ALBERT

Description: Global Navigation Satellite Systems (GNSS) spielen eine wichtige Rolle bei der präzisen Positionierung für Geodäsie und Vermessungstechnik. Der Schlüssel für die präzise GNSS Echtzeitpositionierung ist die sofortige Auflösung der ganzzahligen Mehrdeutigkeit. Jedoch können einige der Bias in Trägerphasenbeobachtungen nicht durch Differenzbildung entweder zwischen Stationen oder Satelliten entfernt werden, so dass die ganzzahlige Natur der Mehrdeutigkeit durch doppelte Differenzbildung zerstört werden kann und somit können die Mehrdeutigkeit nicht als ganze Zahlen festgelegt werden. Zwei typische Biasarten sind die Inter-Frequency-Bias (IFB) in der Prozessierung von GLObal NAvigation Satellite System (GLONASS) Daten und die Inter-System-Bias (ISB) für Integration von mehreren GNSS. Daher ist das Hauptziel dieser Arbeit die Untersuchung, Schätzung und Korrektur dieser Bias in Trägerphasenbeobachtungen um bessere Positionierungsgenauigkeit, Zuverlässigkeit und Verfügbarkeit durch die Verbesserung der Auflösung von Mehrdeutigkeiten zu erreichen. Die geschätzten Parameter der IFB und ISB von Trägerphasen sind normalerweise die IFB Rate und der bruchzahlige Teil von ISB (F-ISB). Die meisten der aktuellen Methoden schätzen die IFB Rate oder F-ISB zusammen mit den nicht-ganzzahligen Mehrdeutigkeiten und brauchen aufgrund ihrer hohen Korrelation meistens relativ lange Beobachtungszeitintervalle. Theoretisch hängt die Leistung der Auflösung von Mehrdeutigkeiten von der Qualität des gegebenen Wertes von IFB Rate/F-ISB ab, wenn die Beobachtungen präzise modelliert werden. Mit anderen Worten, je näher der gegebene Wert von IFB Rate/F-ISB an dem wahren Wert liegt, desto besser ist die Auflösung. Daher kann RATIO in der Festlegung von Mehrdeutigkeiten als Qualifizierungsfaktor des Wertes von IFB-Rate/F-ISB angewendet werden. Aufgrund dieser Tatsache wurde in dieser Arbeit eine neue auf dem Partikelfilter basierende Methode entwickelt, um diese Bias sowohl in Post-Prozessierung als auch im Echtzeit-Modus zu schätzen. Bei dem vorgeschlagenen Verfahren wird die IFB/ISB durch deren Stichproben (d.h. Partikel) repräsentiert, mit den Gewichten die durch die konstruierte Wahrscheinlichkeitsverteilung (Likelihood-Funktion) von dem dazugehörigen RATIO für die gegebenen Stichprobenwerte so festgelegt werden, dass der wahre Biaswert mit dem Partikelfilterverfahren erfolgreich geschätzt werden kann. Die ganzzahlige Natur der Mehrdeutigkeiten in den Modellen mit IFB/ISB-Parametern wird in der Auflösung von Mehrdeutigkeit mit dem gegebenen Wert von IFB Rate/F-ISB vorteilhaft verwendet. Somit kann das neue Verfahren die Konvergenzzeit erheblich verringern und die Zuverlässigkeit der Schätzung ohne a priori Werte erhöhen. Außerdem, für den Fall wenn mehr als ein Bias-Parameter in dem Modell enthalten ist, wurde der mehrdimensionale Partikelfilter-Ansatz entwickelt, um mehr als einen Bias-Parameter gleichzeitig innerhalb der präzisen GNSS-Positionierung abzuschätzen. In diesem Fall sind die oben genannten Vorteile des Verfahrens noch offensichtlicher. In der GLONASS-Datenverarbeitung mit einer Nicht-Null IFB Rate kann das Verfahren die IFB Rate aus den Beobachtungen von einigen wenigen Epochen abschätzen. Mit der geschätzten IFB Rate sind in den Experimenten mit kurzen Basislinien die GLONASS Lösungen mit festgesetzten Mehrdeutigkeiten so genau wie die dazugehörigen GPS Lösungen. Zusätzlich ist das Bias in der geschätzten IFB Rate, wenn das Zustandsrauschen als ein sehr kleiner Wert oder sogar Null festgelegt wird, signifikant, kann aber mit dem regularisierten Partikelfilter (RPF) entfernt werden und die Präzision des geschätzten IFB wird mit neuen Beobachtungen kontinuierlich verbessert. Ein Ansatz für die Anpassung der Anzahl der Teilchen in der Schätzung der IFB Rate wurde auch vorgeschlagen, um die Berechnungslast zu reduzieren, indem die Anzahl der Partikel mit der Standardabweichung der gewichteten Teilchen in Beziehung gesetzt wurde. In der Schätzung der F-ISB in der Integration von mehreren GNSS reduziert das neue auf dem Partikelfilter basierende Verfahren weitgehend die Konvergenzzeit und verbessert die Zuverlässigkeit der F-ISB Schätzung, wenn Satelliten von jedem einzelnen System zu wenige für eine unabhängige Positionierung sind. Aufgrund der periodischen Eigenschaften von ISB, können die F-ISB Partikel in verschiedene Gruppen getrennt werden, was zur Divergenz der Filterung führen kann. Dieses Problem wird durch die Einführung der Cluster-Analyse, die die Gruppen automatisch erkennen kann, so dass sie in der Filterung in eine Gruppe zusammengeführt werden können, erfolgreich gelöst. Die Schätzung der IFB Rate der Phase mit dem neuen Verfahren ermöglicht die Nutzung von GLONASS in Echtzeit für kinematische Positionierung, auch wenn das Bias zwischen den Empfängern groß ist. Die Schätzung des F-ISB der Phase mit dem neuen Verfahren erlaubt, dass die präzise Positionierung mit weniger Satelliten von jedem System durchgeführt wird, als erforderlich für gängige Methoden. Daher erweitert die Schätzung der IFB Rate/F-ISB bedeutend die Anwendung von kinematischen GNSS Echtzeitpositionierung. Es beweist auch, dass die entwickelte neue Methode Bias schnell und genau schätzen kann, und eine neue Art der Biasschätzung in der präzisen GNSS Positionierung einführt.

