Dissertations / Theses on the topic 'Multi-temporal Analysi'

Multi-temporal image analysis has been widely used in many applications such as urban monitoring, disaster management, and agriculture. With the development of the remote sensing technology, the new generation remote sensing satellite images with High/ Very High spatial resolution (HR/VHR) are now available. Compared to the traditional low/medium spatial resolution images, the detailed information of ground objects can be clearly analyzed in the HR/VHR images. Classical methods of multi-temporal image analysis deal with the images at pixel level and have worked well on low/medium resolution images. However, they provide sub-optimal results on new generation images due to their limited capability of modeling complex spatial and spectral information in the new generation products. Although significant number of object-based methods have been proposed in the last decade, they depend on suitable segmentation scale for diverse kinds of objects present in each temporal image. Thus their capability to express contextual information is limited. Typical spatial properties of last generation images emphasize the need of having more flexible models for object representation. Another drawback of the traditional methods is the difficulty in transferring knowledge learned from one specific problem to another. In the last few years, an interesting development is observed in the machine learning/computer vision field. Deep learning, especially Convolution Neural Networks (CNNs) have shown excellent capability to capture object level information and in transfer learning. By 2015, deep learning achieved state-of-the-art performance in most computer vision tasks. Inspite of its success in computer vision fields, the application of deep learning in multi-temporal image analysis saw slow progress due to the requirement of large labeled datasets to train deep learning models. However, by the start of this PhD activity, few works in the computer vision literature showed that deep learning possesses capability of transfer learning and training without labeled data. Thus, inspired by the success of deep learning, this thesis focuses on developing deep learning based methods for unsupervised/semi-supervised multi-temporal image analysis. This thesis is aimed towards developing methods that combine the benefits of deep learning with the traditional methods of multi-temporal image analysis. Towards this direction, the thesis first explores the research challenges that incorporates deep learning into the popular unsupervised change detection (CD) method - Change Vector Analysis (CVA) and further investigates the possibility of using deep learning for multi-temporal information extraction. The thesis specifically: i) extends the paradigm of unsupervised CVA to novel Deep CVA (DCVA) by using a pre-trained network as deep feature extractor; ii) extends DCVA by exploiting Generative Adversarial Network (GAN) to remove necessity of having a pre-trained deep network; iii) revisits the problem of semi-supervised CD by exploiting Graph Convolutional Network (GCN) for label propagation from the labeled pixels to the unlabeled ones; and iv) extends the problem statement of semantic segmentation to multi-temporal domain via unsupervised deep clustering. The effectiveness of the proposed novel approaches and related techniques is demonstrated on several experiments involving passive VHR (including Pleiades), passive HR (Sentinel-2), and active VHR (COSMO-SkyMed) datasets. A substantial improvement is observed over the state-of-the-art shallow methods.

This work explores the multidimensional, ambiguous and temporal characteristics of tonality. The approach relies on interfacing pitch-spaces with time vs. time-scale descriptions. In this combined representation, the spatial and temporal hierarchies of tonality are evidenced simultaneously and in relation to each other. A visual exploration method is proposed for the analysis of tonal context in music works, using a simple model of tonal induction. The method is extended for the analysis of music based on tonal systems beyond the major-minor paradigm. Two perceptual studies are approached from this descriptive framework. The first study evidences the impact of time-scale in a simple mode of tonal induction, and analyses the mathematical artefacts introduced by evaluations in scaled spaces. In the second study, a model of contextual instability is proposed and discussed in relation to the modelling of tonal tension. The analysis and representation methods are then generalised, through a set-class theoretical domain.
Esta tesis analiza la naturaleza ambigua, multidimensional y temporal de la tonalidad. El método propuesto parte de la conexión entre espacios tonales y descripciones en tiempo y escala temporal. Esta representación conjunta pone de manifiesto la relación entre las jerarquías espaciales y temporales de la tonalidad. Utilizando un modelo simple de inducción tonal, se propone un método de exploración visual del contexto tonal en obras musicales. Dos estudios perceptuales son abordados desde el entorno descriptivo propuesto. En el primer estudio, se pone en evidencia el impacto de la escala temporal como parámetro de un modelo simple de inducción tonal, y se analizan los artificios matemáticos introducidos por evaluaciones en espacios escalados dimensionalmente. En el segundo estudio se propone un modelo de inestabilidad contextual, y se analiza en relación al modelado de la tensión tonal. El método de análisis se generaliza, a través de una categorización contextual en set-classes.

The movement of human beings appears to respond to a complex motor system that contains signals at different hierarchical levels. For example, an action such as ``grasping a glass on a table'' represents a high-level action, but to perform this task, the body needs several motor inputs that include the activation of different joints of the body (shoulder, arm, hand, fingers, etc.). Each of these different joints/muscles have a different size, responsiveness, and precision with a complex non-linearly stratified temporal dimension where every muscle has its temporal scale. Parts such as the fingers responds much faster to brain input than more voluminous body parts such as the shoulder. The cooperation we have when we perform an action produces smooth, effective, and expressive movement in a complex multiple temporal scale cognitive task. Following this layered structure, the human body can be described as a kinematic tree, consisting of joints connected. Although it is nowadays well known that human movement and its perception are characterised by multiple temporal scales, very few works in the literature are focused on studying this particular property. In this thesis, we will focus on the analysis of human movement using data-driven techniques. In particular, we will focus on the non-verbal aspects of human movement, with an emphasis on full-body movements. The data-driven methods can interpret the information in the data by searching for rules, associations or patterns that can represent the relationships between input (e.g. the human action acquired with sensors) and output (e.g. the type of action performed). Furthermore, these models may represent a new research frontier as they can analyse large masses of data and focus on aspects that even an expert user might miss. The literature on data-driven models proposes two families of methods that can process time series and human movement. The first family, called shallow models, extract features from the time series that can help the learning algorithm find associations in the data. These features are identified and designed by domain experts who can identify the best ones for the problem faced. On the other hand, the second family avoids this phase of extraction by the human expert since the models themselves can identify the best set of features to optimise the learning of the model. In this thesis, we will provide a method that can apply the multi-temporal scales property of the human motion domain to deep learning models, the only data-driven models that can be extended to handle this property. We will ask ourselves two questions: what happens if we apply knowledge about how human movements are performed to deep learning models? Can this knowledge improve current automatic recognition standards? In order to prove the validity of our study, we collected data and tested our hypothesis in specially designed experiments. Results support both the proposal and the need for the use of deep multi-scale models as a tool to better understand human movement and its multiple time-scale nature.

Multi-temporal image analysis has been widely used in many applications such as urban monitoring, disaster management, and agriculture. With the development of the remote sensing technology, the new generation remote sensing satellite images with High/ Very High spatial resolution (HR/VHR) are now available. Compared to the traditional low/medium spatial resolution images, the detailed information of ground objects can be clearly analyzed in the HR/VHR images. Classical methods of multi-temporal image analysis deal with the images at pixel level and have worked well on low/medium resolution images. However, they provide sub-optimal results on new generation images due to their limited capability of modeling complex spatial and spectral information in the new generation products. Although significant number of object-based methods have been proposed in the last decade, they depend on suitable segmentation scale for diverse kinds of objects present in each temporal image. Thus their capability to express contextual information is limited. Typical spatial properties of last generation images emphasize the need of having more flexible models for object representation. Another drawback of the traditional methods is the difficulty in transferring knowledge learned from one specific problem to another. In the last few years, an interesting development is observed in the machine learning/computer vision field. Deep learning, especially Convolution Neural Networks (CNNs) have shown excellent capability to capture object level information and in transfer learning. By 2015, deep learning achieved state-of-the-art performance in most computer vision tasks. Inspite of its success in computer vision fields, the application of deep learning in multi-temporal image analysis saw slow progress due to the requirement of large labeled datasets to train deep learning models. However, by the start of this PhD activity, few works in the computer vision literature showed that deep learning possesses capability of transfer learning and training without labeled data. Thus, inspired by the success of deep learning, this thesis focuses on developing deep learning based methods for unsupervised/semi-supervised multi-temporal image analysis. This thesis is aimed towards developing methods that combine the benefits of deep learning with the traditional methods of multi-temporal image analysis. Towards this direction, the thesis first explores the research challenges that incorporates deep learning into the popular unsupervised change detection (CD) method - Change Vector Analysis (CVA) and further investigates the possibility of using deep learning for multi-temporal information extraction. The thesis specifically: i) extends the paradigm of unsupervised CVA to novel Deep CVA (DCVA) by using a pre-trained network as deep feature extractor; ii) extends DCVA by exploiting Generative Adversarial Network (GAN) to remove necessity of having a pre-trained deep network; iii) revisits the problem of semi-supervised CD by exploiting Graph Convolutional Network (GCN) for label propagation from the labeled pixels to the unlabeled ones; and iv) extends the problem statement of semantic segmentation to multi-temporal domain via unsupervised deep clustering. The effectiveness of the proposed novel approaches and related techniques is demonstrated on several experiments involving passive VHR (including Pleiades), passive HR (Sentinel-2), and active VHR (COSMO-SkyMed) datasets. A substantial improvement is observed over the state-of-the-art shallow methods.

Every year wildfires threaten or destroy ecological habitats, man-made infrastructure and people’s lives. Additionally millions of dollars are spent each year trying to prevent and control these fires. Ideally if a wildfire can be detected before it rages out of control it can be extinguished and avoid large scale devastation. Traditional manned fire lookout towers are neither cost effective nor particularly efficient at detecting wildfire. It is proposed that temporal filtering can be used to isolate the signals created at the beginnings of potential wildfires. Temporal filtering can remove any background image and any periodic signals created by the camera movement. Once typical signals are analyzed, digital filters can be designed to pass fire signals while blocking the unwanted signals. The temporal filter passes only fire signals and signals generated by moving objects. These objects can be distinguished from each other by analyzing the objects mid and long wave energy profile. This algorithm is tested on 17 data sources and its results analyzed.

Establishing wetland gains and losses, delineating wetland boundaries, and determining their vegetative composition are major challenges that can be improved through remote sensing studies. In this study, we used the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) to separate wetlands from uplands in a study of 870 locations on the Virginia Coastal Plain. We used the first five bands from each of two ASTER scenes (6 March 2005 and 16 October 2005), covering the visible to the short-wave infrared region (0.52-2.185υm). We included GIS data layers for soil survey, topography, and presence or absence of water in a logistic regression model that predicted the location of over 78% of the wetlands. While this was slightly less accurate (78% vs. 86%) than current National Wetland Inventory (NWI) aerial photo interpretation procedures of locating wetlands, satellite imagery analysis holds great promise for speeding wetland mapping, lowering costs, and improving update frequency. To estimate wetland vegetation composition classs of the study locations, we generated a Classification and Regression Tree (CART) model and a Multinomial Logistic Regression (logit) model, and compared their accuracy in separating woody wetlands, emergent wetlands and open water. The overall accuracy of the CART model was 73.3%, while the overall accuracy of the logit model was 76.7%. Although the CART producer's accuracy (correct category classification) of the emergent wetlands was higher than the accuracy from the multinomial logit (57.1% vs. 40.7%), we obtained the opposite result for the woody wetland category (68.7% vs. 52.6%). A McNemar test between the two models and NWI maps showed that their accuracies were not statistically different. We conducted a sub-pixel analysis of the ASTER images to establish canopy cover of forested wetlands. The canopy cover ranged from 0 to 225 m2. We used visble-near-infrared ASTER bands, Delta Normalized Difference Vegetation Index, and a Tasselled Cap transformation in an ordinary linear regression (OLS) model. The model achieved an adjusted-R2 of 0.69 and an RMSE of 2.73% when the canopy cover is less than 16%. For higher canopy cover values, the adjusted-R2 was 0.4 and the RMSE was19.79%. Taken together, these findings suggest that satellite remote sensing, in concert with other spatial data, has strong potential for mapping both wetland presence and type.
Ph. D.

