Among tools that are used to fill knowledge gaps on natural systems, ecological modelling has been widely applied during the last two decades. Ecological models are simple representations of a complex reality. They allow to highlight environmental drivers of species ecological niche and better understand species responses to environmental changes. However, applying models to Southern Ocean benthic organisms raises several methodological challenges. Species presence datasets are often aggregated in time and space nearby research stations or along main sailing routes. Data are often limited in number to correctly describe species occupied space and physiology. Finally, environmental datasets are not precise enough to accurately represent the complexity of marine habitats. Can we thus generate performant and accurate models at the scale of the Southern Ocean ?What are the limits of such approaches ?How could we improve methods to build more relevant models ?In this PhD thesis, three different model categories have been studied and their performance evaluated. (1) Mechanistic physiological models (Dynamic Energy Budget models, DEB) simulate how the abiotic environment influences individual metabolism and represent the species fundamental niche. (2) Species distribution models (SDMs) predict species distribution probability by studying the relationship between species presences and the environment. They represent the species realised niche. (3) Dispersal lagrangian models predict the drift of propagules in water masses. Results show that physiological models can be developed for marine Southern Ocean species to simulate the metabolic variations in link with the environment and predict population dynamics. However, more data are necessary to highlight detailed physiological contrasts between populations and to accurately evaluate models. Results obtained for SDMs suggest that models generated at the scale of the Southern Ocean and future simulations are not relevant, given the lack of data available to characterise species occupied space, the lack of precision and accuracy of future climate scenarios and the impossibility to evaluate models. Moreover, model extrapolate on a large proportion of the projected area. Adding information on species physiological limits (observations, results from experiments, physiological model outputs) was shown to reduce extrapolation and to improve the capacity of models to estimate the species realised niche. Spatial aggregation of occurrence data, which influenced model predictions and evaluation was also succefully corrected. Finally, dispersal models showed an interesting potential to highlight the role of geographic barriers or conversely of spatial connectivity and also the link between species distribution, physiology and phylogeny history. This PhD thesis provides several methodological advice, annoted codes and tutorials to help implement future modelling works applied to Southern Ocean marine species. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

The thesis is divided into two main parts. The first part is focusing to theoretical bases of work, it is a brief introduction to business performance and utilization of modern indicators to measure it with an emphasis on benchmarking. In second part, we deal with the practical application of information acquired and that a specific enterprise. It is an analysis of the current situation with companies in the same field of business. With the help of financial analysis, SWOT matrix identify the strengths and weaknesses of the company. The most important part is devoted to benchmarking based on publicly available data from the financial statements. For these purposes, I especially enjoyed the information from the available statement. Based on the information they are designed adequate opportunities in improving and recommendations for evaluating enterprise.

This thesis deals with the company's performance, its measurement and increase. The theoretical section provides insights from literature and further on this base an analytic and proposal part follows. The analytical part contains particularly introduction of the company, its activities and strategic and financial analysis of the company. The proposal part deals with the implementation of the Balanced Scorecard in the analysed company.

The master thesis is focused on efficiency evaluation of company IN-EKO TEAM, s. r. o. The theoretical part contains the relevant information from professional publications, basic concepts are explained here and models supporting business management, business process modeling models, and models emphasizing self-evaluation are presented. The second part is an efficiency evaluation by using the selected START Plus model. The key part is the suggestion part in which measures will be proposed to improve the efficiency of examined subject.

This master´s thesis deals with the performance measurement of the company CALIBRI, Ltd. The first part of the thesis explains the term performance and describes methods of performance measurement. In the next part is outlined the profile of the company with its current situation. In the practical part, the evaluation of the performance is measured by START model. The last section is devoted to a summary assessment of the situation of the chosen company and suggestion for improving the company´s performance.