Description: Global Navigation Satellite Systems (GNSS) play an important role in precise positioning for geodesy and surveying engineering. The key to the real-time GNSS precise positioning is the instantaneous integer ambiguity resolution. However, some of the biases in carrier phase observations cannot be removed by differencing between either stations or satellites, so the integer nature of the double-differenced ambiguities is destroyed and thus the ambiguities cannot be fixed to integers. Two typical biases are the inter-frequency bias (IFB) in GLObal NAvigation Satellite System (GLONASS) data processing and the inter-system bias (ISB) in multi-GNSS integration. Hence, the main objective of this thesis is the investigation, estimation and correction of these biases in carrier phase observations to achieve better positioning accuracy, reliability and availability through the improvement of its ambiguity resolution. The estimated parameters of the carrier phase IFB and ISB are usually the IFB rate and the fractional ISB (F-ISB), respectively. Most of the current methods estimate IFB rate or F-ISB together with the float ambiguities and usually need observations of relatively long time due to their high correlation. Theoretically, the performance of the ambiguity resolution depends on the quality of the given IFB rate/F-ISB value if the observations are precisely modelled. In other words, the closer the given IFB rate/F-ISB value to the truth value is, the better the resolution will be. Therefore, the RATIO in the ambiguity fixing can be applied as the qualification factor of the IFB rate/F-ISB value. Based on this fact, a new methodology based on particle filter is developed to estimate these biases in both post-processing and real-time mode in this study. In the proposed method, the IFB/ISB is represented by its samples (i.e. particles) with the weights determined by the designed likelihood function of the related RATIO given the sample values, so that the true bias value can be estimated successfully by the particle filter approach. The integer nature of the ambiguities in the models with IFB/ISB parameters is well utilised in the ambiguity resolution with the given IFB rate/F-ISB values. Thus, the new method can significantly reduce the convergence time and increase the reliability of the estimation without a priori values. Besides, when more than one bias parameter is included in the model, the multi-dimensional particle filter approach is developed to estimate more than one bias parameter simultaneously in GNSS precise positioning. In this case, the aforementioned benefits of the method are obviously enlarged. In the GLONASS data processing with a nonzero IFB rate, the method can estimate the IFB rate from observations of a few epochs. With the estimated IFB rate, the GLONASS fixed solutions are as accurate as the GPS fixed solutions in the experiments with short baselines. In addition, the bias in the estimated IFB rate when the state noise is set to a very small value or even zero is significant, but this bias can be removed by utilising the regularized particle filter (RPF) and the precision of the estimated IFB rate is continuously improved by new observations. An approach for adapting the number of particles in the estimation of the IFB rate is also proposed to reduce the calculation burden by relating the number of particles to the standard deviation of the weighted particles. In the estimation of the F-ISB in multi-GNSS integration, the new method based on particle filter largely reduces the convergence time and improves the reliability of F-ISB estimation when satellites from each system are not sufficient for independent positioning. Due to the periodic characteristics of ISB, the F-ISB particles can be separated into different groups leading to the divergence of the filtering. This problem is solved successfully by introducing the cluster analysis method which can detect the groups automatically so that they can be shifted together into one group in the filtering. The estimation of the phase IFB rate with the new method enables the usage of GLONASS in real-time kinematic positioning even when the IFB between receivers is large. The estimation of the phase F-ISB with the new method allows the precise positioning to be carried out with fewer satellites from each system than the number of satellites required by the current methods. Therefore, the IFB rate/F-ISB estimation significantly extends the application of real-time kinematic GNSS positioning. It also proves that the developed new method is capable of estimating biases quickly and accurately, which initiates a new way of bias estimation in GNSS precise positioning.

