Generation of Digital Surface Models (DSMs) from stereo satellite (spaceborne) images is classically performed by Ground Control Points (GCPs) which require site visits and precise measurement equipment. However, collection of GCPs is not always possible and such requirement limits the usage of spaceborne imagery. This study aims at developing a fast, fully automatic, GCP-free workflow for DSM generation. The problems caused by GCP-free workflow are overcome using freely-available, low resolution static DSMs (LR-DSM). LR-DSM is registered to the reference satellite image and the registered LR-DSM is used for i) correspondence generation and ii) initial estimate generation for 3-D reconstruction. Novel methods are developed for bias removal for LR-DSM registration and bias equalization for projection functions of satellite imaging. The LR-DSM registration is also shown to be useful for computing the parameters of simple, piecewise empirical projective models. Recent computer vision approaches on stereo correspondence generation and dense depth estimation are tested and adopted for spaceborne DSM generation. The study also presents a complete, fully automatic scheme for GCPfree DSM generation and demonstrates that GCP-free DSM generation is possible and can be performed in much faster time on computers. The resulting DSM can be used in various remote sensing applications including building extraction, disaster monitoring and change detection.

The capability to observe ocean swell using spaceborne Synthetic Aperture Radar (SAR) has been demonstrated starting with ERS-1 mission in 1992. This dissertation shows how ocean swell properties can be used to combine swell observations of heterogeneous quality and acquired at various times and locations for the observation and forecast of ocean swell fieldsusing ASAR instrument on-board ENVISAT. The first section is a review of how ocean swell spectra can be derived from the SAR complex images of the ocean surface using a quasi-linear transformation. Then, significant swell heights, peak periods and peak directions from in situ measurements are used to assess the accuracy of the SAR observed swell spectra. Using linear propagation in deep ocean, a new swell field reconstruction methodologyis developed in order to gather SAR swell observations related to the same swell field. Propagated from their generation region, these observations render the spatio-temporal properties of the emanating ocean swell fields. Afterwards, a methodology is developed for the exclusion of outliers taking advantage of the swell field consistency. Also, using the irregularly sampled SAR observations, quality controlled estimations of swell field integral parameters are produced on a regular space-time grid. Validation against in situ measurements reveals the dramatic impact of the density of propagated observations on the integral parameters estimated accuracy. Specifically, this parameter is shown to be very dependent on the satellite orbit. Finally, comparisons with the numerical wave model WAVEWATCH-III prove it could potentially benefit from the SAR swell field estimates for assimilation purposes.

La synthèse d'ouverture est une technique interférométrique similaire à la synthèse par rotation de la terre utilisée en radioastronomie où les signaux reçus par une paire de petites antennes sont traités de telle manière à synthétiser une seule grande antenne. Le concept de synthèse d'ouverture a été réadapté pour l'observation de la terre dans le cas de la télédétection de sources étendues de température. L'utilisation de cette technique pour l'observation de la terre a permis de contourner les limitations sur la taille d'antenne en télédétection passive. La fonction de corrélation, ou de visibilité, obtenue en inter-corrélant les signaux reçus par les an- tennes d'un système interférométrique employant une synthèse d'ouverture est définie comme étant la transformée de Fourier de la carte des températures de bril lance de la scène observée. Cette relation est connue sous le nom du théorème de Van Cittert-Zernike pour des observateurs en repos par rapport aux sources de température. La forme classique de ce théorème a été dérivée en inter-corrélant les échantillons temporels instantanés du champ électrique mesurés par différentes antennes. Un nouveau concept basé une interférométrie spatio-temporelle passive a été proposé comme étant la nouvelle génération qui succédera à la mission SMOS (Soil Moisture and Ocean Salinity) opérant dans l'espace depuis Novembre 2009. Celui-ci a pour objectif principal l'amélioration de la résolution spatiale à des ordres pouvant répondre aux applications hydrologiques à l'échelle locale où des résolutions kilométriques sont exigées. Ce concept interférométrique se base sur l'idée d'intégrer le déplacement de l'observateur (l'antenne) et ainsi la variable temps dans le calcul de la fonction de