In the present work, methodology of statistical analysis of change evolved by Kundzewicz and Robson (204) is revised to obtain a robust methodology named as “Comprehensive Aproach” which addresses research gaps of earlier method, as also those found by literature review. Main aspects of the revised method are: 1) importance of graphical representations as first step, in which, if line spectrum has constant spectral density function then time series is random and no need of further trend detection, 2) importance of computation of statistical parameters of data for deciding type of step change test to be used and for cross checking results of exploratory data analysis (EDA), 3) application of EDA, statistical parameters and checking assumption(s) about the data by statistical test(s) is suggested and also results of these steps can be used to cross check results of each other, 4) suggested basis for selection of step change test(s) i.e. evaluation of two aspects of step change viz. detection and location of step change, 5) suggested basis for selection of trend detection tests i.e. evaluation of all four aspects of trend viz. magnitude, statistical significance, beginning and end of trend and nature of trend, 6) evaluation of regional significance is suggested as essential wherever applicable. The revised method i.e. “Comprehensive Approach” is applied for the trend detection of rainfall of seven homogenous rainfall regions and al India at annual, monthly and seasonal temporal scales for three time periods 1901-203, 1948-203 and 1970-203. Between 100 N to 300 N, there was marked increase in precipitation from 190 to 1950s, but decrease after about 1970 (Trenberth et al., 207). Thus starting years of three time periods are selected as 1901, 1948 and 1970. To have similarity of end year, in analysis periods given in chapters 1, 2 and chapters 3, 4; their end years are kept close to each other i.e. end year of analysis periods is 203 in chapters 1, 2 and end year of analysis periods is 204 in chapters 3, 4. Thus 203 are considered as common end year of three time periods. Burn and Elnur (202) sugested that least number of years required for ensuring statistical validity of results of trend detection are 25 years. So in the third time period (1970-203), the duration is 34 years which is more than 25 years. Three time periods are having data of 103 years (1901-203), 56 years (1948-203) and 34 years (1970- 203) so effect of different time durations on trend detection analysis results is studied. Also temporal scales used in trend detection analysis are annual, monthly and seasonal (4 seasons) thus presence of trend is assessed in these main temporal scales. Results of the analysis showed that, statistically significant trends are found in: 1) winter rainfall time series of peninsular India (PENIN) region for the time period 1901-203, 2) pre-monsoon rainfall time series of north west India (NWIND) and central north east India (CNEIN) regions for the time period 1948-203, 3) monsoon rainfall time series of west central India (WCIND) region for the time period 1948-203, 4) August month rainfall time series of north east India (NEIND) region for the time period 1901-203, 5) June month rainfall time series of NEIND region for the time period 1948-203, 6) Also regionally significant trends are detected in pre- monsoon rainfall time series of five homogeneous regions for the time period 1948-203. Regionally significant trends are detected in pre-monsoon rainfall time series of five homogeneous regions for the time period 1948-203. But effect of cross correlation between rainfall time series of stations of subdivisions and between the sub-divisions in a region is not accounted in the field/regional significance evaluation and Hegel et al. (207) suggested that reactions to external forcing in trends of regional precipitation trends exhibit weak signal to noise ratios and likely to exhibit strong variations in space because of dependency of precipitation on geographic parameters like pornography and atmospheric circulation. Thus attribution of precipitation is more difficult. Also Saikranthi et al. (2013) suggested that homogeneity of rainfall zones may change in future. So, attribution of trends detected in pre-monsoon rainfall time series of five homogeneous regions was not possible. The results of statistically significant trends are confirmed by smoothing curves, innovative trend analysis plots and Sen.’s slope estimates. Contributions by present trend detection study on rainfall of homogenous regions by using “Comprehensive Approach” method are: 1) modification of guidelines of statistical analysis of change to evolve a robust method termed as “Comprehensive Approach”, 2) systematic trend detection analysis is performed pertaining to the rainfall of core monsoon India (CORIN) region and homogeneous India (HOMIN) region, which was not done earlier, 3) systematic trend detection analysis is performed on the rainfall of al India and seven homogenous regions concurrently for aforesaid temporal scales and time periods (except regional significance evaluation only for five homogeneous regions), which was not done earlier, 4) Man Kendal test with block bootstrapping approach (MKBBS) test (not effected by serial