Predictive analysis and mapping of indoor radon concentrations in Switzerland

It is estimated that around 230 people die each year due to radon (222Rn) exposure in Switzerland. 222Rn occurs mainly in closed environments like buildings and originates primarily from the subjacent ground. Therefore it depends strongly on geology and shows substantial regional variations. Correct identification of these regional variations would lead to substantial reduction of 222Rn exposure of the population based on appropriate construction of new and mitigation of already existing buildings. Prediction of indoor 222Rn concentrations (IRC) and identification of 222Rn prone areas is however difficult since IRC depend on a variety of different variables like building characteristics, meteorology, geology and anthropogenic factors. The present work aims at the development of predictive models and the understanding of IRC in Switzerland, taking into account a maximum of information in order to minimize the prediction uncertainty. The predictive maps will be used as a decision-support tool for 222Rn risk management. The construction of these models is based on different data-driven statistical methods, in combination with geographical information systems (GIS). In a first phase we performed univariate analysis of IRC for different variables, namely the detector type, building category, foundation, year of construction, the average outdoor temperature during measurement, altitude and lithology. All variables showed significant associations to IRC. Buildings constructed after 1900 showed significantly lower IRC compared to earlier constructions. We observed a further drop of IRC after 1970. In addition to that, we found an association of IRC with altitude. With regard to lithology, we observed the lowest IRC in sedimentary rocks (excluding carbonates) and sediments and the highest IRC in the Jura carbonates and igneous rock. The IRC data was systematically analyzed for potential bias due to spatially unbalanced sampling of measurements. In order to facilitate the modeling and the interpretation of the influence of geology on IRC, we developed an algorithm based on k-medoids clustering which permits to define coherent geological classes in terms of IRC. We performed a soil gas 222Rn concentration (SRC) measurement campaign in order to determine the predictive power of SRC with respect to IRC. We found that the use of SRC is limited for IRC prediction. The second part of the project was dedicated to predictive mapping of IRC using models which take into account the multidimensionality of the process of 222Rn entry into buildings. We used kernel regression and ensemble regression tree for this purpose. We could explain up to 33% of the variance of the log transformed IRC all over Switzerland. This is a good performance compared to former attempts of IRC modeling in Switzerland. As predictor variables we considered geographical coordinates, altitude, outdoor temperature, building type, foundation, year of construction and detector type. Ensemble regression trees like random forests allow to determine the role of each IRC predictor in a multidimensional setting. We found spatial information like geology, altitude and coordinates to have stronger influences on IRC than building related variables like foundation type, building type and year of construction. Based on kernel estimation we developed an approach to determine the local probability of IRC to exceed 300 Bq/m3. In addition to that we developed a confidence index in order to provide an estimate of uncertainty of the map. All methods allow an easy creation of tailor-made maps for different building characteristics. Our work is an essential step towards a 222Rn risk assessment which accounts at the same time for different architectural situations as well as geological and geographical conditions. For the communication of 222Rn hazard to the population we recommend to make use of the probability map based on kernel estimation. The communication of 222Rn hazard could for example be implemented via a web interface where the users specify the characteristics and coordinates of their home in order to obtain the probability to be above a given IRC with a corresponding index of confidence. Taking into account the health effects of 222Rn, our results have the potential to substantially improve the estimation of the effective dose from 222Rn delivered to the Swiss population.

Contribution of the Global Regulator Gene gacA to Persistence and Dissemination of Pseudomonas fluorescens Biocontrol Strain CHA0 Introduced into Soil Microcosms.