corrélation. Ceci engendre la création de nouvelles lignes de base virtuelles entre les positions des antennes à des instants différents, en plus des lignes de base physiques formées entres les positions des antennes instantanées. L'étude de ce concept de corrélation a malheureusement démontré la suppression exacte de l'information additionnelle due aux lignes de base virtuelles par le décalage Doppler induit par le déplacement. Une seconde étude du concept d'interférométrie spatio-temporelle combinée à une nouvelle procé- dure d'imagerie par corrélation dans l'espace fréquentiel, accomplie en inter-corrélant les spectres fréquentiels des champs électriques mesurés par une paire d'antennes séparées d'une distance Δr à bord d'un satellite à une hauteur h, a démontré l'obtention d'une information en 2D en températures de brillance de la scène observée. En plus, le développement théorique de la fonction de corrélation a mis en évidence une relation liant les visibilités aux températures de brillance par l'intermédiaire d'un noyau hautement oscillatoire. L'élément nouveau apporté par la corrélation dans l'espace fréquentiel consiste à exploiter l'informati- on de corrélation acquise par les antennes du satellite pour des fréquences présentant de petites dif- férences et pas seulement l'auto-corrélation. Cette propriété permet une reconstruction en 2D des températures de brillance avec seulement deux antennes / Aperture synthesis is an interferometric technique similar to Earth rotation synthesis employed in radio astronomy in which the signals received by a pair of small antennas are processed in a way to synthesize a single large antenna. The aperture synthetic concept used in radioastronomy was readapted to Earth remote sensing for large thermal sources. Thanks to this technique, limitations on antenna size in passive microwave remote sensing have been overcome. The correlation, or visibility, function obtained by cross-correlating the signals received by the antennas of an interferometric system using aperture synthesis is linked to the brightness temperature map of the observed scene by means of a Fourier-transform law. This is know as the standard form of the Van Cittert-Zernike theorem for fixed observers with respect to sources of temperature. This stan- dard formulation is derived by cross-correlating the instantaneous temporal components of the measured electric fields by different antennas. A new concept based on a passive spatio-temporal interferometry was proposed as the new generation to follow the well-known SMOS (Soil Moisture and Ocean Salinity) mission successfully operating since November 2, 2009. The aim of the proposed concept is a jump in the current achieved geometric resolution to orders capable of meeting the stringent users' needs for the study of hydrological applications in the local scale where sub-kilometric resolutions are required. This interferometric concept is based on the idea of integrating the displacement of the observer (satellite's antenna), and hence the time variable, in the calculation of the correlation function, which yields the creation of virtual baselines between the positions of antennas at different instants, in addition to the physical ones formed between the instantaneous antennas' spatial positions. Sadly, the additional information due to the virtual baseline was shown to be exactly canceled by the induced Doppler shift due to the observer's motion. We show furthermore that when using the aforementioned spatio-temporal interferometric system combined with a revolutionary Fourier Correlation Imaging (FouCoIm) procedure, consisting in cross-correlating, at slightly different frequencies, the Fourier components of the fluctuations of the re- ceived electric fields by a pair of antennas separated by a distance Δr on board of a satellite flying at height h, the 2D position-dependent brightness temperature can be reconstructed. Besides, the analytical derivation of the correlation function gives rise to a relationship linking the measured cor- relations to the position-dependent brightness temperatures by means of a Highly Oscillatory Integral (HOI) kernel. Interestingly, the analytical study of the HOI kernel showed the remarkable property that a corre- lation between both antenna-signals remains within a small frequency interval (different frequencies) outside the simple auto-correlation (same frequency). As a matter of fact, while existing systems had, until now, only considered the simple 1D information contained in the auto-correlation, it appears that the resulting correlation function from this concept bears a 2D information for the measurement of the position-dependent brightness temperature. Based on this, one is capable of reconstructing 2D bright- ness temperatures starting from a simple 1D geometry (two antennas arranged perpendicularly to the flight direction)

No description available.

