No description available.

In a world under the constant impact of global warming, wetlands are decreasing in size all across the globe. As the wetlands are a vital part of preventing global warming, the ability to prevent their shrinkage through restorative measures is critical. Continuously orbiting the Earth are satellites that can be used to monitor the wetlands by collecting images of them over time. In order to determine the size of a wetland, and to register if it is shrinking or not, deep learning models can be used. Especially useful for this task is convolutional neural networks (CNNs). This project uses one type of CNN, a U-Net, to segment vegetated water in satellite data. However, this task requires labeled data, which is expensive to generate and difficult to acquire. The model used therefore needs to be able to generate reliable results even on small data sets. Therefore, pre-training of the network is used with a large-scale natural image segmentation data set called Common Objects in Context (COCO). To transfer the satellite data into RGB images to use as input for the pre-trained network, three different methods are tried. Firstly, the commonly used linear transformation method which simply moves the value of radar data into the RGB feature space. Secondly, two convolutional layers are placed before the U-Net which gradually changes the number of channels of the input data, with weights trained through backpropagation during the fine-tuning of the segmentation model. Lastly, a convolutional auto-encoder is trained in the same way as the convolutional layers. The results show that the autoencoder does not perform very well, but that the linear transformation and convolutional layers methods each can outperform the other depending on the data set. No statistical significance can be shown however between the performance of the two latter. Experimenting with including different amounts of polarizations from Sentinel-1 and bands from Sentinel-2 showed that only using radar data gave the best results. It remains to be determined whether one or both of the polarizations should be included to achieve the best result. / I en värld som ständigt påverkas av den globala uppvärmningen, minskar våtmarkerna i storlek över hela världen. Eftersom våtmarkerna är en viktig del i att förhindra global uppvärmning, är förmågan att förhindra att de krymper genom återställande åtgärder kritisk. Kontinuerligt kretsande runt jorden finns satelliter som kan användas för att övervaka våtmarkerna genom att samla in bilder av dem över tid. För att bestämma storleken på en våtmark, i syfte att registrera om den krymper eller inte, kan djupinlärningsmodeller användas. Speciellt användbar för denna uppgift är konvolutionella neurala nätverk (CNN). Detta projekt använder en typ av CNN, ett U-Net, för att segmentera vegeterat vatten i satellitdata. Denna uppgift kräver dock märkt data, vilket är dyrt att generera och svårt att få tag på. Modellen som används behöver därför kunna generera pålitliga resultat även med små datauppsättning. Därför används förträning av nätverket med en storskalig naturlig bildsegmenteringsdatauppsättning som kallas Common Objects in Context (COCO). För att överföra satellitdata till RGB-bilder som ska användas som indata för det förtränade nätverket prövas tre olika metoder. För det första, den vanliga linjära transformationsmetoden som helt enkelt flyttar värdet av radardatan till RGB-funktionsutrymmet. För det andra två konvolutionella lager placerade före U-Net:et som gradvis ändrar mängden kanaler i indatan, med vikter tränade genom bakåtpropagering under finjusteringen av segmenteringsmodellen. Slutligen tränade en konvolutionell auto encoder på samma sätt som de konvolutionella lagren. Resultaten visar att auto encodern inte fungerar särskilt bra, men att metoderna för linjär transformation och konvolutionella lager var och en kan överträffa den andra beroende på datauppsättningen. Ingen statistisk signifikans kan dock visas mellan prestationen för de två senare. Experiment med att inkludera olika mängder av polariseringar från Sentinell-1 och band från Sentinell-2 visade att endast användning av radardata gav de bästa resultaten. Om att inkludera båda polariseringarna eller bara en är den mest lämpliga återstår fortfarande att fastställa.

