There has been an increasing demand for describing, predicting, and drawing inferences for various environmental processes, such as air pollution and precipitation. Environmental statistics plays an important role in many related applications, such as weather-related risk assessment for urban design and crop growth. However, modeling the spatio-temporal dynamics of environmental data is challenging due to their inherent high variability and nonstationarity. This dissertation is composed of four signi cant contributions to the modeling, simulation, and prediction of spatiotemporal processes using statistical techniques and machine learning algorithms. This dissertation rstly focuses on the Gaussian process emulators of the numerical climate models over a large spatial region, where the spatial process exhibits nonstationarity. The proposed method allows for estimating a rich class of nonstationary Mat ern covariance functions with spatially varying parameters. The e cient estimation is achieved by local-polynomial tting of the covariance parameters. To extend the applicability of this method to large-scale computations, the proposed method is implemented by developing software with high-performance computing architectures for nonstationary Gaussian process estimation and simulation. The developed software outperforms existing ones in both computational time and accuracy by a large margin. The method and software are applied to the statistical emulation of high-resolution climate models. The second focus of this dissertation is the development of spatio-temporal stochastic weather generators for non-Gaussian and nonstationary processes. The proposed multi-site generator uses a left-censored non-Gaussian vector autoregression model, where the random error follows a skew-symmetric distribution. It not only drives the occurrence and intensity simultaneously but also possesses nice interpretations both physically and statistically. The generator is applied to 30-second precipitation data collected at the University of Lausanne. Finally, this dissertation investigates the spatial prediction with scalable deep learning algorithms to overcome the limitations of the classical Kriging predictor in geostatistics. A novel neural network structure is proposed for spatial prediction by adding an embedding layer of spatial coordinates with basis functions. The proposed method, called DeepKriging, has multiple advantages over Kriging and classical neural networks with spatial coordinates as features. The method is applied to the prediction of ne particulate matter (PM2:5) concentrations in the United States.

京都大学 / 0048 / 新制・課程博士 / 博士(工学) / 甲第18969号 / 工博第4011号 / 新制||工||1618(附属図書館) / 31920 / 京都大学大学院工学研究科都市社会工学専攻 / (主査)教授 小池 克明, 教授 松岡 俊文, 教授 田村 正行 / 学位規則第4条第1項該当 / Doctor of Philosophy (Engineering) / Kyoto University / DFAM

In this study, five real-life problem situations were used to explore the challenges of developing and implementing decision support tools for management of an urban water utility in Uganda. The study sought to explore how the degree of adoption of formal decision support tools in practice, generally perceived to be low, could be improved. In the study, an Action Research (AR) approach was used. AR is an inquiry process that involves partnership between researchers and practitioners for the purpose of addressing a real-life problem issue, while simultaneously gener-ating scientific knowledge. Unlike other research methods where the researcher seeks to study organizational phenomena but not to change them, the action researcher attempts to create or-ganizational change and simultaneously to study the process. It is recognized that AR methods provide a potential avenue to improve the practical relevance of Information Systems (IS) re-search. The five cases that were considered in the study involved participatory problem structuring to address water distribution bottlenecks; identification of Non-Revenue Water (NRW) reduction strategies; facilitation of decentralized management of customer accounts; monitoring and con-trol of procurements and expenditure; and geospatial investigation of declining water sales. Dur-ing the study, participation in problem identification was achieved through discussions and brain-storming sessions bringing together top and middle managers within the organization. A number of prototype decision support tools were developed and implemented. Maps and other geovisu-alization tools were also used to inform and enhance the processes of collective problem identifi-cation and structuring. Results of the study emphasized the need for proper problem structuring prior to the formula-tion of actions; the challenge of moving from planning to action; the importance of user in-volvement in the development of tools; and the need to manage IS implementation as part of a holistic, organization-wide change process. The challenges of embedding formal decision support within existing work systems in organizations were highlighted, and recommendations were made on how best to achieve this. The AR approach was found to be useful in bridging the gap be-tween academic research and technological practice, thus supporting the development of IS with immediate and practical benefits to organizations. / QC 20101115

