We present a method to determine the precise shape of a dynamic object from video. This problem is fundamental to computer vision, and has a number of applications, for example, 3D video/cinema post-production, activity recognition and augmented reality. Current tracking algorithms that determine precise shape can be roughly divided into two categories: 1) Global statistics partitioning methods, where the shape of the object is determined by discriminating global image statistics, and 2) Joint shape and appearance matching methods, where a template of the object from the previous frame is matched to the next image. The former is limited in cases of complex object appearance and cluttered background, where global statistics cannot distinguish between the object and background. The latter is able to cope with complex appearance and a cluttered background, but is limited in cases of camera viewpoint change and object articulation, which induce self-occlusions and self-disocclusions of the object of interest. The purpose of this thesis is to model self-occlusion/disocclusion phenomena in a joint shape and appearance tracking framework. We derive a non-linear dynamic model of the object shape and appearance taking into account occlusion phenomena, which is then used to infer self-occlusions/disocclusions, shape and appearance of the object in a variational optimization framework. To ensure robustness to other unmodeled phenomena that are present in real-video sequences, the Kalman filter is used for appearance updating. Experiments show that our method, which incorporates the modeling of self-occlusion/disocclusion, increases the accuracy of shape estimation in situations of viewpoint change and articulation, and out-performs current state-of-the-art methods for shape tracking.

A discrete, compact, low-power sensor readout circuit that can simultaneously handle two current measurements, and one voltage measurement. This work provides a compact, low-power sensor architecture, with the intent for the serial readout to be replaced with a low-power radio frequency transmitter for continuous monitoring. The proposed circuit is highly precise with an average current draw of 21 micro amps for a sampling frequency of once per minute. The target application is livestock health monitoring, which would be done by placing the sensor and circuit inside of a cow's rumen to monitor changes in pH, lactate, and VFA levels to catch metabolic disease early. / Master of Science / A circuit for reading a sensor with a long battery life. It is small so that it can fit in a cow's stomach to determine if it has a disease in its stomach. This diagnosis can be used to adjust its diet.

Esta tesis doctoral llevó a cabo el estudio, desarrollo e implementación de algoritmos para la navegación con sistemas globales de navegación por satélite (GNSS), enfocándose en la determinación precisa de la posición, velocidad y aceleración usando GPS, en modo post-procesado y lejos de estaciones de referencia. Uno de los objetivos era desarrollar herramientas en esta área y hacerlas disponibles a la comunidad GNSS. Por ello el desarrollo se hizo dentro del marco del proyecto preexistente de software libre llamado GPS Toolkit (GPSTk). Una de las primeras tareas realizadas fue la validación de las capacidades de la GPSTk para el procesado del pseudorango, realizando comparaciones con una herramienta de procesamiento de datos probada (BRUS). La gestión de datos GNSS demostró ser un asunto importante cuando se intentó extender las capacidades de la GPSTk al procesamiento de datos obtenidos de las fases de la señal GPS. Por ello se desarrollaron las Estructuras de Datos GNSS (GDS), que combinadas con su paradigma de procesamiento aceleran el proceso de desarrollo de software y reducen errores. La extensión de la GPSTk a los algoritmos de procesado en fase se hizo mediante la ayuda de las GDS, proporcionándose importantes clases accesorias que facilitan el trabajo. Se implementó el procesado de datos Precise Point Positioning (PPP) con ejemplos relativamente simples basados en las GDS, y al comparar sus resultados con otras aplicaciones de reputación ya establecida, se encontró que destacan entre los mejores. También se estudió cómo obtener la posición precisa, en post-proceso, de un receptor GPS a cientos