This part of the thesis is confidential.

Geringfügig überarb. Fassung der Diss. Univ., Diss.--Zürich, 2004.

Not available / text

The eyes are one of the most expressive non-verbal tools a person has and they are able to communicate a great deal to the outside world about the intentions of that person. Being able to decipher these communications through robust and non-intrusive gaze tracking techniques is increasingly important as we look toward improving Human-Computer Interaction (HCI). Traditionally, devices which are able to determine a user's gaze are large, expensive and often restrictive. This work investigates the prospect of using common mobile devices such as tablets and phones as an alternative means for obtaining a user's gaze. Mobile devices now often contain high resolution cameras, and their ever increasing computational power allows increasingly complex algorithms to be performed in real time. A mobile solution allows us to turn that device into a dedicated portable gaze-tracking device for use in a wide variety of situations. This work specifically looks at where the challenges lie in transitioning current state-of-the-art gaze methodologies to mobile devices and suggests novel solutions to counteract the specific challenges of the medium. In particular, when the mobile device is held in the hands fast changes in position and orientation of the user can occur. In addition, since these devices lack the technologies typically ubiquitous to gaze estimation such as infra-red lighting, novel alternatives are required that work under common everyday conditions. A person's gaze can be determined through both their head pose as well as the orientation of the eye relative to the head. To meet the challenges outlined a geometric approach is taken where a new model for each is introduced that by design are completely synchronised through a common origin. First, a novel 3D head-pose estimation model called the 2.5D Constrained Local Model (2.5D CLM) is introduced that directly and reliably obtains the head-pose from a monocular camera. Then, a new model for gaze-estimation is introduced -- the Constrained Geometric Binocular Model (CGBM), where the visual ray representing the gaze from each eye is jointly optimised to intersect a known monitor plane in 3D space. The potential for both is that the burden of calibration is placed on the camera and monitor setup, which on mobile devices are fixed and can be determined during factory construction. In turn, the user requires either no calibration or optionally a one-time estimation of the visual offset angle. This work details the new models and specifically investigates their applicability and suitability in terms of their potential to be used on mobile platforms.

Three methods for claims reserving are compared on two data sets. The first two methods are the commonly used chain ladder method that uses aggregated payments and the relatively new method, double chain ladder, that apart from the payments data also uses the number of reported claims. The third method is more advanced, data on micro-level is needed such as the reporting delay and the number of payment periods for every single claim. The two data sets that are used consist of claims with typically shorter and longer settlement time, respectively. The questions considered are if you can gain anything from using a method that is more advanced than the chain ladder method and if the gain differs from the two data sets. The methods are compared by simulating the reserves distributions as well as comparing the point estimates of the reserve with the real out-of-sample reserve. The results show that there is no gain in using the micro-level method considered. The double chain lad- der method on the other hand performs better than the chain ladder method. The difference between the two data sets is that the reserve in the data set with longer settlement times is harder to estimate, but no difference can be seen when it comes to method choice. / Tre reservsättningsmetoder jämförs på två dataset. De första två metoderna är den välkända chain ladder-metoden som använder sig av aggregerade utbetalningar samt den relativt nya metoden double chain ladder som förutom utbetalningarna använder sig av antalet anmälda skador. Den tredje metoden baseras på mikro-nivå och kräver information om varje enskild skada, såsom anmälningstid och antalet utbetalningsperioder. De två dataseten som används är ett som innehåller skador med typiskt kortare avvecklingstider och ett som innehåller skador med typiskt längre avvecklingstider. Frågorna som behandlas är om man vinner något på att använda en mer avancerad metod än chain ladder och om det skiljer sig åt mellan dataseten. Metoderna jämförs genom simulering av reserven, men också genom att jämföra punktskattningar med den verkliga reserven. Resultaten visar att man I detta fall inte vinner något på att använda mikro-metoden. Double chain ladder å andra sidan presterar bättre än chain ladder. Skillnaden mellan de två dataseten är att det är svårare att estimera reserven när avvecklingstiden är längre, men ingen skillnad ses när det gäller val av metod