Purpose – This study explores how Earth Observation (EO) start-ups with digital services can utilize business model innovation (BMI) to create, deliver, and capture value in the mining industry. Method – The study adopted a qualitative approach with the unit of analysis being two EO service start-ups and the unit of observation being mining companies complemented with other relevant actors. 22 interviews were conducted with mining companies; 2 with mining consultancy companies; 2 with academia; and 2 with the EO case companies. Findings – EO service start-ups can create value in the mining industry by leveraging an artificial intelligence technology value proposition unlocking new values, as well as engaging in trust and relationship building-activities with customers; deliver value by developing customizable delivery plans in line with the heterogenous customer needs, acquiring a specific understanding of customer operations to deliver expected service quality, and addressing customers' different digital maturity; and capture value by developing a flexible revenue model and enable internal and external scaling of service. Theoretical Implications – This study contributes to existing research by examining how digital technologies shape industry-specific phenomena and related BMs, specifically within the novel EO industry by being one of the pioneering studies in this space. It also fills a gap in the literature by examining start-up BMI in general as well as in the EO industry, the intersection of start-ups and digital services, as well as contributing to the broader service innovation literature. It also makes contributions to the specific mechanisms of the BM. Managerial Implications – This study contributes with insights for managers of digital service start-ups, both in general and in the novel EO industry, outlining how these can create, deliver, and capture value in the mining industry as well as other industrial contexts. These can utilize the findings of this study to either redesign their existing BMs or create completely new ones. Limitations and Future Research – One limitation is that the study only collects data from the mining industry which decreases the generalizability of the findings to other industries. Future research could benefit from investigating EO services from an ecosystem or platform perspective including other actors than provider-customer. Finally, future research could also be conducted utilizing a multiple case study including EO service start-ups focusing on multiple industries. / Syfte – Denna studie undersöker hur Earth Observation (EO)-startups med digitala tjänster kan använda affärsmodellsinnovation för att skapa, leverera och fånga värde inom gruvindustrin. Metod – Studien använde en kvalitativ metod med två EO-startups med digitala tjänster som analysenehet. Gruvföretag kompletterat med andra relevanta aktörer intervjuades för att besvara studiens syfte. Totalt genomfördes 22 intervjuer med gruvföretag; 2 med konsultföretag inom gruvindustrin; 2 inom akademin; samt 2 med EO-startupsen. Resultat – EO-startups med digitala tjänster kan skapa värde i gruvindustrin genom att nyttja artificiell intelligens i sitt värdeerbjudande för att tillhandahålla nya värden, samt genom att genomföra aktiviteter riktat mot kunder med syfte att stärka förtroende och relationer; leverera värde genom att utveckla anpassningsbara leveransalternativ i linje med de heterogena kundbehoven, anförskaffa en specifik förståelse för kundernas verksamhet för att leverera tjänsten i linje med vad dessa förväntar sig samt genom att adressera kundernas olika digitala mognadsnivåer; och fånga värde genom att utveckla en flexibel intäktsmodell och möjliggöra intern och extern skalning av tjänsten. Teoretiska implikationer – Denna studie bidrar till befintlig forskning genom att undersöka hur digitala teknologier formar branschspecifika fenomen och relaterade affärsmodeller, specifikt inom EO-industrin genom att vara en av de första studierna inom detta område. Studien fyller också ett gap i litteraturen genom att undersöka affärsmodellsinnovation relaterat till startups generellt samt inom EO-industrin, skärningspunkten mellan startups och digitala tjänster, samt genom att bidra till den generella litteraturen för tjänsteinnovation. Den bidrar också till de specifika mekanismerna i affärsmodeller. Praktiska implikationer – Denna studie bidrar med insikter för chefer och ledare inom digitala tjänste-startups, både generellt samt inom EO-industrin, och beskriver hur dessa kan skapa, leverera och fånga värde inom gruvindustrin såväl som i andra industriella sammanhang. Dessa kan använda studiens resultat för att antingen förändra sina befintliga affärsmodeller eller skapa helt nya. Begränsningar och framtida forskning – En begränsning är att studien endast samlar data från gruvindustrin, vilket minskar generaliserbarheten av resultaten till andra industrier. Framtida forskning kan dra nytta av att undersöka EO-tjänster ur ett ekosystem- eller plattformsperspektiv som inkluderar andra aktörer än leverantör-kund. Slutligen skulle framtida forskning även kunna genomföras med en flerfallstudie som inkluderar EO-tjänste-startups verksamma inom fler industrier.

