<p>A global climate change may have large impacts on water resources on regional and global scales. General circulation models (GCMs) are the most used tools to evaluate climate-change scenarios on a global scale. They are, however, insufficiently describing the effects at the local scale. This thesis evaluates different approaches of statistical downscaling of precipitation from large-scale circulation variables, both concerning the method performance and the optimum choice of predictor variables. </p><p>The analogue downscaling method (AM) was found to work well as “benchmark” method in comparison to more complicated methods. AM was implemented using principal component analysis (PCA) and Teweles-Wobus Scores (TWS). Statistical properties of daily and monthly precipitation on a catchment in south-central Sweden, as well as daily precipitation in three catchments in China were acceptably downscaled.</p><p>A regression method conditioning a weather generator (SDSM) as well as a fuzzy-rule based circulation-pattern classification method conditioning a stochastical precipitation model (MOFRBC) gave good results when applied on Swedish and Chinese catchments. Statistical downscaling with MOFRBC from GMC (HADAM3P) output improved the statistical properties as well as the intra-annual variation of precipitation.</p><p>The studies show that temporal and areal settings of the predictor are important factors concerning the success of precipitation modelling. The MOFRCB and SDSM are generally performing better than the AM, and the best choice of method is depending on the purpose of the study. MOFRBC applied on output from a GCM future scenario indicates that the large-scale circulation will not be significantly affected. Adding humidity flux as predictor indicated an increased intensity both in extreme events and daily amounts in central and northern Sweden.</p>

Wireless sensor networks form an emerging class of computing devices capable of observing the world with an unprecedented resolution, and promise to provide a revolutionary instrument for environmental monitoring. Such a network is composed of a collection of battery-operated wireless sensors, or sensor nodes, each of which is equipped with sensing, processing and wireless communication capabilities. Thanks to advances in microelectronics and wireless technologies, wireless sensors are small in size, and can be deployed at low cost over different kinds of environments in order to monitor both over space and time the variations of physical quantities such as temperature, humidity, light, or sound. In environmental monitoring studies, many applications are expected to run unattended for months or years. Sensor nodes are however constrained by limited resources, particularly in terms of energy. Since communication is one order of magnitude more energy-consuming than processing, the design of data collection schemes that limit the amount of transmitted data is therefore recognized as a central issue for wireless sensor networks. An efficient way to address this challenge is to approximate, by means of mathematical models, the evolution of the measurements taken by sensors over space and/or time. Indeed, whenever a mathematical model may be used in place of the true measurements, significant gains in communications may be obtained by only transmitting the parameters of the model instead of the set of real measurements. Since in most cases there is little or no a priori information about the variations taken by sensor measurements, the models must be identified in an automated manner. This calls for the use of machine learning techniques, which allow to model the variations of future measurements on the basis of past measurements. This thesis brings two main contributions to the use of learning techniques in a sensor network. First, we propose an approach which combines time series prediction and model selection for reducing the amount of communication. The rationale of this approach, called adaptive model selection, is to let the sensors determine in an automated manner a prediction model that does not only fits their measurements, but that also reduces the amount of transmitted data. The second main contribution is the design of a distributed approach for modeling sensed data, based on the principal component analysis (PCA). The proposed method allows to transform along a routing tree the measurements taken in such a way that (i) most of the variability in the measurements is retained, and (ii) the network load sustained by sensor nodes is reduced and more evenly distributed, which in turn extends the overall network lifetime. The framework can be seen as a truly distributed approach for the principal component analysis, and finds applications not only for approximated data collection tasks, but also for event detection or recognition tasks. / Les réseaux de capteurs sans fil forment une nouvelle famille de systèmes informatiques permettant d'observer le monde avec une résolution sans précédent. En particulier, ces systèmes promettent de révolutionner le domaine de l'étude environnementale. Un tel réseau est composé d'un ensemble de capteurs sans fil, ou unités sensorielles, capables de collecter, traiter, et transmettre de l'information. Grâce aux avancées dans les domaines de la microélectronique et des technologies sans fil, ces systèmes sont à la fois peu volumineux et peu coûteux. Ceci permet leurs deploiements dans différents types d'environnements, afin d'observer l'évolution dans le temps et l'espace de quantités physiques telles que la température, l'humidité, la lumière ou le son. Dans le domaine de l'étude environnementale, les systèmes de prise de mesures doivent souvent fonctionner de manière autonome pendant plusieurs mois ou plusieurs années. Les capteurs sans fil ont cependant des ressources limitées, particulièrement en terme d'énergie. Les communications radios étant d'un ordre de grandeur plus coûteuses en énergie que l'utilisation du processeur, la conception de méthodes de collecte de données limitant la transmission de données est devenue l'un des principaux défis soulevés par cette technologie. Ce défi peut être abordé de manière efficace par l'utilisation de modèles mathématiques modélisant l'évolution spatiotemporelle des mesures prises par les capteurs. En effet, si un tel modèle peut être utilisé à la place des mesures, d'importants gains en communications peuvent être obtenus en utilisant les paramètres du modèle comme substitut des mesures. Cependant, dans la majorité des cas, peu ou aucune information sur la nature des mesures prises par les capteurs ne sont disponibles, et donc aucun modèle ne peut être a priori défini. Dans ces cas, les techniques issues du domaine de l'apprentissage machine sont particulièrement appropriées. Ces techniques ont pour but de créer ces modèles de façon autonome, en anticipant les mesures à venir sur la base des mesures passées. Dans cette thèse, deux contributions sont principalement apportées permettant l'applica-tion de techniques d'apprentissage machine dans le domaine des réseaux de capteurs sans fil. Premièrement, nous proposons une approche qui combine la prédiction de série temporelle avec la sélection de modèles afin de réduire la communication. La logique de cette approche, appelée sélection de modèle adaptive, est de permettre aux unités sensorielles de determiner de manière autonome un modèle de prédiction qui anticipe correctement leurs mesures, tout en réduisant l'utilisation de leur radio. Deuxièmement, nous avons conçu une méthode permettant de modéliser de façon distribuée les mesures collectées, qui se base sur l'analyse en composantes principales (ACP). La méthode permet de transformer les mesures le long d'un arbre de routage, de façon à ce que (i) la majeure partie des variations dans les mesures des capteurs soient conservées, et (ii) la charge réseau soit réduite et mieux distribuée, ce qui permet d'augmenter également la durée de vie du réseau. L'approche proposée permet de véritablement distribuer l'ACP, et peut être utilisée pour des applications impliquant la collecte de données, mais également pour la détection ou la classification d'événements.