Southeast Louisiana is susceptible to the impact of subsidence due to natural and anthropogenic processes including sediment compaction and loading, fluid withdrawal, and faulting. Subsidence rates in Southeast Louisiana are higher than anywhere else in the United States, and the impact of subsidence rates on industrial complexes has not been studied. Spatial interpolation methods were analyzed to determine the best fit for subsidence rates and to create a predictive surface for the lower Mississippi River Industrial corridor (LMRIC). Empirical Bayesian kriging, ordinary kriging, universal kriging, and inverse distance weighted interpolation methods were applied to the 2004 National Oceanic and Atmospheric Administration (NOAA) published Technical Report #50 dataset and cross-validation methods were utilized to determine the accuracy of each method. The mean error and root mean square error were calculated for each interpolation method, then used to detect bias and compare the predicted value with the actual observation value. Cross-validation estimates are comparable for each method statistically and visually; however, the results indicate the empirical Bayesian kriging interpolation method is the most accurate of the methods using the lowest mean error and root mean square error scores. Digital elevation models for the years 2025, 2050, and 2075 were developed based on the predictive surface of subsidence rates using the results from the empirical Bayesian kriging interpolation method. Results indicate that by 2025, 31.4% of landmass in the LMRIC will be below 0 m NAVD88, with 40.4% below 0 m NAVD88 by 2050, and 51.8% by 2075. Subsidence rates in the LMRIC range from approximately 16 mm to less than one mm per year. Nine of the 122 industrial complexes located in the LMRIC are estimated to be below 0 m NAVD88 by the year 2075. Limited economic impacts can be inferred based on the number of facilities impacted; however, service disruptions due to subsidence impacting infrastructure surrounding these industrial complexes would have catastrophic economic impacts on a regional, state, and national level.

This study was designed to investigate the relationships between spectral properties of cottonwood plantations and defoliation by the cottonwood leaf beetle (CLB), Chrysomela scripta F., as well as to develop and analyze the spatial structure of CLB and associated defoliation using geospatial information technology. Multispectral imagery data from airborne platforms were acquired using three remote sensing systems: (1) GeoVantage remote sensing system with 450 nm, 550 nm, 650 nm and 850 nm wavelengths; (2) RDACS (real time digital airborne camera system) with 540 nm, 675 nm, 695 nm and 840 nm wavelengths; and (3) Kodak DCS420 digital camera with the 500-810 nm wavelengths (red, green and near infrared). In addition, multispectral and hyperspectral radiometric data were collected using spectroradiometer. Analyzing reflectance values for simulated CLB defoliation indicated that the 0% and 25% defoliation could be differentiated from the 75% defoliation in the near infrared (NIR). Utilizing normalized difference vegetation index and a simple vegetation index, 0% and 25% defoliation could also be separated from 50% and 75% defoliation. Reflectance values for natural defoliation by CLB and various ground covers indicated that NIR was the best indicator for distinguishing different ground cover types. Heavy feeding by CLB could be detected but light or no feeding could not be discriminated from each other using these systems mentioned above. Spectral spatial analysis of a first year rising cottonwood plantation indicated that a wave (hole-effect) variogram model could be used to describe the spatial structure at omnidirection. Within the 1.90 - 2.48 m range, there is a significant spatial autocorrelation of reflectance. Spatial structure of adult and larval populations indicated that spatial dependence varied among dates and directions. Spherical and Gaussian functions provided the best statistical fit for CLB adult and larval spatial distributions, which were aggregated. The signature analysis for hyperspectral data indicated that when comparing the 0% defoliation to the 25, 50 and 75% defoliation, maximum reflectance differences were found near 294 nm in the ultraviolet, 550 nm in the visible spectrum and 764 nm in the near infrared spectrum. The highest reflectance sensitivity occurred between 528-557 nm in the visible spectrum. High sensitivities were also found between 730-740 nm and 930-940 nm in the infrared spectrum. Bands from 892-894 nm were best for separating various defoliation levels. This study has first demonstrated the application of remote sensing combined with GPS and geostatistics to CLB defoliation of cottonwood. The defoliation level of cottonwood can be detected and assessed using different remote sensing systems. Spatial models can be used to map CLB population densities and defoliation rates in landscapes. These maps can be used for site specific CLB management.

