The membrane aerated bioreactor (MABR) is a unique technological innovation where a gas permeable membrane is applied to biological processes. In an MABR, oxygen and other substrates diffuse from the opposite directions into a biofilm, and thus simultaneous chemical oxygen demand (COD) and nitrogen removal can be achieved. However, controlling biofilm thickness, stability, and attachment is challenging. The objectives of this research were to study the effect of oxygen partial pressure on process performance with respect to nitrogen removal and examine the biomass properties in MABRs at different oxygen partial pressures and COD loadings. The conditions within the bioreactors were based on a low hydrodynamic condition (average fluid velocity 22 cm/min along the membrane surface), with the intention of minimizing the impact of the hydrodynamic shear on biomass properties. Simultaneous nitrification and denitrification were achieved in the reactors, and increasing oxygen partial pressure enhanced the total nitrogen removal. The biomass at the membrane-biofilm interface was more porous at a loading of 11.3 kg COD/1000 m2/day (areal porosity about 0.9) as compared with a loading of 22.6 kg COD/1000 m2/day (areal porosity about 0.7), indicating carbon substrate was limiting near the membrane. Long-term (over 30 days) experimental results showed that at the loading of 11.3 kg COD/1000 m2/day, the oxygen partial pressures of 0.59 atm and 0.88 atm caused over 80% of the biomass to become suspended in the bulk phase while at 0.25 atm and 0.41 atm oxygen over 97% of the biomass was immobilized on the membrane. There is a critical oxygen partial pressure that can sustain the biofilm, which increases with an increasing COD loading. The nitrifying population in the reactors was examined by applying fluorescence in situ hybridization (FISH). At the loading of 22.6 kg COD/1000 m2/day, there were 12% beta-proteobacterial ammonia oxidizing bacteria (AOB) and 17%Nitrobacter in homogenized biofilm biomass at 0.59 atm oxygen while there were 7% beta-proteobacterial AOB and 4% Nitrobacter at 0.25 atm oxygen. The ratio of protein to carbohydrate in extracellular polymeric substances (EPS) of the homogenized biomass in the reactor decreased with increasing oxygen partial pressure. Surface characterization of the biomass revealed that the higher the oxygen partial pressure, the lower the biomass hydrophobicity and surface charge. The ratio of EPS protein to carbohydrate in a membrane aerated biofilm decreased when approaching the membrane-biofilm interface. The distribution of nitrifiers and dissolved oxygen profiles inside the biofilm suggested that dual substrate limitations exist, and it was concluded that the membrane aerated biofilm had an aerobic region in the inner layer and an anoxic region in the outer layer. It is proposed that the loss of EPS due to secondary substrate consumption, especially the loss of EPS proteins, at the bottom of the biofilm was responsible for biofilm detachment subjected to a critical oxygen partial pressure.

The use of remote techniques to capture the geometrical characteristics of rock masses has seen increased use and development in recent years. Apart from the obvious improved Health and Safety aspects, remote techniques allow rapid collection of digital data that can be subsequently analysed to provide input parameters for a variety of geomechanical applications. Remote data capture is a new technique used to collect geotechnical data and little independent work has been done concerning the comparative limitations and benefits of photogrammetry and laser scanning. Photogrammetry and laser scanning produce three dimensional digital representations of a studied rock face which can then be mapped for geotechnical data using specialist software. Research conducted at Camborne School of Mines, University of Exeter has focussed on developing robust and flexible methodologies for remote data capture techniques, namely photogrammetry and laser scanning. Geotechnical characterisation for photogrammetry was tested using the CSIRO Sirovision software and laser scanning was used with SplitFX from Split Engineering. A comparative method of assessing the error between orientation measurements was developed based on calculating the pole vector difference between remotely captured and traditionally hand-mapped data. This allowed for testing of the benefits of the remote data capture systems and limitations whilst comparing them with conventional hand-mapping. The thesis also describes the results of detailed comparisons between hand-mapping, photogrammetric and laser scanned data collection for discontinuity orientation, roughness, discontinuity trace lengths and potential end-use applications. During fieldwork in Cornwall, Brighton Cliffs and northern France it was found that remote data capture techniques struggled to collect orientation data from intensely fractured rock masses where features are primarily represented as discontinuity traces. It was found that both photogrammetry and laser scanning produce orientation data comparable to traditionally mapped data, with an average pole vector difference less than 12° from data mapped from the Tremough Campus road cutting to the University of Exeter’s Cornwall Campus. Set analysis on 151 comparable data points yielded a maximum set pole vector difference of 9.8°, where the closest difference was 2.24°. Testing the accuracy of discontinuity trace orientations captured by photogrammetry using the pole vector difference methods indicate that planar derived orientations are more accurate, with an average difference of 16.67° compared to 37.72°. This thesis contains the reviews and analyses of photogrammetry and laser scanning for use in characterising natural and manmade rock slopes. Improved field and post-processing methodologies have been developed to aid the safe, efficient and suitable geotechnical characterisation of rock fracture networks. The continual development and use of remote mapping techniques, whilst supplementing their unique qualities with traditional mapping, have the capability to revolutionise rock mass mapping. Particular development needed is the implementation of ISRM guidelines to standardise photogrammetric and laser scanning fieldwork and post-processing data analysis.