A global climate change may have large impacts on water resources on regional and global scales. General circulation models (GCMs) are the most used tools to evaluate climate-change scenarios on a global scale. They are, however, insufficiently describing the effects at the local scale. This thesis evaluates different approaches of statistical downscaling of precipitation from large-scale circulation variables, both concerning the method performance and the optimum choice of predictor variables. The analogue downscaling method (AM) was found to work well as “benchmark” method in comparison to more complicated methods. AM was implemented using principal component analysis (PCA) and Teweles-Wobus Scores (TWS). Statistical properties of daily and monthly precipitation on a catchment in south-central Sweden, as well as daily precipitation in three catchments in China were acceptably downscaled. A regression method conditioning a weather generator (SDSM) as well as a fuzzy-rule based circulation-pattern classification method conditioning a stochastical precipitation model (MOFRBC) gave good results when applied on Swedish and Chinese catchments. Statistical downscaling with MOFRBC from GMC (HADAM3P) output improved the statistical properties as well as the intra-annual variation of precipitation. The studies show that temporal and areal settings of the predictor are important factors concerning the success of precipitation modelling. The MOFRCB and SDSM are generally performing better than the AM, and the best choice of method is depending on the purpose of the study. MOFRBC applied on output from a GCM future scenario indicates that the large-scale circulation will not be significantly affected. Adding humidity flux as predictor indicated an increased intensity both in extreme events and daily amounts in central and northern Sweden.

Face recognition is a popular research area where there are different approaches studied in the literature. In this thesis, a holistic Principal Component Analysis (PCA) based method, namely Eigenface method is studied in detail and three of the methods based on the Eigenface method are compared. These are the Bayesian PCA where Bayesian classifier is applied after dimension reduction with PCA, the Subspace Linear Discriminant Analysis (LDA) where LDA is applied after PCA and Eigenface where Nearest Mean Classifier applied after PCA. All the three methods are implemented on the Olivetti Research Laboratory (ORL) face database, the Face Recognition Technology (FERET) database and the CNN-TURK Speakers face database. The results are compared with respect to the effects of changes in illumination, pose and aging. Simulation results show that Subspace LDA and Bayesian PCA perform slightly well with respect to PCA under changes in pose / however, even Subspace LDA and Bayesian PCA do not perform well under changes in illumination and aging although they perform better than PCA.