A rearing methodology for dogwood borer was developed, using standardized procedures at each developmental stage. These methods enabled the establishment of a laboratory colony and efficient production of synchronized cohorts of each of its lifestages throughout the year for specific experimental needs. The behavioral repertoire shown by mated female dogwood borer in an apple orchard was characterized and quantified and the diel periodicity with which those behaviors occurred was determined. Mated females were easily recognized, based on their characteristic casting flight directed toward areas below the graft union of apple trees, and were observed more frequently during the late afternoon and evening. Casting flight, probing with the ovipositor, and oviposition were the most frequent behaviors observed, but the duration of those behaviors was relatively short compared with the much lengthier periods of resting behavior that typically occurred within the canopy. Data from a previous, three-year study in two newly planted apple orchards were subjected to geostatistical analyses to examine the temporal and spatial patterns of infestation by larval dogwood borer and to gain further information about the spatial scales at which oviposition occurs. There were moderate to high degrees of aggregation of dogwood borer infestations on neighboring apple trees, with ranges of spatial dependence from 7.50–19.87 m. No directionality was observed in the spatial autocorrelation of infestation and it appears that females utilized oviposition sites equally along and across orchard rows. The aggregated nature of infestations requires that random, independent samples must be taken from a number of sample pairs at distances greater than the range of spatial dependence to ensure that sample data are not autocorrelated. Alternatively, an efficient sampling program for mapping dogwood borer infestation can be achieved by limiting sample points to distances within the range of spatial dependence. These sample points can be used in interpolating algorithms, such as kriging, to predict infestation at unsampled locations in space for use in site-specific pest management programs. The external morphology of male and female dogwood borer antennae and their sensilla were examined using light and scanning electron microscopy to characterize, measure and compare the types, number, and distribution of sensilla. Although the general shape and size of male and female antennae were similar, those from females possessed a greater number of generally smaller antennal flagellomeres. The flagellum of both male and female antennae contained seven sensillum types including auricillica, basiconica, chaetica, coeloconica, squamiformia, styloconica, and three subtypes of sensilla trichoidea. With the exception of sensilla basiconica, which were present in roughly equal numbers on male and female antennae, all other sensillum types were significantly more abundant on female antennae. The antennae of female dogwood borer appear well equipped to perceive olfactory stimuli, based on the types and number of sensilla present. Coupled gas chromatography and electroantennogram detection (GC-EAD) analyses of headspace collections from damaged and undamaged tissues from apple and dogwood trees were conducted to examine and compare the antennal responsiveness of female dogwood borer to host plant volatiles. A total of 16 and 9 compounds from apple and dogwood tissues, respectively, consistently elicited an antennal response in females. There were no differences in the response of antennae from virgin and mated females, and the amplitude of the female response to host odors was greater than that of males. Six compounds were identified from the headspace collections from apple trees, four of which (octanal, nonanal, decanal, and methyl salicylate) were identified from all apple tissues sampled. A novel compound,α-bergamotene, was identified from injured apple bark, from apple burr knots infested with dogwood borer larvae and from larval dogwood borer frass, and appears to be produced by apple trees in response to injury. Another novel compound, methyl-2,4-decadienoate, was identified from infested burr knot tissue and appears to be produced in response to an insect-plant interaction. Two compounds, hexanoic and nonanoic acid, were identified from headspace collections from dogwood trees. Numerous approaches were used to examine the behavioral response of mated female dogwood borer to host plant headspace collections and to individual compounds from those collections that elicited a strong and repeatable antennal response. Under both natural and semi-natural conditions in the field and in laboratory bioassays, neither attraction/orientation or consistent oviposition were documented and it is apparent that correlating the electrophysiological and behavioral responses of mated female dogwood borer to olfactory stimuli from their host plants will require further research on bioassay development. / Ph. D.