Structural and regulatory genes involved in the synthesis of antimicrobial metabolites are essential for the biocontrol activity of fluorescent pseudomonads and, in principle, amenable to genetic engineering for strain improvement. An eventual large-scale release of such bacteria raises the question of whether such genes also contribute to the persistence and dissemination of the bacteria in soil ecosystems. Pseudomonas fluorescens wild-type strain CHA0 protects plants against a variety of fungal diseases and produces several antimicrobial metabolites. The regulatory gene gacA globally controls antibiotic production and is crucial for disease suppression in CHA0. This gene also regulates the production of extracellular protease and phospholipase. The contribution of gacA to survival and vertical translocation of CHA0 in soil microcosms of increasing complexity was studied in coinoculation experiments with the wild type and a gacA mutant which lacks antibiotics and some exoenzymes. Both strains were marked with spontaneous resistance to rifampin. In a closed system with sterile soil, strain CHA0 and the gacA mutant multiplied for several weeks, whereas these strains declined exponentially in nonsterile soil of different Swiss origins. The gacA mutant was less persistent in nonrhizosphere raw soil than was the wild type, but no competitive disadvantage when colonizing the rhizosphere and roots of wheat was found in the particular soil type and during the period studied. Vertical translocation was assessed after strains had been applied to undisturbed, long (60-cm) or short (20-cm) soil columns, both planted with wheat. A smaller number of cells of the gacA mutant than of the wild type were detected in the percolated water and in different depths of the soil column. Single-strain inoculation gave similar results in all microcosms tested. We conclude that mutation in a single regulatory gene involved in antibiotic and exoenzyme synthesis can affect the survival of P. fluorescens more profoundly in unplanted soil than in the rhizosphere.

Learning about Bayesian networks for forensic interpretation : An example based on the "the problem of multiple propositions"

Both, Bayesian networks and probabilistic evaluation are gaining more and more widespread use within many professional branches, including forensic science. Notwithstanding, they constitute subtle topics with definitional details that require careful study. While many sophisticated developments of probabilistic approaches to evaluation of forensic findings may readily be found in published literature, there remains a gap with respect to writings that focus on foundational aspects and on how these may be acquired by interested scientists new to these topics. This paper takes this as a starting point to report on the learning about Bayesian networks for likelihood ratio based, probabilistic inference procedures in a class of master students in forensic science. The presentation uses an example that relies on a casework scenario drawn from published literature, involving a questioned signature. A complicating aspect of that case study - proposed to students in a teaching scenario - is due to the need of considering multiple competing propositions, which is an outset that may not readily be approached within a likelihood ratio based framework without drawing attention to some additional technical details. Using generic Bayesian networks fragments from existing literature on the topic, course participants were able to track the probabilistic underpinnings of the proposed scenario correctly both in terms of likelihood ratios and of posterior probabilities. In addition, further study of the example by students allowed them to derive an alternative Bayesian network structure with a computational output that is equivalent to existing probabilistic solutions. This practical experience underlines the potential of Bayesian networks to support and clarify foundational principles of probabilistic procedures for forensic evaluation.