Globally fish production has continued to increase during recent years at a rate exceeding that of human population growth. However the contribution from capture fisheries has remained largely static since the late 1980s with the increase in production being accounted for by dramatic growth in the aquaculture sector. As of 2012 aquaculture accounted for approximately 42% of total fisheries production and 78% of inland fish production. In view of these figures it is unsurprising that for a number of regions aquaculture represents an important source of both food security and income. The use of Geographical Information Systems (GIS) and spatial data have seen substantial developments in recent years with the help of increasingly affordable computing capacity. From an aquaculture perspective the use of GIS has shown significant potential as a means of combining varied data sources, including those acquired via remote sensing, into models to provide decision support in relation to site selection. A common theme amongst site suitability assessments is the incorporation of climate variables relating to temperature and water availability. These factors in turn can have a significant influence on aquaculture in terms of water availability and quality, and temperature modulated growth performance. There is now a strong consensus that during the 20th century, and especially during recent decades, the earth has experienced a significant warming trend. There is also strong agreement that this warming trend is at least partially a consequence of anthropogenic greenhouse gas emissions and that some degree of further warming is inevitable. While global warming is typically discussed in terms of degrees centigrade of average global temperature increase the full effects in terms of climate changes will be varied both in terms of location and season. The current project focuses on site suitability for aquaculture in relation to changing climate conditions. Significant use is made of GIS and a range of spatial data including remotely sensed data and output from a series of climate models. The project consists of a number of key components: 1. Vulnerability of aquaculture related livelihoods to climate change was assessed at the global scale based on the concept of vulnerability to climate related impacts as a function of sensitivity to climate change, exposure to climate change, and adaptive capacity. Use was made of national level statistics along with gridded climate and population data. Climate change scenarios were supplied using the MAGICC/SCENGEN climate modelling tools. Analysis was conducted for aquaculture in freshwater, brackish, and marine environments with outputs represented as a series of raster images. A number of Asian countries (Vietnam, Bangladesh, Laos, and China) were indicated as most vulnerable to impacts on freshwater production. Vietnam, Thailand, Egypt and Ecuador stood out in terms of brackish water production. Norway and Chile were considered most vulnerable to impacts on Marine production while a number of Asian countries (China, Vietnam, and the Philippines) also ranked highly. 2. Site suitability for pond-based aquaculture was modelled at the global scale using a 10 arcsecond grid. Data from an ensemble of 13 climate models was used to model pond temperature and water availability for rain fed ponds under late 20th century conditions and for a 2°C global warming scenario. Two methods are demonstrated for combining data with a focus on the culture of warm water species. Results suggest both positive and negative impacts in relation to the 2°C warming scenario depending on location and season. Some areas are projected to see negative effects from maximum temperatures during the warmest parts of the year while for many regions there are likely to be potential increases in growth performance during colder months with possible expansion into previously unsuitable areas. 3. Methods for detecting surface water using remotely sensed data were investigated for Bangladesh. Use was made of data from the Moderate-resolution Imaging Spectroradiometer (MODIS) and Landsat ETM+ instruments with accuracy assessed against ground truth data collected in the field. A time series was constructed using all available MODIS data (approximately 13 years with an 8 day temporal resolution) to show areas of: surface water, land, and mixed land and water. The time series was then analysed to produce a layer showing the percentage of the total time series where surface water is indicated thus providing a spatial representation of flood prevalence. 