Oggigiorno, per poter affrontare efficacemente i problemi ambientali su larga scala, la necessità di disporre di mappe di copertura del suolo affidabili e ad alta risoluzione spaziale e temporale è più che mai urgente. Infatti, numerosi contesti potrebbero trarre beneficio da tali prodotti come, ad esempio, il cambiamento climatico, la desertificazione, l'inverdimento dell'artico, la deforestazione, l'urbanizzazione, l'erosione del suolo, il monitoraggio delle foreste, la conservazione della biodiversità, la gestione delle aree urbane, la gestione delle risorse idriche, l'agricoltura, la sicurezza alimentare e molti altri. Siccome le variabili di interesse tendono a cambiare molto rapidamente nel tempo e nello spazio, la disponibilità di mappe di copertura del suolo frequenti e di buona qualità suscita un grande interesse. Negli ultimi anni sono state prodotte diverse mappe tematiche e di copertura del suolo su scala regionale/globale le quali, tuttavia, spesso non soddisfano i requisiti imposti dalle applicazioni; ciò è dovuto principalmente al fatto che i prodotti esistenti sono stati generati da diversi sensori satellitari (ottici, radar o entrambi), diverse strategie di campionamento, diverse legende, diversi protocolli di validazione, ecc. Inoltre, la risoluzione spaziale e/o temporale di tali prodotti è spesso insufficiente per diverse applicazioni. In questo lavoro di tesi è stato studiato come sfruttare dati multitemporali di tipo ottico e SAR (Synthetic Aperture Radar) per caratterizzare un insieme molto ristretto di classi, piuttosto che un'ampia gamma di tipi di copertura del suolo. Il lavoro presentato si concentra sulla vegetazione (tra cui specie arboree, praterie, arbusti ed altri), corpi idrici (tra cui laghi, mari, fiumi ed altri) e colture biologiche (in particolare, pratiche di agricoltura biologica). Per quanto riguarda la vegetazione, la letteratura scientifica offre numerose metodologie consolidate, finalizzate alla mappatura delle coperture vegetative. Al contrario, gli approcci che sfruttano i sensori SAR come principale fonte di dati sono decisamente più rari. Per questo motivo, parte di questa tesi è dedicata all'analisi della potenzialità dei dati SAR multitemporali nel caratterizzare diversi tipi di vegetazione naturale. Per quanto riguarda la mappatura dei corpi idrici, la letteratura tecnica fornisce diverse soluzioni basate sia su dati ottici che SAR. Tuttavia, la maggioranza delle metodologie analizzate presentano alcune limitazioni legate principalmente alla mancanza di automatismo degli algoritmi, l'impossibilità di utilizzare il modello in altre regioni di interesse, alla risoluzione spaziale relativamente bassa ed altri. Dal momento che la comunità sul cambiamento climatico necessita di informazioni tempestive relative allo stato dei corpi idrici a livello non solo locale/regionale ma anche globale, e che sia indipendente dalle condizioni meteorologiche delle diverse aree del mondo, in questa tesi si propone una metodologia volta a mappare i corpi idrici sfruttando sequenze temporali di dati SAR, che sia in grado di superare le limitazioni più gravi presenti negli approcci esistenti. Infine, per quanto concerne la caratterizzazione dei terreni agricoli biologici, occorre rilevare e monitorare diversi aspetti, tra cui le operazioni di diserbo, le attività di fertilizzazione e le tecniche di lavorazione del terreno. A tal fine, sia i dati ottici multitemporali che i dati SAR vengono sfruttati per costruire piccoli blocchi che faranno parte di un sistema di monitoraggio dell'agricoltura biologica più complesso, volto a migliorare la trasparenza e la tracciabilità all'interno della catena di approvvigionamento alimentare biologica. In generale, i risultati hanno dimostrato che le sequenze temporali di dati SAR e multispettrali possono essere impiegate con successo nella classificazione dei diversi tipi di copertura del suolo di cui sopra.
Nowadays, the need for reliable, timely, high-resolution land cover maps is more than urgent if large-scale environmental problems are to be tackled effectively. Many different contexts would in fact benefit from such products, such as climate change, desertification, arctic greening, deforestation, urbanization, soil erosion, forest monitoring, conservation of biodiversity, urban area management, water resources management, agriculture, food security and many others. Due to the fact that the involved variables tend to change very rapidly in time and space, the availability of frequent and good quality global land cover products raises great interest. Several regional/global thematic and land cover maps have been delivered and other are expected, but they often do not meet the specific requirements of various applications; this is mainly due to the fact that all the existing products have been generated from different satellite sensors (optical, radar or both), different sampling strategies, different types of mapped land cover types, different validation protocols, etc. Moreover, the spatial and/or temporal resolution of these products is often insufficient for some applications. In this thesis work, we investigated how to leverage multitemporal optical and SAR data to characterize a very small set of classes rather than a full range of land cover types. Our work focuses on vegetation (including tree species, grasslands, shrublands and others), water bodies (including lakes, seas, rivers and others) and organic croplands (specifically, organic farming practices). Regarding vegetation, the technical literature offers numerous well-established methodologies aimed at mapping vegetated land covers. On the contrary, approaches that use SAR sensors as the main source of data are definitely more scarce. For this reason, part of this thesis work will be devoted to analyze the potential of multitemporal SAR data to characterize several types of natural vegetation. Regarding mapping of water bodies, the scientific literature provides several solutions based on optical and SAR data. However, almost all the analyzed methodologies have some limitations, mainly related to lack of automatism, impossibility to use the proposed method in other regions of interest, relatively low spatial resolution and others. Given the climate change community's need for timely information on the status of water bodies at the global level regardless of weather conditions, in this thesis a methodology aimed at mapping water bodies using sequences of SAR data, that is able to overcome the most severe limitations of the existing methodologies, is proposed. Finally, to characterize organic farmland, several aspects must be detected and monitored, including weed-killer operations, fertilization activities and tillage techniques. To do so, both multitemporal optical and SAR data are exploited to build small detection blocks, that will be part of a more complex organic farming monitoring system aimed at improving transparency and traceability within the organic food supply chain. In general, results showed that SAR and multispectral time series can be successfully employed to classify these land cover types.

In recent years, advances in small unmanned aerial vehicle (UAV) technology have transformed the use cases of these aircraft from hobby flying to industrial and business applications. These maneuverable, easily deployed tools can be retrofitted with a myriad of sensors and equipment, which make them suitable to perform a variety of specialized tasks. With increasing UAV capabilities, the function of small UAVs can be extended from pure monitoring or surveillance to the dual objective of monitoring an environment for events and addressing the events in some way. This thesis seeks to explore a subdomain of the dual objective problem described, referred to in this thesis as the multi-UAV persistent search and retrieval task with stochastic target appearance (PSR-STA), in which UAVs continuously search an area over a long period of time for targets of interest, which appear according to a probabilistic model, to retrieve and deliver them to a collector location. The advent of high-speed computers and agent-based modeling theory enable the simulation of multi-UAV PSR-STA. However, it can be complicated to combine parts of multi-UAV PSR-STA such as motion models and multi-UAV coordination into one integrated system, and even after they are combined successfully, it is difficult to analyze the system except with simple comparison tools. This thesis 1) proposes a framework that builds a foundation for understanding how to simulate and analyze multi-UAV PSR-STA through prescribing important design decisions and methods for simulation and 2) identifies metrics, analysis tools, and trends related to overall system effectiveness for multi-UAV PSR-STA. A case study of multi-UAV park cleanup is implemented where many simulations with input parameters chosen by a latin hypercube design of experiments are examined, algorithms for choosing the locations of collectors and charging stations based on probabilistic information are proposed, and the differences in effectiveness between four coverage search patterns are analyzed. Measures are highlighted that provide insight into performance variability over time and space. Line charts and the discrete Fourier transform are used to understand temporal patterns inherent in the data. Principal component analysis is used to analyze relevant spatial patterns in effectiveness, and a random forest surrogate model with a profiler is used to explore the non-linear influence of input parameters on the spatial patterns. The trellis chart or figure of figures method is presented for visualizing spatial and temporal data across many simulations. A second set of experiments based on the park cleanup case study are performed and examined to verify the benefits of these methods.

ABSTRACT DECISION TREE CLASSIFICATION OF MULTI-TEMPORAL IMAGES FOR FIELD-BASED CROP MAPPING Sencan, Seç
il M. Sc., Department of Geodetic and Geographic Information Technologies Supervisor: Assist. Prof. Dr. Mustafa Tü
rker August 2004, 125 pages A decision tree (DT) classification approach was used to identify summer (August) crop types in an agricultural area near Karacabey (Bursa), Turkey from multi-temporal images. For the analysis, Landsat 7 ETM+ images acquired in May, July, and August 2000 were used. In addition to the original bands, NDVI, PCA, and Tasselled Cap Transformation bands were also generated and included in the classification procedure. Initially, the images were classified on a per-pixel basis using the multi-temporal masking technique together with the DT approach. Then, the classified outputs were applied a field-based analysis and the class labels of the fields were directly entered into the Geographical Information System (GIS) database. The results were compared with the classified outputs of the three dates of imagery generated using a traditional maximum likelihood (ML) algorithm. It was observed that the proposed approach provided significantly higher overall accuracies for the May and August images, for which the number of classes were low. In May and July, the DT approach produced the classification accuracies of 91.10% and 66.15% while the ML classifier produced 84.38% and 63.55%, respectively. However, in August nearly the similar overall accuracies were obtained for the ML (70.82%) and DT (69.14%) approaches. It was also observed that the use of additional bands for the proposed technique improved the separability of the sugar beet, tomato, pea, pepper, and rice classes.

Natural hazard related to the volcanic activity represents a potential risk factor, particularly in the vicinity of human settlements. Besides to the risk related to the explosive and effusive activity, the instability of volcanic edifices may develop into large landslides often catastrophically destructive, as shown by the collapse of the northern flank of Mount St. Helens in 1980. A combined approach was applied to analyse slope failures that occurred at Stromboli volcano. SdF slope stability was evaluated by using high-resolution multi-temporal DTMMs and performing limit equilibrium stability analyses. High-resolution topographical data collected with remote sensing techniques and three-dimensional slope stability analysis play a key role in understanding instability mechanism and the related risks. Analyses carried out on the 2002–2003 and 2007 Stromboli eruptions, starting from high-resolution data acquired through airborne remote sensing surveys, permitted the estimation of the lava volumes emplaced on the SdF slope and contributed to the investigation of the link between magma emission and slope instabilities. Limit Equilibrium analyses were performed on the 2001 and 2007 3D models, in order to simulate the slope behavior before 2002-2003 landslide event and after the 2007 eruption. Stability analyses were conducted to understand the mechanisms that controlled the slope deformations which occurred shortly after the 2007 eruption onset, involving the upper part of slope. Limit equilibrium analyses applied to both cases yielded results which are congruent with observations and monitoring data. The results presented in this work undoubtedly indicate that hazard assessment for the island of Stromboli should take into account the fact that a new magma intrusion could lead to further destabilisation of the slope, which may be more significant than the one recently observed because it will affect an already disarranged deposit and fractured and loosened crater area. The two-pronged approach based on the analysis of 3D multi-temporal mapping datasets and on the application of LE methods contributed to better understanding volcano flank behaviour and to be prepared to undertake actions aimed at risk mitigation.

Natural hazard related to the volcanic activity represents a potential risk factor, particularly in the vicinity of human settlements. Besides to the risk related to the explosive and effusive activity, the instability of volcanic edifices may develop into large landslides often catastrophically destructive, as shown by the collapse of the northern flank of Mount St. Helens in 1980. A combined approach was applied to analyse slope failures that occurred at Stromboli volcano. SdF slope stability was evaluated by using high-resolution multi-temporal DTMMs and performing limit equilibrium stability analyses. High-resolution topographical data collected with remote sensing techniques and three-dimensional slope stability analysis play a key role in understanding instability mechanism and the related risks. Analyses carried out on the 2002–2003 and 2007 Stromboli eruptions, starting from high-resolution data acquired through airborne remote sensing surveys, permitted the estimation of the lava volumes emplaced on the SdF slope and contributed to the investigation of the link between magma emission and slope instabilities. Limit Equilibrium analyses were performed on the 2001 and 2007 3D models, in order to simulate the slope behavior before 2002-2003 landslide event and after the 2007 eruption. Stability analyses were conducted to understand the mechanisms that controlled the slope deformations which occurred shortly after the 2007 eruption onset, involving the upper part of slope. Limit equilibrium analyses applied to both cases yielded results which are congruent with observations and monitoring data. The results presented in this work undoubtedly indicate that hazard assessment for the island of Stromboli should take into account the fact that a new magma intrusion could lead to further destabilisation of the slope, which may be more significant than the one recently observed because it will affect an already disarranged deposit and fractured and loosened crater area. The two-pronged approach based on the analysis of 3D multi-temporal mapping datasets and on the application of LE methods contributed to better understanding volcano flank behaviour and to be prepared to undertake actions aimed at risk mitigation.

Landslide hazard assessment of densely forested, remote, and difficult to access areas can be rapidly accomplished with airborne light detection and ranging (lidar) data. An evaluation of geomorphic change by lidar-derived digital elevation models (DEMs) coupled with geotechnical soils analysis, aerial photographs, ground measurements, precipitation data, and numerical modeling can provide valuable insight to the reactivation process of unstable landslides. A landslide was selected based on previous work by Mickleson (2011) and Burns et al. (2010) that identified the Madrone Landslide with significant volumetric changes. This study expands on previous work though an evaluation of the timing and causation of slope failure of the Madrone Landslide. The purpose of this study was to evaluate landslide morphology, precipitation data, historical aerial photographs, ground crack measurements, geotechnical properties of soil, numerical modeling, and elevation data (with multi-temporal lidar data), to determine the conditions associated with failure of the Madrone Landslide. To evaluate the processes involved and timing of slope failure events, a deep seated potentially unstable landslide, situated near the contact of Eocene sedimentary and volcanic rocks, was selected for a detailed analysis. The Madrone Landslide (45.298383/-123.338796) is located in Yamhill County, about 12 kilometers west of Carlton, Oregon. Site elevation ranges from 206 meters (m) North American Vertical Datum (NAVD-88) near the head scarp to 152 m at the toe. The landslide is composed of two parts, an upper more recent rotational slump landslide and a lower much older earth flow landslide. The upper slide has an area of 2,700 m2 with a head scarp of 5-7 m and a volume of 15,700 m3. The lower earth flow has an area of 2300 m2, a head scarp of 15 m, and a volume of 287,500 m3. Analysis of aerial photographs indicates the lower slide probably originated between 1956 and 1963. The landslide is located at a geologic unit contact of Eocene deep marine sedimentary rock and intrusive volcanic rock. The landslide was instrumented with 20 crack monitors established across ground cracks and measured periodically. Field measurements did not detect ground crack displacement over a 15 month period. Soil samples indicate the soil is an MH soil with a unit weight of 12 kN/m3 and residual friction angle of 28φ'r which were both used as input for slope stability modeling. Differential DEMs from lidar data were calculated to generate a DEM of Difference (DoD) raster to identify and quantify elevation changes. Historical aerial photograph review, differential lidar analysis, and precipitation data suggest the upper portion of the landslide failed as a result of the December 2007 storm.