[pt] Na área de Exploração e Produção (EeP) de petróleo, o estudo minucioso das características de um reservatório é imperativo para a criação de modelos de simulação que representem adequadamente as suas propriedades petrofísicas. A disponibilidade de um modelo adequado é fundamental para a obtenção de previsões acertadas acerca da produção do reservatório, e isso impacta diretamente a tomada de decisões gerenciais. Devido às incertezas inerentes ao processo de caracterização, ao longo da vida produtiva do reservatório, periodicamente o seu modelo de simulação correspondente precisa ser ajustado. Todavia, a tarefa de ajustar as propriedades do modelo se traduz em um problema de otimização complexo, onde o número de variáveis envolvidas é tão maior quanto maior for a quantidade de blocos que compõem a malha do modelo de simulação. Na maioria das vezes esses ajustes envolvem processos empíricos que demandam elevada carga de trabalho do especialista. Esta pesquisa investiga e avalia uma nova técnica computacional híbrida, que combina Algoritmos Genéticos e Geoestatística Multiponto, para a otimização de propriedades em modelos de reservatórios. Os resultados obtidos demonstram a robustez e a confiabilidade da solução proposta, uma vez que, diferentemente das abordagens tradicionalmente adotadas, é capaz de gerar modelos que não apenas proporcionam um ajuste adequado das curvas de produção, mas também que respeitam as características geológicas do reservatório. / [en] In the Exploration and Production (EeP) of oil, the detailed study of reservoir characteristics is imperative for the creation of simulation models that adequately represent their petrophysical properties. The availability of an appropriate model is fundamental to obtaining accurate predictions about the reservoir production. In addition, this impacts directly the management decisions. Due to the uncertainties inherent in the characterization process, along the productive period of the reservoir, its corresponding simulation model needs to be matched periodically. However, the task of matching the model properties represents a complex optimization problem. In this case, the number of variables involved increases with the number of blocks that make up the grid of the simulation model. In most cases these matches involve empirical processes that take too much time of an expert. This research investigates and evaluates a new hybrid computer technique, which combines Genetic Algorithms and Multipoint Geostatistics, for the optimization of properties in reservoir models. The results demonstrate the robustness and reliability of the proposed solution. Unlike traditional approaches, it is able to generate models that not only provide a proper match of the production curves, but also satisfies the geological characteristics of the reservoir.

学位プログラム名: 京都大学大学院思修館 / 京都大学 / 新制・課程博士 / 博士(総合学術) / 甲第25517号 / 総総博第39号 / 新制||総総||7(附属図書館) / 京都大学大学院総合生存学館総合生存学専攻 / (主査)准教授 趙 亮, 教授 長山 浩章, 教授 関山 健 / 学位規則第4条第1項該当 / Doctor of Philosophy / Kyoto University / DGAM

Das Ziel der vorliegenden Arbeit war die Erstellung eines dreidimensionalen Untergrundmodells der Region Göttingen basierend auf einer geotechnischen Klassifikation der unkosolidierten Sedimente. Die untersuchten Materialen reichen von Lockersedimenten bis hin zu Festgesteinen, werden jedoch in der vorliegenden Arbeit als Boden, Bodenklassen bzw. Bodenkategorien bezeichnet. Diese Studie evaluiert verschiedene Möglichkeiten durch geostatistische Methoden und Simulationen heterogene Untergründe zu erfassen. Derartige Modellierungen stellen ein fundamentales Hilfswerkzeug u.a. in der Geotechnik, im Bergbau, der Ölprospektion sowie in der Hydrogeologie dar. Eine detaillierte Modellierung der benötigten kontinuierlichen Parameter wie z. B. der Porosität, der Permeabilität oder hydraulischen Leitfähigkeit des Untergrundes setzt eine exakte Bestimmung der Grenzen von Fazies- und Bodenkategorien voraus. Der Fokus dieser Arbeit liegt auf der dreidimensionalen Modellierung von Lockergesteinen und deren Klassifikation basierend auf entsprechend geostatistisch ermittelten Kennwerten. Als Methoden wurden konventionelle, pixelbasierende sowie übergangswahrscheinlichkeitsbasierende Markov-Ketten Modelle verwendet. Nach einer generellen statistischen Auswertung der Parameter wird das Vorhandensein bzw. Fehlen einer Bodenkategorie entlang der Bohrlöcher durch Indikatorparameter beschrieben. Der Indikator einer Kategorie eines Probepunkts