Coastal habitats are highly sensitive to change and highly diverse. Degrading environmental conditions have led to a global decline in biodiversity through loss, modification and fragmentation of habitats, triggering an increased effort to conserve these ecosystems. Remote sensing is important tool for filling in critical information gaps for monitoring habitats, yet significant barriers exist for operational use within the ecological and conservation communities. Reporting on both extent and condition of habitats are critical to fulfil policy requirements, specifically the ECs Habitat’s Directive. This study focuses on the use of Very High Resolution (VHR) optical imagery for retrieving parameters to identifyassociations that can separate habitat boundaries for extent mapping down to species level for indicators of condition, with a focus on operational use. The Earth Observation Data for Habitat Monitoring (EODHaM) system was implemented using Worldview-2 data from two periods (July and September), in situ data and local ecological knowledge for two sites in Wales, Kenfig Burrows SAC and Castlemartin SSSI. The system utilises the Food and Agricultural Organisation’s (FAO) Land Cover Classification System (LCCS) but translations between land cover and habitat schemes are not straight forward and need special consideration that are likely to be site specific. Limitations within therule-based method of the EODHaM system were identified and therefore augmented with machine learning based classification algorithms creating a hybrid method of classification generating accurate (>80% overall accuracy) baseline maps with a more automated and repeatable method. Quantitative methods of validation traditionally used within the remote sensing community do not consider spatial aspects of maps. Therefore, qualitative assessments carried out in the field were used in addition to error matrices, overall accuracy and the kappa coefficient. This required input from ecologists and site specialists, enhancing communication and understanding between the different communities. Generating baseline maps required significant amount of training data and updating baselines through change detection methods is recommended for monitoring. An automated, novel map-to-image change detection was therefore implemented. Natural and anthropogenic changes were successfully detected from Worldview-2 and Sentinel-2 data at Kenfig Burrows. An innovative component of this research was the development of methods, which were demonstrated to be transferable between both sites and increased understanding between remote sensing scientist and ecologist. Through this approach, a more operational method for monitoring site specific habitats through satellite data is proposed, with direct benefits for conservation, environment and policy.

The Indian Ocean Experiment (INDOEX) had as a primary objective determining the radiative forcing due to anthropogenic aerosols over climatologically significant space and time scales: the Indian Ocean during the winter monsoon, January-March. During the winter monsoon, polluted, low-level air from the Asian subcontinent blows over the Arabian Sea and Indian Ocean. As part of INDOEX, aerosol optical depths were derived from Advanced Very High Resolution Radiometer (AVHRR) data for the cloud-free ocean regions. The AVHRR radiances were first calibrated using the interior zone of the Antarctic and Greenland ice sheets, which proved to be radiometrically stable calibration targets. Optical depths were derived by matching the observed radiances to radiances calculated for a wide range of optical depths and viewing geometry. Optical depths derived with the AVHRR were compared with those derived with NASA's Aerosol Robotic Network (AERONET) CIMEL instrument at the Center for Clouds, Chemistry, and Climate's Kaashidhoo Observatory, as well as with other surface and shipboard observations taken in the INDOEX region. The retrieved and surface-based optical depths agreed best for a new 2-channel, 2- aerosol model scheme in which the AVHRR observations at O·64 and O·84 microns were used to determine relative amounts of marine and polluted continental aerosol and then the resulting aerosol mixture was used to derive the optical depths. Broadband radiative transfer calculations for the mixture of marine and polluted continental aerosols were combined with the 0·64 and 0·84-micron AVHRR radiances to determine the radiative forcing due to aerosols in the INDOEX region. Monthly composites of aerosol optical depth and top of the atmosphere, surface, and atmospheric radiative forcing were derived from calibrated AVHRR radiances for January-March 1996-2000. An inter-annual variability in the magnitude and spatial extent of high value regions is noted for derived optical depths and radiative forcing, with highest values reached in 1999, particularly in the Bay of Bengal which during the IFP was covered by plumes from Indochina. Frequency distributions of the optical depth for 1⁰ x 1⁰ latitude-longitude regions are well represented by gamma distribution functions. The day-to-day and year-to-year variability of the optical depth for such regions is correlated with the long term average optical depth. Interannual variability of the monthly mean optical depths for such regions is found to be as large as the day to day. / Graduation date: 2001