4. A land cover data set was produced using 9 Landsat ETM+ scenes to cover the majority of Bangladesh. 10 different classification routines were evaluated including a decision tree approach unique to the current study. Classification results were assessed against two sets of ground control points produced: one based on field collected ground truth data and the other using a stratified random sampling procedure in association with visual analysis of high resolution true colour satellite images and ETM+ composites. The most accurate classifications were provided by the decision tree method developed for the current study and a Multi-Layer Perceptron (MLP) neural network based classifier. 5. Site suitability for pond-based aquaculture within Bangladesh was assessed using a GIS in combination with the ETM+ based land cover data, the MODIS based surface water time series, and components of the global site suitability assessment including modelled pond temperature data. Assessments were made based on late 20th century conditions and a 2°C global warming scenario. The MODIS surface water time series was also used to show the effects of storm surge flooding in relation to cyclone Aila that struck Bangladesh on 25th May 2009. The south and east of the country were considered most suitable for aquaculture due to more favourable cold season temperatures and higher water balance values. The north west of the country was considered least favourable due to higher maximum modelled pond temperatures and lower water balance values. The effect of the 2°C warming scenario was to enhance these trends. To date the potential spatial implications of changing climate for aquaculture has been significantly under researched. In this respect the current study provides a highly useful indication of where aquaculture related livelihoods may be vulnerable. In addition valuable and unique insights are provided into the distribution of areas of both potential increased, as well as decreased, suitability for existing aquaculture and further aquaculture development.

Globale Urbanisierungsprozesse und eine Zunahme der räumlichen Konzentration von exponierten Elementen wie Menschen, Gebäude, Infrastruktur und ökonomische Werte induzieren ein ungekanntes Risiko in erdbebengefährdeten Regionen. Wenn keine Abschwächung des Risikos erfolgt werden dramatische Folgen in der Zukunft erwartet. Diese umfassen eine beispiellose Anzahl an Todesopfer, enorme ökonomische und ökologische Verluste und Ausfälle bezüglich kritischer Infrastruktur und Versorgung etc. Um derartige Gefährdungen abzuschwächen sind detaillierte Informationen über seismisches Risiko notwendig. Die seismische Verwundbarkeit von Siedlungsarealen ist dabei als zentrale, konstituierende Komponente von seismischem Risiko zu berücksichtigen. In diesem Zusammenhang ist es von besonderem Interesse das Verhalten von Gebäudeinventaren unter einem bestimmten Erdbebeneinfluss abschätzen zu können. Das Hauptziel der Arbeit war es maßgeschneiderte Methoden zu entwickeln, die eine Bewertung der seismischen Vulnerabilität von Siedlungsräumen, basierend auf Fernerkundungsdaten, durchführbar machen. Es wurden Methoden aus dem Bereich des maschinellen Lernens adaptiert, um Verwundbarkeitsstufen von Gebäuden und homogenen Siedlungsstrukturen zu bestimmen. Hierfür wurden Merkmale aus Fernerkundungsdaten abgeleitet und mit in situ Informationen verknüpft. Wir verwenden verschiedene Ensembles von Fernerkundungssensoren, um die urbane Morphologie umfassend zu charakterisieren. Empirische Ergebnisse, die für die erdbebengefährdeten Städte Padang (Indonesien) und Istanbul (Türkei) generiert werden konnten, bestätigen die Durchführbarkeit der entwickelten Verfahren. Zukünftige Arbeiten können daran anknüpfen und beispielsweise empirische Erkenntnisse in weiteren Fallstudien anzweifeln, eine Verbesserung der Methodik vornehmen, Konzepte und Ansätze auf andere Sensorsysteme oder Datenquellen übertragen oder Daten und Methoden im Rahmen von holistischen Risikobewertungsstrategien anwenden. / Global urbanization processes and increasing spatial concentration of exposed elements such as people, buildings, infrastructure, and economic values in earthquake prone regions