Résumé : Les eaux souterraines ont un impact majeur sur la vie terrestre, les besoins domestiques et le climat. Elles sont aussi un maillon essentiel du cycle hydrologique. Au Canada par exemple, plus de 30 % de la population est tributaire des eaux souterraines pour leur alimentation en eau potable. Ces ressources subissent de nombreuses pressions sous l’influence de certains facteurs comme la salinisation, la contamination et l’épuisement. La variabilité du climat et la demande croissante sur ces ressources imposent l'amélioration de nos connaissances sur les eaux souterraines. L’objectif principal du projet de recherche est d’exploiter les données d’anomalies (TWS) de la mission Gravity Recovery And Climate Experiment (GRACE) pour localiser, quantifier et analyser les variations des eaux souterraines à travers les bassins versants du Bas-Mackenzie, du Saint-Laurent, du Nord-Québec et du Labrador. Il s’agit aussi d’analyser l’influence des cycles d’accumulation et de fonte de neige sur les variations du niveau des eaux souterraines. Pour estimer les variations des eaux souterraines, la connaissance des autres paramètres du bilan hydrologique est nécessaire. Ces paramètres sont estimés à l’aide des sorties du modèles de surface CLM du Système Global d’Assimilation des Données de la Terre (GLDAS). Les données GRACE qui ont été utilisées sont celles acquises durant la période allant de mars 2002 à août 2012. Les résultats ont été évalués à partir d’enregistrements de niveaux piézométriques provenant de 1841 puits localisés dans les aquifères libres du bassin des réseaux de suivi des eaux souterraines au Canada. Les valeurs de rendements spécifiques des différents types d’aquifères de chaque puits et celles des variations mensuelles du niveau d’eau dans ces puits ont été utilisées pour estimer les variations des anomalies des eaux souterraines in-situ. L’étude de corrélation entre les variations des anomalies des eaux souterraines estimées à partir de la combinaison GRACE-GLDAS et celles issues de données in-situ révèle des concordances significatives avec des valeurs de 𝑅=0,46 et 𝑅 = 0,50 entre ΔGW et ΔGWPiézo, respectivement pour le bassin du Bas-Mackenzie et celui du Saint-Laurent. Quant à la combinaison de tous les bassins, elle s’est caractérisée par une corrélation un peu plus faible (𝑅=0,44). Les valeurs des coefficients de corrélation calculée avec des données indépendantes de SWE venant des produits GlobSnow ; R = -0,68 sur le bassin du Bas-Mackenzie, R = -0,76 sur le bassin versant du Nord-du-Québec et du Labrador, R = -0,51 sur le bassin versant du Saint-Laurent et R = -0,52 pour les trois bassins réunis, montrent que ΔGW est très sensible à la quantité d’eau dans la neige, avec des spécificités pour chaque bassin versant étudié. / Abstract : Groundwater is an important component of the hydrological cycle. In Canada, more than 30% of the population relies on the main source of water for domestic use. These resources are under pressure factors such as salinization, contamination and exhaustion. Our knowledge on groundwater needs improvement because of climate variability and an increasing demand for these resources. The main objective of this project is to use the Gravity Recovery And Climate Experiment (GRACE) mission data (TWS) anomalies to locate, quantify and analyze the groundwater changes in Mackenzie St. Lawrence, North-Quebec and Labrador basins. Also this work aims to analyze the influence of accumulation cycles and snowmelt on groundwater level changes. For this groundwater changes estimation, it is necessary to estimate other water balance parameters. These parameters are estimated by using the outputs of the land surface model CLM of the Global Land Data Assimilation System (GLDAS) Data Products. The TWS data used are from March 2002 to august 2012. The results were evaluated by using the groundwater levels records from 1841 wells located in the free aquifers of different Canadian groundwater monitoring networks. The Specific yields values of the different types of aquifers from each well and the monthly changes in the water level in these wells were used to estimate changes in the in-situ groundwater anomalies. The correlation study between changes in groundwater anomalies estimated from the combination GRACE-GLDAS and those from in-situ data reveals significant matches with values of R = 0,46 and R = 0,50 between ΔGW and ΔGWPiézo respectively for the Lower-Mackenzie and the St. Lawrence basins. The combination of all the basins is characterized by a slightly weaker correlation (R = 0, 44).The correlation coefficients calculated with an independent data coming from SWE GlobSnow products are R = -0, 68 for Lower-Mackenzie River Basin, R = -0, 76 for Nord-du-Québec and Labrador basin, R = -0, 51 for St. Lawrence basin and R = -0, 52 for the three basins taken together. These correlations show that ΔGW is very sensitive to the snowmelt water amount, but with specificities for each studied basin.