de kilómetros de la estación de referencia más cercana y usando tasas de datos arbitrarias (una limitación del método PPP). Las ventajas aportadas por las GDS permitieron la implementación de un procesado semejante a un PPP cinemático basado en una red de estaciones de referencia, estrategia bautizada como Precise Orbits Positioning (POP) porque sólo necesita órbitas precisas para trabajar y es independiente de la información de los relojes de los satélites GPS. Los resultados de este enfoque fueron muy similares a los del método PPP cinemático estándar, pero proporcionando soluciones de posición con una tasa mayor y de manera más robusta. La última parte se enfocó en la implementación y mejora de algoritmos para determinar con precisión la velocidad y aceleración de un receptor GPS. Se hizo énfasis en el método de las fases de Kennedy debido a su buen rendimiento, desarrollando una implementación de referencia y demostrando la existencia de una falla en el procedimiento propuesto originalmente para el cálculo de las velocidades de los satélites. Se propuso entonces una modificación relativamente sencilla que redujo en un factor mayor que 35 el RMS de los errores 3D en velocidad. Tomando ideas de los métodos Kennedy y POP se desarrolló e implementó un nuevo procedimiento de determinación de velocidad y aceleración que extiende el alcance. Este método fue llamado Extended Velocity and Acceleration determination (EVA). Un experimento usando una aeronave ligera volando sobre los Pirineos mostró que tanto el método de Kennedy (modificado) como el método EVA son capaces de responder ante la dinámica de este tipo de vuelos. Finalmente, tanto el método de Kennedy modificado como el método EVA fueron aplicados a una red en la zona ecuatorial de Sur América con líneas de base mayores a 1770 km. En este escenario el método EVA mostró una clara ventaja tanto en los promedios como en las desviaciones estándar para todas las componentes de la velocidad y la aceleración. / This Ph.D. Thesis focuses on the development of algorithms and tools for precise GPS-based position, velocity and acceleration determination very far from reference stations in post-process mode. One of the goals of this thesis was to develop a set of state-of-the-art GNSS data processing tools, and make them available for the research community. Therefore, the software development effort was done within the frame of a preexistent open source project called the GPSTk. Therefore, validation of the GPSTk pseudorange-based processing capabilities with a trusted GPS data processing tool was one of the initial task carried out in this work. GNSS data management proved to be an important issue when trying to extend GPSTk capabilities to carrier phasebased data processing algorithms. In order to tackle this problem the GNSS Data Structures (GDS) and their associated processing paradigm were developed. With this approach the GNSS data processing becomes like an assembly line, providing an easy and straightforward way to write clean, simple to read and use software that speeds up development and reduces errors. The extension of GPSTk capabilities to carrier phase-based data processing algorithms was carried out with the help of the GDS, adding important accessory classes necessary for this kind of data processing and providing reference implementations. The performance comparison of these relatively simple GDS-based source code examples with other state-of-the art Precise Point Positioning (PPP) suites demonstrated that their results are among the best. Furthermore, given that the GDS design is based on data abstraction, it allows a very flexible handling of concepts beyond mere data encapsulation, including programmable general solvers, among others. The problem of post-process precise positioning of GPS receivers hundreds of kilometers away from nearest reference station at arbitrary data rates was dealt with, overcoming an important limitation of classical post-processing strategies like PPP. The advantages of GDS data abstraction regarding solvers were used to implement a kinematic PPP-like processing based on a network of stations. This procedure was named Precise Orbits Positioning (POP) because it is independent of precise clock information and it only needs precise orbits to work. The results from