In recent years, the celebration activities of various types of folk-custom festivals in Taiwan have already been getting more and more attention from civilians. Festivities throughout the whole island are traditionally accompanied by loud and brightly colored firework displays. Among these activities, the firework display during the Chinese Lantern Festival in Kaohsiung City is one of the largest festivals in Taiwan every year. Therefore, it is important to investigate the influences of firework displays on ambient air quality during the Chinese Lantern Festival in Kaohsiung City. Field measurement of ambient gaseous pollutants and particulate matter (PM) was conducted on February 9-12, 2009, the Chinese Lantern Festival, in Kaohsiung City. Moreover, three kinds of firework powders obtained from the same factory producing Kaohsiung Lantern Festival fireworks were burned in a combustion chamber to determine the physicochemical properties of firework aerosols. Metallic elements were analyzed with an inductively coupled plasma-atomic emission spectrometer (ICP-AES). Ionic species and carbonaceous contents in the PM samples were analyzed with an ion chromatography (IC) and an elemental analyzer (EA), respectively. Finally, the source identification and apportionment of PM were analyzed by principal component analysis (PCA), enrichment factor (EF), and receptor modeling (CMB). For inorganic gaseous pollutants, the concentration peaks of NO, NO2, O3, CO were observed during the firework periods, and the concentration peak of NO was approximately 8.8 times higher than those during the non-firework periods. This study further revealed that, even at nighttime, ambient O3 could be reduced dramatically during the firework periods, whenas NO2 concentration increased concurrently, due to titration effects resulting from the prompt reaction of NO with O3 to form NO2 and O2. For organic gaseous pollutants, the concentration peak of toluene during the firework periods was approximately 2.2-4.1 times higher than those during the non-firework periods. Several metallic elements of PM during the firework display periods were obviously higher than those during the non-firework periods. On February 10, the concentrations of Mg, K, Pb, and Sr in PM2.5 were 10 times higher than those during the non-firework periods. Besides, the Cl-/Na+ ratio was slightly smaller than 1 in Kaohsiung Harbor, but it was approximately 3 during the firework display periods since Cl- came form chlorine content in firework aerosols at this time. Moreover, OC/EC ratio increased up to 2.8. In addition to the analysis of gaseous pollutants and PM during the Chinese Lantern Festival in Kaohsiung City, this study burned firework powders in a self-designed combustion chamber to measure the physicochemical properties of firework aerosols. In the results, K, Mg, Cl-, OC were major contents (<10%) in the aerosols produced from the burning firework powders. Moreover, Cl-/Na+ and OC/EC ratio were 15.0~23.4 and 2.9~3.2, respectively. Consequently, Cl-/Na+ and OC/EC ratio can be used as two important indicators of firework displays. Results obtained from PCA and CMB receptor modeling showed that the major sources of aerosols during the firework display periods were firework displays, motor/diesel vehicle exhanst, soil dusts, and marine aerosols. Besides, the firework displays on February 10 contributed approximately 25.2% and 16.6% of PM10 at two sampling sites, respectively.