Saltmarshes provide a diverse range of ecosystem services and are protected under a number of international designations. Nevertheless they are generally declining in extent in the United Kingdom and North West Europe. The drivers of this decline are complex and poorly understood. When considering mitigation and management for future ecosystem service provision it will be important to understand why, where, and to what extent decline is likely to occur. Few studies have attempted to forecast saltmarsh morphodynamics at a system level over decadal time scales. There is no synthesis of existing knowledge available for specific site predictions nor is there a formalised framework for individual site assessment and management. This project evaluates the extent to which machine learning model approaches (boosted regression trees, neural networks and Bayesian networks) can facilitate synthesis of information and prediction of decadal-scale morphological tendencies of saltmarshes. Importantly, data-driven predictions are independent of the assumptions underlying physically-based models, and therefore offer an additional opportunity to crossvalidate between two paradigms. Marsh margins and interiors are both considered but are treated separately since they are regarded as being sensitive to different process suites. The study therefore identifies factors likely to control morphological trajectories and develops geospatial methodologies to derive proxy measures relating to controls or processes. These metrics are developed at a high spatial density in the order of tens of metres allowing for the resolution of fine-scale behavioural differences. Conventional statistical approaches, as have been previously adopted, are applied to the dataset to assess consistency with previous findings, with some agreement being found. The data are subsequently used to train and compare three types of machine learning model. Boosted regression trees outperform the other two methods in this context. The resulting models are able to explain more than 95% of the variance in marginal changes and 91% for internal dynamics. Models are selected based on validation performance and are then queried with realistic future scenarios which represent altered input conditions that may arise as a consequence of future environmental change. Responses to these scenarios are evaluated, suggesting system sensitivity to all scenarios tested and offering a high degree of spatial detail in responses. While mechanistic interpretation of some responses is challenging, process-based justifications are offered for many of the observed behaviours, providing confidence that the results are realistic. The work demonstrates a potentially powerful alternative (and complement) to current morphodynamic models that can be applied over large areas with relative ease, compared to numerical implementations. Powerful analyses with broad scope are now available to the field of coastal geomorphology through the combination of spatial data streams and machine learning. Such methods are shown to be of great potential value in support of applied management and monitoring interventions.

Zum Aufbau eines flächendeckenden Monitors der Siedlungs- und Freiraumentwicklung in Deutschland werden geeignete Geodaten benötigt. Ausgehend von den raum- und umweltplanerischen Zielstellungen eines Flächenmonitorings in Verbindung mit dem Anliegen der laufenden Raumbeobachtung ergeben sich dafür grundlegende Anforderungen. Verfügbare Datenquellen werden im Hinblick auf ihre potenzielle Eignung vorgestellt und verglichen: das Amtliche Topographisch-Kartographische Informationssystem ATKIS (insbesondere das Basis-DLM), das Amtliche Liegenschaftskataster-Informationssystem ALKIS (basierend auf der automatisierten Liegenschaftskarte ALK und dem Liegenschaftsbuch ALB) sowie klassifizierte Luft- und Satellitenbilddaten. Erkennbare Datenprobleme werden im Hinblick auf die Berechnung von Indikatoren diskutiert und mit Fallbeispielen illustriert. Außerdem wird eine Lösung für die administrative Bezugsgeometrie des Monitors vorgestellt.

Les zones côtières ne sont prises en compte dans les budgets globaux de CO2 atmosphérique que depuis peu. Il s’avère que bien qu’elles ne représentent globalement que de faibles superficies, les flux de carbone et de nutriments y sont très significatifs à l’échelle globale. On sait peu de chose sur le comportement des écosystèmes lagunaires vis-à-vis du CO2 et, encore moins des zones intertidales où les échanges avec l’atmosphère ont lieu alternativement avec l’eau et le sédiment. Les objectifs de cette étude ont été d’une part, d’établir le bilan de carbone échangé entre la lagune d’Arcachon, l’atmosphère et le milieu terrestre, et d’autre part de mettre en relation ces flux avec la production nette de l’écosystème (NEP) afin de mieux caractériser le statut métabolique de celle-ci ainsi que les facteurs environnementaux clés. Pour cela, nous avons mis en place pour la première fois et à différentes saisons et stations, des mesures directes de flux de CO2 par Eddy Corrélation, une méthode fonctionnant en continu pendant l’immersion et l’émersion. En parallèle, les apports de carbone terrestre sous ses différentes formes ont été quantifiés par un suivi annuel sur 9 rivières alimentant la lagune. L’export total de carbone par le bassin versant à travers les eaux de surface des rivières est estimé à 116 t C km-2 an-1 dont 39% est