this approach were very similar (as expected) to the standard kinematic PPP processing strategy, but yielding a higher positioning rate. Also, the network-based processing of POP seems to provide additional robustness to the results, even for receivers outside the network area. The last part of this thesis focused on implementing, improving and testing algorithms for the precise determination of velocity and acceleration hundreds of kilometers away from nearest reference station. Special emphasis was done on the Kennedy method because of its good performance. A reference implementation of Kennedy method was developed, and several experiments were carried out. Experiments done with very short baselines showed a flaw in the way satellite velocities were computed, introducing biases in the velocity solution. A relatively simple modification was proposed, and it reduced the RMS of 5-min average velocity 3D errors by a factor of over 35. Then, borrowing ideas from Kennedy method and the POP method, a new velocity and acceleration determination procedure named EVA was developed and implemented that greatly extends the effective range. An experiment using a light aircraft flying over the Pyrenees showed that both the modified-Kennedy and EVA methods were able to cope with the dynamics of this type of flight. Finally, both modified-Kennedy and EVA method were applied to a challenging scenario in equatorial South America, with baselines over 1770 km, where EVA method showed a clear advantage in both averages and standard deviations for all components of velocity and acceleration. Lloc i

ITC/USA 2015 Conference Proceedings / The Fifty-First Annual International Telemetering Conference and Technical Exhibition / October 26-29, 2015 / Bally's Hotel & Convention Center, Las Vegas, NV / The use of differential GPS post-procesing for precise trajectography computation has been widely used since early 90s. Up to recent dates, installation of a GPS receiver in a well known position (base station) has been mandatory. Operating range from this base station varies from 50 km up to 100 km, depending on the accuracy required, which impose single or dual frequency GPS technique. Nowadays, the huge amount of GPS base stations continuous logging data worldwide have allowed to improve the error models a lot. Using these precise models, it is possible to achieve centimeter accuracy in GPS trajectography by using only one GPS receiver without range to a base station restrictions. This technique is called Precise Point Positioning (PPP). The performance results for PPP obtained after a real 10 flights campaign will be presented.

Utvecklingen av satellitbaserad positionsbestämning gör det idag befogat att begära låga osäkerheter med GNSS. Det är idag möjligt att uppnå osäkerheter kring centimetern. Bäst mätosäkerhet ger relativ mätning som sker med stöd av antingen enkelstations- eller nätverks-RTK. I Sverige erbjuder Lantmäteriet med sitt SWEPOS ett tätt referensnätverk som förser användaren med korrektionsdata oavsett position inom Sveriges gränser. Dock är det inte alla länder som kan erbjuda denna positionstjänst. Geografiskt stora länder har mycket svårt att skapa ett referensnät, det skulle betyda flera tusen stationer och gör det till en ekonomisk fråga. Det är bl.a. ur den synpunkten andra metoder har växt fram. En av dessa är Precise Point Positioning (PPP). Enligt G. Hedling (personlig kommunikation, 18 mars 2015) har PPP fått en väl etablering inom jordbruket samt på maritima gruv- och oljeplattformar. Metoden är lämplig vid stora öppna ytor och när avståndet till närmsta referensstation är stor. PPP använder sig av absolut positionering och kan mäta både statiskt och kinematiskt och resultat kan fås i realtid och genom efterberäkning. Det ligger i Lantmäteriets intresse att testa kinematisk PPP i Sverige och den här studien testar kinematisk PPP i realtid med programvaran BNC 2.11 och med korrektioner från International GPS Service (IGS). Enligt Bisnath & Gao (2009) erhålls decimeterosäkerhet med kinematisk PPP och för att bestämma dess tillförlitlighet har i den här studien koordinatavvikelse beräknats mellan BNC och enkelstations-RTK med stöd från SWEPOS. Koordinaterna från enkelstations-RTK har vid testerna angivits som de sanna koordinaterna, genom ett statiskt test