In recent years, the rapid development of economy and industry in Xiamen Bay causes serious environmental problems, particularly poor air quality and visibility impairment. There are no large-scale industrial emission sources in Kinmen Island, however, its ambient air quality is always the poorest in Taiwan. Moreover, ambient air quality monitoring data showed that PM10 concentrations varied in daytime and at nighttime. Consequently, this study tired to ascertain the potential causes for this phenomenon. This study selected ten particulate matter (PM) sampling sites at Xiamen Bay, including five sites at Kinmen Island and five sites at metro Xiamen. Particulate matter sampling was conducted in daytime (8:00-17:00) and at nighttime (17:00-8:00), which included regular and intensive sampling. Regular sampling was conducted to collect PM10 with high-volume samplers three times a month from April 2009 to April 2010, while intensive sampling was conducted to collect fine (PM2.5) and coarse (PM2.5-10) particles with dichotomous samplers and particle size distribution with a MOUDI at site B2 for consecutive 5 days in the spring and winter of 2009~2010. After sampling, the physicochemical properties of PM, including mass concentrations, particle size distribution, water- soluble ionic species, metallic elements, and carbonaceous contents were further analyzed. The level of atmospheric PM is affected by meteorological condition, thus PM10 concentrations in winter and fall was much higher than those in spring and summer. Results from backward trajectories showed that the concentrations of PM10 blown from the north were generally higher than those from the south. Furthermore, t-test analysis indicated that PM10 concentrations in daytime and at nighttime at site B3 were significantly different (p-value<0.05). During the intensive sampling periods, PM10 concentrations were mainly affected by coarse particles compared to fine particles. The highest concentration for fine and coarse particle modes occurred at the size ranges of 0.32~0.56 £gm and 3.2~5.6 £gm, respectively. The most abundant water-soluble ionic species of PM10 were secondary inorganic aerosols (SO42-, NO3-, and NH4+) which accounted for 85% of total ions. The daytime and nighttime PM10 concentration ratios (D/N) for Mg, K, Ca, Cr, Mn, Fe, Zn, Al, Cu, As, and V were in the same order of magnitude, however, the D/N ratios of Cd, Pb, Ni, and Ti in spring and summer varied higher than an order of magnitude, indicating that the emission sources of PM were different in daytime and at nighttime. Correlation analysis of OC and EC showed that OC and EC at nighttime had a higher correlation than those in daytime, while OC and EC had a higher correlation in Kinmen Island than those in metro Xiamen, indicating carbonaceous sources must be different in summer and winter at Xiamen Bay. Enrichment factor analysis revealed that ceramic industry, stone processing, and cement industry had higher correlation with PM10 concentration than utility power plants. Crustal dusts consisted of road dusts, farmland dusts, and constructive dusts, while biomass burning was not a negligible sources. Results obtained from PCA and CMB receptor modeling showed that major sources of PM in Xiamen Bay were secondary inorganic aerosols, fuel and biomass burning, marine aerosols, vehicular exhansts, and soil dusts. Besides, stone processing, cement industry, ceramic industry, and utility power plants had the highest contribution in winter. Their contributions in daytime and at nighttime were 38% and 45%, respectively.

This thesis presents content-related and construct-related validity evidence for science sub-tests within Student Selection Test (SST) in Turkey via underlying the content, cognitive processes, item characteristics, factorial structure, and group differences based on high school type. A total number of 126,245 students were present in the research from six type of school in the data of SST 2006. Reliability Analysis, Item Analysis, Principle Component Analysis (PCA) and one-way ANOVA have been carried out to evaluate the content-related and construct-related evidence of validity of SST. SPSS and ITEMAN programs were used to conduct the above-mentioned analyses. According to the results of content analysis, science items in the SST 2006 found to be measuring various cognitive processes under knowledge, understanding and problem solving cognitive domains. Those items loaded under three factors according to PCA findings which were measuring very close dimensions. Moreover, a threat to validity was detected via one-way ANOVA due to significant mean difference across high school types.

This study investigated the concentration distributions of polyclic aromatic hydrocarbons (PAHs) in the sediment cores collected from Fang-Liao submarine canyon. Chemical fingerprinting techniques and statistical analysis were applied to delineate the possible sources of the PAHs in deposited sediment core samples. It is noteworthy that all cores were not dated; therefore the deposition age could not estimate from the depth of deposition directly. The average concentrations of polyclic aromatic hydrocarbons (£U51PAHs) were found ranged from 229 to 638 (ng/g dry wt) in the sediment cores in Fang-Liao submarine canyon. In addition, the low molecular weight PAHs (2-3 ring PAHs) were found dominant in the PAH composition pattern of most samples. Total PAH concentrations were significantly correlated with total organic carbon (TOC) in all the sediment cores. Compared with sediment quality guidelines (SQGs), the PAH concentrations of all sediment samples were lower than those outlined in the criteria, that suggests no evident adverse biological effects caused by PAHs. Results also showed that total PAH concentration of surface sediments (0-2 cm) decreased with the water depth. Identification of PAHs sources suggests that all up-cores were dominated by petrogenic sources, but all down-cores except for S17 and S18 were dominated by pyrogenic sources or mixed sources. In contrast, biogenic sources were found dominant in S17 and S18 as they were characterized by higher ratio of perylene/£Upenta-PAHs(%). Compared with literature, the sediment cores of Fang-Liao submarine canyon were moderately polluted with PAHs. Analysis of diagnostic ratios and hierarchical cluster analysis (HCA) as well as principal component analysis (PCA) all indicate PAHs sources of Fang-Liao submarine canyon were mainly from petroleum and petroleum combustion sources for site of S3,S5,S7,S8 and S17; while pyrogenic or mixed sources for site of S1,S2,S9,S18 and S33.