O desenvolvimento de uma nova geração de sensores nos últimos 20 anos consolidou as imagens de sensoriamento remoto como uma importante fonte de dados para estudos ambientais e fenômenos geográficos em larga escala. É grande a variedade de resoluções (espacial, temporal e espectral) das imagens de sensoriamento remoto, desde pancromáticas até imagens polarimétricas. Apesar da grande experiência em coleta, armazenamento e distribuição de imagens e da diversidade de ferramentas computacionais para processamento e análise, ainda é difícil de se encontrar sistemas que apresentem um ambiente integrado para transformar imagens multi-temporais e de diversas resoluções em informação útil. Tendo em vista este panorama, a contribuição desta tese é dupla. Em primeiro lugar, propomos e implementamos uma nova ferramenta, seguindo os padrões de código-fonte aberto (\textit{Free and Open Source Software - FOSS}) , para integrar métodos de análise de imagens com técnicas de mineração de dados, visando produzir um ambiente computacional extensível e focado no usuário, aplicado à extração de informações e à descoberta de conhecimento em grandes bases de dados geométricos. Esta ferramenta é chamada GeoDMA - \textit{Geographic Data Mining Analyst} (mineração de dados geográficos). GeoDMA integra técnicas de sementação de imagens, extração e seleção de atributos, classificação, métricas da ecologia da paisagem, métodos de análise multi-temporal para detecção de mudanças e classificação por métodos de árvores de decisão adaptados à mineração de dados espaciais. O sistema agrega imagens de sensoriamento remoto com outros tipos de dados geográficos através do acesso a bancos de dados locais ou remotos. GeoDMA também provê métodos de simulação para avaliar a acurácia dos modelos, e ferramentas para análise espaço-temporal, incluindo um esquema de visuação de perfis temporais que auxilia os usuários a descrever padrões em eventos cíclicos. Em segundo lugar, desenvolvemos um novo método para analizar dados espaço-temporais baseados na transformação dos perfis em coordenadas polares, o que permite a geração de um novo conjunto de atributos que aumenta a acurácia da classificação de imagens multi-temporais. O sistema GeoDMA foi construído como uma extensão do SIG Terra View, e por isso mapas temáticos e demais resultados são produzidos rapidamente, aproveitando-se das funcionalidades deste SIG. Para demonstrar as ferramentas do GeoDMA, cinco (5) casos de estudo, aplicados em diferentes contextos de detecção de uso e cobertura da terra, foram realizados usando dados de diferentes domínios. A avaliação destes experimentos, do ponto de vista do usuário, mostrou que a ferramenta obteve os resultados com um nível de integração não encontrado em sistemas semelhantes.
The deployment of a new generation of sensors over the last 20 years has made satellite remotely sensed imagery a very important source of spatial data available for environmental studies of large-scale geographic phenomena. The variety of spatial, temporal and spectral resolutions for remote sensing images is large, ranging from panchromatic images to polarimetric radar images. Despite the great experience in image data gathering and distribution and a diversity of image processing and analysis toolboxes, it is still difficult to find image analysis systems that provide a straightforward fully integrated environment to transform multi-temporal and multiresolution satellite image data into meaningful information. Taking this into account, the contribution of this thesis is two-fold. Firstly, we propose and implement a new toolbox, developed under the Free and Open Source Software (FOSS) foundation, for integrating remote sensing imagery analysis methods with data mining techniques producing a user-centered, extensible, rich computational environment for information extraction and knowledge discovery over large geographic databases. The toolbox is called GeoDMA - Geographic Data Mining Analyst. It integrates techniques of segmentation, feature extraction, feature selection, classification, landscape metrics and multi-temporal methods for change detection and analysis with decision-tree based strategies adapted for spatial data mining. It gathers remotely sensed imagery with other geographic data types using access to local or remote databases. GeoDMA provides simulation methods to assess the accuracy of process mo dels as well as tools for spatio-temporal analysis, including a visualization scheme for temporal profiles that helps users to describe patterns in cyclic events. Secondly, we develop a new approach for analyzing spatio-temporal data based on a polar coordinates transformation that allows creating a new set of features which improves the classification accuracy of multi-temporal image databases. As GeoDMA was built on top of Terra View GIS, thematic maps and other results can be produced rapidly, taking advantage of the basic GIS functionalities. To demonstrate the features of GeoDMA toolbox, five (5) case studies, applied in contexts of land use and land cover change, were carried out in different application domains. Evaluations of these experiments pointed out that the GeoDMA toolbox achieved results with a level of integration, from a user perspective, that could not be found elsewhere.

L’Arganier [Argania spinosa (L.) Skeels] est un arbre de la famille des Sapotacées, endémique du sud-ouest marocain. C’est un arbre multi-usages, qui constitue une ressource primordiale pour les populations de cet espace semi-aride et aride du Maroc. Il constitue la clef-de-voûte de l’agro-écosystème traditionnel de l’arganeraie reposant sur un équilibre entre ressources et exploitation humaine, et joue également un rôle important dans la lutte contre la désertification et l’érosion. Actuellement, la menace de dégradation de l’arganeraie est une préoccupation majeure aussi bien pour la population que pour les scientifiques. On assiste en effet depuis plusieurs décennies à une diminution du couvert arboré, à la fois en surface occupée et en densité d’arbres. Face à cette préoccupation, nous avons étudié l’espace multidimensionnel de l’arganeraie en cherchant à identifier les principales caractéristiques de cet espace, ainsi que les facteurs responsables de sa dégradation. Ensuite, nous avons dévoilé l’originalité de cet espace à partir de son organisation sociale et spatiale, ainsi que le mode de fonctionnement et de gestion de ce territoire. Dans la deuxième partie nous avons montré l’apport de la télédétection spatiale et des systèmes d’information géographique pour la caractérisation de l’occupation du sol et l’identification des changements à partir d’un suivi diachronique, en utilisant une série d’images SPOT, Landsat, Google Earth, Ikonos. Nous avons également testé la possibilité d'évaluer la densité des arganiers à partir des images à haute résolution spatiale Ikonos et Google Earth. Nous concluons à la nécessité d’un suivi de ce territoire afin de pouvoir évaluer les changements et prendre les mesures d’aménagement et de protection nécessaires
The Argan [Argania spinosa (L.) Skeels] is a species of tree endemic to the calcareous semi-desert Sous valley of southwestern Morocco. It is the sole species in the genus Argania (family of Sapotaceae). It is a multi-purpose tree, and the main resource provider for the population of this semi-arid and arid area (source of forage, oil, timber and fuel). Argan is the keystone species of the traditional agro-ecosystem of the Berber society, ensuring a meta-stable equilibrium between resource availability and anthropic use; it plays a major role in preventing erosion and desertification damages.Currently, in spite of the Biosphere Reserve label attributed by UNESCO in 1998, the threat of degradation of the sparse Argan forest is a main concern for both local population and scientists. Since several decades, a decrease of extension area of the species and of tree density has been observed. According to this preoccupation, we have studied the multidimensional space of the Argan forest, in view of identifying its main features and the potential drivers of degradation processes. Then the originality of this area has been demonstrated through the assessment of its social and spatial organization, and of land-use and management practices.In the second part, we have shown the possible use of remotely sensed data and of Geographic Information Systems for surveying land-use/land-cover and for monitoring changes through a multi-temporal analysis of satellite images: SPOT, Landsat, Ikonos and Google Earth imagery. The evaluation of tree density has been performed through object-oriented classification of high spatial resolution satellite imagery (Ikonos, Google Earth). In conclusion, we recommend the effective use of a monitoring system to follow environmental changes in the Argan tree area, and to produce the detailed information needed for implementation of management and conservation strategies ensuring a sustainable development of the area

My doctoral research dissertation focuses on two aspects of functional data analysis (FDA): FDA under spatial interdependence and FDA for multi-level data. The first part of my thesis focuses on developing modeling and inference procedure for functional data under spatial dependence. The methodology introduced in this part is motivated by a research study on inequities in accessibility to financial services. The first research problem in this part is concerned with a novel model-based method for clustering random time functions which are spatially interdependent. A cluster consists of time functions which are similar in shape. The time functions are decomposed into spatial global and time-dependent cluster effects using a semi-parametric model. We also assume that the clustering membership is a realization from a Markov random field. Under these model assumptions, we borrow information across curves from nearby locations resulting in enhanced estimation accuracy of the cluster effects and of the cluster membership. In a simulation study, we assess the estimation accuracy of our clustering algorithm under a series of settings: small number of time points, high noise level and varying dependence structures. Over all simulation settings, the spatial-functional clustering method outperforms existing model-based clustering methods. In the case study presented in this project, we focus on estimates and classifies service accessibility patterns varying over a large geographic area (California and Georgia) and over a period of 15 years. The focus of this study is on financial services but it generally applies to any other service operation. The second research project of this part studies an association analysis of space-time varying processes, which is rigorous, computational feasible and implementable with standard software. We introduce general measures to model different aspects of the temporal and spatial association between processes varying in space and time. Using a nonparametric spatiotemporal model, we show that the proposed association estimators are asymptotically unbiased and consistent. We complement the point association estimates with simultaneous confidence bands to assess the uncertainty in the point estimates. In a simulation study, we evaluate the accuracy of the association estimates with respect to the sample size as well as the coverage of the confidence bands. In the case study in this project, we investigate the association between service accessibility and income level. The primary objective of this association analysis is to assess whether there are significant changes in the income-driven equity of financial service accessibility over time and to identify potential under-served markets. The second part of the thesis discusses novel statistical methodology for analyzing multilevel functional data including a clustering method based on a functional ANOVA model and a spatio-temporal model for functional data with a nested hierarchical structure. In this part, I introduce and compare a series of clustering approaches for multilevel functional data. For brevity, I present the clustering methods for two-level data: multiple samples of random functions, each sample corresponding to a case and each random function within a sample/case corresponding to a measurement type. A cluster consists of cases which have similar within-case means (level-1 clustering) or similar between-case means (level-2 clustering). Our primary focus is to evaluate a model-based clustering to more straightforward hard clustering methods. The clustering model is based on a multilevel functional principal component analysis. In a simulation study, we assess the estimation accuracy of our clustering algorithm under a series of settings: small vs. moderate number of time points, high noise level and small number of measurement types. We demonstrate the applicability of the clustering analysis to a real data set consisting of time-varying sales for multiple products sold by a large retailer in the U.S. My ongoing research work in multilevel functional data analysis is developing a statistical model for estimating temporal and spatial associations of a series of time-varying variables with an intrinsic nested hierarchical structure. This work has a great potential in many real applications where the data are areal data collected from different data sources and over geographic regions of different spatial resolution.