har det undersökts om det är motiverat. Utifrån den statiska mätningen har även intialiseringstiden kunnat utredas, alltså den tid det tar för PPP att konvergera. Efterberäkningstjänsten CSRS-PPP har också testats och jämförts mot kända koordinater vid den statiska mätningen.Studien visar att efter närmare en timmes observation avviker PPP under 2 dm i plan mot enkelstations-RTK. Den visar också att 15-30 minuters konvergeringstid är nödvändig för att erhålla osäkerheter på några decimeter. Några av de faktorer som påverkar resultatet är bl.a. jonosfärstörning. högt PDOP-värde och antal processerade satelliter i mjukvarorna, hur mycket är svårt att säga. Vid en tappad signal krävs en ny omintialisering på flera tiotals minuter. Studien visar också att det är lämpligt att använda enkelstations-RTK som sanning. Vid den statiska mätningen avviker enkelstations-RTK kring centimetern mot den kända punktens koordinater, vilket anses godtagbart. CSRS-PPP uppvisar bra resultat och är inte mycket sämre än det resultat enkelstations-RTK redovisar. / Today it´s possible to achieve low uncertainties when surveying with GNSS. You can expect uncertainties around centimeter-level. The best results are achieved when using relative-surveying with corrections from single-station- or network-RTK. The Swedish mapping, cadastral and land registration authority (Lantmäteriet) is providing a well-developed network of reference stations. The network, called SWEPOS, offers corrections for its users independent of position within the Swedish borders. Far from all nations has the ability or the financial resources to create such an expanded network. Instead, other methods for satellite surveying have been developed, including Precise Point Positioning (PPP). According to G. Hedling (personal communication, 18 march 2015) PPP is well-established in the agriculture and in the maritime mining- and oil-industry. The method is suitable in open areas and it is independently of nearby reference stations. PPP is using what’s called absolute-surveying. The surveying is performed either kinematic or static and the results can be obtained thru post-processing or in real-time. “Lantmäteriet” has interest in testing kinematic PPP in Sweden and for this thesis kinematic PPP in real-time is tested with BNC 2.11 software and corrections is given from the International GPS Service (IGS). According to Bisnath & Gao (2009) it is possible to achieve uncertainties in decimeter-level with kinematic PPP. To determine the reliability of PPP the deviation has been calculated against single-station-RTK. The single-station-RTK coordinates have in this study been used as the “truth” and in an additional test using static measurements it has been investigated if that’s correct. From the static test the initialization time for PPP as well as the quality of the post-processing service CSRS-PPP has been studied.The results show that after nearly an hour of observation the deviation between PPP and single-station-RTK are below 2 dm for the level-coordinates. The initialization time of 15-30 minutes is necessary to achieve uncertainties of a few decimeters. Elements that are affecting the results are disturbance in the ionosphere, high PDOP and number of processed satellites in the software. In which extent it’s not possible to determine. When the signal is lost between rover and satellites a re-initialization of 15-30 minutes is needed. It also shows that it is reasonable to use single-station-RTK as the “truth”. Single-station-RTK deviates a proximately one centimeter in relation to known coordinates. The post-processing service CSRS-PPP gives remarkably good results not far from what single-station-RTK offers.