Im Einzugsgebiet des Wairau River auf der Südinsel Neuseelands sind in den vergangenen Jahrzehnten klimatische Veränderungen beobachtet worden. Beeinflusst durch eine Abnahme der jährlichen Niederschläge sinken auch die Gebietsabflüsse. Der Wairau Aquifer, der für die Bewässerung Neuseelands größtem Weinanbaugebiet genutzt wird, ist maßgeblich von dem Abflussverhalten des Wairau River abhängig und verzeichnet einen Rückgang der Grundwasserstände. Die Simulation der hydrologischen Prozesse im Einzugsgebiet ist daher notwendig, um das Verständnis für die Zusammenhänge zu erhöhen. Um die Prozesse abzubilden, wurde in dieser Masterarbeit die Modular Assessment of Rainfall–Runoff Models Toolbox (MARRMoT) verwendet. Ziel der Anwendung von MARRMoT war, ein Modell oder ein Modellensemble zu finden, das für die Simulation der Niederschlags-Abfluss-Beziehungen im Einzugsgebiet Wairau River geeignet ist. Zu diesem Zweck wurden in einem ersten Szenario 42 Modelle aus MARRMoT auf das Einzugsgebiet Wairau River mit dem Gütekriterium Kling-Gupta-Effizienz (KGE) kalibriert. In weiteren Szenarien wurden die Kalibrierungen mit dem KGE mit invers transformierten Durchflüssen (KGEi) sowie mit einer Mischform der beiden Kriterien (KGEm) realisiert. Die Bewertung der Modellperformance erfolgte anhand ausgewählter Gütekriterien. Durch die Auswertung konnten drei MARRMoT-Modelle ermittelt werden, die in allen drei Kalibrierszenarien eine hohe Performance gezeigt haben und für die weitere Anwendung im Einzugsgebiet Wairau River geeignet sind. Ein Teilziel dieser Arbeit war zudem die Korrektur von simulierten Niederschlagsdaten aus dem unbeobachteten Quellgebiet des Wairau River, die in vorherigen Arbeiten als fehlerhaft identifiziert wurden (Ludwig 2018). Dafür wurde eine höhenabhängige Korrekturvorschrift ermittelt. Die anschließende Simulation des Gebiets hat jedoch gezeigt, dass die Korrektur ohne eine bessere Datengrundlage nicht erfolgreich ist.:1. Einführung 2. Beschreibung des Untersuchungsgebiets 2.1 Klima 2.2 Einzugsgebietsgrenzen 2.3 Geologie 2.4 Boden 2.5 Landnutzung 2.6 Aquifer 3. Datengrundlage 3.1 Meteorologische Daten 3.2 Hydrologische Daten 4. Analyse und Korrektur von Niederschlagsdaten 4.1 Problematik im Teileinzugsgebiet Dip Flat 4.2 Aufbereitung der Daten 4.3 Erstellung einer Korrekturvorschrift 5. Anwendung der MARRMo Toolbox 5.1 Grundlagen von MARRMoT 5.2 Betrachtung ausgewählter MARRMoT-Modelle 5.3 Gütekriterien 5.4 Kalibrierung und Evaluierung 5.4.1 Grundlagen 5.4.2 Automatische Kalibrierung mit MARRMoT 5.4.3 Kalibrierszenarien 6. Analyse und Diskussion der Ergebnisse 6.1 Niederschlagskorrektur und Datengrundlage 6.1.1 Erstellung der Korrekturvorschrift 6.1.2 Unsicherheiten der Niederschlagskorrektur 6.1.3 Analyse der Datengrundlage für das Gesamteinzugsgebiet 6.1.4 Zusammenfassung 6.2 Anwendung der MARRMo Toolbox 6.2.1 Ergebnisse der Kalibrierszenarien 6.2.2 Detaillierte Auswertung ausgewählter Modelle 6.2.3 Unsicherheiten zwischen und innerhalb der MARRMoT Modelle 6.2.4 Zusammenfassung 6.3 Zusammenführung der Ergebnisse 7. Fazit und Ausblick 8. Literaturverzeichnis / In the catchment area of the Wairau River on the South Island of New Zealand, climatic changes were observed in the past decades. Influenced by a decrease in precipitation, the runoff is reduced. The Wairau Aquifer, which is used for irrigation of New Zealand's largest wine-growing region, is significantly dependent on the runoff of the Wairau River and recorded a decline in groundwater levels. The simulation of the hydrological processes in the catchment area is therefore necessary. For that reason in this Master’s thesis the Modular Assessment of Rainfall Runoff Model Toolbox (MARRMoT) was used in order to find a model or model ensemble suitable for the simulation of precipitation-runoff relationships in the Wairau River catchment. In a first scenario, 42 models from MARRMoT were calibrated for the Wairau River catchment using the efficiency criteria Kling-Gupta-Efficiency (KGE). In further scenarios, calibrations using the KGE with inverse transformed flows (KGEi) as well as a mixed form of the two criteria (KGEm) were realized. The evaluation of the model performance was based on selected performance criteria. The evaluation led to three MARRMoT models, which demonstrated high performance in all three calibration scenarios and are suitable for further use in the Wairau River catchment. A secondary objective of this work was the correction of simulated precipitation data from the unobserved headwaters of the catchment, which were identified as faulty in previous work (Ludwig 2018). For this purpose, a height-dependent correction rule was determined. Subsequent simulation of the area, however, has shown that the correction was unsuccessful and needs a more solid data foundation.:1. Einführung 2. Beschreibung des Untersuchungsgebiets 2.1 Klima 2.2 Einzugsgebietsgrenzen 2.3 Geologie 2.4 Boden 2.5 Landnutzung 2.6 Aquifer 3. Datengrundlage 3.1 Meteorologische Daten 3.2 Hydrologische Daten 4. Analyse und Korrektur von Niederschlagsdaten 4.1 Problematik im Teileinzugsgebiet Dip Flat 4.2 Aufbereitung der Daten 4.3 Erstellung einer Korrekturvorschrift 5. Anwendung der MARRMo Toolbox 5.1 Grundlagen von MARRMoT 5.2 Betrachtung ausgewählter MARRMoT-Modelle 5.3 Gütekriterien 5.4 Kalibrierung und Evaluierung 5.4.1 Grundlagen 5.4.2 Automatische Kalibrierung mit MARRMoT 5.4.3 Kalibrierszenarien 6. Analyse und Diskussion der Ergebnisse 6.1 Niederschlagskorrektur und Datengrundlage 6.1.1 Erstellung der Korrekturvorschrift 6.1.2 Unsicherheiten der Niederschlagskorrektur 6.1.3 Analyse der Datengrundlage für das Gesamteinzugsgebiet 6.1.4 Zusammenfassung 6.2 Anwendung der MARRMo Toolbox 6.2.1 Ergebnisse der Kalibrierszenarien 6.2.2 Detaillierte Auswertung ausgewählter Modelle 6.2.3 Unsicherheiten zwischen und innerhalb der MARRMoT Modelle 6.2.4 Zusammenfassung 6.3 Zusammenführung der Ergebnisse 7. Fazit und Ausblick 8. Literaturverzeichnis