MOLECULAR MECHANISMS OF A ROOT SYSTEM TO SOIL ADAPTATION

Contrairement aux animaux, les plantes sont des organismes sessiles qui ne possèdent pas de mécanismes de fuite quand les conditions environnementales ne sont plus optimales. Les plantes sont physiquement ancrées à l'endroit où elles ont germées et aux conditions environnementales qui parfois peuvent être extrêmes. Les possibilités d'acclimatation de différentes espèces, parfois même de groupes de plantes au sein d'une même espèce, peuvent varier mais repose sur une adaptation génétique de la plante. L'adaptation est un long processus qui repose sur l'apparition spontanée de mutations génétiques, leur mise à l'épreuve face aux conditions environnementales, et dans le cas où la mutation a un impact positif sur la survie dans cet habitat particulier, elle sera maintenue dans une population donnée de plantes. De telles populations, appelées écotypes, sont le matériel de départ pour la découverte de gènes qui induisent un bénéfice pour la plante dans un environnement donné. La plante la plus étudiée en biologie moléculaire est Arabidopsis thaliana, l'arabette des prés. Dans une étude précédente, les racines d'écotypes naturels d'Arabidopsis ont été comparées et un écotype, Uk-1, avait le système racinaire le plus particulier. Cet écotype possède des racines beaucoup plus courtes et plus ramifiées que tous les autres écotypes. Des analyses plus poussées ont montré qu'une seule mutation dans un gène était la cause de ce phénotype, le gène BREVIS RADIX (BRX), mot latin signifiant 'racine courte'. Bien que l'on connaisse le gène BRX, on connaît finalement peu de choses sur son importance adaptative. Dans cette étude, nous avons montré que la mutation dans le gène BRX rend la plante plus résistante aux sols acides. Dans l'optique de mieux comprendre cette valeur adaptative du mutant brx, nous avons analysé dans quels tissus le gène BRX jouait un rôle important. Nous avons pu mettre en évidence que BRX est important pour le développement du protophloème. Le protophloème est un élément du système vasculaire de la plante. En général, les plantes supérieures possèdent deux systèmes de transport à longue distance. L'un d'eux, appelé xylème, transporte l'eau et les nutriments absorbés du sol par les racines vers les feuilles. Les feuilles sont le siège du processus de photosynthèse au cours duquel sont produits des sucres qui devront être distribués partout dans les autres parties de la plante. Le tissu cellulaire chargé de livrer les produits de la photosynthèse, ainsi que les régulateurs de croissance, est le phloème. Ce dernier regroupe le métaphloème et le protophloème. Le protophloème est essentiel pour la livraison des sucres synthétisés ainsi que des signaux de croissance aux pointes des racines, centres organogéniques responsables de la production de nouvelles cellules durant la phase de croissance de la racine. La structure du protophloème peut être décrite comme des tubes continus, vides et résistants, faits de cellules spécialisées qui permettent un transport efficace et rapide. Nous avons montré que dans les mutants brx ces canaux de transports sont discontinus car certaines cellules n'ont pas terminé leur cycle de différenciation. Ces cellules obstruent le conduit ce qui fait que les sucres et les signaux de croissance, comme l'auxine, ne peuvent plus être transportés aux méristèmes. En conséquence, la prolifération de l'activité des méristèmes est compromise, ce qui explique les racines courtes. Au lieu d'être délivré aux méristèmes, l'auxine se concentre en amont des méristèmes où cela provoque l'apparition de nouvelles racines branchées et, très probablement, l'activation des pompes à protons. Sur des sols acides, la concentration en ion H+ est très élevée. Ces ions entrent dans les cellules de la racine par diffusion et perturbent notablement la croissance des racines et de la plante en général. Si les cellules de la racine possédaient des pompes à protons hyperactives, elles seraient capable d'évacuer le surplus d'ions H+ en dehors de la cellule, ce qui leur assurerait de meilleures chances de survie sur sols acides. De fait, le mutant brx est capable d'acidifier le milieu de culture dans lequel il est cultivé plus efficacement que la plante sauvage. Ce mutant est également capable de donner plus de progéniture sur ce type de milieu de croissance que les plantes sauvages. Finalement, nous avons trouvé d'autres mutants brx en milieu naturel poussant sur sols acides, ce qui suggère fortement que la mutation du gène BRX est une des causes de l'adaptation aux sols acides. -- Plants as sessile organisms have developed