exporté à la lagune sous forme organique dissoute (DOC) du fait de la prédominance de podzols dans le bassin versant. La forte minéralisation de la matière organique terrestre dans les sols et eaux souterraines sursature largement les eaux en CO2 et l’export sous forme de carbone inorganique dissoute (DIC) représente environ 21%. La formulation d’un modèle mathématique, le « StreamCO2-DEGAS », basé sur les mesures de pCO2, de concentrations et de compositions isotopiques en DIC a permis de montrer que 43% de l’export total de carbone était dégazé sous forme de CO2 depuis les rivières vers l’atmosphère, réduisant alors le flux net entrant dans la lagune à 66 t C km-2 an-1. Concernant la mesure de flux verticaux, l’analyse cospectrale ainsi que les résultats obtenus en adéquation avec les contrôles physiques et biologiques aux différentes échelles tidale, diurne et saisonnières, ont permis de valider la méthode de l’Eddy Covariance en zone intertidale. Sur l’ensemble de la période de mesures, les flux de CO2 étaient faibles, variant entre -13 et 19 µmol m-2 s-1. Des puits de CO2 atmosphérique à marée basse le jour ont été systématiquement observés. Au contraire, pendant l’immersion et à marée basse la nuit, des flux positifs ou négatifs ou proche de zéro ont été observés suivant la saison et la station étudiées. L’analyse concomitante des flux de CO2 et des images satellites du platier à marée basse le jour a clairement permis de discriminer l’importance relative des deux cycles métaboliques distincts des principaux producteurs primaires avec (1) les herbiers de Zostera noltii à cycle annuel long, dominant la NEP en été et en automne à la station la plus centrale et (2) les communautés microphytobenthiques, dominant la production primaire brute (PPB) au printemps à la même station et en automne au fond du bassin. Un recyclage rapide de cette production durant l’immersion et l’émersion a aussi clairement été mis évidence. Au vue des différents résultats, la technique d’Eddy Covariance utilisée en zone intertidale laisse envisager d’intéressantes perspectives en termes de connaissances sur les budgets de carbone et les processus écologiques et biogéochimiques dans la zone côtière. / The coastal zone is only taken into account since recently in global carbon budgeting efforts. Although covering globally modest surface areas, carbon and nutrient fluxes in the coastal zone appear significant at the global scale. However, little is known about the CO2 behaviour in lagoons and even less in intertidal zones where exchanges with the atmosphere occur alternatively with the water and the sediment. The purposes of this work are, on one hand, to establish the carbon budget between the Arcachon lagoon, the atmosphere and the terrestrial watershed and on the other hand, to link these fluxes with the net ecosystem production (NEP) and better characterize its metabolic status along with the relevant environmental factors. For the first time, CO2 flux measurements by Eddy Correlation have been carried out at different seasons and stations in the tidal flat. In parallel, the total terrestrial carbon export from river waters has been quantified throughout a complete hydrological cycle in nine watercourses flowing into the lagoon. The total carbon export from the watershed through surface river waters is estimated at 116 t C km-2 yr-1 on which 39% is exported to the lagoon as dissolved organic carbon (DOC) owing to the predominance of podzols in the watershed. Intense organic matter mineralization in soils and groundwaters largely over-saturate river waters in CO2 on which export accounts for 21% as dissolved inorganic carbon (DIC). The mathematical “StreamCO2-DEGAS” model formulation based on water pCO2, DIC concentrations and isotopic composition measurements permits to show that 43% of the total carbon export was degassed as CO2 from the riverine surface waters to the atmosphere, lowering then this latter to 66 t C km-2 yr-1. With respect to the CO2 flux measurements in the lagoon, cospectral analysis and the well accordance of results with physical and biological controls at the tidal, diurnal and seasonal time scales permit to validate the Eddy Correlation technique over tidal coastal zone. CO2 fluxes with the atmosphere, during each period, were generally weak and ranged between -13 and 19 µmol m-2 s-1. Low tide and daytime conditions were always characterized by an uptake of atmospheric CO2. In contrast, during the immersion and during low tide at night, CO2 fluxes where either positive or negative, or close to zero, depending on the season and the site. The concomitant analysis of CO2 fluxes with satellite images of the lagoon at low tide during the day clearly discriminate the relative importance of the two distinct metabolic carbon cycling involving the main primary producers, i.e. (1) the Zostera noltii seagrass meadow predominance on the NEP in autumn and summer in the more central station, with an annual cycling and (2) the microphytobenthos community predominance on the gross primary production (GPP) in spring at the same station and in autumn in the inner part of the bay where a rapid carbon cycling during the immersion and the emersion was clearly highlighted. The different results obtained with the Eddy Correlation technique over tidal flats opens interesting perspectives on the knowledge of the carbon budget and the biogeochemical and ecological processes within the coastal zone.