Face detection refers to a number of techniques that identify faces in images and videos. As part of the senior project exercise at Pomona College, I explore the process of face detection using a JavaScript library called CLMtrackr. CLMtrackr works in any browser and detects faces within the video stream captured by a webcam. The focus of this paper is to explore the shortcomings in the inclusivity of the CLMtrackr library and consequently that of face detection. In my research, I have used two datasets that contain human faces with diverse backgrounds, in order to assess the accuracy of CLMtrackr. The two datasets are the MUCT and PPB. In addition, I investigate whether skin color is a key factor in determining face detection's success, to ascertain where and why a face might not be recognized within an image. While my research and work produced some inconclusive results due to a small sample size and a couple outliers in my outputs, it is clear that there is a trends toward the CLMtrackr algorithm recognizing faces with lighter skin tones more often than darker ones.

Forests perform many important ecological functions and provide numerous environmental services to humanity. Currently forests are under ever increasing pressures from humans through deforestation, changes in land use, and anthropogenic climate change. Managed forests play an important role in supplying forest products to the global population, necessitating the need to predict how forests will respond to climate change and how this will influence future wood product supplies. In this dissertation I used loblolly pine (Pinus taeda), the most extensively cultivated tree species in the United States, as a study system to simulate how climate change and forest management could alter the dynamics of managed forests in the future. Using the land component (the Community Land Model) of the widely used Community Earth System Model (CESM), I developed and validated a set of tools necessary to simulate the loblolly pine plantation system using the vegetation demography model embedded in CESM (FATES). This included developing a representation of loblolly pine using data from the literature, which was better able to capture forest growth and development observed in field studies than FATES's existing conifer tree representation. I added the ability to simulate several aspects of forest management not previously supported in FATES by creating the Vegetation Management Module, which I showed was able to realistically reproduce the common management practice of stand thinning. I used these new tools to perform simulations of how loblolly pine will grow across the Southeastern United States until the end of the 21st century, under the high and low climate change scenarios developed by the scientific community in the Coupled Model Intercomparison Project Phase 6 (CMIP6). Our experiments show that loblolly pine productivity may as much as double by the end of the century, with total wood harvest over that period increasing by almost half. I also showed that different management activities had significant effects on loblolly plantation yields, with mid-rotation stand thinning having an effect under both climate scenarios on par with increases due to the extreme climate change scenario SSP5 RCP8.5. I showed that these changes in wood yields could decrease the forest area in the Southeast required to meet the wood product demands over the rest of the century. These changes in plantation productivity could interact with socioeconomic factors to drive changes in land use and carbon storage in the Southeastern U.S. This work increases our understanding of how managed forests in the U.S.\ will be affected by climate change and how our management choices modulate that response. The techniques and tools developed here open up new areas of research into the role of forest management in the climate system. / Doctor of Philosophy / Forests benefit humans by regulating Earth's climate and by providing natural resources such as wood. In the Southeastern United States forestry is an important industry. Tree farms of southern pine trees produce a large percentage of the region's wood. Predicting how forests will grow in the future is important for planning and making investments. However, the burning of fossil fuels has increased carbon dioxide in the atmosphere and is changing Earth's climate. This is affecting how fast trees grow and how much wood can be harvested from forests. The methods that foresters have traditionally used to predict how trees will grow in the future do not account for climate change, and thus may not be as accurate in the future. An alternative is to use the computer models that scientists have developed to predict both how global climate will change in the future and how forests are influenced by climate. These computer programs can be used to predict how natural forests will grow in the future, but aren't set up to predict managed forests well. I made changes to one of these programs to make it possible to simulate the managed loblolly pine forests of the Southeastern United States. First, I tested these changes to make sure that simulated forests grew like real forests do today. Then I simulated how pine forests in Southeastern United States could grow over the next century with climate change. I found that pine forests will grow faster and allow more wood to be harvested as carbon dioxide in the atmosphere increases. If climate changes are extreme, loblolly forests could produce 70\% more wood than today by the end of the 21st century. I also showed that the manner in which forests were managed in simulations changes the amount of wood they produced, with some management practices increasing wood harvested by 50\% over the rest of the century. Because climate change could increase the amount of wood that can be produced from a fixed area of forest, I investigated how this might change the area of forest plantation in the Southeastern United States. Based on projections of demand for wood for the rest of the century I calculated how much loblolly pine forest would be needed to produce this wood over the next century. I found that increases in forest productivity due to climate change and forest management could decrease the forest area required to grow the wood we need. This could change how we use forests in the Southeastern United States, which in turn could have impacts on the climate.