correlation) is used for trend detection of serially correlated data and Man Kendal (MK) test is used for trend detection of serially uncorrelated data. Sen.’s slope is used for evaluation of trend magnitude, 5) evaluation of field/regional significance of trends in rainfall over five homogenous regions is performed, which was not done earlier, 6) Location of beginning, end and progress of trend in rainfall of all India and seven homogenous regions concurrently is performed, which was not done earlier. As mentioned aforesaid, attribution of regionally significant trends detected in pre-monsoon rainfall time series of five homogeneous regions for the time period 1948-203 was not possible because of non-accounting of effects of cross correlation, attribution of rainfall is difficult and homogeneity of rainfall zones may change in future as discussed above in detail. So a thorough investigation about trends in rainfall, three temperatures (minimum, mean and maximum) and stream flow at regional (basin) scale was proposed to be ascertained. As Tapi basin is exposed to occurrence of heavy floods (Joshi and Shah, 2014) and it is climatically sensitive (Bhamare and Agone, 201; Gosain et al. 206; Deshpande et al., 2016), it is considered as study area. The trend detection analysis of gridded data (chapter 4) and regional time series (chapter 3) of rainfall and three temperatures data (1971-204) along with that for station data of stream flow (1979-204) of five gauging stations (chapter 4) is carried out using “Comprehensive Approach” for all temporal scales. Common available end year of data of rainfall, temperature and stream flow was 204 as data after 204 was not available for stream flow for all five gauging stations. Also data of rainfall (0.50 x 0.50) was available from year 1971, which was common starting year among data of rainfall and three temperatures. Also common starting year of stream flow data was 1979. Because of unavailability of rainfall data (0.50 x 0.50) before 1971, the three time periods used in chapters 1 and 2 are not used in chapters 3 and 4, thus only one time period is used for rainfall and three temperatures (1971-204) and stream flow (1979-204). The analysis has shown the presence of regionally significant rends in the gridded data of annual mean temperature (Tmean) and winter Tmean over Tapi basin apart from significant trends found in regional time series of annual Tmean and winter Tmean of Tapi basin. Monthly, winter and pre- monsoon stream flow volume time series have also shown regionally significant trends over five gauging stations of Tapi basin. Main contributions of the trend detection analysis of hydro- climatic variables of Tapi basin are: 1) grid wise, regional scale and station wise trend detection of three temperatures, rainfall and stream flow respectively is performed, which was not done earlier, 2) regional significance evaluation of gridded data (rainfall and three temperatures) and station data of stream flow (five stream flow gauging stations) is performed, which was not done earlier, 3) all four aspects of trend of hydro-climatic variables are evaluated, which was not done earlier, 4) systematic trend detection study of gridded, regional and station data of hydro-climatic variables is performed in present study which was not done earlier. After detection of regionally significant trends, next step is finding the causal factors through attribution study. Once causal factors of climate change observed in given variable are found, then remedial measures can be carried out for minimizing the effect of these factors on climate change observed in given variable. There are three main methods of attribution found in literature viz. finger printing, optimal finger printing and artificial neural network (ANN) model. In finger printing method only the leading empirical orthogonal function (EOF) is used, so this method is conservative. In optimal finger printing, multivariate regression is used, which has certain assumptions which are difficult to be fulfilled in the case of climate studies as climate is essentially a non-linear dynamic system. ANN being non-linear in nature provides the required solution for the attribution problem related to climate. Attribution of regionally significant trends detected in monthly, winter and pre-monsoon stream flow volume time series of five gauging stations of Tape basin is not performed because five gauging stations were not representative of entire Tapi basin and two out of the five gauging stations have missing data greater than 15%. Number of significant monotonically increasing trends are more in winter gridded Tmean data as compared to annual gridded Tmean data. Thus attribution analysis of winter gridded Tmean data has given first priority followed by attribution of annual gridded Tmean data. ANN model is developed for the attribution of climate change observed in gridded data of winter Tmean and annual Tmean in three steps: 1) input variable selection (IVS) based on partial mutual information (PMI), 2) data splitting using k-means clustering method and Neyman allocation, 3) ANN model formulation by using best training algorithm among Levenberg-Marquardt (LM) algorithm, scaled conjugate gradient (SCG) algorithm and