different mechanisms to cope with the complex environmental conditions in which they live. Adaptation is the process through which traits evolve by natural selection to functionally improve in a given environmental context. An adaptation to the environment is characterized by the genetic changes in the entire populations that have been fixed by natural selection over many generations. BREVIS RADIX (BRX) gene was found through natural Arabidopsis accessions screen and was characterized as a root growth regulator since loss-of-function mutants exhibit arrested post-embryonic primary root growth in addition to a more branched root system. Although brx loss-of-function causes a complete alteration in root architecture, BRX activity is only required in the root vasculature, in particular in protophloem cell file. Protophloem is a part of the phloem transport network and is responsible for delivery of photo-assimilates and growth regulators, coming from the shoot through mature phloem component - metaphloem, to the all plant primary meristems. In order to perform its function, protophloem is the first cell file to differentiate within the root meristem. During this process, protophloem cells undergo a partial programmed cell death, during which they build a thicker cell wall, degrade nucleus and tonoplast while plasma membrane stays functional. Interestingly, protophloem cells enter elongation process only after differentiation into sieve elements is completed. Here we show that brx mutants fail to differentiate protophloem cell file properly, a phenotype that can be distinguished by a presence of a "gap" cells, non-differentiated cells between two flanking differentiated cells. Discontinuity of protophloem differentiation in brx mutants is considered to be a consequence of local hyperactivity of CLAVATA3/EMBRYO SURROUNDING REGION 45 (CLE45) - BARELY ANY MERISTEM 3 (BAM3) signaling module. Interestingly, a CLE45 activity, most probably at the level of receptor binding, can be modulated by apoplastic pH. Altogether, our results imply that the activity of proton pumps, expressed in non-differentiated cells of protophloem, must be maintained under certain threshold, otherwise CLE45-BAM3 signaling pathway will be stimulated and in turn protophloem will not differentiate. Based on vacuolar morphology, a premature cell wall acidification in brx mutants stochastically prevents the protophloem differentiation. Only after protophloem differentiates, proton pumps can be activated in order to acidify apoplast and to support enucleated protophloem multifold elongation driven by surrounding cells growth. Finally, the protophloem differentiation failure would result in an auxin "traffic jam" in the upper parts of the root, created from the phloem-transported auxin that cannot be efficiently delivered to the meristem. Physiologically, auxin "leakage" from the plant vasculature network could have various consequences, since auxin is involved in the regulation of almost every aspect of plant growth and development. Thus, given that auxin stimulates lateral roots initiation and growth, this scenario explains more branched brx root system. Nevertheless, auxin is considered to activate plasma membrane proton pumps. Along with this, it has been shown that brx mutants acidify media much more than the wild type plants do, a trait that was proposed as an adaptive feature of naturally occurring brx null alleles in Arabidopsis populations found on acidic soils. Additionally, in our study we found that most of accessions originally collected from acidic sampling sites exhibit hypersensitivity to CLE45 treatment. This implies that adaptation of plants to acidic soil involves a positive selection pressure against upstream negative regulators of CLE45-BAM3 signaling, such as BRX. Perspective analysis of these accessions would provide more profound understanding of molecular mechanisms underlying plant adaptation to acidic soils. All these results are suggesting that targeting of the factors that affect protophloem differentiation is a good strategy of natural selection to change the root architecture and to develop an adaptation to a certain environment. -- Les plantes comme organismes sessiles ont développé différents mécanismes pour s'adapter aux conditions environnementales complexes dans lesquelles elles vivent. L'adaptation est le processus par lequel des traits vont évoluer via la sélection naturelle vers une amélioration fonctionnelle dans un contexte environnemental donné. Une adaptation à l'environnement est caractérisée par des changements génétiques dans des populations entières qui ont été fixés par la