Several decades of intensive dryland-farming in the Gadarif Region, located in the Eastern part of Sudan, has led to rapid landuse/landcover (LULC) changes mainly due to agricultural expansion, government policies and environmental calamities such as drought. The study area represents part of the African Sahel. The fundamental goal of the thesis was to assess land degradation and the impact of agriculture expansion on land cover, soil and crops production. To analyse and to monitor the LULC changes, multi-temporal Landsat data of the years 1979, 1989 and 1999 and ASTER data of the year 2009 covering an area of approximately 1200 km² were used. For this a post-classification comparison technique was applied to detect LULC changes from satellite images. Six LULC classes were identified during the classification scheme, namely cultivated land, fallow land, woodland, bare land, settlement and water. For the four dates of satellite images the overall classification accuracy ranged from 86 % to 92 %. During the three decades of the study period an extensive change of LULC patterns occurred. The cultivated areas increased significantly, covering 81 % of the previous woodland in the period 1979 – 2009. Fallow land only increased during the period 1989 – 1999. Over the three decades, urban expansion continuously increased covering an area of 23, 21 and 27 km² for the periods 1979 – 1989, 1989 – 1999 and 1999 – 2009 respectively. The detailed LULC map of the study area was obtained by using a dual polarisation (HH and HV) TerraSAR-X data of the year 2009. The different LULCs of the study area were analysed by employing an object-oriented classification approach. For that purpose, multi-resolution segmentation of the Definiens Software was used for creating the image objects. Using the feature-space optimisation tool the attributes of the TerraSAR-X images were optimised in order to obtain the best separability among classes for the LULC mapping. In addition to the classes that have been obtained by the optical data, the following LULCs resulted from SAR data: harvested land, rock, settlement 1 (local-roof buildings) and settlement 2 (concrete roof buildings). The backscattering coefficients for some classes were different along HH and HV polarisation. The best separation distance of the tested spectral, shape and textural features showed different variations among the discriminated LULC classes. An overall accuracy of 84 % with a kappa value of 0.82 was resulted from the classification scheme. Accuracy differences among the classes were kept minimal. For more than six decades in the Gadarif Region mechanised dryland farming is practised. As a result, due to continuous conventional tillage, extensive woodcutting and over-grazing, serious soil degradation occurred. To discuss the impact of LULC changes on the selected soil properties, three main LULC types were chosen to be investigated, namely: cultivated land, fallow land and woodland. In addition to the reference soil profiles, soil samples were also collected at two depths from ten sample plots for each of the LULC type. For these soil samples, various soil properties such as texture, bulk density (BD), organic matter (OM), soil pH, electrical conductivity (EC), sodium adsorption ratio (SoAR), phosphorous (P) and potassium (K) were analysed. Laboratory tests proved that soil properties were significantly affected by LULC changes. Within the different LULC types, clay content in the surface layers (0 – 5 and 5 – 15 cm) varied from 59 % to 65 %, whereas silt fractions ranged from 27 % to 37 %. Soil BD, OM and P were significantly different (p < 0·05) across the three LULC types. Soil pH was significantly different between cultivated land and woodland on one side and between fallow land and woodland on the other side. EC and SoAR values of fallow land were found to be significantly different (p < 0·05) from woodland. The dryland vertisol of the Gadarif Region in Sudan produced more than one-third of the national production of sorghum – the main food stuff in the country. Soil compaction has been recognised as one of the major problems in crop production worldwide. Soil strength and infiltration rate are important variables for understanding and predicting the soil processes. The effects of three different landuse systems (cultivated land, fallow land and woodland) on soil compaction and infiltration rate were investigated at two sites of the study area. Site 1 represents the older one of the two. The soil penetration resistance (SPR) was measured in three depths using a manually operated cone penetrometer. Infiltration rate was measured in the field using a double-ring infiltrometer. Following the cone-penetrometer sampling, soil samples were collected to determine the variables that affect SPR and infiltration rate vs. particle size, dry BD, volumetric moisture content (VMC) and organic carbon (OC) content. Field measurements and soil samples were collected for each landuse type. The measured infiltration rate data were inserted into the Kostiakov Model in order to predict the cumulative soil water infiltration. Soil compaction for the cultivated land was 65 % larger in comparison to woodland. Woodland areas showed an increase in the infiltration rate by 87 % and 74 % compared to cultivated and fallow land respectively. Both study sites showed an increase in the dry BD when SPR is increasing, while VMC decreases with increasing SPR. Also, low OC contents were observed to be associated with high SPR values. For Site 1 the average coefficient of determination (R²) for the infiltration data fit to the Kostiakov Model were 0.65, 0.73 and 0.84 for cultivated land, fallow land and woodland respectively. However, for Site 2 they were 0.63, 0.76 and 0.78. In the Gadarif Region agriculture is the main activity and practised in many forms with a variety of environmental effects and consequences. Continuous ploughing of the cultivated land coupled with inproper soil management has contributed to soil deterioration when the landuse changed from woodland to cultivated and fallow land. Therefore, the development of sustainable landuse practises in the dryland-farming of the study area need to be improved in order to reduce the amount of soil degradation in the future
Mehrere Jahrzehnte intensiven Trockenfeldbaus in der Region von Gadarif, welche sich im östlichen Teil des Sudans befindet, führten hauptsächlich aufgrund von landwirtschaftlicher Expansion, politischen Beschlüssen der Regierung und Naturkatastrophen wie Trockenheit zu einer raschen Veränderung der Landnutzung und Landbedeckung. Das wesentliche Ziel dieser Dissertation war es, die Degradation des Landes, sowie die Auswirkungen von landwirtschaftlicher Expansion auf die Landbedeckung, den Boden und den Pflanzenbau im Untersuchungsgebiet, welches Teile der afrikanischen Sahelzone beinhaltet, abzuschätzen. Zur Analyse und Beobachtung der Veränderungen der Landnutzung und Landbedeckung wurden multi-temporale Landsat-Daten der Jahre 1979, 1989 und 1999 sowie ASTER-Daten aus dem Jahr 2009 genutzt, welche eine Fläche von schätzungsweise 1200 km² abdecken. Um Veränderungen von Landnutzung und Landbedeckung aus Satellitenbilddaten zu bestimmen, wurde ein auf Post-Klassifikation basierendes Vergleichsverfahren angewandt. Sechs Landnutzungs- und Landbedeckungsklassen, welche die Namen bewirtschaftetes Land, brach liegendes Land, Waldgebiet, Ödland, besiedeltes Land und Wasserfläche tragen, wurden während des Klassifikationsprozesses bestimmt. Für die vier Aufnahmezeitpunkte der Satellitendaten lag die allgemeine Klassifikationsgenauigkeit zwischen 86 % und 92 %. Während des dreißigjährigen Untersuchungszeitraums fand eine beträchtliche Veränderung der Landnutzungs- und Landbedeckungsstruktur statt. Bewirtschaftete Flächen nahmen in ihrem Anteil signifikant zu und bedeckten innerhalb des Zeitraums von 1979 bis 2009 81 % der früheren Waldgebiete. Der Anteil von brach liegendem Land nahm lediglich während des Zeitraums von 1989 bis 1999 zu. Besiedelte Gebiete breiteten sich über die drei Jahrzehnte kontinuierlich aus und wuchsen innerhalb des Zeitraums von 1979 bis 1989 um eine Fläche von 23 km², sowie um 21 km² zwischen 1989 und 1999 und um 27 km² in dem Zeitabschnitt 1999 – 2009. Eine detaillierte Karte zur Landnutzung und Landbedeckung des Untersuchungsgebiets wurde mittels der Nutzung dual polarisierter (HH und HV) TerraSAR-X Daten aus dem Jahr 2009 erzeugt. Die verschiedenen Landnutzungen und Landbedeckungen im Beobachtungsgelände wurden durch die Anwendung eines objektorientierten Klassifikationsansatzes analysiert. Um Bildobjekte zu erzeugen, wurde für diesen Zweck die auf einer mehrfachen Auflösung basierende Segmentierung der Software Definiens genutzt. Das Werkzeug Feature Space Optimisation wurde für die Optimierung der Attribute der TerraSAR-X Bilder angewandt, damit eine ideale Unterscheidungsfähigkeit entlang der Klassen für die Kartierung der Landnutzungen und Landbedeckungen erreicht werden kann. Zusätzlich zu jenen Klassen, welche mittels optischer Daten abgeleitet wurden, ergaben sich aus SAR-Daten noch die nachfolgenden Landnutzungen und Landbedeckungen: Abgeerntetes Land, Fels, Besiedlung 1 (Gebäude mit landestypischer Bedachung) und Besiedlung 2 (Gebäude mit Betondach). Die Koeffizienten der Rückstreuung entlang der Polarisationen HH und HV waren für einige Klassen unterschiedlich. Der günstigste Trennungsabstand der getesteten spektralen, formgebenden und texturalen Features ergab verschiedene Abweichungen zwischen den bestimmten Klassen der Landnutzung und Landbedeckung. Die Klassifikationsmaßnahmen ergaben eine Gesamtgenauigkeit von 84 % mit einem Kappa-Wert von 0.82. Genauigkeitsunterschiede entlang der Klassen wurden minimal gehalten. Seit über sechs Jahrzehnten wird in der Region Gadarif maschinenbetriebener Trockenfeldbau ausgeübt. In Folge dessen fand eine beträchtliche Abholzung und Überweidung sowie eine schwerwiegende Bodendegradation aufgrund des stetigen konventionellen Feldbaus statt. Um die Auswirkungen der Veränderung von Landnutzung und Landbedeckung auf die ausgewählten Bodenbeschaffenheiten auszuwerten, wurden drei Haupttypen der Landnutzung und Landbedeckung für die weitere Untersuchung ausgewählt: Bewirtschaftetes Land, brach liegendes Land, und Waldgebiet. Zusätzlich zu den Referenzbodenprofilen wurden außerdem für jeden Landnutzungs- und Landbedeckungstyp auf je zehn Probeflächen Bodenproben in zwei Tiefen entnommen. Bei diesen Bodenproben wurden zahlreiche Bodeneigenschaften analysiert, wie etwa Textur, Bodendichte (BD), organischer Materialgehalt (OM), pH-Wert des Bodens, elektrische Leitfähigkeit (EC), Adsorptionsgeschwindigkeit von Natrium (SoAR), Phosphorgehalt (P) sowie Kaliumgehalt (K). Labortests ergaben, dass die Bodeneigenschaften signifikant durch die Veränderungen der Landnutzung und Landbedeckung beeinflusst werden. Innerhalb der verschiedenen Landnutzungs- und Landbedeckungstypen variierte der Tongehalt in den Deckschichten (0 – 5 cm und 5 – 15 cm) zwischen 59 % und 65 %, wohin gegen sich die Lehmanteile von 27 % bis 37 % bewegten. Bodendichte, organischer Materialgehalt und Phosphorgehalt zeigten signifikant unterschiedliche Werte bei den drei Typen der Landnutzung und Landbedeckung (p < 0.05). Der pH-Wert des Bodens war signifikant verschieden zwischen bewirtschaftetem Land und Waldgebiet zum einen, und zwischen brach liegendem Land und Waldgebiet zum anderen. Die Werte der elektrischen Leitfähigkeit und der Adsorptionsgeschwindigkeit von Natrium bei brach liegendem Land erwiesen sich als maßgeblich verschieden zu jenen von Waldgebieten (p < 0.05). Auf dem Trockenland-Vertisolboden der Region Gadarif im Sudan wurde mehr als ein Drittel der nationalen Hirseproduktion erwirtschaftet – dem Haupternährungserzeugnis des Landes. Bodenverdichtung erwies sich als eines der weltweiten Hauptprobleme für den Pflanzenbau. Bodenfestigkeit und Versickerungsrate sind wichtige Variabeln, um Bodenprozesse verstehen und vorhersagen zu können. Die Auswirkungen der drei verschiedenen Landnutzungssysteme (bewirtschaftetes Land, brach liegendes Land und Waldgebiet) auf die Bodenverdichtung und Versickerungsrate wurden an zwei Standorten im Beobachtungsgebiet untersucht. Standort 1 ist der ältere der beiden. Der Widerstand der Bodenpenetration (SPR) wurde in drei Tiefen durch eine manuell angewandte Rammsonde gemessen. Mittels der Nutzung eines Doppelring-Infiltrometers ist die Versickerungsrate im Feld gemessen worden. Im Anschluss an die Probenentnahme mittels Rammsonden wurden Bodenproben gesammelt, um jene Variabeln bestimmen zu können, welche den Widerstand der Bodenpenetration sowie der Versickerungsrate im Vergleich zur Partikelgröße, zur trockenen Bodendichte, zum volumetrischen Feuchtigkeitsgehalt (VMC) und zum organischen Karbongehalt (OC) beeinflussen. Für jeden Landnutzungstypen wurden Feldmessungen durchgeführt und Bodenproben entnommen. Die gemessenen Daten der Versickerungsrate wurden in das Kostiakov-Modell eingespeist, um die gesamte Bodenwasserversickerung vorhersagen zu können. Die Bodenverdichtung bei bewirtschaftetem Land war 65 % stärker als bei Waldgebiet. Für Waldgebietsflächen wurde eine Zunahme der Versickerungsrate um 87 % verglichen mit bewirtschaftetem Land und um 74 % im Vergleich zu brach liegendem Land aufgezeigt. Beide Untersuchungsstandorte zeigten eine Zunahme in der trockenen Bodendichte für den Fall, dass der Widerstand der Bodenpenetration zunimmt, während der volumetrische Feuchtigkeitsgehalt mit zunehmendem Bodenpenetrationswiderstand abnimmt. Ebenso wurde beobachtet, dass ein geringer organischer Karbongehalt in Verbindung zu hohen Widerstandswerten der Bodenpenetration steht. Bei Standort 1 passte der durchschnittliche Bestimmungskoeffizient (R²) der Versickerungsrate zum Kostiakov-Modell mit den Werten 0.65 für bewirtschaftetes Land, 0.73 für brach liegendes Land und 0.84 für Waldgebiet. Für Standort 2 indessen ergaben die Werte 0.63, 0.76 und 0.78. Landwirtschaft, die in vielen Formen ausgeübt wird, ist die Haupttätigkeit in der Region Gadarif, und geht mit verschiedenartigsten Umweltauswirkungen und Konsequenzen einher. Kontinuierliche Feldbestellung des bewirtschafteten Landes, verbunden mit ungeeigneter Bodenbewirtschaftung, hat sich seit jenem Zeitpunkt, als sich die Landnutzung von Waldgebiet zu bewirtschaftetem und brach liegendem Land änderte, zu Bodenschädigung geführt. Daher muss die Entwicklung nachhaltiger Landnutzungspraktiken beim Trockenfeldbau im Untersuchungsgebiet verbessert werden, damit in Zukunft der Umfang der Bodendegradation verringert werden kann

Thesis (M.A.)--Georgia State University, 2005.
Title from title screen. Zhi-Yong Yin, committee chair; Paul Knapp, Truman Hartshorn, committee members. Electronic text (135 p. : col. ill., col. maps)) : digital, PDF file. Description based on contents viewed Aug. 2, 2007. Includes bibliographical references (p. 125-133).

Land cover change detection in urban areas provides valuable data on loss of forest and agricultural land to residential and commercial development. Using Landsat 5 Thematic Mapper (1991) and Landsat 7 ETM+ (2000) imagery of Gwinnett County, GA, change images were obtained using image differencing of Normalized Difference Vegetation Index (NDVI), principal components analysis (PCA), and Tasseled Cap-transformed images. Ground truthing and accuracy assessment determined that land cover change detection using the NDVI and Tasseled Cap image transformation methods performed best in the study area, while PCA performed the worst of the three methods assessed. Analyses on vegetative and vegetation changes from 1991- 2000 revealed that these methods perform well for detecting changes in vegetation and/or vegetative characteristics but do not always correspond with changes in land use. Gwinnett County lost an estimated 13,500 hectares of vegetation cover during the study period to urban sprawl, with the majority of the loss coming from forested areas.

Land degradation is one of the most important drivers of landscape change around the globe. This dissertation examines land use-land cover change within a mosaic landscape in Eastern Terai, India, and shows evidence of anthropogenic factors contributing to landscape change. Land use and land cover change were examined within the Alipurduar Subdivision, a representative of the Eastern Terai landscape and the Jaldapara Wildlife Sanctuary, a protected area nested within Alipurduar through the use of multi-temporal satellite data over the past 28 years (1978 - 2006). This study establishes the potential of remote sensing technology to identify the drivers of landscape change; it provides an assessment of how regional drivers of landscape change influence the change within smaller local study extents and provides a methodology to map different types of grassland and monitor their loss within the region. The Normalized Difference Vegetation Index (NDVI) and a Normalized Difference Dry Index (NDDI) were found instrumental in change detection and the classification of different grasslands found inside the park based on their location, structure, and composition. Successful spectral segregation of different types of grasslands and their direct association with different grassland specialist species (e.g., hispid hare, hog deer, Bengal florican) clearly showed the potential of remote sensing technology to efficiently monitor these grasslands and assist in species conservation. Temporal analysis provided evidence of the loss of dense forest and grasslands within both study areas with a considerably higher rate of loss outside the protected area than inside. Results show a decline of forest from 40% in 1978 to 25% in 2006 across Alipurduar. Future trends project forest cover and grassland within Alipurduar to reduce to 15% and 5%, respectively. Within the Alipurduar, deforestation due to growth of tea industry was the primary driver of change. Flooding changed the landscape, but more intensely inside the wildlife preserve. Change of the river course inside Jaldapara during the flood of 1968 significantly altered the distribution of grassland inside the park. Unless, the direction of landscape change is altered, future trends predict growth of the tea industry within the region, increased forest loss, and homogenization of the landscape.