Broyden, Fletcher, Goldfarb, and Shano (BFGS) algorithm and optimum number of hidden neurons (varying from 1 to 3) corresponding to performance in terms of mean squared error (MSE) and to use these in final ANN model formulation with computation of performance evaluation measures (PEMs). Aforesaid third step is repeated for 50 iterations for each input forcing and given target output to minimize any random variation due to reinitialization of training algorithms. Also random variations due to initialization of ANN model are minimized by keeping initial weights and biases equal to zero. Final PEMs evaluated were the averages of 50 iterations as mentioned aforesaid. Target outputs used in two ANN attribution models are time series of regional winter Tmean and regional annual Tmean. Also in some cases of ANN model formulations, network parameters are kept less than number of data points in the training set for minimizing overriding. Inputs for ANN model were circulation indices and regional, global and national scale input variables. The inputs selected by PMI based input selection (PMIS) algorithm in the step of IVS of both ANN attribution models are seen to be subjected to natural and anthropogenic forcing, which undisputedly shows significant role of anthropogenic activities in observed climate change in aforesaid two gridded temperature variables. Also ranking of input forcing is performed in both the ANN attribution models according to their final PEM values. In the case of ANN attribution model for regional winter Tmean time series, dominant role of natural (‘nat’) input forcing is found behind the given climate change as compared to anthropogenic (‘anth’) input forcing. Among ‘anth’ inputs, effect of land cover (‘Landcover’) input forcing is found to be dominant as compared to green house gases (‘GHgases’) input forcing. Among ‘Landcover’ inputs, urban landcover input was found to be one of the important inputs. In the case of ANN attribution model for regional annual Tmean time series, dominant role of ‘anth’ input forcing is found behind the given climate change as compared to ‘nat’ input forcing. Among ‘anth’ inputs, there is dominant role of ‘Landcover’ input forcing as compared to ‘GHgases’ input forcing. Among ‘Landcover’ inputs, urban landcover input was found to be one of the important inputs. Contributions of attribution study are: 1) checking of input independence and significance by using PMI IVS method, which was not performed earlier, 2) division of data in such a way that al patterns of whole data are present in training, testing and validation subsets and the statistical properties of these subsets are similar to each other and to whole data, which was not performed earlier, 3) using LM, SCG and BFGS algorithms which are converging fatly as compared to Windrow-Hof algorithm and gradient descent algorithm. Also these three algorithms are les liable to be get stuck in local minima, 4) using land cover data as input forcing to ANN model used for attribution of climate change, which was not done earlier.

Terrestrial evapotranspiration (ET) is a significant component of the energy and water balances at the land surface. However, direct, continuous measurements of ET are spatially limited and only available since the 1990s. Due to this lack of observations, detecting and attributing long-term regional trends in ET remains difficult. This dissertation aims to alleviate the data limitation and detect long-term trends by developing a method to infer ET from data collected at common weather stations, which are spatially and temporally abundant. The methodology used to infer ET from historical meteorological data is based on an emergent relation between the land surface and atmospheric boundary layer. We refer to this methodology as the Evapotranspiration from Relative Humidity at Equilibrium method, or the “ETRHEQ method”. In the first section of this dissertation, we develop the ETRHEQ method for use at common weather stations and demonstrate the utility of the method at twenty eddy covariance sites spanning a wide range of climate and plant functional types. Next, we apply the ETRHEQ method at historical weather stations across the continental U.S. and show that ET estimates obtained via the ETRHEQ method compare well with watershed scale ET, as well as ET estimates from land surface models. From 1961 to 1997, we find negligible or increasing trends in summertime ET over the central U.S. and the west coast and negative trends in the eastern and western U.S. From 1998 to 2014, we find a sharp decline in summertime ET across the entire U.S. We show that this decline is consistent with decreasing transpiration associated with declines in humidity. Lastly, we assess the sensitivity of ET to perturbations in soil moisture and humidity anticipated with climate change. We demonstrate that the response of ET to changing humidity and soil moisture is strongly dependent on the biological and hydrological state of the surface, particularly the degree of water stress and vegetation fraction. In total, this dissertation demonstrates the utility of the ETRHEQ method as a means to estimate ET from weather station data and highlights the critical role of vegetation in modulating ET variability.