sélection naturelle sur plusieurs générations. Le gène BREVIS RADIX (BRX) a été identifié dans le crible d'une collection d'accessions naturelles d'Arabidopsis et a été caractérisé comme un régulateur de la croissance racinaire étant donné que le mutant perte-de-fonction montre une croissance racinaire primaire arrêtée au stade post-embryonnaire et présente de plus un système racinaire plus ramifié que la plante sauvage. Bien que le mutant perte-de-fonction brx cause une altération complète de l'architecture racinaire, l'activité de BRX n'est requise que dans la vascularisation racinaire, en particulier au niveau du protophloème. Le protophloème est un composant du réseau de transport du phloème et est responsable du transit des dérivés de la photosynthèse ainsi que des régulateurs de croissances, venant de la partie aérienne par le phloème mature (métaphloème) vers tous les méristèmes primaires de la plante. Pour pouvoir réaliser sa fonction, le protophloème est la première file de cellules à se différencier à l'intérieur du méristème de la racine. Pendant ce processus, les cellules du protophloème subissent une mort cellulaire programmée partielle durant laquelle elles épaississent leur paroi cellulaire, dégradent le noyau et le tonoplaste tandis que la membrane plasmique demeure fonctionnelle. De manière intéressante, les cellules du protophloème entament le processus d'allongement seulement après que la différenciation en tubes criblés soit complète. Ce travail montre que le mutant brx est incapable de mener à bien la différenciation de la file de cellules du protophloème, phénotype qui peut être visualisé par la présence de cellules 'trous', de cellules non différenciées entourées de deux cellules différenciées. La discontinuité de la différenciation du phloème dans le mutant brx est considérée comme la conséquence de l'hyperactivité localisée du module de signalisation CLA VA TA3/EMBRYO SURROUNDING REGION 45 (CLE45) - BARELY ANY MERISTEM 3 (BAM3). De manière intéressante, l'activité de CLE45, très probablement au niveau de la liaison avec le récepteur, peut être modulé par le pH apoplastique. Pris ensemble, nos résultats impliquent que l'activité des pompes à protons, actives dans les cellules non différenciées du protophloème, doit être maintenue en dessous d'un certain seuil autrement la cascade de signalisation CLE45-BAM3 serait stimulée, en conséquence de quoi le protophloème ne pourrait se différencier. D'après la morphologie vacuolaire, une acidification prématurée de la paroi cellulaire dans le mutant brx empêche la différenciation du protophloème de manière stochastique. Une fois que le protophloème se différencie, les pompes à protons peuvent alors être activées afin d'acidifier l'apoplaste et ainsi faciliter l'allongement des cellules énuclées du protophloème, entraînées par la croissance des cellules environnantes. Finalement, la différenciation défectueuse du protophloème produit une accumulation d'auxine dans la partie supérieure de la racine car le phloème ne peut plus acheminer efficacement l'auxine au méristème. Physiologiquement, la 'fuite' d'auxine à partir du réseau vasculaire de la plante peut avoir des conséquences variées puisque l'auxine est impliquée dans la régulation de la majorité des aspects de la croissance et développement de la plante. Etant donné que l'auxine stimule l'initiation et développement des racines latérales, ce scénario pourrait expliquer le système racinaire plus ramifié du mutant brx. En plus, l'auxine est considérée comme un activateur des pompes à protons. Par ailleurs, nous avons montré que les mutants brx ont la capacité d'acidifier le milieu plus efficacement que les plantes sauvages, une caractéristique des populations sauvages <¥Arabidopsis poussant sur des sols acides et contenant les allèles délétés brx. De plus, dans nos résultats nous avons mis en évidence que la plupart des accessions collectées originellement sur des sites acidophiles montre une hypersensibilité au traitement par CLE45. Ceci implique que l'adaptation des plantes aux sols acides repose sur la pression de sélection positive à rencontre des régulateurs négatifs de CLE45- BAM3, situés en amont de la cascade, tel le produit du gène BRX. Les analyses de ces accessions pourraient aboutir à une meilleure compréhension des mécanismes moléculaires responsables de l'adaptation des plantes aux sols acides. Tous nos résultats suggèrent que le ciblage des facteurs affectant la différenciation du protophloème serait une stratégie gagnante dans la sélection naturelle pour changer l'architecture de la racine et ainsi s'adapter efficacement à un nouvel environnement.