Precise Point Positioning (PPP) ermöglicht eine präzise Positionsbestimmung mittels globaler Satellitennavigationssysteme (Global Navigation Satellite System, GNSS) ohne die direkte Verwendung der Beobachtungsdaten von regionalen Referenzstationen. Die wesentlichste Einschränkung von PPP im Vergleich zu differenziellen Auswertetechniken (Real-Time Kinematic, RTK) ist die deutlich längere Konvergenzzeit. Voraussetzung für die Verkürzung der Konvergenzzeit ist die Festsetzung der geschätzten Mehrdeutigkeiten auf ganzzahlige Werte. Die Mehrdeutigkeitslösung verlangt ein robustes funktionales Modell und beruht auf einem zweistufigen Mehrdeutigkeitsfestsetzungsverfahren, welches frei von ionosphärischen Einflüssen 1. Ordnung ist. Die sowohl auf Code- als auch auf Phasenbeobachtungen basierende Melbourne-Wübbena-Linearkombination erlaubt hierbei eine einfache Festsetzung der Widelane-Mehrdeutigkeiten. Infolgedessen kann zur Berechnung der ionosphären-freien Linearkombination die im Vergleich zur Wellenlänge der ionosphären-freien Linearkombination deutlich größere Narrowlane-Wellenlänge verwendet werden. Zur Stabilisierung des im Normalfall lediglich auf den Beobachtungsdaten des amerikanischen Global Positioning System (GPS) beruhenden funktionalen Modells können die Beobachtungsdaten des russischen GLObal’naya NAvigatsioannaya Sputnikovaya Sistema (GLONASS) beitragen. Aufgrund der Technik, die GLONASS zur Identifizierung der einzelnen Satelliten einsetzt (Frequency Division Multiple Access, FDMA), unterscheiden sich die Frequenzen der einzelnen Satelliten. Die leicht unterschiedlichen Frequenzen erschweren die Modellierung und Korrektion der instrumentell bedingten Signalverzögerungen (z. B. Fractional-Cycle Biases (FCB)). Vor diesem Hintergrund kann das konventionelle Mehrdeutigkeitsfestsetzungsverfahren nur bedingt für GLONASS verwendet werden. Die Untersuchung der instrumentell bedingten GLONASS-Signalverzögerungen sowie die Entwicklung einer alternativen Methode zur Festsetzung der GLONASS-Mehrdeutigkeiten mit dem Ziel einer kombinierten GPS/GLONASS-Mehrdeutigkeitslösung sind die Schwerpunkte der vorliegenden Arbeit. Die entwickelte alternative Mehrdeutigkeitsfestsetzungsstrategie baut auf der puren Widelane-Linearkombination auf, weshalb globale Ionosphärenmodelle unabdingbar sind. Sie eignet sich sowohl für GLONASS als auch für GPS und zeigt gleichwertige Ergebnisse für beide GNSS, wenngleich im Vergleich zur konventionellen Methode mit geringeren Mehrdeutigkeitsfestsetzungsquoten zu rechnen ist. / Precise Point Positioning (PPP) allows for accurate Global Navigation Satellite System (GNSS) based positioning without the immediate need for observations collected by regional station networks. The fundamental drawback of PPP in comparison to differential techniques such as Real-Time Kinematic (RTK) is a significant increase in convergence time. Among a plurality of different measures aiming for a reduction of convergence time, fixing the estimated carrier phase ambiguities to integer values is the key technique for success. The ambiguity resolution asks for a robust functional model and rests upon a two-stage method ruling out first-order ionospheric effects. In this context the Melbourne-Wübbena linear combination of dual-frequency carrier phase and code measurements leverages a simple resolution of widelane ambiguities. As a consequence the in comparison to the wavelength of the ionosphere-free linear combination significantly longer narrowlane wavelength can be used to form the ionosphere-free linear combination. By default the applied functional model is solely based on observations of the Global Positioning System (GPS). However measurements from the GLObal’naya NAvigatsioannaya Sputnikovaya Sistema (GLONASS) can contribute to improve the model’s stability significantly. Due to the technique used by GLONASS to distinguish individual satellites (Frequency Division Multiple Access, FDMA), the signals broadcast by those satellites differ in their frequencies. The resulting slightly different frequencies constitute a barricade for both modelling and correcting any device-dependent signal delays, e.g. fractional-cycle biases (FCB). These facts limit the applicability of the conventional ambiguity-fixing approach when it comes to GLONASS signals. The present work puts a focus both on investigating the device-dependent GLONASS signal delays and on developing an alternative method for fixing GLONASS ambiguities with the ultimate objective of a combined GPS/GLONASS ambiguity resolution. The alternative ambiguity resolution strategy is based on the pure widelane linear combination, for which reason ionospheric corrections are indispensable. The procedure is applicable for GLONASS in the first instance but reveals equivalent results for both GPS and GLONASS. The disadvantage relative to the conventional approach is the reduced ambiguity fixing success rate.