This dissertation is an empirical modeling investigation of the impact of forest harvesting, climate change and topography on stream nitrate fluxes in the Turkey Lakes Watershed (TLW), Ontario, Canada. Data used for this study include topography (DEM), climate (mean monthly temperature and total monthly precipitation), wet nitrogen deposition (total monthly nitrate-N and ammonium-N), nitrate water samples and streamflow in 13 headwater catchments within the TLW. First, a paired-watershed approach was used to examine the impact of forest harvesting intensity on stream water nitrate fluxes by developing transfer function noise (TFN) models that related monthly stream water nitrate fluxes of three treatment catchments to those of one control catchment. Second, TFN models were also developed to relate monthly stream nitrate fluxes in 13 catchments to the temperature, precipitation and wet nitrogen deposition to examine the spatially varying responses of stream nitrate fluxes to changes in climate and bulk deposition. Third, geographically weighted regression (GWR) was introduced to model the spatial and temporal relationships between topography and stream nitrate fluxes in 13 headwater catchments. The results showed that there existed a new phenomenon of clustered wave-up and wave-down of the stream nitrate increases caused by clearcut and selectioncut at the monthly scale, respectively. This phenomenon was never reported by previous studies because it was not possible to be identified with ordinary least squares (OLS) regression at an annual scale. There also existed significant responses of stream nitrate fluxes to wet nitrogen deposition in all catchments at the monthly scale over a long-term record between 1982 and 2003. These responses were previously thought to be lower and masked by the impact of climate variations. There further existed significant spatial and seasonal variability of the relationships between topography and stream nitrate fluxes across space and over time. This variability was largely ignored in previous studies with possibly misleading interpretation on the empirical relations. / Thesis (Ph.D, Geography) -- Queen's University, 2012-12-31 22:37:39.137

Malgré des conditions sèches qui prédominent depuis les années 1970, l’Afrique de l’Ouest a subi au cours des deux dernières décennies des épisodes d’inondations sévères qui ont provoqué de nombreux décès et dommages socio-économiques. L’émergence de ce nouveau problème montre une nouvelle facette de la sensibilité de cette région aux changements hydro-climatiques, appelant à une meilleure caractérisation de l’aléa inondation, des processus qui le génèrent, ainsi que la mise en place de méthodes permettant de projeter les évolutions futures de cet aléa pour mieux s’en prémunir.Dans ce contexte, la thèse cherche à répondre à trois questions principales :1) L’augmentation des dommages liés aux inondations s’est-elle accompagnée d’une intensification des crues extrêmes en Afrique de l’Ouest?2) Comment modéliser les orages de mousson, premier facteur de génération du ruissellement, afin d’explorer l’impact de leurs caractéristiques sur les crues?3) Compte tenu des changements climatiques à l’œuvre dans la région, à quelles tendances hydro-climatiques peut-on s’attendre dans le futur ?Dans un premier temps, on évalue l’évolution des crues en Afrique de l’Ouest au cours des soixante dernières années en utilisant de méthodes basées sur la théorie de valeurs extrêmes. Les résultats montrent une augmentation forte des événements hydrologiques extrêmes depuis les années 1970s dans les sous-bassins Sahéliens du fleuve Niger et depuis les années 1980s dans les sous-bassins soudano-guinéens du fleuve Sénégal. Les niveaux de retour calculés à partir des modèles non-stationnaires dépassent ceux qui ont été calculés avec un modèle stationnaire avec plus de 95% de certitude pour les périodes de retour les plus courtes (<10 ans).On présente ensuite des développements récents apportés à un simulateur stochastique d’orages de mousson à meso-échelle (StochaStorm). Ils incluent: une modélisation de l’occurrence de ces orages, la représentation explicite des valeurs de pluie extrêmes et une amélioration du schéma temporel d’intensité infra-événementielle. Implémenté et évalué à partir des donnés haute-résolution de l’observatoire AMMA-CATCH, le générateur montrent de très bonnes capacités à reproduire les propriétés des orages, confirmant son potentiel pour des études d’impact hydrologique.Enfin, une chaîne de modélisation est élaborée afin de proposer des projections hydrologiques pour le futur sur un bassin sahélien de meso-échelle (Dargol, 7000 km²). L’originalité de cette chaîne provient de la prise en compte du continuum d’échelles entre climat global et impact local à travers la représentation du régime des pluies à l’échelle des orages de mousson, dont les propriétés d’occurrence et d’intensité ont des impacts majeurs sur la réponse hydrologique. La chaîne de modélisation inclut le modèle