Organic matter processing and soil evolution in a braided river system

Traditionally, braided river research has considered flow, sediment transport processes and, recently, vegetation dynamics in relation to river morphodynamics. However, if considering the development of woody vegetated patches over a time scale of decades, we must consider the extent to which soil forming processes, particularly related to soil organic matter, impact the alluvial geomorphic-vegetation system. Here we quantify the soil organic matter processing (humification) that occurs on young alluvial landforms. We sampled different geomorphic units, ranging from the active river channel to established river terraces in a braided river system. For each geomorphic unit, soil pits were used to sample sediment/soil layers that were analysed in terms of grain size (<2mm) and organic matter quantity and quality (RockEval method). A principal components analysis was used to identify patterns in the dataset. Results suggest that during the succession from bare river gravels to a terrace soil, there is a transition from small amounts of external organic matter supply provided by sedimentation processes (e.g. organic matter transported in suspension and deposited on bars), to large amounts of autogenic in situ organic matter production due to plant colonisation. This appears to change the time scale and pathways of alluvial succession (bio-geomorphic succession). However, this process is complicated by: the ongoing possibility of local sedimentation, which can serve to isolate surface layers via aggradation from the exogenic supply; and erosion which tends to create fresh deposits upon which organic matter processing must re-start. The result is a complex pattern of organic matter states as well as a general lack of any clear chronosequence within the active river corridor. This state reflects the continual battle between deposition events that can isolate organic matter from the surface, erosion events that can destroy accumulating organic matter and the early ecosystem processes necessary to assist the co-evolution of soil and vegetation. A key question emerges over the extent to which the fresh organic matter deposited in the active zone is capable of significantly transforming the local geochemical environment sufficiently to accelerate soil development.