induce seismic risk at a uniquely high level. This situation, when left unmitigated, is expected to cause unprecedented death tolls, enormous economic and ecological losses, and critical infrastructure and service failures, etc., in the future. To mitigate those perils requires detailed knowledge about seismic risks. As an important constituent element of seismic risk, the seismic vulnerability of the built environment has to be assessed. In particular, it is crucial to know about the behavior of the building inventory under a certain level of ground shaking. The main goal of the thesis was to develop and evaluate tailored methods and procedures that allow for a viable seismic vulnerability assessment of the built environment with remote sensing data. In particular, methods from the machine learning domain were adapted to estimate vulnerability levels of buildings and homogeneous urban structures based on features derived from remote sensing and by incorporation of in situ knowledge. To this purpose we deploy ensembles of earth observation sensors to exhaustively characterize the urban morphology. Empirical results, obtained for the earthquake prone cities Padang (Indonesia) and Istanbul (Turkey), confirm the viability of the approaches. Overall, this thesis provides some promising results, which show that remote sensing has a high capability to contribute to a rapid screening assessment of the seismic vulnerability of buildings and urban structures. Further work can build upon these results and may challenge empirical findings in further case studies, enhance developed and applied methods, transfer concepts and approaches to other sensor systems and data sources, or apply data and methodologies within integrative and holistic risk assessment strategies.

<div> <div> <p>Global Navigation Satellite System Reflectometry (GNSS-R) has presented a great potential as an important approach for ocean remote sensing. Numerous studies have demonstrated that the shape of a code-correlation waveform of forward-scattered Global Positioning System (GPS) signals may be used to measure ocean surface roughness and related geophysical parameters such as wind speed. Recent experiments have extended the reflectometry technique to transmissions from communication satellites. Due to the high power and frequencies of these signals, they are more sensitive to smaller scale ocean surface features, which makes communication satellites a promising signal of opportunity (SoOp) for ocean remote sensing. Recent advancements in fundamental physics are represented by the new scattering model and bistatic radar function developed by Voronovich and Zavorotny based on the SSA (Small Slope Approximation). This new model allows the partially coherent scattering in low wind conditions to be correctly described, which overcomes the limitations of diffuse scattering inherited in the conventional KA-GO (Kirchhoff Approximation-Geometric Optics) model. Furthermore, exploration and practice using spaceborne platforms have become a primary research focus, which is highlighted by the launch of CYGNSS (Cyclone Global Navigation Satellite System) in 2016. CYGNSS is a NASA (National Aeronautics and Space Administration) Earth Venture Mission consisting of an 8 micro-satellite constellation of GNSS-R instruments designed to observe tropical cyclones.</p><p>However, in spite of the significant achievements made in the past 10 years, there are still a variety of challenges to be addressed currently in the ocean reflectometry field. To begin with, the airborne demonstration experiments conducted previously for S-band reflectometry provided neither sufficient amount of data nor the desired scenarios to assess high wind retrieval performance of S-band signals. The current L-band empirical model function theoretically does not also apply to S-band reflectometry. With respect to scattering models, there have been no results of actual data processing so far to verify the performance of the SSA model, especially on low wind retrievals. Lastly, the conventional model fitting methods for ocean wind retrievals were proposed for airborne missions, and new approaches will need to be developed to satisfy the requirement of spaceborne systems.