Conselho Nacional de Desenvolvimento Científico e Tecnológico
The monitoring of the use and coverage is very important when studying determined regions, just because it helps knowing the environmental reality and contributes to solve problems that can probably appear. This research was done from the images of ALOS and LANDSAT satellites. Its main objective was to have a multi-temporal analysis of Arroio Grande micro watershed, central region of Rio Grande do Sul. The specific purposes were to identify and to quantify the different classes of land use found in this micro watershed along the 1987, 1998, 2002, 2005, 2007 and 2009 periods, as well as cross the land use information to show the forest coverage changes during the 22 years of analysis. The software SPRING 5.1.7 was employed to classify the supervised images through Bhattacharya, a sorter algorithm, and the map spatial analysis was done through the Spatial Language of Algebraic Geoprocessing program with the same computational application. The classes of land use as forest, field, agriculture, irrigated agriculture, exposed soil and water layer were observed in the images of each year in this analysis. These classes were utilized in the spatial analysis of the forest coverage in which forest monitoring parameters have been defined (forest maintenance and regeneration, deforestation). In this research the principal results that have been noticed were the increase of 17,98% on the distributed forest coverage, mainly in the areas of bigger declination, and the reduction of 16,32% on the field area. The analysis of the spatial forest coverage has presented stability with the landscape, in a gradual progression, because the area of forest maintenance, found in these 22 years, was 12.252,60ha, the forest regeneration was 4.389,12ha and only 1.853,82ha of deforested area.
O monitoramento do uso e cobertura da terra faz-se importante no estudo de determinadas regiões, pois auxilia no conhecimento da realidade ambiental e contribui na busca por soluções de problemas que possam se apresentar. A partir do uso de imagens dos satélites, ALOS e LANDSAT, foi realizada esta pesquisa com o objetivo principal de fazer uma análise multitemporal na microbacia do Arroio Grande, região central do Rio Grande do Sul, cujos objetivos específicos foram: Identificar e quantificar as diferentes classes de uso da terra encontradas na microbacia nos períodos de 1987, 1998, 2002, 2005, 2007 e 2009; além de cruzar as informações de uso da terra, evidenciando a cobertura florestal que sofreu alterações no decorrer dos 22 anos de análise. Para tanto, foram utilizados o software SPRING 5.1.7 para a classificação supervisionada das imagens, com a adoção do algoritmo classificador Bhattacharya, e a análise espacial dos mapas com a programação LEGAL do mesmo aplicativo computacional. Para esta análise, foram observadas as classes de uso do solo: floresta, campo, agricultura, agricultura irrigada, solo exposto e lâmina d água, nas imagens de cada ano. Estas classes foram empregadas na análise espacial da cobertura florestal em que foram definidos parâmetros para o monitoramento florestal (manutenção florestal, regeneração florestal e desmatamentos). Os principais resultados notados, nesta pesquisa, foram o aumento de 17,98% na cobertura florestal distribuída, principalmente, nas áreas de maiores declividade, e a redução de 16,32% sobre a área de campo. Quanto à análise espacial da cobertura florestal, esta mostrou-se em estabilidade com a paisagem, e em gradual progressão, pois a área de manutenção florestal encontrada, nestes 22 anos, foi de 12.252,60 ha, a regeneração florestal foi de 4.389,12 ha e apenas 1.853,82 ha de área desmatada.

The development of modern information technology has enabled collecting data of unprecedented size and complexity. Examples include web text data, microarray & proteomics, and data from scientific domains (e.g., meteorology). To learn from these high dimensional and complex data, traditional machine learning techniques often suffer from the curse of dimensionality and unaffordable computational cost. However, learning from large-scale high-dimensional data promises big payoffs in text mining, gene analysis, and numerous other consequential tasks. Recently developed sparse learning techniques provide us a suite of tools for understanding and exploring high dimensional data from many areas in science and engineering. By exploring sparsity, we can always learn a parsimonious and compact model which is more interpretable and computationally tractable at application time. When it is known that the underlying model is indeed sparse, sparse learning methods can provide us a more consistent model and much improved prediction performance. However, the existing methods are still insufficient for modeling complex or dynamic structures of the data, such as those evidenced in pathways of genomic data, gene regulatory network, and synonyms in text data. This thesis develops structured sparse learning methods along with scalable optimization algorithms to explore and predict high dimensional data with complex structures. In particular, we address three aspects of structured sparse learning: 1. Efficient and scalable optimization methods with fast convergence guarantees for a wide spectrum of high-dimensional learning tasks, including single or multi-task structured regression, canonical correlation analysis as well as online sparse learning. 2. Learning dynamic structures of different types of undirected graphical models, e.g., conditional Gaussian or conditional forest graphical models. 3. Demonstrating the usefulness of the proposed methods in various applications, e.g., computational genomics and spatial-temporal climatological data. In addition, we also design specialized sparse learning methods for text mining applications, including ranking and latent semantic analysis. In the last part of the thesis, we also present the future direction of the high-dimensional structured sparse learning from both computational and statistical aspects.

An integrating approach, including knowledge about whole systems of processes, is essential in order to reach both development and environmental protection goals. In this thesis Geographical Information Systems (GIS) are suggested as a tool to realise such integrated models. The main hypothesis in this work is that several natural technical and social systems that share a time-space can be compared and analysed in a GIS. My first objective was to analyze how GIS can support research, planning, and, more specifically, bring a broad scattering of competence together in an interdisciplinary process. In this process GIS was ivestigated as a tool to achieve models that give us a better overview of a problem, a better understanding for the processes involved, aid in foreseeing conflicts between interests, find ecological limits and assist in choosing countermeasures and monitor the result of different programs. The second objective concerns the requirement that models should be comparable and possible to include in other models and that they can be communicated to planners, politicians and the public. For this reason the possibilities to communicate the result and model components of multidimensional and multi-temporal data are investigated. Four examples on the possibilities and problems when using GIS in interdisciplinary studies are presented. In the examples, water plays a central role as a component in questions about development, management and environmental impact. The first articles focus on non-point source pollutants as a problem under growing attention when the big industrial and municipal point sources are brought under control. To manage non-point source pollutants, detailed knowledge about local conditions is required to facilitate precise advices on land use. To estimate the flow of metals and N(itrogen) in an area it is important to identify the soil moisture. Soil moisture changes over time but also significantly in the landscape according to several factors. Here a method is presented that calculate soil moisture over large areas. Man as a hydrologie factor has to be assessed to also understand the relative importance of anthropogen processes. To offer a supplement to direct measurements and add anthropogen factors, a GIS model is presented that takes soil-type, topography, vegetation, land-use, agricultural drainage and relative position in the watershed into account. A method to analyse and visualise development over time and space in the same model is presented in the last empirical study. The development of agricultural drainage can be discussed as a product of several forces here analyzed together and visualized with help of colour coded "Hyper pixels" and maps. Finally a discussion concerning the physiological and psychological possibilities to communicate multidimensional phenomena with the help of pictures and maps is held. The main conclusions in this theses are that GIS offer the possibilities to develop distributed models, e.g., models that calculate effects from a vide range of factors in larger areas and with a much higher spatial resolution than has been possible earlier. GIS also offer a possibility to integrate and communicate information from different disciplines to scientists, decision makers and the public.

Diss. (sammanfattning) Umeå : Umeå universitet, 1993, härtill 6 uppsatser.

digitalisering@umu

Nous abordons dans ces travaux l’identification non intrusive des charges des bâtiments résidentiels ainsi que la prédiction de leur état futur. L'originalité de ces travaux réside dans la méthode utilisée pour obtenir les résultats voulus, à savoir l'analyse statistique des données(algorithmes de classification). Celle-ci se base sur des hypothèses réalistes et restrictives sans pour autant avoir de limitation sur les modèles comportementaux des charges (variations de charges ou modèles) ni besoin de la connaissance des changements d'état des charges. Ainsi, nous sommes en mesure d’identifier et/ou de prédire l'état des charges consommatrices d'énergie (et potentiellement contrôlables) en se basant uniquement sur une phase d'entrainement réduite et des mesures de puissance active agrégée sur un pas de mesure de dix minutes, préservant donc la vie privée des habitants.Dans cette communication, après avoir décrit la méthodologie développée pour classifier les charges et leurs états, ainsi que les connaissances métier fournies aux algorithmes, nous comparons les résultats d’identification pour cinq algorithmes tirés de l'état de l'art et les utilisons comme support d'application à la prédiction. Les algorithmes utilisés se différencient par leur capacité à traiter des problèmes plus ou moins complexe (notamment la prise en compte de relations entre les charges) et se ne révèlent pas tous appropriés à tout type de charge dans le bâtiment résidentiel
Smart metering is one of the fundamental units of a smart grid, as many further applicationsdepend on the availability of fine-grained information of energy consumption and production.Demand response techniques can be substantially improved by processing smart meter data to extractrelevant knowledge of appliances within a residence. The thesis aims at finding generic solutions for thenon-intrusive load monitoring and future usage prediction of residential loads at a low sampling rate.Load monitoring refers to the dis-aggregation of individual loads from the total consumption at thesmart meter. Future usage prediction of appliances are important from the energy management point ofview. In this work, state of the art multi-label temporal classification techniques are implemented usingnovel set of features. Moreover, multi-label classifiers are able to take inter-appliance correlation intoaccount. The methods are validated using a dataset of residential loads in 100 houses monitored over aduration of 1-year

In recent years, higher education institutes have shifted towards managerial organisational models. Some observers see this as a sign of our neoliberal times, with obsession for rankings, performance indicators and resource allocation. The result is that academic work is more competitive nowadays. Rankings and quantitative analysis of research output are more and more crucial for hiring, promotion and funding allocations. Chapter 2 touches upon these themes and suggests the fruitfulness of cross fertilisation between sociology and science studies. To study this hyper-competitive context, we designed a complex research project to answer different questions regarding multi-faceted aspects of the subject. Our main question was to find what factors drive research collaboration and productivity. These factors are helping some researchers be more successful than others in current evaluation based system. We have employed two sets of data to achieve this goal. One national and one international, both considering the case of sociologists. For individual research productivity measurement and to explore correlates of this productivity (Chapter 3) and macro level policy effect analysis (Chapter 4), we reconstructed the full publication list of all currently hired Italian sociologists on available data. We looked into their research productivity and how they have reacted to the ANVUR national policies by taking into account their embeddedness in different academic contexts. Our aim in Chapter 3 was to explain individual research productivity with organisational embeddedness and we found that male scientists, those working more internationally, and those working with a similar group of coauthors were more productive but not necessarily more cited by other members of the community.In Chapter 4, we analysed the effects of the Italian national research assessment exercise (VQR 2004-2010 by ANVUR) on research productivity and publication behavior of sociologists. Results showed that ANVUR had a limited influence on research productivity. Indeed most differences in individual research productivity of Italian sociologists were due to individual characteristics. Academics who experienced a promotion after 2010 were the most prolific authors.To explore the structural and societal effects on research productivity of sociologists in a more competitive arena at the international level, in Chapter 5, we reconstructed gender, background information and coauthorship networks of all published authors in two top sociology journals, i.e., the American Journal of Sociology (AJS) and the American Sociological Review (ASR). We expected that examining the élite of our community could reveal interesting patterns, especially to understand certain implications of the hyper-competitive academic culture. We found that white male authors affiliated to US institutes were over-represented in these journals. We also found that male authors tended to work more in team and found trace of significant gender and ethnicity penalties. In Chapter 6 we looked into research communities formation and evolution through the time among Italian sociologists. We aimed to investigate if being a member of these communities would inspire different patterns of scientific collaboration among Italian sociologists. We used a sophisticated multi-level design by using temporal community detection. We found the two largest and most stable research communities among Italian sociologists who were political and economic sociologists. We further explored the underlying mechanisms and processes of coauthorship tie existence in multi-level exponential random graph models (ERGMs) trying to take individual, community and macro levels into account in one integrated framework. We found that the collaboration ties were mainly driven by research focus while preferential attachment was also at work and highly prolific researchers attracted further coauthorship ties. In Chapter 7, we conclude by emphasising that academic work has changed drastically in 21st century. Scientific collaboration is a multi-faceted phenomenon and any effort at studying it only with one or two approaches or with one observational unit would yield reductionistic results. That was the main reason behind our effort to investigate this phenomenon from different points of views. Finally, in Appendices Chapter, how to access the data and R and Python scripts developed during this research project is described and an Annotated bibliography on different aspects of academic work is provided.