Rapid prototyping systems have advanced significantly with respect to material capabilities, fabrication speed, and surface quality. However, build jobs are still manually activated one at a time. The result is non-productive machine time whenever an operator is not at hand to make a job changeover. A low-cost auxiliary system, named Continuous Layered Manufacturing (CLM), has been developed to automatically load and unload the FDM 1600 rapid prototyping system (Stratasys, Inc.). The modifications made to the FDM 1600 system are minimal. The door to the FDM 1600 build chamber is removed, and the .SML build files that are used to drive the FDM 1600 are modified at both ends to facilitate synchronized operation between the two systems. The CLM system is capable of running three consecutive build jobs without operator intervention. As long as an operator removes finished build jobs, and adds new build trays before at most every three build jobs, the FDM can operate near indefinitely. The impact of the CLM system on the productivity of the FDM 1600 rapid prototyping system is demonstrated by the expected reduction from the customary eight weeks down to a future three and one-half weeks required to complete the typical forty build jobs during a semester in the course ME 4644 Introduction to Rapid Prototyping at Virginia Tech. / Master of Science

Diese Dissertation beschäftigt sich mit regionaler Klimamodellierung über Südamerika, der Analyse von Modellsensitivitäten bezüglich Wolkenparametrisierungen und der Entwicklung neuer Methoden zur Modellevaluierung mithilfe von Klimanetzwerken. Im ersten Teil untersuchen wir Simulationen mit dem COnsortium for Small scale MOdeling model in CLimate Mode (COSMO-CLM) und stellen die erste umfassende Evaluierung dieses dynamischen regionalen Klimamodells über Südamerika vor. Dabei untersuchen wir insbesondere die Abhängigkeit simulierter tropischer Niederschläge von Parametrisierungen subgitterskaliger cumuliformer und stratiformer Wolken und finden starke Sensitivitäten bezüglich beider Wolkenparametrisierungen über Land. Durch einen simultanen Austausch der entsprechenden Schemata gelingt uns eine beträchtliche Reduzierung von Fehlern in klimatologischen Niederschlags- und Strahlungsmitteln, die das COSMO-CLM über tropischen Regionen für lange Zeit charakterisierten. Im zweiten Teil führen wir neue Metriken für die Evaluierung von Klimamodellen bezüglich räumlicher Kovariabilitäten ein. Im Kern bestehen diese Metriken aus Unähnlichkeitsmaßen für den Vergleich von simulierten mit beobachteten Klimanetzwerken. Wir entwickeln lokale und globale Unähnlichkeitsmaße zum Zwecke der Darstellung lokaler Unähnlichkeiten in Form von Fehlerkarten sowie der Rangordnung von Modellen durch Zusammenfassung lokaler zu globalen Unähnlichkeiten. Die neuen Maße werden dann für eine vergleichende Evaluierung regionaler Klimasimulationen mit COSMO-CLM und dem Statistical Analogue Resampling Scheme über Südamerika verwendet. Dabei vergleichen wir die sich ergebenden Modellrangfolgen mit solchen basierend auf mittleren quadratischen Abweichungen klimatologischer Mittelwerte und Varianzen und untersuchen die Abhängigkeit dieser Rangfolgen von der betrachteten Jahreszeit, Variable, dem verwendeten Referenzdatensatz und Klimanetzwerktyp. / This dissertation is about regional climate modeling over South America, the analysis of model sensitivities to cloud parameterizations, and the development of novel model evaluation techniques based on climate networks. In the first part we examine simulations with the COnsortium for Small scale MOdeling weather prediction model in CLimate Mode (COSMO-CLM) and provide the first thorough evaluation of this dynamical regional climate model over South America. We focus our analysis on the sensitivity of simulated tropical precipitation to the parameterizations of subgrid-scale cumuliform and stratiform clouds. It is shown that COSMO-CLM is strongly sensitive to both cloud parameterizations over tropical land. Using nondefault cumulus and stratus parameterization schemes we are able to considerably reduce long-standing precipitation and radiation biases that have plagued COSMO-CLM across tropical domains. In the second part we introduce new performance metrics for climate model evaluation with respect to spatial covariabilities. In essence, these metrics consist of dissimilarity measures for climate networks constructed from simulations and observations. We develop both local and global dissimilarity measures to facilitate the depiction of local dissimilarities in the form of bias maps as well as the aggregation of those local to global dissimilarities for the purposes of climate model intercomparison and ranking. The new measures are then applied for a comparative evaluation of regional climate simulations with COSMO-CLM and the STatistical Analogue Resampling Scheme (STARS) over South America. We compare model rankings obtained with our new performance metrics to those obtained with conventional root-mean-square errors of climatological mean values and variances, and analyze how these rankings depend on season, variable, reference data set, and climate network type.