasing understanding of how eDNA, electrofishing and sampling fishing may be used to examine fish biodiversity. It also aims at understanding if fish indexes developed within the Water Framework Directive reflect biodiversity, as well as the potential of eDNA data to serve as input to these indexes. This was done by using three different approaches. Firstly, in order to establish which of the methods eDNA, electrofishing and sampling fishing is more suitable to measure the different dimensions of biodiversity (species richness, species evenness and genetic diversity), a literature review comparing the different methods was carried out. It was found that eDNA yields a more detailed results for species richness, electrofishing yields better results for species evenness and sampling fishing is outperformed by eDNA and electrofishing alike. Both electrofishing and sampling fishing may collect data for genetic diversity analysis, however electrofishing outperforms sampling fishing with regards to amount of species caught, making electrofishing a more suitable data collection method. Secondly, in order to gain insight on practical usage of eDNA, a case study of Spjutmo (Dalarna county) was reviewed. It was established that eDNA generated more detailed information of species richness in the case of Spjutmo (as compared to electrofishing). The relative abundance data generated by the eDNA study might be seen as a measure of species evenness. However, electrofishing yielded data which may serve as input to species evenness indices. To the best knowledge of the author, none of the methods generated data on genetic diversity in this specific case. Officials from the energy company Fortum and the county board of Dalarna were also interviewed in order to get insight on what potential they see for eDNA to contribute to environmental monitoring. Both officials point at the ability to estimate abundance as a desired feature, hence a better understanding of what the relative abundance results indicates is wanted. The two interviews indicate that this understanding is an important feature to develop in order to make metabarcoding studies effective in current environmental monitoring. Thirdly, in order to understand if fish indexes developed within the Water Framework Directive reflect biodiversity, a literature review was performed. It was found that, all but one of the compared indexes incorporates or somewhat incorporates species richness. However, only five indexes are indicative or somewhat indicative of species richness. Species evenness is incorporated or somewhat incorporated by two indexes, which are also indicative or somewhat indicative of species evenness. None of the indexes incorporate or indicate genetic diversity. Within the third literature review, the potential of eDNA data to serve as input to current fish-based indexes developed within the Water Framework Directive, was studied. It was found that eDNA data may serve as input to only one index in its present form. However, five indexes also use proportional information (e.g. proportion of tolerant species), which possibly could be provided by eDNA data. The index where usage of eDNA data is currently possible uses presence-absence information. / Målet med denna rapport är att bidra till kunskapsläget kring miljöövervakning genom att öka förståelsen för hur eDNA, elfiske och provfiske kan användas för att undersöka fisk biodiversitet. Målet är också att förstå om fisk-index utvecklade inom ramen för det Europeiska vattendirektivet reflekterar biodiversitet samt om data från eDNA kan utgöra input till dessa index. För att uppfylla dessa mål användes tre metoder. För att etablera vilken av metoderna eDNA, elfiske och provfiske är mer lämpad att mäta de olika dimensionerna av biodiversitet (artrikedom, distribution av arter och genetisk diversitet), genomfördes en litteraturstudie. Slutsatsen kunde dras att eDNA mäter artrikedom med högst noggrannhet, elfiske mäter distribution av arter mer detaljerat och att provfiske överträffas av både eDNA och elfiske i alla dimensioner. Både elfiske och provfiske kan samla data för analys av genetisk diversitet, men elfiske överträffar provfiske gällande hur många arter som fångas, vilket gör elfiske mer lämpligt som metod att samla in data för genetisk analys. För att få praktisk insikt i ett fall där eDNA använts, granskades en fallstudie från Spjutmo (i Dalarnas län). eDNA genererade mer detaljerad information om artrikedom än elfiske i detta fall. Datan genererad av eDNA kring relativ abundans mellan arter skulle kunna tolkas som ett mått på distribution av arter. Data genererad av elfiske kan å andra sidan användas som input till olika index för distribution av arter. Författaren veterligen, genererade varken eDNA eller elfiske mått på genetisk diversitet i detta specifika fall. Två personer, en från Fortum och en från länsstyrelsen Dalarna intervjuades också för att få insikt i deras syn på potentialen av att använda eDNA som ett miljöövervakningsverktyg. Båda intervjupersonerna pekade på att en bättre förståelse av de relativa abundansvärdena indikerar är önskad. Båda intervjupersonerna pekade på att det är en viktig aspekt för att metabarcoding studier ska vara effektiva i nuvarande miljöövervakning. För att förstå om fisk-index utvecklade för EU’s vattendirektiv reflekterar biodiversitet, genomfördes en komparativ litteraturstudie av index. Alla index förutom ett inkorporerar eller delvis inkorporerar artrikedom. Bara fem indikerar eller delvis indikerar artrikedom. Distribution av arter inkorporeras eller delvis inkorporeras av två index som också indikerar eller delvis indikerar distribution av arter. Inom den komparativa litteraturstudien av index, studerades även potential att fungera som input av data genererad av eDNA till indexen. Data genererad av eDNA kan i dagsläget fungera som input till ett av indexen. Fem index använder någon form av proportionell data (t.ex. proportion av toleranta arter), som möjligen skulle kunna ges av eDNA. Indexet till vilket det är möjligt att använda eDNA data använder närvarande-ej närvarande information som input.