Machine learning for geospatial data : algorithms, software tools and case studies

Résumé Cette thèse est consacrée à l'analyse, la modélisation et la visualisation de données environnementales à référence spatiale à l'aide d'algorithmes d'apprentissage automatique (Machine Learning). L'apprentissage automatique peut être considéré au sens large comme une sous-catégorie de l'intelligence artificielle qui concerne particulièrement le développement de techniques et d'algorithmes permettant à une machine d'apprendre à partir de données. Dans cette thèse, les algorithmes d'apprentissage automatique sont adaptés pour être appliqués à des données environnementales et à la prédiction spatiale. Pourquoi l'apprentissage automatique ? Parce que la majorité des algorithmes d'apprentissage automatiques sont universels, adaptatifs, non-linéaires, robustes et efficaces pour la modélisation. Ils peuvent résoudre des problèmes de classification, de régression et de modélisation de densité de probabilités dans des espaces à haute dimension, composés de variables informatives spatialisées (« géo-features ») en plus des coordonnées géographiques. De plus, ils sont idéaux pour être implémentés en tant qu'outils d'aide à la décision pour des questions environnementales allant de la reconnaissance de pattern à la modélisation et la prédiction en passant par la cartographie automatique. Leur efficacité est comparable au modèles géostatistiques dans l'espace des coordonnées géographiques, mais ils sont indispensables pour des données à hautes dimensions incluant des géo-features. Les algorithmes d'apprentissage automatique les plus importants et les plus populaires sont présentés théoriquement et implémentés sous forme de logiciels pour les sciences environnementales. Les principaux algorithmes décrits sont le Perceptron multicouches (MultiLayer Perceptron, MLP) - l'algorithme le plus connu dans l'intelligence artificielle, le réseau de neurones de régression généralisée (General Regression Neural Networks, GRNN), le réseau de neurones probabiliste (Probabilistic Neural Networks, PNN), les cartes auto-organisées (SelfOrganized Maps, SOM), les modèles à mixture Gaussiennes (Gaussian Mixture Models, GMM), les réseaux à fonctions de base radiales (Radial Basis Functions Networks, RBF) et les réseaux à mixture de densité (Mixture Density Networks, MDN). Cette gamme d'algorithmes permet de couvrir des tâches variées telle que la classification, la régression ou l'estimation de densité de probabilité. L'analyse exploratoire des données (Exploratory Data Analysis, EDA) est le premier pas de toute analyse de données. Dans cette thèse les concepts d'analyse exploratoire de données spatiales (Exploratory Spatial Data Analysis, ESDA) sont traités selon l'approche traditionnelle de la géostatistique avec la variographie expérimentale et selon les principes de l'apprentissage automatique. La variographie expérimentale, qui étudie les relations entre pairs de points, est un outil de base pour l'analyse géostatistique de corrélations spatiales anisotropiques qui permet de détecter la présence de patterns spatiaux descriptible par une statistique. L'approche de l'apprentissage automatique pour l'ESDA est présentée à travers l'application de la méthode des k plus proches voisins qui est très simple et possède d'excellentes qualités d'interprétation et de visualisation. Une part importante de la thèse traite de sujets d'actualité comme la cartographie automatique de données spatiales. Le réseau de neurones de régression généralisée est proposé pour résoudre cette tâche efficacement. Les performances du GRNN sont démontrées par des données de Comparaison d'Interpolation Spatiale (SIC) de 2004 pour lesquelles le GRNN bat significativement toutes les autres méthodes, particulièrement lors de situations d'urgence. La thèse est composée de quatre chapitres : théorie, applications, outils logiciels et des exemples guidés. Une partie importante du travail consiste en une collection de logiciels : Machine Learning Office. Cette collection de logiciels a été développée durant les 15 dernières années et a été utilisée pour l'enseignement de nombreux cours, dont des workshops internationaux en Chine, France, Italie, Irlande et Suisse ainsi que dans des projets de recherche fondamentaux et appliqués. Les cas d'études considérés couvrent un vaste spectre de problèmes géoenvironnementaux réels à basse et haute dimensionnalité, tels que la pollution de l'air, du sol et de l'eau par des produits radioactifs et des métaux lourds, la classification de types de sols et d'unités hydrogéologiques, la cartographie des incertitudes pour l'aide à la décision et l'estimation de risques naturels (glissements de terrain, avalanches). Des outils complémentaires pour l'analyse exploratoire des données et la visualisation ont également été développés en prenant soin de créer une interface conviviale et facile à l'utilisation. Machine Learning for geospatial data: algorithms, software tools and case studies Abstract The thesis is devoted to the analysis, modeling and visualisation of spatial environmental data using machine learning algorithms. In a broad sense machine learning can be considered as a subfield of artificial intelligence. It mainly concerns with the development of techniques and algorithms that allow computers to learn from data. In this thesis machine learning algorithms are adapted to learn from spatial environmental data and to make spatial predictions. Why machine learning? In few words most of machine learning algorithms are universal, adaptive, nonlinear, robust and efficient modeling tools. They can find solutions for the classification, regression, and probability density modeling problems in high-dimensional geo-feature spaces, composed of geographical space and additional relevant spatially referenced features. They are well-suited to be implemented as predictive engines in decision support systems, for the purposes of environmental data mining including pattern recognition, modeling and predictions as well as automatic data mapping. They have competitive efficiency to the geostatistical models in low dimensional geographical spaces but are indispensable in high-dimensional geo-feature spaces. The most important and popular machine learning algorithms and models interesting for geo- and environmental sciences are presented in details: from theoretical description of the concepts to the software implementation. The main algorithms and models considered are the following: multi-layer perceptron (a workhorse of machine learning), general regression neural networks, probabilistic neural networks, self-organising (Kohonen) maps, Gaussian mixture models, radial basis functions networks, mixture density networks. This set of models covers machine learning tasks such as classification, regression, and density estimation. Exploratory data analysis (EDA) is initial and very important part of data analysis. In this thesis the concepts of exploratory spatial data analysis (ESDA) is considered using both traditional geostatistical approach such as_experimental variography and machine learning. Experimental variography is a basic tool for geostatistical analysis of anisotropic spatial correlations which helps to understand the presence of spatial patterns, at least described by two-point statistics. A machine learning approach for ESDA is presented by applying the k-nearest neighbors (k-NN) method which is simple and has very good interpretation and visualization properties. Important part of the thesis deals with a hot topic of nowadays, namely, an automatic mapping of geospatial data. General regression neural networks (GRNN) is proposed as efficient model to solve this task. Performance of the GRNN model is demonstrated on Spatial Interpolation Comparison (SIC) 2004 data where GRNN model significantly outperformed all other approaches, especially in case of emergency conditions. The thesis consists of four chapters and has the following structure: theory, applications, software tools, and how-to-do-it examples. An important part of the work is a collection of software tools - Machine Learning Office. Machine Learning Office tools were developed during last 15 years and was used both for many teaching courses, including international workshops in China, France, Italy, Ireland, Switzerland and for realizing fundamental and applied research projects. Case studies considered cover wide spectrum of the real-life low and high-dimensional geo- and environmental problems, such as air, soil and water pollution by radionuclides and heavy metals, soil types and hydro-geological units classification, decision-oriented mapping with uncertainties, natural hazards (landslides, avalanches) assessments and susceptibility mapping. Complementary tools useful for the exploratory data analysis and visualisation were developed as well. The software is user friendly and easy to use.