Description: Soil water supply and atmospheric humidity conditions play an important role in controlling plants' stomatal behavior and water use efficiency. Based on the fluxnet2015 dataset around the world, the water use efficiency (WUE) and carbon use efficiency (CUE) of five plant functional types(PFTs), namely crops (CRO), deciduous broad-leaved forest (DBF), evergreen coniferous forest (ENF), grassland (GRA) and shrub (SAV) are estimated under different level of water vapor pressure (VPD) and soil water content (SWC). The results showed that high VPD and the low SWC limit the level of GPP for all vagetation types, while high SWC could offset the negative effects caused by high VPD to some extent for certain types of vegetation (CRO, DBF). In general, the increase of VPD in vegetation dominated the changes in WUE and CUE. The major change of WUE caused by the rise of VPD was at about 50 μmol/mol, Compared to VPD, the effect of SWC limitation is smaller, and the change of WUE dominated by the decline of SWC is at 30μmol/mol. It is worth noting that the changes of CUE of SAV is different from other vegetation types in that the rise of VPD and the decrease of SWC promote the increase of CUE of SAV to some extent. The study could help to provide understanding for the role of plants in the carbon and water cycle of the ecosystem under extreme climate and build an efficient water management schedule.

Description: A gravity sensor onboard an aircraft always measures the sum of all accelerations acting on it. To separate the accelerations caused by the movement of the aircraft from the total measured acceleration, the movement, including position, velocity and acceleration of the aircraft, must be measured independently. Nowadays, this is possible using GNSS. Obviously, this means that the kinematic acceleration must be measured or derived from GNSS measurements as accurately as the intended gravity survey. Compared to the traditional airborne gravimetry, the determination of positions and velocities from GNSS is a big challenge for the special HALO and similar jet aircrafts, which are characterized by high-altitude and long-range flying capabilities. A strategy of integrated GNSS Doppler velocity determination based on a combination of robust estimation and Helmert’s Variance Components Estimation (VCE) is proposed in this study to fulfill the requirements for such an aircraft. This strategy is tested by processing GNSS Doppler data recorded onboard the HALO aircraft. The velocity obtained has been applied in the data processing of the GEOHALO airborne gravimetry campaign of 2012. The results show that the proposed strategy improves GNSS Doppler velocity determination accuracy and allows the subtraction of the kinematic vertical accelerations from the GEOHALO airborne gravimetry records.