climatique CP4-Africa, unique modèle à convection explicite fournissant des simulations de long terme en Afrique ; une méthode de débiaisage statistique; le simulateur Stochastorm ; et un modèle pluie-débit spécifiquement adapté aux processus hydrologique sahéliens. La chaine est évaluée sur une période de contrôle 1997-2006 puis utilisée pour des projections futures montrant une hausse par un facteur 1,5 des débits maximum annuels et un doublement des volumes moyens annuels à l’horizon 2100.Les résultats ont des implications majeures notamment pour l’ingénierie hydrologique. Les méthodes actuellement utilisées pour appréhender les risques hydrologiques dans la région ne prennent pas en compte la non-stationnarité hydro-climatique risquant de sous-évaluer l’aléa hydrologique et sous-dimensionner les ouvrages hydrauliques utilisés pour s’en protéger. La thèse suggère aussi quelques pistes afin mieux définir les trajectoires hydrologiques passées et futures en incluant, au-delà des précipitations, les changements sociétaux et environnementaux, leurs interactions et rétroactions dans les approches de modélisation. / The semi-arid regions of West Africa are known for their dry conditions which have predominated since the 1970s. In recent years, however, West Africa has witnessed a series of severe flooding events which caused widespread fatalities and socioeconomic damages. The emergence of this new problem demonstrates the sensitivity of the region to changes in the hydroclimatic system and calls for an improved characterization of flood hazard and the mechanisms that generate it. It also signals the need to develop projections for how flood hazard may evolve in the future in order to inform appropriate adaptation measures.In this context, the following PhD thesis seeks to answer three main questions:1) Is there a significant trend in extreme streamflow in West Africa, or are the documented flooding events isolated incidences?2) How can one model mesoscale convective systems, the primary driver of runoff in the region, in order to explore the properties of precipitation that drive streamflow?3) Based on potential climate change in the region, what trends might be observed in streamflow in the future?First, changes in extreme hydrological events West Africa over the past 60 years are evaluated by applying non-stationary methods based on extreme value theory. Results show a strong increasing trend in extreme hydrological events since the 1970s in the Sahelian Niger River basin and since the 1980s in the Sudano-Guinean catchments in the Senegal River basin. Return levels calculated from non-stationary models are determined to exceed those calculated from a stationary model with over 95% certainty for shorter return periods (<10 years).Next, recent developments are presented for a stochastic precipitation simulator (Stochastorm) designed for modeling mesoscale convective storms, the main rainfall source in the Sahel. Developments include a model for storm occurrence, the explicit representation of extreme rainfall values, and an improvement in the modeling of sub-event intensities. Using high-resolution data from the AMMA-CATCH observatory, simulation outputs were confirmed to realistically represent key characteristics of MCSs, showing the simulator’s potential for use in impact studies.Finally, a modeling chain for producing future hydrological projections is developed and implemented in a Sahelian river basin (Dargol, 7000km2). The chain is original as it is the first attempt in West Africa to encompass the continuum of scales from global climate to convective storms, whose properties have major impacts on hydrological response and as a result local flood risk. The modeling chain components include the convection-permitting regional climate model (RCM) CP4-Africa, the only RCM (to date) explicitly resolving convection and providing long-term simulations in Africa; a bias correction approach; the stochastic precipitation generator Stochastorm; and a rainfall-runoff model specifically developed for Sahelian hydrological processes. The modeling chain is evaluated for a control period (1997-2006) then for future projections (ten years at the end of the 21st century). Hydrological projections show that peak annual flow may become 1.5-2 times greater and streamflow volumes may double or triple on average near the end of the 21st century compared to 1997-2006 in response to projected changes in precipitation.The results raise critical issues notably for hydrological engineering. Current methods used to evaluate flood risk in the region do not take non-stationarity into account, leading to a major risk of underestimating potential floods and undersizing the hydraulic infrastructure designed for protecting against them. It is also suggested to not only consider rainfall changes but also societal and environmental changes, interactions, and feedbacks in order to better attribute past hydrological hazards and their future trajectories to related causes.