A rating curve provides a functional relationship between water height (i.e. stage) and discharge at a specified cross-section in a river. Used in combination with a time series of stage, rating curves become one of the central components for generating continuous records of streamflow. Since developing and maintaining rating curves can be time consuming, hydraulic models have shown potential to reduce the effort required for developing rating curves. A central challenge with modeling procedures, however, is the acquisition of accurate stream channel and floodplain topography. From this perspective, this thesis focuses on the real-world application of close-range remote sensing techniques such as laser-based ranging technologies (i.e. Light detection and ranging or LiDAR) or acoustic based ranging technologies (i.e. acoustic Doppler current profiler or ADCP) to capture topographic information for hydraulic modeling applications across various spatial scales. First, a review of the current LiDAR literature was carried out to identify potential ways to take full advantage of these novel data and technologies in the future. This was followed by four interconnected studies whereby: (i) a low-cost custom laser scanning system was designed to capture grain size distributions for a small stream; (ii) synthetically thinned airborne laser scanning (ALS) data was applied in a physically-based hydraulic modelling framework to develop rating curves; (iii) low-resolution national-scale ALS was coupled with ADCP bathymetry to be used in conjunction with a hydraulic model to develop rating curves; and (iv) the impact of measurement uncertainties on generating rating curves with a hydraulic model were investigated. This thesis highlights the potential of close-range remote sensing techniques for capturing accurate stream channel topography and derive from these data, the necessary parameters required for hydraulic modeling applications. / En avbördningskurva tillhandahåller ett funktionellt förhållande mellan vattendjup (dvs. vattenstånd) och flöde vid ett specifikt tvärsnitt i ett vattendrag. Avbördningskurvan blir en central komponent för generering av kontinuerliga tidsserier av vattenföring från tidsserier av vattenstånd. Eftersom det är tidskrävande att utveckla och underhålla avbördningskurvor erbjuder hydrauliska modeller attraktiva möjligheter att minska den insats som krävs för att utveckla avbördningskurvorna. En central utmaning för sådana modelleringsförfaranden är emellertid tillgången till noggrann topografidata av strömfåran och de omgivande stränderna. Den här avhandlingen fokuserar på tillämpningen av fjärranalystekniker för avståndsmätning på nära håll, såsom laserbaserade teknik (dvs. Light detection and ranging eller LiDAR) och akustisk baserat teknik (dvs. acoustic Doppler current profiler eller ADCP), för att fånga topografisk information för hydraulisk modellering av vattendrag i olika rumsliga skalor. Först presenteras en litteraturstudie av den nuvarande LiDAR-litteratur för att identifiera potentiella sätt att dra full nytta av dessa nya data och tekniker i framtiden. Detta följs av fyra sammanlänkade studier: (i) tillämpning av ett lågkostnads-laseravsökningssystem för att fånga kornstorleksfördelningar i ett litet vattendrag, (ii) syntetiskt förtunnad flygburen laserskanningsdata (ALS) applicerad i en fysiskt baserad hydraulisk modell för att utveckla avbördningskurvor, (iii) lågupplösta ALS från Svensk nationell höjdmodell kopplade med ADCP-batymetri för att ta fram en avbördningskurva med en hydraulisk modell, och (iv) undersökning av effekterna av osäkerheter på mätdata för att generera avbördningskurvor med en hydraulisk modell. Denna avhandling belyser potentialen för fjärranalystekniker för avståndsmätning på nära håll, för att fånga strömfårans exakta topografi och ifrån dessa data härleda de parametrar som krävs för hydrauliska modelleringstillämpningar. / <p>At the time of the doctoral defense, the following paper was unpublished and had a status as follows: Paper 5: Manuscript.</p>