Cette thèse constitue une analyse de la dynamique du bois mort en rivière par une approche de bilan ligneux multi-scalaire dans une rivière semi-alluviale de région froide: la rivière Saint-Jean, Gaspé. L’étude de la variabilité spatiale et temporelle de la dynamique du bois mort a été réalisée grâce à une approche méthodologique combinant quatre années de terrain et par l’analyse de documents historiques. Les rivières de la péninsule gaspésienne produisent et transportent annuellement de grandes quantités de bois mort. Cette production provient à la fois de la puissance spécifique élevée des cours d’eau et de berges composées de sédiments peu cohésifs et comportant une ripisylve arborescente généralement dense. Jusqu’en 2015, le delta de la rivière Saint-Jean comportait plusieurs embâcles de bois de très grandes tailles. Ces embâcles se mettent en place depuis les années 1960, ils représentent une opportunité unique de quantifier et d’appliquer un budget ligneux pour cerner les variables clefs liées à la dynamique du bois mort à multiples échelles spatiales et temporelles.Nos résultats montrent que la quasi-totalité des bois morts en rivière est produite par la migration latérale et par l’influence de la morphologie. Au niveau des accumulations dans le corridor fluvial, deux zones accumulent la majorité des bois et ces premiers résultats montrent une mobilité importante qui peut varier fortement d'une année à l'autre. Au niveau de la mobilité, l’analyse d’imagerie vidéo de trois évènements différents ont montré que l’intensité (nombre de bois par minute) du transport de bois pouvait être jusqu’à dix fois supérieures durant un évènement avec débâcle mécanique glacielle, comparativement à un évènement hydroclimatique en eau libre.Finalement, nous avons réalisé une analyse par bilans ligneux et une analyse de la trajectoire éco-hydrophormologique sur une période de plus de 50 ans. Nous avons quantifié chaque composante (intrant, accumulée et sortie) d'un budget ligneux à multiples échelles spatio-temporelles. Autant à l’échelle interannuelle, qu’à l’échelle décennale, la dynamique du bois mort comporte des périodes où l’entrée de bois, le stockage et la mobilité diffèrent selon les contextes éco-hydromorphologiques. La trajectoire éco-hydromorphologiques suggère une augmentation de la dynamique fluviale en raison d'un changement significatif dans l'hydrologie conduisant à une plus grande production et mobilité des bois morts et par l'augmentation des volumes accumulés dans le corridor de la RSJ depuis 2004
This thesis is an analysis of the dynamics of large wood in river by a multi-scale large wood budget approach in a semi-alluvial river of cold regions: the Saint-Jean River, Gaspé. The study of the spatial and temporal variability of the dynamics of large wood was carried out through a methodological approach combining four years of field and by analyzing historical documents. The rivers of the Gaspé Peninsula produce annually and carry large amounts of large wood. This production comes from the high specific power of rivers and by banks composed of noncohesive sediment and having a generally dense riparian forest tree. Until 2015, the Delta of the Saint-Jean River had several very large jams. These jams are put in place since the 1960s, they represent a unique opportunity to quantify and apply a wood budget and to identify key variables related to the dynamics of large wood at multiple spatial and temporal scales.Our results show that almost all large wood in river is produced by lateral migration and by the influence of the morphology. For the accumulations in the river corridor, two areas accumulate the majority of wood and these first results show a significant mobility, which can fluctuate substantially from year to year. In terms of mobility, video analysis of three different events showed that the intensity of the transport (number of large wood per minute) can be higher to ten times during an event with mechanical ice-breakup, compared to an open water.hydroclimatic event.Finally, we conducted an analysis by large wood budget and analysis of eco-hydromorphological trajectory over more than 50 years. We quantified each component (input, output and accumulated) of a large wood budget at multiple spatial and temporal scales. At the interannual scale or decadal, scale, the dynamics of large wood have periods when the input, storage and mobility of large wood differ according to eco-hydromorphological contexts. The eco-hydromorphological trajectory suggests an increase in river dynamics due to a significant change in the hydrology, resulting in higher production and mobility of large wood and increased in volumes accumulated in the corridor of the Saint-Jean River since the last decade

Fine grained sediment deposition in urban environments during natural hazard events can impact critical infrastructure and properties (urban terrain) leading to reduced social and economic function and potentially adverse public health effects. Therefore, clean-up of the sediments is required to minimise impacts and restore social and economic functionality as soon as possible. The strategies employed to manage and coordinate the clean-up significantly influence the speed, cost and quality of the clean-up operation. Additionally, the physical properties of the fine grained sediment affects the clean-up, transport, storage and future usage of the sediment. The goals of the research are to assess the resources, time and cost required for fine grained sediment clean-up in an urban environment following a disaster and to determine how the geotechnical properties of sediment will affect urban clean-up strategies. The thesis focuses on the impact of fine grained sediment (<1 mm) deposition from three liquefaction events during the Canterbury earthquake sequence (2010-2011) on residential suburbs and transport networks in Christchurch. It also presents how geotechnical properties of the material may affect clean-up strategies and methods by presenting geotechnical analysis of tephra material from the North Island of New Zealand. Finally, lessons for disaster response planning and decision making for clean-up of sediment in urban environments are presented. A series of semi-structured interviews of key stakeholders supported by relevant academic literature and media reports were used to record the clean-up operation coordination and management and to make a preliminary qualification of the Christchurch liquefaction ejecta clean-up (costs breakdown, time, volume, resources, coordination, planning and priorities). Further analysis of the costs and resources involved for better accuracy was required and so the analysis of Christchurch City Council road management database (RAMM) was done. In order to make a transition from general fine sediment clean-up to specific types of fine disaster sediment clean-up, adequate information about the material properties is required as they will define how the material will be handled, transported and stored. Laboratory analysis of young volcanic tephra from the New Zealand’s North Island was performed to identify their geotechnical properties (density, granulometry, plasticity, composition and angle of repose). The major findings of this research were that emergency planning and the use of the coordinated incident management system (CIMS) system during the emergency were important to facilitate rapid clean-up tasking, management of resources and ultimately recovery from widespread and voluminous liquefaction ejecta deposition in eastern Christchurch. A total estimated cost of approximately $NZ 40 million was calculated for the Christchurch City clean-up following the 2010-2011 Canterbury earthquake sequence with a partial cost of $NZ 12 million for the Southern part of the city, where up to 33% (418 km) of the road network was impacted by liquefaction ejecta and required clearing of the material following the 22 February 2011 earthquake. Over 500,000 tonnes of ejecta has been stockpiled at Burwood landfill for all three liquefaction inducing earthquake events. The average cost per kilometre for the event clean-up was $NZ 5,500/km (4 September 2010), $NZ 11,650/km (22 February 2011) and $NZ 11,185/km (13 June 2011). The duration of clean-up time of residential properties and the road network was approximately two to three months for each of the three liquefaction ejecta events; despite events volumes and spatial distribution of ejecta. Interviews and quantitative analysis of RAMM data revealed that the experience and knowledge gained from the Darfield earthquake (4 September 2010) clean-up increased the efficiency of the following Christchurch earthquake induced liquefaction ejecta clean-up events. Density, particle size, particle shape, clay content and moisture content, are the important geotechnical properties that need to be considered when planning for a clean-up method that incorporates collection, transport and disposal or storage. The geotechnical properties for the tephra samples were analysed to increase preparedness and reaction response of potentially affected North Island cities from possible product from the active volcanoes in their region. The geotechnical results from this study show that volcanic tephra could be used in road or construction material but the properties would have to be further investigated for a New Zealand context. Using fresh volcanic material in road, building or flood control construction requires good understanding of the material properties and precaution during design and construction to extra care, but if well planned, it can be economically beneficial.

Esta pesquisa de doutorado faz uma avaliação correlacional entre a evolução espacial da rede de monitoramento de qualidade das águas interiores do Estado de São Paulo, através do número de pontos de amostragem e sua respectiva densidade espacial ao longo dos 30 anos de existência da mesma e o crescimento populacional, através da densidade populacional dos municípios inseridos nas 22 unidades de gerenciamento de recursos hídricos do Estado de São Paulo. Analisa também se estes pontos mantiveram a capacidade de representar o status da qualidade da água em função do crescimento populacional, e se é necessário expandir ou adensar a rede em determinadas regiões. Esta comparação ficou mais racional com o uso dos recursos das geotecnologias e da análise multicritério aplicada ao planejamento e gerenciamento de recursos hídricos, com a construção do SIG SP_WATERNET e através da criação de um índice de avaliação do monitoramento que relaciona as 22 unidades espacialmente e ao longo do período estudado. Este índice pode ressaltar o grau de abrangência e de vulnerabilidade da rede de monitoramento das águas interiores superficiais no Estado de São Paulo.
This Ph.D. research makes a correlational evaluation between the spatial evolution of the monitoring network of inland surface waters of State of São Paulo, through the number of sampling points and its density throughout the 30 years of existence of same and the population growth (urbanization) and the population density in the 22 units of water management units of the São Paulo State. It also analyzes if these points had kept the capacity to represent the status of the quality of the water in function of the population growth, and if it is necessary to expand or to rearrange the network in some regions. This comparison was more rational with the use of the resources of the geo-information applied for water resources planning and management, with the construction of SIG SP_WATERNET and with the proposal of an multi-criteria evaluation monitoring index concerning the 22 water management units throughout a studied period. This index can point out the coverage or the vulnerability of the monitoring efficiency of the superficial waters network of São Paulo State.

Au 20ème siècle, les massifs montagneux, dont les Alpes, ont connu un réchauffement significatif avec une augmentation des températures deux fois plus importante que la moyenne mondiale. Un tel réchauffement altère les composantes de la cryosphère. Elle induit, par exemple, un passage des précipitations solides aux précipitations liquides, des phases de fonte des neiges plus fréquentes et plus intenses, ainsi qu’une forte diminution de la quantité de neige et une réduction de la durée de la couverture neigeuse. Aux horizons 2050–2100, les modèles climatiques prévoient que l'épaisseur du manteau neigeux sera considérablement réduite et que les propriétés de la neige, et notamment la stabilité du manteau neigeux, seront modifiées. Ces changements devraient entraîner des modifications importantes dans l'activité des avalanches. Parallèlement, l’afforestation induite par la déprise agro-sylvo-pastorale, la démocratisation des sports d'hiver et l'urbanisation des versants ont profondément modifié les paysages de montagne depuis le milieu du 18ème siècle, de même que l'exposition des individus. Dans ce contexte, une documentation précise de l'activité passée des avalanches est cruciale pour mettre en évidence et comprendre les impacts du réchauffement climatique sur l'activité avalancheuse. Jusqu'à présent, cette documentation s'appuyait sur des chroniques historiques ou des observations systématiques. Cependant, les premières sont souvent discontinues et axées sur des événements catastrophiques, tandis que les secondes se limitent à la seconde moitié du 20ème siècle, excluant toute comparaison avec des périodes climatiques distinctes – tel que les phases froides du Petit Âge Glaciaire, par exemple. Sur les versants forestiers, l'approche dendrogéomorphique apparaît être un complément fiable aux archives historiques et aux séries d'observations systématiques, car elle permet de reconstruire l'activité passée des avalanches, en continu, à l‘échelle des plusieurs siècles, avec une résolution annuelle. Pourtant, jusqu'à présent, même si de nombreuses reconstructions locales ont été proposées, la fiabilité de l’approche a été peu souvent analysée et aucune chronologie régionale – cruciale pour distinguer les interférences potentielles entre l'activité des avalanches, les fluctuations climatiques et les changements socio-économiques – n'a été développée dans les Alpes. Dans cette thèse de doctorat, des avancées méthodologiques significatives ont été réalisées afin (1) d’améliorer la détection des avalanches dans les cernes de croissance, (2) d’éliminer les non-stationnarités liées à la diminution du nombre d'arbres au cours du temps dans les reconstructions et (3) d’agréger les reconstructions locales en chronologies régionales. Sur la base de ces développements, des chronologies régionales pluriséculaires homogénéisées ont été développées (4) pour 10 couloirs d’avalanche de la vallée de Goms (Valais, Alpes suisses, 1880-2014) et (5) 11 couloirs du massif du Queyras (Alpes françaises, 1560-2016). Ces dernières ont été confrontées aux fluctuations climatiques et aux changements d'occupation du sol. À Goms, l'absence de signal climatique clair dans la chronologie régionale souligne les interférences induites par les non-stationnarités locales et démontre qu’une stratégie d'échantillonnage à l'échelle régionale devra nécessairement constituer un préalable au développement d’une chronologie robuste. Dans le massif du Queyras, la forte diminution de l'activité avalancheuse observée au cours du 20ème siècle est attribuée au réchauffement climatique et au processus d’afforestation des versants
For the 20th century, high mountain areas, such as the Alps, have undergone a significant warming with temperature increase twice as much as the global average. Such warming strongly alters the cryosphere components. It induces, for example, a shift from solid to liquid precipitation, more frequent and more intense snowmelt phases or a strong decrease in the amount and duration of snow cover, especially at the elevation of the snowrain transition. In the future, climate models forecast that snow depth will be significantly reduced and that snow properties such as snow stability will be modified. These changes in snow cover characteristics and amounts are expected to induce significant changes in snow avalanches activity. At the same time, afforestation induced by the abandonment of agriculture and grazing, the democratization of winter recreation activities and the urban sprawl on the slopes have strongly modified the mountain landscapes since the mid-18th century as well as the exposition of individuals to snow avalanches. In this context, a precise documentation of past snow avalanche activity is crucial to decipher and to understand the impacts of the undergoing climate warming on the snow avalanche activity. To date, this documentation usually relies on historical chronicles or systematic observations. However, the firsts are often discontinuous and focused on catastrophic events. The seconds are limited to the second half of the 20th century thus precluding a comparison from climatically distinct period. On forested paths, the dendrogeomorphic approach is theoretically a reliable approach complement to historical archives and series of systematic observation to infer past snow avalanche activity. Yet, so far, the robustness of this approach has been poorly questioned and no regional chronology, crucial to disentangle potential interferences between snow avalanche activity, climate fluctuations and socio-economic changes, has been developed in the Alps. In this PhD thesis proposes new methodological frameworks to (1) detect avalanche events from tree-ring series, (2) remove non-stationarities related to the decreasing number of trees over time in the reconstruction and (3) aggregate locals reconstructions in regional chronologies. Based on these development, homogenized multicentennial regional chronologies developed (4) for 10 paths of the Goms Valley (Valais canton, Swiss Alps, 1880-2014) and (5) 11 paths from the Queyras Massif (French Alps, 1560-2016) are confronted to climatic fluctuations and land use changes. At Goms, the absence of clear climatic signal in the regional chronology evidence the interference with local nonstationarities and question the need for a sampling strategy at the regional scale to create a robust chronology. In the Queyras massif, the strong decrease of avalanche activity observed over the 20th century is attributed to global warming and to the afforestation process

S tím, jak se neustále vyvíjejí nové technologie pro energeticky náročná průmyslová odvětví, stávající zařízení postupně zaostávají v efektivitě a produktivitě. Tvrdá konkurence na trhu a legislativa v oblasti životního prostředí nutí tato tradiční zařízení k ukončení provozu a k odstavení. Zlepšování procesu a projekty modernizace jsou zásadní v udržování provozních výkonů těchto zařízení. Současné přístupy pro zlepšování procesů jsou hlavně: integrace procesů, optimalizace procesů a intenzifikace procesů. Obecně se v těchto oblastech využívá matematické optimalizace, zkušeností řešitele a provozní heuristiky. Tyto přístupy slouží jako základ pro zlepšování procesů. Avšak, jejich výkon lze dále zlepšit pomocí moderní výpočtové inteligence. Účelem této práce je tudíž aplikace pokročilých technik umělé inteligence a strojového učení za účelem zlepšování procesů v energeticky náročných průmyslových procesech. V této práci je využit přístup, který řeší tento problém simulací průmyslových systémů a přispívá následujícím: (i)Aplikace techniky strojového učení, která zahrnuje jednorázové učení a neuro-evoluci pro modelování a optimalizaci jednotlivých jednotek na základě dat. (ii) Aplikace redukce dimenze (např. Analýza hlavních komponent, autoendkodér) pro vícekriteriální optimalizaci procesu s více jednotkami. (iii) Návrh nového nástroje pro analýzu problematických částí systému za účelem jejich odstranění (bottleneck tree analysis – BOTA). Bylo také navrženo rozšíření nástroje, které umožňuje řešit vícerozměrné problémy pomocí přístupu založeného na datech. (iv) Prokázání účinnosti simulací Monte-Carlo, neuronové sítě a rozhodovacích stromů pro rozhodování při integraci nové technologie procesu do stávajících procesů. (v) Porovnání techniky HTM (Hierarchical Temporal Memory) a duální optimalizace s několika prediktivními nástroji pro podporu managementu provozu v reálném čase. (vi) Implementace umělé neuronové sítě v rámci rozhraní pro konvenční procesní graf (P-graf). (vii) Zdůraznění budoucnosti umělé inteligence a procesního inženýrství v biosystémech prostřednictvím komerčně založeného paradigmatu multi-omics.