Thesis (Doctor of Engineering in Electrical Engineering)--Cape Peninsula University of Technology, 2017. / Remote sensing provides a way of frequently observing broad land surfaces. The availability of various earth observation data and their potential exploitation in a wide range of socioeconomic applications stimulated the rapid development of remote sensing technology. Much of the research and most of the publications dealing with remote sensing in the solar spectral domain focus on analysing and interpreting the spectral, spatial and temporal signatures of the observed areas. However, the angular signatures of the reflectance field, known as surface anisotropy, also merit attention. The current research took an exploratory approach to the land surface anisotropy described by the RPV model parameters derived from the MISR-HR processing system (denoted as MISR-HR anisotropy data or MISR-HR RPV data), over a period of 14+ years, for three typical terrestrial surfaces in the Western Cape Province of South Africa: a semi-desert area, a wheat field and a vineyard area. The objectives of this study were to explore (1) to what extent spectral and directional signatures of the MISR-HR RPV data may vary in time and space over the different targets (landscapes), and (2) whether the observed variations in anisotropy might be useful in classifying different land surfaces or as a supplementary method to the traditional land cover classification method. The objectives were achieved by exploring the statistics of the MISR-HR RPV data in each spectral band over the different land surfaces, as well as seasonality and trend in these data. The MISR-HR RPV products were affected by outliers and missing values, both of which influenced the statistics, seasonality and trend of the examined time series. This research proposes a new outlier detection method, based on the cost function derived from the RPV model inversion process. Removed outliers and missing values leave gaps in a MISR-HR RPV time series; to avoid introducing extra biases in the statistics of the anisotropy data, this research kept the gaps and relied on gap-resilient trend and seasonality detection methods, such as the Mann-Kendal trend detection and Lomb-Scargle periodogram methods. The exploration of the statistics of the anisotropy data showed that RPV parameter rho exhibited distinctive over the different study sites; NIR band parameter k exhibits prominent high values for the vineyard area; red band parameter Theta data are not that distinctive over different study sites; variance is important in describing all three RPV parameters. The explorations on trends also demonstrated interesting findings: the downward trend in green band parameter rho data for the semi-desert and vineyard areas; and the upward trend in blue band parameters k and Theta data for all the three study sites. The investigation on seasonality showed that all the RPV parameters had seasonal variations which differed over spectral bands and land covers; the results confirmed expectations in previous literature that parameter varies regularly along the observation time, and also revealed seasonal variations in the parameter rho and Theta data. The explorations on the statistics and seasonality of the MISR-HR anisotropy data show that these data are potentially useful for classifying different landscapes. Finally, the classification results demonstrated that both red band parameter rho data and NIR band parameter k data could successfully separate the three different land surfaces in this research, which fulfilled the second primary objective of this study. This research also demonstrated a classification method using multiple RPV parameters as the classification signatures to discriminate different terrestrial surfaces; significant separation results were obtained by this method.

Nous présentons dans cette thèse en premier lieu la méthode de Bootstrap par permutation appliquée à la méthode des blocs maxima utilisée en théorie des valeurs extrêmes (TVE) univariée. La méthode est basée sur un échantillonnage particulier des données en utilisant les rangs des blocs maxima dont la distribution est présentée et introduite dans les simulations. Elle amène à une réduction de la variance des paramètres de la loi GEV et des quantiles estimés. En second lieu, on s’intéresse au cas où les observations sont indépendantes mais non identiquement distribuées en TVE. Cette variation dans la distribution est quantifiée en utilisant une fonction dite « skedasis function » notée c qui représente la fréquence des extrêmes. Ce modèle a été introduit par Einmahl et al. dans le papier « Statistics of heteroscedastic extremes ». On étudie plusieurs modèles paramétriques pour c (log-linéaire, linéaire, log-linéaire discret) ainsi que les résultats de consistance et de normalité asymptotique du paramètre θ représentant la tendance. Le test θ =0 contre θ ≠0 est interprété alors comme un test de détection de tendance dans les extrêmes. Nous illustrons nos résultats dans une étude par simulation qui montre en particulier que les tests paramétriques sont en général plus puissants que les tests non paramétriques pour la détection de la tendance, d’où l’utilité de notre travail. Nous discutons en plus le choix du seuil k en appliquant la méthode de Lepski. Enfin, nous appliquons la méthodologie sur les données de températures minimales et maximales dans la région de Fort Collins, Colorado durant le 20ème siècle afin de détecter la présence d’une tendance dans les extrêmes sur cette période. En troisième lieu, on dispose d’un jeu de données de précipitation journalière maximale sur 24 ans dans 40 stations. On réalise une prédiction spatio-temporelle des quantiles correspondants à un niveau de retour de 20 ans pour les précipitations mensuelles dans chaque station. Nous utilisons des modèles de GEV en introduisant des covariables dans les paramètres. Le meilleur modèle est choisi en termes d’AIC et par la méthode de validation croisée. Pour chacun des deux modèles choisis, nous estimons les quantiles extrêmes. Finalement, on applique la TVE unvariée et bivariée sur les vitesses du vent et la hauteur des