Modelling of selected forest geometric parameters from airborne laser scanning data Abstract The main aim of this work is to approximate the shape of a tree crown with mathematically describable 3D shape based on airborne laser scanning (ALS) data. And consequently derive geometrical parameters describing the tree from this model. Included in the work is a custom designed algorithm based on angular segmentation. Measured results of this algorithm are then compared to an algorithm based on RANSAC and field measurement. The first part of this work describes airborne laser scanning, its use to derive characteristics of forest stands and individual trees and the theory of tree crown modelling. The next part contains a description of both algorithms and presentation of results and field measurements. The conclusion summarizes and evaluates the outputs of the custom angular segmentation algorithm and discusses its possible modifications. Keywords: airborne laser scanning, tree height, crown width, crown height, crown cover, crown volume, crown shape, RANSAC

The Generalization of the Digital Terrain Model Based on the TIN Abstract This diploma thesis deals with the up to now way and the possibilities of the digital terrain model generalization based on the TIN (the triangulate irregular network). New suitable way of the generalization of the digital terrain model procured from laser scanning data is proposed on the base of the existing generalization methods designated for digital models. Laser scanning data is characterized by a high areal density so the basic requirement is computing speed, maintaining the terrain features, such as a ridge, valley, steep hill, saddle, depression … and so on. The proposed algorithm is compared with the results of suggested algorithms and results from the generalization by the geographic software, such as Atlas DMT and ArcGIS.

Coregistration of airborne laser scanning data and aerial images Abstract This thesis is dealing with the co-registration of aerial laser scanning and aerial images. Theoretical part with research of current methods puts emphasis on methods suitable for remote sensing datasets. Part of the thesis is about pre-processing data for co-registration and DSM production. Selection of co- registration methods for remote sensing is based on previous researches. Selected co-registration methods are applied on datasets from EuroSDR research project and ČÚZK dataset. Application is realised by programming codes and functions that were created for this purpose in Matlab. Possibilities of usage, advantages and disadvantages of methods are being mentioned in the next parts of the thesis with emphasis on time of the computation and final accuracy. The function programmed in Matlab allows comparison of co-registration methods and allows the user to decide which of the co- registration methods to use on input datasets. Discussion section describes the possibilities of method extensions and problematic parts across the whole co-registration process. Keywords: co-registration, laser, scanning, images, photogrammetry, remote sensing, coordinate, image matching.

Weld quality is essential when manufacturing fatigue-loaded structures. Defective welds are problematic, causing increased lead times, repairs and breakdowns. Over- processing as a result of unnecessary safety margins and poor quality is another issue, leading to a more expensive process and heavier products. Knowledge about what variation in weld quality can be expected in the process enables more efficient problem solving and use of resources. Reduced variation equals increased quality. Consequently, reduced variation is both a manufacturer and customer gain. This thesis studies in what way variation in welding production can be mapped and presented, to guide towards the right improvement actions. A literature study addresses welding, weld quality, measurement methods and variation. Two empirical studies using laser scanning equipment along with staff interviews will then be conducted to develop the method to capture variation in welding. In the journey towards machine learning and elimination of operator decisions, knowledge and understanding of variation in the process is necessary. Concrete results of the empirical studies gave new and valuable information to the company. Further, the method to map, analyse and display variation was believed to be useful in several ways, both at the case company but also at other plants. The results show that knowledge about variation could have a large financial effect. By identifying the areas of over-processing and deficient quality, the process can be optimized to increase productivity. While technical issues such as equipment to collect data can be barriers, soft issues like competence, a common understanding, and visualisation of variation seem just as essential.