ist eins wenn die Kategorie vorhanden ist bzw. null wenn sie nicht vorhanden ist. Zwischenstadien können ebenfalls definiert werden. Beispielsweise wird ein Wert von 0.5 definiert falls zwei Kategorien vorhanden sind, der genauen Anteil jedoch nicht näher bekannt ist. Um die stationären Eigenschaften der Indikatorvariablen zu verbessern, werden die initialen Koordinaten in ein neues System, proportional zur Ober- bzw. Unterseite der entsprechenden Modellschicht, transformiert. Im neuen Koordinatenraum werden die entsprechenden Indikatorvariogramme für jede Kategorie für verschiedene Raumrichtungen berechnet. Semi-Variogramme werden in dieser Arbeit, zur besseren Übersicht, ebenfalls als Variogramme bezeichnet. IV Durch ein Indikatorkriging wird die Wahrscheinlichkeit jeder Kategorie an einem Modellknoten berechnet. Basierend auf den berechneten Wahrscheinlichkeiten für die Existenz einer Modellkategorie im vorherigen Schritt wird die wahrscheinlichste Kategorie dem Knoten zugeordnet. Die verwendeten Indikator-Variogramm Modelle und Indikatorkriging Parameter wurden validiert und optimiert. Die Reduktion der Modellknoten und die Auswirkung auf die Präzision des Modells wurden ebenfalls untersucht. Um kleinskalige Variationen der Kategorien auflösen zu können, wurden die entwickelten Methoden angewendet und verglichen. Als Simulationsmethoden wurden "Sequential Indicator Simulation" (SISIM) und der "Transition Probability Markov Chain" (TP/MC) verwendet. Die durchgeführten Studien zeigen, dass die TP/MC Methode generell gute Ergebnisse liefert, insbesondere im Vergleich zur SISIM Methode. Vergleichend werden alternative Methoden für ähnlichen Fragestellungen evaluiert und deren Ineffizienz aufgezeigt. Eine Verbesserung der TP/MC Methoden wird ebenfalls beschrieben und mit Ergebnissen belegt, sowie weitere Vorschläge zur Modifikation der Methoden gegeben. Basierend auf den Ergebnissen wird zur Anwendung der Methode für ähnliche Fragestellungen geraten. Hierfür werden Simulationsauswahl, Tests und Bewertungsysteme vorgeschlagen sowie weitere Studienschwerpunkte beleuchtet. Eine computergestützte Nutzung des Verfahrens, die alle Simulationsschritte umfasst, könnte zukünftig entwickelt werden um die Effizienz zu erhöhen. Die Ergebnisse dieser Studie und nachfolgende Untersuchungen könnten für eine Vielzahl von Fragestellungen im Bergbau, der Erdölindustrie, Geotechnik und Hydrogeologie von Bedeutung sein.

Résumé : Les barrages induisent un impact potentiellement important sur l'environnement et la société par la modification des écosystèmes, le déplacement des populations en amont et surtout par la présence d'un risque imposé aux populations situées à l'aval. Même si les ruptures de ces ouvrages ont une probabilité d'occurrence faible, la cause principale des accidents, à côté des submersions par des crues imprévues, a toujours été la rupture des fondations (ICOLD 1993) suite à des problèmes d'érosion, ou d'insuffisance de résistance au cisaillement le long des discontinuités rocheuses, des joints de levées ou de l'interface barrage-fondation. La fréquence de ruptures par défaut de résistance au cisaillement des discontinuités des fondations des barrages en béton est estimée à 21 % (ICOLD 1995). Les fondations rocheuses présentent souvent une variabilité verticale et horizontale des propriétés de résistance au cisaillement des discontinuités. Une variabilité verticale peut être matérialisée par une moyenne de la propriété de résistance qui varie en profondeur et dont la prise en compte réduit la variance de cette propriété de résistance. Aussi, la rupture des fondations rocheuses par défaut de résistance au cisaillement des discontinuités est un mécanisme qui mobilise les propriétés ponctuelles de résistance au cisaillement présentes sur cette discontinuité. La résistance au cisaillement mobilisée le long de cette discontinuité présente parfois une variance moins élevée que celle de la résistance ponctuelle au cisaillement (échelle de laboratoire). Ces phénomènes de variabilité spatiale ne sont pas pris en compte dans la justification du comportement en cisaillement des fondations ni dans les travaux de recherches publiées dans la littérature. L'objectif de la thèse est d'élaborer une démarche d'analyse de la variabilité spatiale de la résistance au cisaillement qui prend en compte cette réduction de la dispersion. Cette démarche a été élaborée dans le contexte de l'état de l'art actuel, avec la combinaison des impératifs