Toxic compounds, such as pesticides, are routinely tested against a range of aquatic, avian and mammalian species as part of the registration process. The need for reducing dependence on animal testing has led to an increasing interest in alternative methods such as in silico modelling. The QSAR (Quantitative Structure Activity Relationship)-based models are already in use for predicting physicochemical properties, environmental fate, eco-toxicological effects, and specific biological endpoints for a wide range of chemicals. Data plays an important role in modelling QSARs and also in result analysis for toxicity testing processes. This research addresses number of issues in predictive toxicology. One issue is the problem of data quality. Although large amount of toxicity data is available from online sources, this data may contain some unreliable samples and may be defined as of low quality. Its presentation also might not be consistent throughout different sources and that makes the access, interpretation and comparison of the information difficult. To address this issue we started with detailed investigation and experimental work on DEMETRA data. The DEMETRA datasets have been produced by the EC-funded project DEMETRA. Based on the investigation, experiments and the results obtained, the author identified a number of data quality criteria in order to provide a solution for data evaluation in toxicology domain. An algorithm has also been proposed to assess data quality before modelling. Another issue considered in the thesis was the missing values in datasets for toxicology domain. Least Square Method for a paired dataset and Serial Correlation for single version dataset provided the solution for the problem in two different situations. A procedural algorithm using these two methods has been proposed in order to overcome the problem of missing values. Another issue we paid attention to in this thesis was modelling of multi-class data sets in which the severe imbalance class samples distribution exists. The imbalanced data affect the performance of classifiers during the classification process. We have shown that as long as we understand how class members are constructed in dimensional space in each cluster we can reform the distribution and provide more knowledge domain for the classifier.

Efforts to control surface runoff and erosion from forest roads at stream crossings are critical for the protection of aquatic ecosystems in forests. In this research, annual and event-based sediment delivery rates were estimated for reopened legacy roads at forest stream crossings in the Virginia Piedmont physiographic region, USA. Sediment delivery rates were compared among stream crossing approaches with diverse intensities of best management practice (BMP) implementations for surface cover and water control. Model predictions from the Water Erosion Prediction Project (WEPP) were compared to field observations of surface runoff and sediment delivery to evaluate model performance. Annual sediment delivery rates from reopened (bare) legacy road approaches to stream crossings were 7.5 times higher than those of completely graveled approaches. Sediment delivery rates ranged from 34 to 287 Mg ha⁻¹ year⁻¹ for the bare approaches and from 10 to 16 Mg ha⁻¹ year⁻¹ for the graveled approaches. Event-based surface runoff and associated total suspended solids (TSS) concentrations were compared among a succession of gravel surfacing treatments that represented increasing intensities of BMP implementations on reopened approaches. The three treatments were No Gravel (10-19% cover), Low Gravel (34-60% cover), and High Gravel (50-99% cover). Median TSS concentration of surface runoff for the No Gravel treatment (2.84 g L⁻¹) was greater than Low Gravel (1.10 g L⁻¹) and High Gravel (0.82 g L⁻¹) by factors of 2.6 and 3.5, respectively. WEPP predictions of event-based sediment yield show clear differences among the different road surface treatments, but prediction intervals were wide, reflecting substantial prediction uncertainty. These findings show that reopened legacy roads and associated stream crossing approaches can deliver significant quantities of sediment if roads are not adequately closed or maintained and that corrective best management practices (BMPs), such as gravel and appropriate spacing of water control structures, can reduce sediment delivery to streams. Watershed management decisions that hinge upon WEPP predictions of sediment yield from forest roads must necessarily take into account a wide range of potential erosion rates for specific management scenarios. / Ph. D.