On the diurnal soil water content dynamics during evaporation using dielectric methods

The water content dynamics in the upper soil surface during evaporation is a key element in land-atmosphere exchanges. Previous experimental studies have suggested that the soil water content increases at the depth of 5 to 15 cm below the soil surface during evapo- ration, while the layer in the immediate vicinity of the soil surface is drying. In this study, the dynamics of water content profiles exposed to solar radiative forcing was monitored at a high temporal resolution using dielectric methods both in the presence and absence of evaporation. A 4-d comparison of reported moisture content in coarse sand in covered and uncovered buckets using a commercial dielectric-based probe (70 MHz ECH2O-5TE, Decagon Devices, Pullman, WA) and the standard 1-GHz time domain reflectometry method. Both sensors reported a positive correlation between temperature and water content in the 5- to 10-cm depth, most pronounced in the morning during heating and in the afternoon during cooling. Such positive correlation might have a physical origin induced by evaporation at the surface and redistribution due to liquid water fluxes resulting from the temperature- gradient dynamics within the sand profile at those depths. Our experimental data suggest that the combined effect of surface evaporation and temperature-gradient dynamics should be considered to analyze experimental soil water profiles. Additional effects related to the frequency of operation and to protocols for temperature compensation of the dielectric sensors may also affect the probes' response during large temperature changes.

Robust Nonlinear Mapping of Soil Contamination Using Support Regression

Spatial data analysis mapping and visualization is of great importance in various fields: environment, pollution, natural hazards and risks, epidemiology, spatial econometrics, etc. A basic task of spatial mapping is to make predictions based on some empirical data (measurements). A number of state-of-the-art methods can be used for the task: deterministic interpolations, methods of geostatistics: the family of kriging estimators (Deutsch and Journel, 1997), machine learning algorithms such as artificial neural networks (ANN) of different architectures, hybrid ANN-geostatistics models (Kanevski and Maignan, 2004; Kanevski et al., 1996), etc. All the methods mentioned above can be used for solving the problem of spatial data mapping. Environmental empirical data are always contaminated/corrupted by noise, and often with noise of unknown nature. That's one of the reasons why deterministic models can be inconsistent, since they treat the measurements as values of some unknown function that should be interpolated. Kriging estimators treat the measurements as the realization of some spatial randomn process. To obtain the estimation with kriging one has to model the spatial structure of the data: spatial correlation function or (semi-)variogram. This task can be complicated if there is not sufficient number of measurements and variogram is sensitive to outliers and extremes. ANN is a powerful tool, but it also suffers from the number of reasons. of a special type ? multiplayer perceptrons ? are often used as a detrending tool in hybrid (ANN+geostatistics) models (Kanevski and Maignank, 2004). Therefore, development and adaptation of the method that would be nonlinear and robust to noise in measurements, would deal with the small empirical datasets and which has solid mathematical background is of great importance. The present paper deals with such model, based on Statistical Learning Theory (SLT) - Support Vector Regression. SLT is a general mathematical framework devoted to the problem of estimation of the dependencies from empirical data (Hastie et al, 2004; Vapnik, 1998). SLT models for classification - Support Vector Machines - have shown good results on different machine learning tasks. The results of SVM classification of spatial data are also promising (Kanevski et al, 2002). The properties of SVM for regression - Support Vector Regression (SVR) are less studied. First results of the application of SVR for spatial mapping of physical quantities were obtained by the authorsin for mapping of medium porosity (Kanevski et al, 1999), and for mapping of radioactively contaminated territories (Kanevski and Canu, 2000). The present paper is devoted to further understanding of the properties of SVR model for spatial data analysis and mapping. Detailed description of the SVR theory can be found in (Cristianini and Shawe-Taylor, 2000; Smola, 1996) and basic equations for the nonlinear modeling are given in section 2. Section 3 discusses the application of SVR for spatial data mapping on the real case study - soil pollution by Cs137 radionuclide. Section 4 discusses the properties of the modelapplied to noised data or data with outliers.

Soil diffuse degassing and thermal energy fluxes from the southern Lakki plain, Nisyros (Greece)

Two diffuse soil CO2 flux surveys from the southern Lakki plain show that CO2 is mainly released from the hydrothermal explosion craters. The correspondence between high CO2 fluxes and elevated soil temperatures suggests that a flux of hot hydrothermal fluids ascends towards the surface. Steam mostly condenses near the surface and the heat given off is conductively transferred to the atmosphere through the soil, accompanied by a large CO2 flux. Tt was calculated, that 68 t d(-1) of hydrothermal CO2 are released through the total surveyed area of similar to1.3 km(2) Admitting that a steam flux of 2200 t d(-1) accompanies this CO2 flux, the thermal energy released through steam condensation amounts to 58 MW.