碩士
國立彰化師範大學
地理學系
106
Land degradation and climate change are global issues in recent years. There are many landslides in Taiwan due to its fragile geology, abundant rainfall and high population density. The frequency of extreme precipitation and the rainfall intensity are increasing in Taiwan due to the global climate change. It may result in mudslide and landslides. Landslides may become high-risk areas where productivity decline. This study discusses whether land degradation trigged by landslide occurs in Taiwan or not. Meanwhile, we will discover the factors that affect the landslides. This study analyzed land degradation through satellite images, geographic information systems(GIS), and Normalized Difference Vegetation Index(NDVI). It’s defined as the stable area where the change rate of NDVI is between -0.05~0.05. Besides, those are the degraded area where the decrease rate of NDVI is more than 0.05 per ten years. At last, we will discuss the differences of environmental factors among each research areas. The results show that the landslides often occur in the Sansia Fm, the windward slope, and the elevation below 2000 meters. Landslide often occurred in these area after earthquakes and typhoons disturbance. The land become bare after disasters, but vegetation will recover in two years. It means that the environment of Taiwan mountain area is highly resilient.

Tese de Doutoramento em Informática
Forest and agriculture ecosystems are prone to disturbances caused by human action or natural effects. For instance, climate change is projected to be a key influence on vegetation across the globe. Regarding agriculture, primary climate vectors with a significant impact include temperature, moisture stress, and radiation. Within this context, it is of foremost importance to monitor crops along time, as well as to detect pests, diseases, assess and control irrigation demands. Regular monitoring activities will enable timely measures that may trigger field interventions that are used to preserve health status of crops, achieving both time and economic gains, while assuring a more sustainable activity. Within this scope, precision agriculture (PA) techniques appear as an effective alternative to the traditional agronomy practices. In fact, the technological advances that promote PA are able to enhance support when making decisions, resulting in agronomical processes upgraded by employing site or plant specific management operations. In this regard, the capabilities of unmanned aerial vehicles (UAVs) to provide flexible, efficient, non-destructive, and non-invasive means of acquiring data on agricultural crops and the various agro-environmental factors of the parcel, can be used for PA applications. The high- temporal, radiometric and spatial resolutions achieved by UAV-based aerial imagery make possible to foresee new and important advances in PA practices. In this study it is presented the development of a management support system for the agriculture and forestry sectors, based on the analysis of multi-temporal data obtained through different sensors coupled to UAVs. With a continuous monitoring, it is intended to monitor the vegetative development and to identify, in an early and (semi)automatic way, potential issues, allowing their localized mitigation, through methodologies and algorithms developed for this purpose. To meet these main objectives, two important agricultural crops from the region of Trás-osMontes and Alto Douro (Portugal) economy, were identified: the grapevine (Vitis vinifera L.); and the European chestnut (Castanea Sativa Mill.). Both of these crops have a high socioeconomic relevance for the population of this region and represent an important share of national production. Thus, the work is divided into two parts, one focuses on monitoring chestnut stands and the other focuses on vineyards. The several differences among these two species in the planting typology and their geometry, make the approaches to each of the sectors also different. However, this fact will allow the adaptation of the proposed methodologies to almost all agricultural species, regardless of the type and the way they are arranged, in a grid or in rows. Although there are several approaches to detect and monitor vegetation through aerial imagery, most of them remain dependent of manual extraction of vegetation parameters. This work presents automatic methods that allow—with none or few parametrization—the individual detection of the trees/grapevines and their multi-temporal analysis. The approach for tree detection was applied to several chestnut stands, allowing the automatic estimation of several parameters, such as the number of trees, the canopy coverage, tree height, and crown diameter. A novel methodology that enables the identification of phytosanitary issues from multitemporal analysis of chestnut stands, using UAV-based multispectral imagery, was also developed and it is presented in this thesis. This approach not only allows the absence or presence of phytosanitary issues but also the identification and the classification of biotic or abiotic factors affecting the trees. The developed methodology proved to be effective in automatically detecting and classifying phytosanitary issues in chestnut trees throughout the growing season. Likewise, methods to automatically estimate and extract grapevine vegetation parameters are also proposed. A full pipeline for vineyards management was developed. First, a methodology able to differentiate grapevine canopy between inter-row vegetation cover and soil, and to identify independent vine row was built. Then, the outputs were provided but the former methods were used to create a multi-temporal data analysis of vineyards, enabling the monitoring of vegetation dynamics of a given vineyard plot along the growing season. This way, areas with canopy management operation needs, and with different vigour levels, are identified. The approaches proposed enable to fully exploit the advantages offered from the UAV-based multi-sensor data (RGB, multispectral and thermal infrared), by performing multitemporal analysis of vineyards both at the plot and at the plant scales. Individual grapevine detection permits the estimation of geometrical and biophysical parameters, as well as missing grapevine plants. Thus, the developed methodologies proved to be very effective and can be used in a single epoch, analyzing the data from one individual flight campaign to estimate different parameters (depending on the used sensors), both at parcel-level and at the plant-level. In terms of agricultural plot, the canopy coverage, the estimation of the number of trees/grapevines, and the estimation of other vegetation and bare soil can be reached, as well as mean values of the species under analysis. Regarding the plant-level monitoring, geometrical and biophysical parameters as height, canopy volume, crown diameter, temperature and vegetation indices that correlate with yield, biomass, leaf density and phytosanitary issues are also possible to estimate. Combining data from different flight campaigns, allows a multi-temporal analysis to be performed. Moreover, this multi-temporal analysis can be carried out over a single vegetative cycle and/or over different agricultural years, allowing, in any case, to obtain important management information. Hence, the original methods presented in this work have shown to be effective and have proved that their potential goes beyond vegetation detection, since they can be employed in an operational routine for the automatic monitoring of vineyard plots and chestnut stands. Thus, this work can be seen as an important contribution towards the substitution of time-consuming and costly field campaigns for managing plantations in a quicker and more sustainable way.

碩士
國立臺灣海洋大學
河海工程學系
107
Coastal areas are highly variable and fragile, and are susceptible to natural or human factors, resulting in changes in topography and geomorphology. Therefore, rapid mastery and long-term accumulation of topographic information are important for exploring terrain change mechanisms. In recent years, with the advancement of the UAV photogrammetry algorithm, it is possible to rapidly produce a high-precision, high-resolution numerical surface model, which can compensate for the mobility of the general-station theodolite on the coast monitoring and consumes a large labor cost. In this study, by applying the monitoring data of the UAV for many periods, the image feature points are matched by the aerial photogrammetry image technology and the aerial triangulation method to obtain the relative soil sandographic point, and the virtualized base station is used for the networked real-time dynamic positioning. The ground control point measurement coordinates are used to obtain the actual coordinates of the earth and sand to obtain the actual earth sand topographic points, and then compare the image matching point cloud with the direct measurement results. It is known that the volume change of the intrusion and the relationship between the coastal power and the spatial distribution of the intrusion are integrated through the geographic information system. The results of the topographic changes in Yancheng Beach for many periods show that the variation of the 0m line from May 20th to May 2019 in Yancheng Beach is between -24.82 and 24.49 meters, and the weekly change rate is between -3.05 and 1.29. Between m/week. The results of this study show that the application of UAV imaging technology can improve the efficiency of traditional manual sampling and reduce the cost of using indirect measurement observations. It can reduce the error in measurement and save the cost of field measurement. It can help to understand the changes of coastal terrain through multi-period topographic data. Characteristics, in a short period of time to understand the changes in terrain caused by waves, currents, tides and other seas, the spatial differences in the spatial changes.

abstract: Thousands of high-resolution images are generated each day. Detecting and analyzing variations in these images are key steps in image understanding. This work focuses on spatial and multitemporal visual change detection and its applications in multi-temporal synthetic aperture radar (SAR) images. The Canny edge detector is one of the most widely-used edge detection algorithms due to its superior performance in terms of SNR and edge localization and only one response to a single edge. In this work, we propose a mechanism to implement the Canny algorithm at the block level without any loss in edge detection performance as compared to the original frame-level Canny algorithm. The resulting block-based algorithm has significantly reduced memory requirements and can achieve a significantly reduced latency. Furthermore, the proposed algorithm can be easily integrated with other block-based image processing systems. In addition, quantitative evaluations and subjective tests show that the edge detection performance of the proposed algorithm is better than the original frame-based algorithm, especially when noise is present in the images. In the context of multi-temporal SAR images for earth monitoring applications, one critical issue is the detection of changes occurring after a natural or anthropic disaster. In this work, we propose a novel similarity measure for automatic change detection using a pair of SAR images acquired at different times and apply it in both the spatial and wavelet domains. This measure is based on the evolution of the local statistics of the image between two dates. The local statistics are modeled as a Gaussian Mixture Model (GMM), which is more suitable and flexible to approximate the local distribution of the SAR image with distinct land-cover typologies. Tests on real datasets show that the proposed detectors outperform existing methods in terms of the quality of the similarity maps, which are assessed using the receiver operating characteristic (ROC) curves, and in terms of the total error rates of the final change detection maps. Furthermore, we proposed a new similarity measure for automatic change detection based on a divisive normalization transform in order to reduce the computation complexity. Tests show that our proposed DNT-based change detector exhibits competitive detection performance while achieving lower computational complexity as compared to previously suggested methods.
Dissertation/Thesis
Doctoral Dissertation Electrical Engineering 2014

博士
國立中央大學
太空科學研究所
105
Drought indices can be used to evaluate drought detection using meteorological measurements data of the temperature and precipitation. Moreover, the satellite-based data provides spatial and temporal patterns for the regional-scale drought occurrences. This dissertation is to investigate the drought detection in relation to climatic condition over Mongolia by using satellite remote sensing imagery, which was acquired by the Moderate Resolution Imaging Spectroradiometer (MODIS). The drought index was evaluated from the MODIS data acquired during May to August from 2000 to 2013 using the Drought Severity Index-2 (DSI2) and Integrated Drought Severity Index (IDSI) methods. These indices were empirically calculated by standardized characteristics of the MODIS two-band Enhanced Vegetation Index (EVI2), Land Surface Temperature (LST), Evapotranspiration (ET), and Potential Evapotranspiration (PET) data. DSI is based on the monthly standardized ET/PET ratio and EVI2 index. The modification of DSI2, IDSI was calculated by standardization of the sum of separately monthly standardized ET/PET ratio and EVI2/LST ratio. Consequently, the ratio between EVI2 and LST was calculated by parameter features and integrated into the DSI2. In addition, fourteen-year summer monthly data for air temperature, precipitation, and soil moisture content of in-situ measurements data from the meteorological and agricultural stations were analyzed. The climatological variables anomaly of in situ measurements was also calculated by standardized anomaly to compare to the DSI2 and IDSI at the eighteen stations. The multi-temporal of all MODIS data were processed using supervised classification. A standardized anomaly method was also calculated by both MODIS and in situ measurement data. Therefore, the linear spectral mixture analysis (LSMA) and the threshold value of change vector analysis (CVA) were used for drought-indices classes. A statistical analysis and Pearson correlation coefficients (r) for the DSI2 versus the climatological anomaly and the IDSI versus the climatological anomaly were computed for the study period. From the standardized anomaly analysis of in situ measurements, it was shown that the wettest years were 2003 and 2011–2013, while the driest years were 2001, 2002, 2007, and 2009; the rest of the years were normal years. Generally speaking, dry weather implies lower rainfall and higher temperature, so that drought occurred in the years 2002 and 2007. By contrast, wet weather accompanies higher precipitation and lower temperature, such as the years 2003, 2012, and 2013. For the improvement of the parameters of DSI that is the ratio between MODIS EVI2 and LST, the results showed that the vegetation-temperature feature space was well-defined. This indicated a wide range of surface wetness and drought in the study area. The validation results of EVI2/LST ratio were carried out by comparing EVI2/LST values with monthly rainfall throughout the study area. The comparison results were revealed with good agreement and sensitivity between EVI2/LST ratio and rainfall data. Moreover, ET/PET ratio results found that the relationship between the ET/PET ratio and precipitation has a similar variation in different conditions. It is indicating that the ET/PET ratio reveals a good parameter for detecting wet and drought conditions. The comparison results between DSI2 and IDSI demonstrated that the IDSI gave slightly better classification results than the DSI2. The modification of DSI2 results was found that IDSI dynamics revealed the spatiotemporal occurrence of dry (2001, 2002, 2007 and 2009) and wet (2003 and 2011–2013) periods as shown in time series analysis of in situ measurements. From a detailed spatial analysis of IDSI dynamics, it was found that the wettest and drought occurred in 2003 and 2007 and occupied the largest region of the study area by about 60% and 67% as compared to other years. The relationships between remotely sensed and in situ based data indicated that the correlation for IDSI versus climatological anomaly is higher than DSI2 versus climatological anomaly. Correlation coefficients obtained over the eighteen measurement stations between the IDSI and climatological anomaly (r = 0.84) show a good agreement between the satellite-derived and measured anomalies. This dissertation has demonstrated merits of using MODIS data for studying drought variability in relation to climatic characteristics, and is important for drought monitoring in agricultural management and development, and one of an input parameter for drought.