<br></p><p>The research described in this thesis is mainly focused on the development, application and evaluation of new models and algorithms for ocean wind remote sensing. The first part of the thesis studies the extension of reflectometry methods to the general class of SoOps. The airborne reception of commercial satellite S-band transmissions is demonstrated under both low and high wind speed conditions. As part of this effort, a new S-band geophysical model function (GMF) is developed for ocean wind remote sensing using S-band data collected in the 2014 NOAA (National Oceanic and Atmospheric Administration) hurricane campaign. The second part introduces a dual polarization L- and S-band reflectometry experiment, performed in collaboration with Naval Research Lab (NRL), to retrieve and analyze surface winds and compare the results with CYGNSS satellite retrievals and NOAA data buoy measurements. The problems associated with low wind speed retrieval arising from near specular surface reflections are studied. Results have shown improved wind speed retrieval accuracy using bistatic radar cross section (BRCS) modeled by the SSA when compared with KA-GO, in the cases of low to medium diffuse scattering. The last part focuses on the contributions to the NASA-funded spaceborne CYGNSS project. It shows that the accuracy of CYGNSS ocean wind retrieval is improved by an Extended Kalman Filter (EKF) algorithm. Compared with the baseline observable methods, preliminary results showed promising accuracy improvement when the EKF was applied to actual CYGNSS data.<br><br></p></div></div>

L’imagerie satellite radar propose un point de vue intéressant pour l’étude et la compréhension des océans. Là où l’altimétrie, reconnue et utilisée mondialement, a su s’imposer comme une source de données majeure, les observations de houle issues du SAR (de l’anglais « Synthetic Aperture Radar ») restent encore largement sous exploitées. L’objet de cette thèse est de promouvoir l’utilisation de ces données en proposant un modèle pour l’analyse et la prévision de la houle à l’échelle du globe qui soit indépendant des modèles numériques classiques comme Wavewatch-III. Ce travail s’inscrit dans une logique de pérennisation de la mesure de houle depuis l’espace avec le lancement dans les trois années à venir des trois missions satellites Sentinel-1 A et B et CFOSAT. Un des principaux résultats de ce travail est la capacité de la méthode développée à fournir une information plus précise que celle des modèles existants. Cette méthode permet également une meilleure caractérisation des mesures utilisées en entrée et des pistes d’amélioration de ces dernières sont dégagées pour les futures activités de calibration/validation. Ces travaux ouvrent également des perspectives sur les possibilités d’assimilation des sorties de ce nouveau modèle dans les modèles numériques classiques. / The capability to observe ocean swell using spaceborne Synthetic Aperture Radar (SAR) has been demonstrated starting with ERS-1 mission in 1992. This dissertation shows how ocean swell properties can be used to combine swell observations of heterogeneous quality and acquired at various times and locations for the observation and forecast of ocean swell fieldsusing ASAR instrument on-board ENVISAT. The first section is a review of how ocean swell spectra can be derived from the SAR complex images of the ocean surface using a quasi-linear transformation. Then, significant swell heights, peak periods and peak directions from in situ measurements are used to assess the accuracy of the SAR observed swell spectra. Using linear propagation in deep ocean, a new swell field reconstruction methodologyis developed in order to gather SAR swell observations related to the same swell field. Propagated from their generation region, these observations render the spatio-temporal properties of the emanating ocean swell fields. Afterwards, a methodology is developed for the exclusion of outliers taking advantage of the swell field consistency. Also, using the irregularly sampled SAR observations, quality controlled estimations of swell field integral parameters are produced on a regular space-time grid. Validation against in situ measurements reveals the dramatic impact of the density of propagated observations on the integral parameters estimated accuracy. Specifically, this parameter is shown to be very dependent on the satellite orbit. Finally, comparisons with the numerical wave model WAVEWATCH-III prove it could potentially benefit from the SAR swell field estimates for assimilation purposes.