Precise Point Positioning (PPP) ermöglicht eine präzise Positionsbestimmung mittels globaler Satellitennavigationssysteme (Global Navigation Satellite System, GNSS) ohne die direkte Verwendung der Beobachtungsdaten von regionalen Referenzstationen. Die wesentlichste Einschränkung von PPP im Vergleich zu differenziellen Auswertetechniken (Real-Time Kinematic, RTK) ist die deutlich längere Konvergenzzeit. Voraussetzung für die Verkürzung der Konvergenzzeit ist die Festsetzung der geschätzten Mehrdeutigkeiten auf ganzzahlige Werte. Die Mehrdeutigkeitslösung verlangt ein robustes funktionales Modell und beruht auf einem zweistufigen Mehrdeutigkeitsfestsetzungsverfahren, welches frei von ionosphärischen Einflüssen 1. Ordnung ist. Die sowohl auf Code- als auch auf Phasenbeobachtungen basierende Melbourne-Wübbena-Linearkombination erlaubt hierbei eine einfache Festsetzung der Widelane-Mehrdeutigkeiten. Infolgedessen kann zur Berechnung der ionosphären-freien Linearkombination die im Vergleich zur Wellenlänge der ionosphären-freien Linearkombination deutlich größere Narrowlane-Wellenlänge verwendet werden. Zur Stabilisierung des im Normalfall lediglich auf den Beobachtungsdaten des amerikanischen Global Positioning System (GPS) beruhenden funktionalen Modells können die Beobachtungsdaten des russischen GLObal’naya NAvigatsioannaya Sputnikovaya Sistema (GLONASS) beitragen. Aufgrund der Technik, die GLONASS zur Identifizierung der einzelnen Satelliten einsetzt (Frequency Division Multiple Access, FDMA), unterscheiden sich die Frequenzen der einzelnen Satelliten. Die leicht unterschiedlichen Frequenzen erschweren die Modellierung und Korrektion der instrumentell bedingten Signalverzögerungen (z. B. Fractional-Cycle Biases (FCB)). Vor diesem Hintergrund kann das konventionelle Mehrdeutigkeitsfestsetzungsverfahren nur bedingt für GLONASS verwendet werden. Die Untersuchung der instrumentell bedingten GLONASS-Signalverzögerungen sowie die Entwicklung einer alternativen Methode zur Festsetzung der GLONASS-Mehrdeutigkeiten mit dem Ziel einer kombinierten GPS/GLONASS-Mehrdeutigkeitslösung sind die Schwerpunkte der vorliegenden Arbeit. Die entwickelte alternative Mehrdeutigkeitsfestsetzungsstrategie baut auf der puren Widelane-Linearkombination auf, weshalb globale Ionosphärenmodelle unabdingbar sind. Sie eignet sich sowohl für GLONASS als auch für GPS und zeigt gleichwertige Ergebnisse für beide GNSS, wenngleich im Vergleich zur konventionellen Methode mit geringeren Mehrdeutigkeitsfestsetzungsquoten zu rechnen ist. / Precise Point Positioning (PPP) allows for accurate Global Navigation Satellite System (GNSS) based positioning without the immediate need for observations collected by regional station networks. The fundamental drawback of PPP in comparison to differential techniques such as Real-Time Kinematic (RTK) is a significant increase in convergence time. Among a plurality of different measures aiming for a reduction of convergence time, fixing the estimated carrier phase ambiguities to integer values is the key technique for success. The ambiguity resolution asks for a robust functional model and rests upon a two-stage method ruling out first-order ionospheric effects. In this context the Melbourne-Wübbena linear combination of dual-frequency carrier phase and code measurements leverages a simple resolution of widelane ambiguities. As a consequence the in comparison to the wavelength of the ionosphere-free linear combination significantly longer narrowlane wavelength can be used to form the ionosphere-free linear combination. By default the applied functional model is solely based on observations of the Global Positioning System (GPS). However measurements from the GLObal’naya NAvigatsioannaya Sputnikovaya Sistema (GLONASS) can contribute to improve the model’s stability significantly. Due to the technique used by GLONASS to distinguish individual satellites (Frequency Division Multiple Access, FDMA), the signals broadcast by those satellites differ in their frequencies. The resulting slightly different frequencies constitute a barricade for both modelling and correcting any device-dependent signal delays, e.g. fractional-cycle biases (FCB). These facts limit the applicability of the conventional ambiguity-fixing approach when it comes to GLONASS signals. The present work puts a focus both on investigating the device-dependent GLONASS signal delays and on developing an alternative method for fixing GLONASS ambiguities with the ultimate objective of a combined GPS/GLONASS ambiguity resolution. The alternative ambiguity resolution strategy is based on the pure widelane linear combination, for which reason ionospheric corrections are indispensable. The procedure is applicable for GLONASS in the first instance but reveals equivalent results for both GPS and GLONASS. The disadvantage relative to the conventional approach is the reduced ambiguity fixing success rate.