A silvicultura urbana é um dos elementos essenciais à manutenção da qualidade de vida nos grandes centros urbanos. A existência de uma ampla rede de arvores distribuída ao longo das vias e dos espaços públicos atua promovendo a qualidade do ar, a conservação de água, o conforto térmico, acústico e psicológico dos cidadãos. Florestas urbanas são capazes de amenizar as emissões dos Gases do Efeito Estufa (GEE), tais como o CO2, atuando como sumidouros. Visto sua importância, novas aplicações de ferramentas de sensoriamento remoto têm surgido para auxiliar no planejamento e implantação da silvicultura urbana. O sistema de escaneamento a laser aerotransportado LiDAR (Light Detection And Ranging em inglês), gera uma representação em três dimensões do objeto alvo por meio de uma nuvem de pontos georreferenciados. O cruzamento com sensores de altíssima resolução espectral proporciona analises mais aprofundadas do objeto, podendo-se extrair diversas métricas florestais tais como altura, área basal, e até mesmo espécie. O trabalho teve como objetivo verificar a contribuição do uso de informações derivadas da nuvem de pontos LiDAR, na identificação e classificação das seis espécies florestais mais frequentes do Parque Municipal Américo Renné Giannetti em Belo Horizonte, Minas Gerais, Brasil, a fim de auxiliar no planejamento e manejo da silvicultura urbana. Para tanto, por meio de classificação supervisionada, cruzou-se informações de levantamento de campo, segmentação de copas, pontos de topo de copa de árvore extraídos da nuvem LiDAR, e uma imagem multiespectral WordlView-2. A acurácia da classificação foi medida por análise da exatidão global do processo e por meio do índice Kappa. Os pontos de topo de copa de árvore derivados da nuvem LiDAR contribuíram para a localização e classificação das classes referentes às espécies florestais, quando comparados ao mesmo processo sem estes pontos. A segmentação das copas executada pelo programa eCognition facilitou o lançamento das amostras treinamento e teste. O classificador ECHO conseguiu melhores valores de acurácia e índice Kappa, frente aos outros classificadores do programa Multispec. O uso de informações provenientes da nuvem de pontos LiDAR se mostrou promissor em imagens multiespectrais de ambiente florestal urbano, aumentando a acurácia geral da classificação supervisionada. / Urban forestry is a key element to maintaining the quality of life in urban centers. The existence of a broad network of trees distributed along roads and public spaces acts to promote air quality, water conservation, thermal comfort, acoustic and psychological citizens. Urban forests are able to mitigate the emissions of Greenhouse Gases (GHG) such as CO2, acting as sinks. Since its importance, new applications of remote sensing tools have emerged to assist in planning and implementation of urban forestry. The laser scanning system airborne LiDAR (Light Detection And Ranging), generates a three-dimensional representation of the target object through a cloud of points georeferenced. The crossing with very high resolution sensors provides more in-depth analysis of the object and can be extracted several forest metrics such as height, basal area, and even species. The study aimed to verify the contribution of LiDAR derived points in the identification and classification of six most common tree species in Parque Municipal Americo Renne Giannetti, Belo Horizonte, Minas Gerais, Brazil, in order to assist urban forestry planning and management. Through supervised classification, field survey information, segmented areas, LiDAR treetop points, and a multispectral WordlView-2 image were crossed together. The classification accuracy was measured by analyzing overall accuracy and Kappa index. The LiDAR treetop points contributed to location and classification of tree species\' classes, when compared to the same process without these points. The segmentation of crowns performed by eCognition program facilitated the launch of training and test samples. ECHO classifier showed the best accuracy and Kappa index in comparison to other Multispec program classifiers. The aggregation of LiDAR data showed promise in urban forest multispectral images, increasing supervised classification overall accuracy.