Earth Observation is the gathering of information about planet Earth’s system via Remote Sensing technologies for monitoring land cover types and their changes. Through the years, image classification techniques have been widely studied and employed to extract useful information from Earth Observation data such as satellite imagery. One of the most attractive use cases is the monitoring of polar regions, that recently observed some dramatic changes due to global warming. Indeed drifting ice caps and icebergs represent threats to ship activities and navigation in polar areas, and the risk of collision with land-derived ice highlights the need to design a robust and automatic Sea-Ice classification for delivering up-to- date and accurate information. To achieve this goal, satellite data such as Sentinel-1 Synthetic Aperture Radar images from the European Union’s Copernicus program can be given in input to a Deep Learning classifier based on Convolutional Neural Networks capable of giving the content categorization of such images as output. For the task at hand, the availability of labeled data is generally scarce, there- fore the problem of learning with limited labeled data must be faced. There- fore, this work aims at leveraging the broader pool of unlabeled satellite data available to open up new classification solutions. This thesis proposes a Semi-Supervised Learning approach based on Generative Adversarial Networks. Such an architecture takes in input both labeled and unlabeled data and outputs the classification results exploiting the knowledge retrieved from both the data sources. Its classification performance is evaluated and it is later compared with the Supervised Learning approach and the Transfer Learning approach based on pre-trained networks. This work empirically proves that the Semi-Supervised Generative Adversarial Networks approach outperforms the Supervised Learning method, improving its Overall Accuracy by at least 5% in configurations with less than 100 training labeled samples available in the use cases under evaluation, achieving performance comparable to the Transfer Learning approach and even over- coming it under specific experimental configurations. Further analyses are then performed to highlight the effectiveness of the proposed solution. / Jordobservation är samlingen av information om jordklotets system via fjärravkänningstekniker för övervakning av landskapstyper och deras förändringar. Under årens lopp har bildklassificeringstekniker studerats och använts för att extrahera användbar information från jordobservationsdata som satellitbilder. Ett av de mest attraktiva användningsfallen är övervakningen av polära regioner, som nyligen observerade några dramatiska förändringar på grund av den globala uppvärmningen. Driftande istäcken och isberg representerar ett verkligt hot mot fartygsaktiviteter och navigering inom polära områden, och risken för kollision med land-baserad is belyser behovet av att utforma en robust och automatisk Hav-Is-klassificering för att leverera aktuell och korrekt information. För att uppnå detta mål kan satellitdata som Sentinel-1 Synthetic Aperture Radar-bilder från Europeiska unionens Copernicus-program ges som input till en Deep Learning-klassificerare baserad på Convolutional Neural Networks som kan ge innehållskategorisering av sådana bilder som output. För den aktuella uppgiften är tillgängligheten av märkt data i allmänhet otillräcklig, därför måste problemet med inlärning med begränsad mängd märkt data ställas inför rätta. Därav syftar detta arbete till att utnyttja den bredare samlingen av omärkt satellitdata som finns tillgänglig för att öppna nya klassificeringslösningar. Denna avhandling föreslår en Semi-Supervised Learning-strategi baserad på Generative Adversarial Networks. En sådan arkitektur tar som input både märkt och omärkt data, och matar ut klassificeringsresultat som utnyttjar den kunskap som hämtats från båda datakällorna. Dess klassificeringsprestanda ut- värderas och jämförs senare med tillvägagångssättet Supervised Learning och metoden Transfer Learning baserat på förtränade nätverk. Detta arbete bevisar empiriskt att Semi-Supervised Generative Adversarial Network överträffar metoden Supervised Learning och förbättrar dess totala noggrannhet med minst 5% i konfigurationer med mindre än 100 tränings- märkta prover tillgängliga i användningsfallen under utvärdering, vilket uppnår prestanda som både är jämförbar med Transfer Learning-metoden och överlägsen jämte den under specifika experimentella konfigurationer. Ytterligare analyser utförs sedan för att belysa effektiviteten hos den föreslagna lösningen.