vagues dans une région au Liban en vue de protéger la plateforme pétrolière qui y sera installée de ces risques environnementaux. On applique d’abord la théorie univariée sur la vitesse du vent et la hauteur des vagues séparément en utilisant la méthode des blocs maximas pour estimer les paramètres de la GEV et les niveaux de retour associés à des périodes de retour de 50, 100 et 500 années. Nous passons ensuite à l’application de la théorie bivariée afin d’estimer la dépendance entre les vents et les vagues extrêmes et d’estimer des probabilités jointes de dépassement des niveaux de retour univariés. Nous associons ces probabilités jointes de dépassement à des périodes de retour jointes et nous les comparons aux périodes de retour marginales. / We firstly present in this thesis the permutation Bootstrap method applied for the block maxima (BM) method in extreme value theory. The method is based on BM ranks whose distribution is presented and simulated. It performs well and leads to a variance reduction in the estimation of the GEV parameters and the extreme quantiles. Secondly, we build upon the heteroscedastic extremes framework by Einmahl et al. (2016) where the observations are assumed independent but not identically distributed and the variation in their tail distributions is modeled by the so-called skedasis function. While the original paper focuses on non-parametric estimation of the skedasis function, we consider here parametric models and prove the consistency and asymptotic normality of the parameter estimators. A parametric test for trend detection in the case where the skedasis function is monotone is introduced. A short simulation study shows that the parametric test can be more powerful than the non-parametric Kolmogorov-Smirnov type test, even for misspecified models. We also discuss the choice of threshold based on Lepski's method. The methodology is finally illustrated on a dataset of minimal/maximal daily temperatures in Fort Collins, Colorado, during the 20th century. Thirdly, we have a training sample data of daily maxima precipitation over 24 years in 40 stations. We make spatio-temporal prediction of quantile of level corresponding to extreme monthly precipitation over the next 20 years in every station. We use generalized extreme value models by incorporating covariates. After selecting the best model based on the Akaike information criterion and the k-fold cross validation method, we present the results of the estimated quantiles for the selected models. Finally, we study the wind speed and wave height risks in Beddawi region in the northern Lebanon during the winter season in order to protect the oil rig that will be installed. We estimate the return levels associated to return periods of 50, 100 and 500 years for each risk separately using the univariate extreme value theory. Then, by using the multivariate extreme value theory we estimate the dependence between extreme wind speed and wave height as well as joint exceedance probabilities and joint return levels to take into consideration the risk of these two environmental factors simultaneously.

The anthropogenic impact on earth’s systems has rapidly increased since the middle of the last century and today it is hard to find a stream that is not influenced by human activities. The understanding of causes to changes is an important knowledge for future water management and planning and of that reason climatic and anthropogenic impact on streamflow changes in Sweden were explored and quantified. In the first step trends and abrupt changes in annual streamflow were detected and verified with the non- parametric Mann-Kendall’s and Pettitt’s test, all performed as moving window tests. In the second step HBV, a climatic driven rainfall-runoff model, was used to attribute the causes of the detected changes. Detection and attribution of changes were performed on several catchments in order to investigate regional patterns. On one hand using smaller window sizes, period higher number of detected positive and negative trends were found. On the other hand bigger window sizes resulted in positive trends in more than half of the catchments and almost no negative trends. The detected changes were highly dependent on the investigated time frame, due to periodicity, e.g. natural variability in streamflow. In general the anthropogenic impact on streamflow changes was smaller than changes due to temperature and streamflow. In median anthropogenic impact could explain 7% of the total change. No regional differences were found which indicated that anthropogenic impact varies more between individual catchments than following a regional pattern. / Sedan mitten av förra århundradet har den antropogena påverkan på jordens system ökat kraftigt. Idag är det svårt att hitta ett vattendrag som inte är påverkat av mänsklig aktivitet. Att förstå orsakerna bakom förändringarna är en viktig kunskap för framtida vattenplanering och av denna anledning undersöktes och kvantiferades den antropogen och klimatpåverkan på flödesförändringar i svenska vattendrag. I arbetets första steg användes de Mann-Kendalls och Pettitts test för att lokalisera och verifiera förändringar i årligt vattenflöde. Alla test var icke parametriska och utfördes som ett glidande fönster. I nästa steg undersöktes orsakerna till förändringar med hjälp av HBV, en klimatdriven avrinningsmodell. Ett större antal avrinningsområden undersöktes för att upptäcka regionala mönster och skillnader. Perioder med omväxlande positiva och negativa trender upptäcktes med mindre fönsterstorlekar, medan större fönster hittade positiva trender i mer än hälften av områdena och knappt några negativa trender hittades. De detekterade förändringarna var på grund av periodicitet i årligt vattenflöde till stor grad beroende på det undersöka tidsintervallet. Generellt var den antropogena påverkan större påverkan från nederbörd och temperatur, med ett medianvärde där 7 % av den totala förändringen kunde förklaras med antropogen påverkan. Inga regionala skillnader i antropogen påverkan kunde identifieras vilket indikerar att den varierar mer mellan individuella områden än följer ett regionalt mönster.