Toxic compounds, such as pesticides, are routinely tested against a range of aquatic, avian and mammalian species as part of the registration process. The need for reducing dependence on animal testing has led to an increasing interest in alternative methods such as in silico modelling. The QSAR (Quantitative Structure Activity Relationship)-based models are already in use for predicting physicochemical properties, environmental fate, eco-toxicological effects, and specific biological endpoints for a wide range of chemicals. Data plays an important role in modelling QSARs and also in result analysis for toxicity testing processes. This research addresses number of issues in predictive toxicology. One issue is the problem of data quality. Although large amount of toxicity data is available from online sources, this data may contain some unreliable samples and may be defined as of low quality. Its presentation also might not be consistent throughout different sources and that makes the access, interpretation and comparison of the information difficult. To address this issue we started with detailed investigation and experimental work on DEMETRA data. The DEMETRA datasets have been produced by the EC-funded project DEMETRA. Based on the investigation, experiments and the results obtained, the author identified a number of data quality criteria in order to provide a solution for data evaluation in toxicology domain. An algorithm has also been proposed to assess data quality before modelling. Another issue considered in the thesis was the missing values in datasets for toxicology domain. Least Square Method for a paired dataset and Serial Correlation for single version dataset provided the solution for the problem in two different situations. A procedural algorithm using these two methods has been proposed in order to overcome the problem of missing values. Another issue we paid attention to in this thesis was modelling of multi-class data sets in which the severe imbalance class samples distribution exists. The imbalanced data affect the performance of classifiers during the classification process. We have shown that as long as we understand how class members are constructed in dimensional space in each cluster we can reform the distribution and provide more knowledge domain for the classifier.

Avaliou-se o desempenho de um modelo numérico de previsão do tempo (GFS - Global Forecast System – antigo AVN – AViatioN model - do Centro Nacional para Previsão Ambiental – NCEP) no prognóstico de variáveis meteorológicas temperatura, déficit de pressão de vapor do ar, saldo de radiação e velocidade do vento, e da evapotranspiração de referência calculada pelos métodos de Thornthwaite (1948) e de Penman-Monteith (Allen et al., 1998). O desempenho foi avaliado por comparação com dados provenientes de uma estação meteorológica, situada em Piracicaba, São Paulo. A temperatura e o déficit de pressão de vapor do ar foram os elementos melhor prognosticados, com desempenho "muito bom" e "bom", de acordo com o índice de desempenho proposto por Camargo e Sentelhas (1997), para no máximo quatro e três dia de antecedência, respectivamente, durante o período seco. Para o período úmido, somente o prognóstico do déficit de pressão de vapor do ar para o primeiro dia mostrou-se "bom". Os prognósticos de saldo de radiação e velocidade do vento foram ruins para ambos os períodos. Em decorrência do bom desempenho do modelo para prognosticar a temperatura, verificou-se que a estimativa de ETo pelo método de Thornthwaite teve boa concordância com o calculado a partir dos dados da estação meteorológica, com antecedência de até três dias para o período seco. Para o úmido, este fato foi observado apenas para o primeiro dia de antecedência. A concordância entre os valores estimados pelo modelo e a partir da estação para o método de Penman-Monteith foi muito baixa, em conseqüência do desempenho do modelo de previsão do tempo em prognosticar o saldo de radiação e a velocidade do vento. / The performance of a numeric weather forecast model (GFS- Forecast System, former AVN - AvatioN model, National Center for Environmental Prediction-NCEP) was evaluated for predicting weather variables, like air temperature and vapor pressure deficit, net radiation and wind speed, as well as reference evapotranspiration calculated by Thornthwaite (1948) and Penman-Monteith (Allen et al., 1948) methods, by the comparison with data obtained by an automatic weather station, in Piracicaba, State of São Paulo, Brazil. Temperature and vapor pressure deficit were the variables predicted with the best accuracy, with a "very good" and "good" performance, according to the index of confidence proposed by Camargo and Sentelhas (1997), for the maximum of four and three days in advance, respectively, during the dry season. For the wet season, only vapor pressure deficit was predicted with a "good" performance of the model. The predictions of net radiation and wind speed were very poor for both seasons. As the weather forecast model predicted temperature well, ETo estimated by Thornthwaite method showed a good agreement with ETo values estimated by observed data from the weather station, with till three days in advance for the dry season. For the wet season, such agreement was observed just for one day in advance. When ETo estimated by Penman-Monteith method with data from the weather forecast model and from weather station were compared any agreement was observed, which was caused by the poor performance of the numeric weather forecast model to predict net radiation and wind speed.