A silvicultura urbana é um dos elementos essenciais à manutenção da qualidade de vida nos grandes centros urbanos. A existência de uma ampla rede de arvores distribuída ao longo das vias e dos espaços públicos atua promovendo a qualidade do ar, a conservação de água, o conforto térmico, acústico e psicológico dos cidadãos. Florestas urbanas são capazes de amenizar as emissões dos Gases do Efeito Estufa (GEE), tais como o CO2, atuando como sumidouros. Visto sua importância, novas aplicações de ferramentas de sensoriamento remoto têm surgido para auxiliar no planejamento e implantação da silvicultura urbana. O sistema de escaneamento a laser aerotransportado LiDAR (Light Detection And Ranging em inglês), gera uma representação em três dimensões do objeto alvo por meio de uma nuvem de pontos georreferenciados. O cruzamento com sensores de altíssima resolução espectral proporciona analises mais aprofundadas do objeto, podendo-se extrair diversas métricas florestais tais como altura, área basal, e até mesmo espécie. O trabalho teve como objetivo verificar a contribuição do uso de informações derivadas da nuvem de pontos LiDAR, na identificação e classificação das seis espécies florestais mais frequentes do Parque Municipal Américo Renné Giannetti em Belo Horizonte, Minas Gerais, Brasil, a fim de auxiliar no planejamento e manejo da silvicultura urbana. Para tanto, por meio de classificação supervisionada, cruzou-se informações de levantamento de campo, segmentação de copas, pontos de topo de copa de árvore extraídos da nuvem LiDAR, e uma imagem multiespectral WordlView-2. A acurácia da classificação foi medida por análise da exatidão global do processo e por meio do índice Kappa. Os pontos de topo de copa de árvore derivados da nuvem LiDAR contribuíram para a localização e classificação das classes referentes às espécies florestais, quando comparados ao mesmo processo sem estes pontos. A segmentação das copas executada pelo programa eCognition facilitou o lançamento das amostras treinamento e teste. O classificador ECHO conseguiu melhores valores de acurácia e índice Kappa, frente aos outros classificadores do programa Multispec. O uso de informações provenientes da nuvem de pontos LiDAR se mostrou promissor em imagens multiespectrais de ambiente florestal urbano, aumentando a acurácia geral da classificação supervisionada. / Urban forestry is a key element to maintaining the quality of life in urban centers. The existence of a broad network of trees distributed along roads and public spaces acts to promote air quality, water conservation, thermal comfort, acoustic and psychological citizens. Urban forests are able to mitigate the emissions of Greenhouse Gases (GHG) such as CO2, acting as sinks. Since its importance, new applications of remote sensing tools have emerged to assist in planning and implementation of urban forestry. The laser scanning system airborne LiDAR (Light Detection And Ranging), generates a three-dimensional representation of the target object through a cloud of points georeferenced. The crossing with very high resolution sensors provides more in-depth analysis of the object and can be extracted several forest metrics such as height, basal area, and even species. The study aimed to verify the contribution of LiDAR derived points in the identification and classification of six most common tree species in Parque Municipal Americo Renne Giannetti, Belo Horizonte, Minas Gerais, Brazil, in order to assist urban forestry planning and management. Through supervised classification, field survey information, segmented areas, LiDAR treetop points, and a multispectral WordlView-2 image were crossed together. The classification accuracy was measured by analyzing overall accuracy and Kappa index. The LiDAR treetop points contributed to location and classification of tree species\' classes, when compared to the same process without these points. The segmentation of crowns performed by eCognition program facilitated the launch of training and test samples. ECHO classifier showed the best accuracy and Kappa index in comparison to other Multispec program classifiers. The aggregation of LiDAR data showed promise in urban forest multispectral images, increasing supervised classification overall accuracy.

Above-Ground Biomass (AGB) assessment using remote sensing has been an active area of research since the 1970s. However, improvements in the reported accuracy of wide scale studies remain relatively small. Therefore, there is a need to improve error analysis to answer the question: Why is AGB assessment accuracy still under doubt? This project aimed to develop and implement a systematic quantitative methodology to analyse the uncertainty of remotely sensed AGB, including all perceptible error types and reducing the associated costs and computational effort required in comparison to conventional methods. An accuracy prediction tool was designed based on previous study inputs and their outcome accuracy. The methodology used included training a neural network tool to emulate human decision making for the optimal trade-off between cost and accuracy for forest biomass surveys. The training samples were based on outputs from a number of previous biomass surveys, including 64 optical data based studies, 62 Lidar data based studies, 100 Radar data based studies, and 50 combined data studies. The tool showed promising convergent results of medium production ability. However, it might take many years until enough studies will be published to provide sufficient samples for accurate predictions. To provide field data for the next steps, 38 plots within six sites were scanned with a Leica ScanStation P20 terrestrial laser scanner. The Terrestrial Laser Scanning (TLS) data analysis used existing techniques such as 3D voxels and applied allometric equations, alongside exploring new features such as non-plane voxel layers, parent-child relationships between layers and skeletonising tree branches to speed up the overall processing time. The results were two maps for each plot, a tree trunk map and branch map. An error analysis tool was designed to work on three stages. Stage 1 uses a Taylor method to propagate errors from remote sensing data for the products that were used as direct inputs to the biomass assessment process. Stage 2 applies a Monte Carlo method to propagate errors from the direct remote sensing and field inputs to the mathematical model. Stage 3 includes generating an error estimation model that is trained based on the error behaviour of the training samples. The tool was applied to four biomass assessment scenarios, and the results show that the relative error of AGB represented by the RMSE of the model fitting was high (20-35% of the AGB) in spite of the relatively high correlation coefficients. About 65% of the RMSE is due to the remote sensing and field data errors, with the remaining 35% due to the ill-defined relationship between the remote sensing data and AGB. The error component that has the largest influence was the remote sensing error (50-60% of the propagated error), with both the spatial and spectral error components having a clear influence on the total error. The influence of field data errors was close to the remote sensing data errors (40-50% of the propagated error) and its spatial and non-spatial Overall, the study successfully traced the errors and applied certainty-scenarios using the software tool designed for this purpose. The applied novel approach allowed for a relatively fast solution when mapping errors outside the fieldwork areas.