La dissémination d'information explore les chemins pris par l'information qui est transmise dans un réseau social, afin de comprendre et modéliser les relations entre les utilisateurs de ce réseau, ce qui permet une meilleur compréhension des relations humaines et leurs dynamique. Même si la priorité de ce travail soit théorique, en envisageant des aspects psychologiques et sociologiques des réseaux sociaux, les modèles de dissémination d'information sont aussi à la base de plusieurs applications concrètes, comme la maximisation d'influence, la prédication de liens, la découverte des noeuds influents, la détection des communautés, la détection des tendances, etc. Cette thèse est donc basée sur ces deux facettes de la dissémination d'information: nous développons d'abord des cadres théoriques mathématiquement solides pour étudier les relations entre les personnes et l'information, et dans un deuxième moment nous créons des outils responsables pour une exploration plus cohérente des liens cachés dans ces relations. Les outils théoriques développés ici sont les modèles de dynamique d'opinions et de dissémination d'information, où nous étudions le flot d'informations des utilisateurs dans les réseaux sociaux, et les outils pratiques développés ici sont un nouveau algorithme de détection de communautés et un nouveau algorithme de détection de tendances dans les réseaux sociaux / Our aim in this Ph. D. thesis is to study the diffusion of information as well as the opinion dynamics of users in social networks. Information diffusion models explore the paths taken by information being transmitted through a social network in order to understand and analyze the relationships between users in such network, leading to a better comprehension of human relations and dynamics. This thesis is based on both sides of information diffusion: first by developing mathematical theories and models to study the relationships between people and information, and in a second time by creating tools to better exploit the hidden patterns in these relationships. The theoretical tools developed in this thesis are opinion dynamics models and information diffusion models, where we study the information flow from users in social networks, and the practical tools developed in this thesis are a novel community detection algorithm and a novel trend detection algorithm. We start by introducing an opinion dynamics model in which agents interact with each other about several distinct opinions/contents. In our framework, agents do not exchange all their opinions with each other, they communicate about randomly chosen opinions at each time. We show, using stochastic approximation algorithms, that under mild assumptions this opinion dynamics algorithm converges as time increases, whose behavior is ruled by how users choose the opinions to broadcast at each time. We develop next a community detection algorithm which is a direct application of this opinion dynamics model: when agents broadcast the content they appreciate the most. Communities are thus formed, where they are defined as groups of users that appreciate mostly the same content. This algorithm, which is distributed by nature, has the remarkable property that the discovered communities can be studied from a solid mathematical standpoint. In addition to the theoretical advantage over heuristic community detection methods, the presented algorithm is able to accommodate weighted networks, parametric and nonparametric versions, with the discovery of overlapping communities a byproduct with no mathematical overhead. In a second part, we define a general framework to model information diffusion in social networks. The proposed framework takes into consideration not only the hidden interactions between users, but as well the interactions between contents and multiple social networks. It also accommodates dynamic networks and various temporal effects of the diffusion. This framework can be combined with topic modeling, for which several estimation techniques are derived, which are based on nonnegative tensor factorization techniques. Together with a dimensionality reduction argument, this techniques discover, in addition, the latent community structure of the users in the social networks. At last, we use one instance of the previous framework to develop a trend detection algorithm designed to find trendy topics in a social network. We take into consideration the interaction between users and topics, we formally define trendiness and derive trend indices for each topic being disseminated in the social network. These indices take into consideration the distance between the real broadcast intensity and the maximum expected broadcast intensity and the social network topology. The proposed trend detection algorithm uses stochastic control techniques in order calculate the trend indices, is fast and aggregates all the information of the broadcasts into a simple one-dimensional process, thus reducing its complexity and the quantity of necessary data to the detection. To the best of our knowledge, this is the first trend detection algorithm that is based solely on the individual performances of topics