碩士
國立臺灣海洋大學
海洋事務與資源管理研究所
102
Fisheries resources are important food sources for humans. Due to global change and overfishing caused the decreasing of fisheries resource. Marine Protection Areas (MPA) and other conservation measures are considered as important tools for marine resource conversation by the IUCN (International Union for Conservation of Nature and Natural Resources) and UNEP (United Nations Environment Programme). Northern coastal waters of Taiwan have abundant marine resource and considered as important fishing grounds. However, there are limited conservation measures in this area. Therefore, this study will be focus on the biodiversity of Taiwan's northern waters fishery species, and propose conservation management recommendations.This study area is between 120。E~127。E；24。N~31。N. Logbooks of 754 vessels from January to December 2010 were collected for analysis. The fishing effort data and CPUE (CPUE，kg/voyage) catch data were estimated by 0.25*0.25 degree. The Shannon’s index was calculated by the catch amount by species. Seasonal catch/effort and biodiversity index value were drawn by Mapinfo. Preliminary results showed scombridae, other scads, other skipjacks, oceanic, neritic squids bonito, skipjack tuna are the major fisheries in northern Taiwan. Catch in furst quarter was the highest, followed by fourth quarter , second quarter, and first quarter. Major target species includes mackerels, oceanic bonito, neritic, squids, dorado, hairtail, skipjack tuna etc. Nevertheless, species diversity were higher in the summer within 122。E~124。E;24。N~26。N and the autumn within 122。E~123。E; 26。N~27。N. This study indicated that the amount of catch and species diversity varied by season. In conclusion, it is recommended fishery knowledge of conservation measures such as ecosystem-based management approach should be considered in the summer and autumn. Thus, there are four mainly complements blew context (1) gilled network fisheries activities should be monitored and limited fishing days if necessary. (2) High fishing effort, catch, species, CPUE, and species diversity should be deemed to be fishing hotspots and take more conservation measures. (3) The government and fishermen should work together to achieve consensus on the conservation of fisheries resources and to strengthen cooperation with fishermen fishery resources, thereby effectively control the temporal characteristics of various types of coastal waters along the northern Taiwan.

碩士
明新科技大學
土木工程與環境資源管理系碩士班
102
To obtain a large area of high resolution digital elevation model (DEM) in a short period of time by using LiDAR systems are nowadays largely used for quantitative analyses and modeling in geology, coastal erosion, and geomorphology. High-quality DEMs are required for the accurate morphometric and volumetric measurement of land features. Terrain changes due to changes in surface topography has been an important issue in many research areas, including land reclamation, orogeny, shoreline change, river bed change, land subsidence, landslides and erosion. In this study, two DEM datasets taken on 2005 and 2011 in Turtle island, an active volcano island in Taiwan, are used to derive shadow relief maps, differential DEM map, slope map, aspect map and for PIV analysis. Subsequently the change analysis is performed. Results of change analysis show that a fault feature occurred in the northern slope surface of Turtle island, and the rock falls at coastal cliff tends to be spreading out at the turtle head. Moreover, the deep-cut geomorphometric feature can also be found after coastal erosion. The observations of these results can be inferred that the morphometric features can be clearly recognized using the terrain data produced form the LiDAR systems. Therefore, good results of coastal change erosion can be achieved using multi-temporal LiDAR data. Further studies can be conducted to improve a quantitive estimation of the change volumes through thorough understanding of problem of sensor characteristics, parameters of data acquisition, survey datum, and others.

碩士
國立中興大學
土木工程學系所
100
It is to state the Different Period of Construction Zone Monitoring Network with Different Adjustment Mode Result Analysis. We will propose the better method while proceeding the studying in details onto this analysis. Finally we will choose the best solution as our conclusion. The study found that the first reference to the classic fixed adjustment，T-test after test to find the displacement point I point a settlement 14.4mm，Not yet reached the value of building subsidence security management point of warning value，But its obvious deformation better reflects the purpose of construction safety management。

Dissertation submitted in partial fulfilment of the requirements for the degree of Master of Science in Geospatial Technologies
In recent years, data mining algorithms are increasingly applied to optimise the classification process of remotely sensed imagery. Random Forest algorithms have shown high potential for land cover mapping problems yet have not been sufficiently tested on their ability to process and classify multi-temporal data within one classification process. Additionally, a growing amount of geospatial data is freely available online without having their usability assessed, such as EUROSTAT´s LUCAS land use land cover dataset. This study provides a comparative analysis of two land cover classification approaches using Random Forest on open-access multi-spectral, multi-temporal Sentinel-2A/B data. A classification system composed of six classes (sealed surfaces, non-vegetated unsealed surfaces, water, woody, herbaceous permanent, herbaceous periodic) was designed for this study. Ten images of ten bands plus NDVI each, taken between November 2016 and October 2017 in Central Portugal, were processed in R using a pixel-based approach. Ten maps based on single month data were produced. These were then used as input data for the classifier to create a final map. This map was compared with a map using all 100 bands at once as training for the classifier. This study concluded that the approach using all bands produced maps with 11% higher, yet overall low accuracy of 58%. It was also less time-consuming with about 5 hours to over 15 hours of work for the multi-temporal predictions. The main causes for the low accuracy identified by this thesis are uncertainties with EUROSTAT´s Land Use/Cover Area Statistical Survey (LUCAS) training data and issues with the accompanying nomenclature definition. Additional to the comparison of the classification approaches, the usability of LUCAS (2015) is tested by comparing four different variations of it as training data for the classification based on 100 bands. This research indicates high potential of using Sentinel-2 imagery and multi-temporal stacks of bands to achieve an averaged land cover classification of the investigated time span. Moreover, the research points out lower potential of the multi-map approach and issues regarding the suitability of using LUCAS open-access data as sole input for training a classifier for this study. Issues include inaccurate surveying and a partially long distance between the marked point and the actual observation point reached by the surveyors of up to 1.5 km. Review of the database, additional sampling and ancillary data appears to be necessary for achieving accurate results.

碩士
國立中央大學
遙測科技碩士學位學程
105
Tropical forests represent the terrestrial ecosystem with the highest biodiversity. Despite the importance of tropical forest to human species survival, these are frequently at risk, mainly because of the dynamic use of the land. The objective of this study is to assess and analyze the deforestation process in the nucleus zone of the Bosawas Biosphere Reserve, in Nicaragua, Central America, by means of remote sensing image classification and interpretation. Landsat images from 2011, 2015 and 2016 were classified using an object-based approach. The resulting land-cover classification maps were then utilized to conduct a deforestation assessment, which revealed a loss in forest cover, going from 6206.77 km2 in 2011, to 5825.14 km2 in 2015. Moreover, with applying GIS (Geographic Information System), this study incorporated environmental factors into the Maximum Entropy model (MaxEnt) to assess areas most susceptible to deforestation, including clear-cut and vegetated areas. Deforestation patches from 2015-2016 were used to validate the models. The Area Under the Curve of Receiver Operating Curve (AUC) for the clear-cut patch model was 0.71, and 0.88 for the vegetated patch model. The results indicate that MaxEnt is a reliable method to identify the areas more likely to be affected by deforestation within the Bosawas nucleus zone. Results show that distance to soil and distance to other types of vegetation are the most important variables for the clear-cut and vegetated patch models, respectively. This indicates that the two types of patches have different deforestation dynamics because they represent different stages of the deforestation process. Clear-cut patches belong to the early stages of deforestation, they are clustered and diffuse, and their appearance can be related to the availability of purchasable lands in the area. On the other side, vegetated patches represent more advanced phases of the deforestation process, and they tend to expand through the years. In both models, however, variables including distance to communities, roads and rivers do not show significant contribution to deforestation, meaning that accessibility is not a vital factor in the deforestation process of Bosawas. In addition to analyze the relationship between deforestation and environmental factors, to study the characteristics of the patches (e.g. morphology, type, evolution) is also suggested an important approach to broaden the understanding of the complex issue of deforestation in tropical rainforest.

Natural hazard related to the volcanic activity represents a potential risk factor, particularly in the vicinity of human settlements. Besides to the risk related to the explosive and effusive activity, the instability of volcanic edifices may develop into large landslides often catastrophically destructive, as shown by the collapse of the northern flank of Mount St. Helens in 1980. A combined approach was applied to analyse slope failures that occurred at Stromboli volcano. SdF slope stability was evaluated by using high-resolution multi-temporal DTMMs and performing limit equilibrium stability analyses. High-resolution topographical data collected with remote sensing techniques and three-dimensional slope stability analysis play a key role in understanding instability mechanism and the related risks. Analyses carried out on the 2002–2003 and 2007 Stromboli eruptions, starting from high-resolution data acquired through airborne remote sensing surveys, permitted the estimation of the lava volumes emplaced on the SdF slope and contributed to the investigation of the link between magma emission and slope instabilities. Limit Equilibrium analyses were performed on the 2001 and 2007 3D models, in order to simulate the slope behavior before 2002-2003 landslide event and after the 2007 eruption. Stability analyses were conducted to understand the mechanisms that controlled the slope deformations which occurred shortly after the 2007 eruption onset, involving the upper part of slope. Limit equilibrium analyses applied to both cases yielded results which are congruent with observations and monitoring data. The results presented in this work undoubtedly indicate that hazard assessment for the island of Stromboli should take into account the fact that a new magma intrusion could lead to further destabilisation of the slope, which may be more significant than the one recently observed because it will affect an already disarranged deposit and fractured and loosened crater area. The two-pronged approach based on the analysis of 3D multi-temporal mapping datasets and on the application of LE methods contributed to better understanding volcano flank behaviour and to be prepared to undertake actions aimed at risk mitigation.

Visual analytics of large amounts of spatio-temporal data is challenging due to the overlap and clutter from movements of multiple objects. A common approach for analyzing such data is to consider how groups of items cluster and move together in space and time. However, most methods for showing Spatio-temporal Cluster (STC) properties, concentrate on a few dimensions of the cluster (e.g. the cluster movement direction or cluster density) and many other properties are not represented. Furthermore, while representing multiple attributes of clusters in a single view existing methods fail to preserve the original shape of the cluster or distort the actual spatial covering of the dataset. In this thesis, I propose a simple yet effective visualization, FlockViz, for showing multiple STC data dimensions in a single view by preserving the original cluster shape. To evaluate this method I develop a framework for categorizing the wide range of tasks involved in analyzing STCs. I conclude this work through a controlled user study comparing the performance of FlockViz with alternative visualization techniques that aid with cluster-based analytic tasks. Finally the exploration capability of FlockViz is demonstrated in some real life data sets such as fish movement, caribou movement, eagle migration, and hurricane movement. The results of the user studies and use cases confirm the advantage and novelty of the novel FlockViz design for visual analytic tasks.

碩士
國立交通大學
應用化學系所
95
Flow injection analysis (FIA) is widely used as an automatic analysis technology. The FIA signals often appear in asymmetric tailing peaks. According to the traditional theory, the tailing phenomenon occurs because the fluid moves in the form of laminar flow that results in dispersion in the FIA manifolds. Furthermore, the defects of the hardware itself also contributes to the fluid change, thus to the skewness of the peak shape. Recently a hypothesis of temporal effect is proposed to explain the tailing phenomenon. Because the spatial distribution of the sample in the FIA manifolds changes with time, the recorded signal from a traditional single-channel detector is actually the accumulation result of the constant signal variation of the sample in the manifold. Accordingly, the recorded signal as a function of time is different from the true spatial distribution of the sample in the manifold. This phenomenon results in tailing peak, even though the spatial distribution of the sample appears considerably symmetric. This research scrutinizes the spatial distribution of the sample in the FIA manifolds and the peak shape recorded as a function of time simultaneously using a multi-channel detector. Obvious difference is observed by comparing peak shapes between spatial and temporal profiles. This study confirms that the temporal effect is the major factor for the skewed FIA peaks. In addition, the theoretical foundation of the temporally convoluted Gaussian equation proposed in the literature is verified. Furthermore, the method for calculating the dispersion coefficient using the temporal signal is validated by the experimental spatial signal.

Dissertation presented as the partial requirement for obtaining a Master's degree in Geographic Information Systems and Science
Realizou-se uma investigação sobre a aplicação do Modelo de Mistura Linear (MML) numa série multi-temporal. Depois de aplicado a uma sequência anual completa de imagens MODIS com 500 metros de resolução especial, o MML gerou um conjunto de imagens fracção com uma variação intra-anual da abundância de componentes puros vegetação, solo e sombra na mistura do pixel. No pré-processamento das imagens de MOD09A1 utilizou-se uma abordagem do Compósito do Valor Máximo, para obter os dados de base para input do modelo de MML. Baseado em 36 imagens fracção, caracterizaram-se 304 amostras, correspondendo cada amostra a uma de onze classes de ocupação do solo definidas neste estudo. Estabeleceu-se uma relação entre estas classes e o CLC2000. Para a classificação das unidades de ocupação do solo, baseadas na variação do perfil temporal dos componentes puros no pixel das amostras, recorreu-se à utilização de uma rede neuronal de mapas autoorganizados (SOM). A análise da qualidade dos resultados resultou da construção da matriz de convergência, sendo calculados os índices de exatidão global, do produtor e do utilizador para avaliação de resultados. O trabalho reporta uma análise descritiva dos resultados obtidos segundo a metodologia proposta e apresenta-se como um estudo preliminar para análise de mistura espectrais, numa vertente multi-temporal, para caracterização da ocupação do solo.