suivants : • La démarche est basée sur une méthodologie expérimentale i)-alternative à l'essai de cisaillement direct (coûteux en temps et en argent) pouvant être relativement simple à mettre en œuvre ii)- qui permet de générer des données quantitatives de cisaillement suffisamment abondantes pour analyser la variabilité spatiale de la résistance au cisaillement. • Cette démarche a été mise en œuvre au contexte des outils numériques et des méthodes de justification de la résistance à l'effort tranchant d'une fondation rocheuse de barrage-poids. La méthodologie expérimentale définie dans la thèse a fait l'objet d'une validation par étude comparative avec 35 essais de cisaillement direct sur des joints à différents degrés de rugosité et d’altération, prélevés d'une fondation de barrage-poids en béton. Il a été développé une démarche d'analyse de la variabilité spatiale de la résistance au cisaillement des joints. Cette démarche repose sur l'utilisation des données géo référencées de paramètres de cisaillement caractérisant la même famille de discontinuité le long d'un forage vertical. Afin de valider la pertinence et l’applicabilité de cette démarche, nous l’avons implémentée dans un cas d'étude de fondation rocheuse de barrage. A l'aide d'un code de calcul numérique basé sur la méthode des éléments finis, une étude a permis d'illustrer la pertinence de l'analyse de la variabilité spatiale de la résistance au cisaillement des joints dans une étude de stabilité au glissement des fondations rocheuses de barrage. Les résultats montrent que la prise en compte de la variabilité spatiale des paramètres de cisaillement permet d'augmenter la marge de sécurité mesurée à travers une hausse du coefficient de sécurité et une baisse significative de la probabilité de défaillance. // Abstract : Dams induce a potentially significant impact on the environment and society by changing ecosystems, by populations’ relocation upstream and especially by the presence of an imposed risk on populations located downstream. Although dam failures have a low probability, the main cause of accidents, near submersions by unexpected flood has been foundations failures (ICOLD 1993) due to problems of erosion, or insufficient shear strength along the rock discontinuities or the interface between dam and foundation. Among dam failures, 21% are related to lack of shear strength along the discontinuities of the foundations (ICOLD 1995). It is well known that rock, as soil, is a complex engineering material formed y natural process, which induces vertical and horizontal variability. Plus, the sliding process of a rock foundation discontinuity is a mechanism that mobilizes points shear properties of this discontinuity. The variability of shear properties averaged over the sheared discontinuities is less than that of their point shear properties. This is known as the average effect in spatial variability. These phenomena of spatial variability are not taken into account in the standards of stability analysis of dam foundations sliding or research work published in the literature. The aim of the thesis is to develop an analysis of the spatial variability of shear strength that takes into account the reduction of the variability by scaling up effect. This approach was developed in the context of the current state of the art. An experimental method, which aimed to investigate in a simple way the shear strength of joints, was suggested in the thesis and has been validated by comparative study with more than thirty direct shear tests on joints taken from a foundation of concrete gravity dam. It was developed a process to analyze the spatial variability of the shear strength of the joints using the input parameters of the model of Barton and Choubey (1977). This approach is based on the use of geo -referenced data on these parameters characterizing the family of discontinuity along a vertical borehole. To validate the relevance and applicability of this approach, we have implemented it in a case study of a rock dam foundation. Using finite element software, a study illustrates the relevance of the analysis of the spatial variability of the shear strength of the joints in the assessment of stability against rock dam foundations sliding. The results show that taking into account the spatial variability of shear parameters can increase the margin of safety measured through an increase in the safety factor and a significant decrease in the probability of failure.