Στην παρούσα εργασία θα αξιολογηθούν κάποιες αυτόματες μέθοδοι διαχωρισμού ακουστικών βιοσημάτων τα οποία λαμβάνονται από συστοιχία πιεζοηλεκτρικών αισθητήρων σε χαμηλές συχνότητες. Πιο συγκεκριμένα αρχικά θα οριστεί το πρόβλημα το οποίο μας ζητείται να επιλύσουμε και θα γίνουν αναφορές στη διαδρομή των δύο σημαντικότερων μεθόδων διαχωρισμού , της PCA και της ICA. Εν συνεχεία θα γίνει αναφορά στα βιοσήματα τόσο ως προς την προέλευση όσο και ως προς τα σημαντικότερα χαρακτηριστικά τους , η γνώση των οποίων διευκολύνει κατά πολύ τόσο τη διαδικασία του διαχωρισμού όσο και την αξιολόγηση της τελευταίας. Σε επόμενο κεφάλαιο θα γίνει εκτενής αναφορά στους πιεζοηλεκτρικούς αισθητήρες και τον τρόπο με τον οποίο κωδικοποιούν τα βιοσήματα με στόχο την περαιτέρω επεξεργασία τους. Στο μεγαλύτερο τμήμα της εργασίας αυτής ωστόσο θα αναλυθούν οι δύο τεχνικές διαχωρισμού , PCA και ICA και θα γίνει νύξη στους σημαντικότερους αλγορίθμους των παραπάνω (FastICA). Τέλος, θα γίνει εφαρμογή των μεθόδων αυτών τόσο σε τεχνητά όσο και σε πραγματικά σήματα και ανάλυση των αποτελεσμάτων που θα εξαχθούν. / In this diploma thesis some automatic acoustic bio-signal separation techniques are going to be evaluated. The signals used are taken from an array of piezoelectric sensors at low frequencies. To be more specific we are going to set the problem and make a brief report of the main historical facts about PCA and ICA. Furthermore, we are going to analyze both the origin and the most significant characteristics of bio-signals. This knowledge is going to provide us with a much easier separation procedure and a robust evaluation. Additionally not only piezoelectric sensors are going to be analyzed but also PCA and ICA will be resolved too. Main algorithms of both techniques will be mentioned. In conclusion those methods will be applied both on artificial and real data in order to draw some useful conclusions.

This thesis describes the application of multivariate methods in analyses of genomic DNA sequences, gene expression and protein synthesis, which represent each of the steps in the central dogma of biology. The recent finalisation of large sequencing projects has given us a definable core of genetic data and large-scale methods for the dynamic quantification of gene expression and protein synthesis. However, in order to gain meaningful knowledge from such data, appropriate data analysis methods must be applied. The multivariate projection methods, principal component analysis (PCA) and partial least squares projection to latent structures (PLS), were used for clustering and multivariate calibration of data. By combining results from these and other statistical methods with interactive visualisation, valuable information was extracted and further interpreted. We analysed genomic sequences by combining multivariate statistics with cytological observations and full genome annotations. All oligomers of di- (16), tri- (64), tetra- (256), penta- (1024) and hexa-mers (4096) of DNA were separately counted and normalised and their distributions in the chromosomes of three Drosophila genomes were studied by using PCA. Using this strategy sequence signatures responsible for the differentiation of chromosomal elements were identified and related to previously defined biological features. We also developed a tool, which has been made publicly available, to interactively analyse single nucleotide polymorphism data and to visualise annotations and linkage disequilibrium. PLS was used to investigate the relationships between weather factors and gene expression in field-grown aspen leaves. By interpreting PLS models it was possible to predict if genes were mainly environmentally or developmentally regulated. Based on a PCA model calculated from seasonal gene expression profiles, different phases of the growing season were identified as different clusters. In addition, a publicly available dataset with gene expression values for 7070 genes was analysed by PLS to classify tumour types. All samples in a training set and an external test set were correctly classified. For the interpretation of these results a method was applied to obtain a cut-off value for deciding which genes could be of interest for further studies. Potential biomarkers for the efficacy of radiation treatment of brain tumours were identified by combining quantification of protein profiles by SELDI-MS-TOF with multivariate analysis using PCA and PLS. We were also able to differentiate brain tumours from normal brain tissue based on protein profiles, and observed that radiation treatment slows down the development of tumours at a molecular level. By applying a multivariate approach for the analysis of biological data information was extracted that would be impossible or very difficult to acquire with traditional methods. The next step in a systems biology approach will be to perform a combined analysis in order to elucidate how the different levels of information are linked together to form a regulatory network.