Description: The South China Block hosts a variety of U and HREE mineralisation styles. The Yushui Cu deposit is located at a sedimentary unconformity and is enriched in HREEs and U. U-Pb ages of uraninite and xenotime indicate that the HREE miner- alisation is epigenetic and formed at ca. 223 ± 1 Ma. Ore petrography, elemental map- ping, and Nd isotope data suggest that HREEs and U were leached from the footwall sandstone and transported to the Cu deposit via oxidised basinal brines. U-Pb ages of detrital xenotime and zircon from the sandstone show that this sedimentary sequence was mainly derived from Silurian S-type granites, which were emplaced during Gondwana amalgamation. Rapid erosion formed clastic sedimentary rocks that con- tain accessory HREE-U minerals which could be remobilised by younger mineralising events. S-type granite magmatism during the final assembly of Gondwana established the crustal metal reservoir which was repeatedly tapped over geological history, including the modern formation of regolith hosted HREE deposits in South China. Given the global distribution of analogous S-type granites in other terranes globally, our study has exploration implica- tions outside of China. This will be enlightening for finding new HREE deposits, which is vital to support the transition to a low carbon footprint energy.

Description: The Hadoop Distributed File System (HDFS) has become an important data repository in the enterprise as the center for all business analytics, from SQL queries and machine learning to reporting. At the same time, enterprise data warehouses (EDWs) continue to support critical business analytics. This has created the need for a new generation of a special federation between Hadoop-like big data platforms and EDWs, which we call the hybrid warehouse. There are many applications that require correlating data stored in HDFS with EDW data, such as the analysis that associates click logs stored in HDFS with the sales data stored in the database. All existing solutions reach out to HDFS and read the data into the EDW to perform the joins, assuming that the Hadoop side does not have efficient SQL support. In this article, we show that it is actually better to do most data processing on the HDFS side, provided that we can leverage a sophisticated execution engine for joins on the Hadoop side. We identify the best hybrid warehouse architecture by studying various algorithms to join database and HDFS tables. We utilize Bloom filters to minimize the data movement and exploit the massive parallelism in both systems to the fullest extent possible. We describe a new zigzag join algorithm and show that it is a robust join algorithm for hybrid warehouses that performs well in almost all cases. We further develop a sophisticated cost model for the various join algorithms and show that it can facilitate query optimization in the hybrid warehouse to correctly choose the right algorithm under different predicate and join selectivities.

Description: Signal propagation in complex networks drives epidemics, is responsible for information going viral, promotes trust and facilitates moral behavior in social groups, enables the development of misinformation detection algorithms, and it is the main pillar supporting the fascinating cognitive abilities of the brain, to name just some examples. The geometry of signal propagation is determined as much by the network topology as it is by the diverse forms of nonlinear interactions that may take place between the nodes. Advances are therefore often system dependent and have limited translational potential across domains. Given over two decades worth of research on the subject, the time is thus certainly ripe, indeed the need is urgent, for a comprehensive review of signal propagation in complex networks. We here first survey different models that determine the nature of interactions between the nodes, including epidemic models, Kuramoto models, diffusion models, cascading failure models, and models describing neuronal dynamics. Secondly, we cover different types of complex networks and their topologies, including temporal networks, multilayer networks, and neural networks. Next, we cover network time series analysis techniques that make use of signal propagation, including network correlation analysis, information transfer and nonlinear correlation tools, network reconstruction, source localization and link prediction, as well as approaches based on artificial intelligence. Lastly, we review applications in epidemiology, social dynamics, neuroscience, engineering, and robotics. Taken together, we thus provide the reader with an up-to-date review of the complexities associated with the network’s role in propagating signals in the hope of better harnessing this to devise innovative applications across engineering, the social and natural sciences as well as to inspire future research.