Laser altimetry and satellite gravity surveys indicate that the St Elias Icefields are currently losing mass and are among the largest non-polar sea level contributors in the world. However, a poor understanding of glacier dynamics in the region is a major hurdle in evaluating regional variations in ice motion and the relationship between changing surface conditions and ice flux. This study combines in-situ dGPS measurements and advanced Radarsat-2 (RS-2) processing techniques to determine daily and seasonal ice velocities for the Kaskawulsh Glacier from summer 2009 to summer 2011. Three permanent dGPS stations were installed along the centreline of the glacier in 2009, with an additional permanent station on the South Arm in 2010. The Precise Point Positioning (PPP) method is used to process the dGPS data using high accuracy orbital reconstruction. RS-2 imagery was acquired on a 24-day cycle from January to March 2010, and from October to March 2010-2011 in a combination of ultra-fine and fine beam modes. Seasonal velocity regimes are readily identifiable in the dGPS results, with distinct variations in both horizontal velocity and vertical motion. The Spring Regime consists of an annual peak in horizontal velocity that corresponds closely with the onset of the melt season and progresses up-glacier, following the onset of melt at each station. The Summer Regime sees variable horizontal velocity and vertical uplift, superimposed on a long-term decline in motion. The Fall Regime sees a gradual slowing at all stations with little variation in horizontal velocity or vertical position. Rapid but short accelerations lasting up to 10 days were seen in the Winter regimes in both 2010 and 2011, occurring at various times throughout each regime. These events initiated at the Upper Station and progress down-glacier at propagation speeds up to 16,380 m day-1 and were accompanied by vertical uplift lasting for similar periods. Three velocity maps, one from the winter of 2010 and two from the fall/winter of 2011, produced from speckle tracking were validated by comparison with dGPS velocity, surface flow direction, and bedrock areas of zero motion, with an average velocity error of 2.0% and average difference in orientation of 4.3º.

O problema de controlar o nível significância do teste FBST(Full BayesianSignificantTest) é estudado no contexto de modelos bayesianos para densidade. Assim, é mostrado um método bayesiano que trabalha com estimação da densidade e como deve ser conduzido o FBST com este método quando deseja-se testar se uma população pode ser dita de determinada distribuição ou testar igualdade de duas populações. Para isso é apresentada a definição do e-valor modificado que é uma maneira alternativa de cálculo da medida de evidência do FBST. Por fim, é feito um estudo de simulação com diferentes distribuições de densidade e analisado o comportamento da função poder do teste nos casos de uma e duas populações. / The problem of controlling the significan celevelof the FBST (FullBayesianSignificantTest) test is studied in the contex of Bayesian models for density, thus, a Bayesian method is shown that works with density estimation estimation and how the FBST should be conducted in that situation with this method when it is desired to test if one population has certain density distribution or equality test of two populations. For this, a modified e-value definition is presented that is an alternative to calculate the FBST measure. At end a simulation study with different density distributions and analysis the power function of the test in cases of one and two populations.

This thesis investigates the major errors and processes affecting the performance of a viable, standalone point positioning technique known as single frequency Precise Point Positioning (PPP). The PPP processing utilises both single frequency code and carrier phase GPS observables. The mathematical model implemented is known as the code and quasi-phase combination. Effective measures to improve the quality of the positioning solutions are assessed and proposed. The a priori observations sigma (or standard deviation) ratio in the sequential least squares adjustment model plays a significant role in determining the accuracy and precision of the estimated solutions, as well as the solutions convergence time. An