Accurate and up-to-date road information is essential for both effective urban planning and disaster management. Today, very high resolution (VHR) imagery acquired by airborne and spaceborne imaging sensors is the primary source for the acquisition of spatial information of increasingly growing road networks. Given the increased availability of the aerial and satellite images, it is necessary to develop computer-aided techniques to improve the efficiency and reduce the cost of road extraction tasks. Therefore, automation of image-based road extraction is a very active research topic. This thesis deals with the development and implementation aspects of a semi-automatic road extraction strategy, which includes two key approaches: multidirectional and single-direction road extraction. It requires a human operator to initialize a seed circle on a road and specify a extraction approach before the road is extracted by automatic algorithms using multiple vision cues. The multidirectional approach is used to detect roads with different materials, widths, intersection shapes, and degrees of noise, but sometimes it also interprets parking lots as road areas. Different from the multidirectional approach, the single-direction approach can detect roads with few mistakes, but each seed circle can only be used to detect one road. In accordance with this strategy, a RoadModeler prototype was developed. Both aerial and GeoEye-1 satellite images of seven different types of scenes with various road shapes in rural, downtown, and residential areas were used to evaluate the performance of the RoadModeler. The experimental results demonstrated that the RoadModeler is reliable and easy-to-use by a non-expert operator. Therefore, the RoadModeler is much better than the object-oriented classification. Its average road completeness, correctness, and quality achieved 94%, 97%, and 94%, respectively. These results are higher than those of Hu et al. (2007), which are 91%, 90%, and 85%, respectively. The successful development of the RoadModeler suggests that the integration of multiple vision cues potentially offers a solution to simple and fast acquisition of road information. Recommendations are given for further research to be conducted to ensure that this progress goes beyond the prototype stage and towards everyday use.

Accurate and up-to-date road information is essential for both effective urban planning and disaster management. Today, very high resolution (VHR) imagery acquired by airborne and spaceborne imaging sensors is the primary source for the acquisition of spatial information of increasingly growing road networks. Given the increased availability of the aerial and satellite images, it is necessary to develop computer-aided techniques to improve the efficiency and reduce the cost of road extraction tasks. Therefore, automation of image-based road extraction is a very active research topic. This thesis deals with the development and implementation aspects of a semi-automatic road extraction strategy, which includes two key approaches: multidirectional and single-direction road extraction. It requires a human operator to initialize a seed circle on a road and specify a extraction approach before the road is extracted by automatic algorithms using multiple vision cues. The multidirectional approach is used to detect roads with different materials, widths, intersection shapes, and degrees of noise, but sometimes it also interprets parking lots as road areas. Different from the multidirectional approach, the single-direction approach can detect roads with few mistakes, but each seed circle can only be used to detect one road. In accordance with this strategy, a RoadModeler prototype was developed. Both aerial and GeoEye-1 satellite images of seven different types of scenes with various road shapes in rural, downtown, and residential areas were used to evaluate the performance of the RoadModeler. The experimental results demonstrated that the RoadModeler is reliable and easy-to-use by a non-expert operator. Therefore, the RoadModeler is much better than the object-oriented classification. Its average road completeness, correctness, and quality achieved 94%, 97%, and 94%, respectively. These results are higher than those of Hu et al. (2007), which are 91%, 90%, and 85%, respectively. The successful development of the RoadModeler suggests that the integration of multiple vision cues potentially offers a solution to simple and fast acquisition of road information. Recommendations are given for further research to be conducted to ensure that this progress goes beyond the prototype stage and towards everyday use.

La spectroscopie est un outil puissant et non-destructif permettant, d’après les spectres, de remonter à des grandeurs physiques importantes telles que la concentration, la température, la pression, … du gaz en question.Les travaux présentés dans ce manuscrit sont consacrés à l’analyse des spectres infrarouges et Raman à très haute résolution de l’éthylène pour des applications atmosphériques et planétologiques.Cette thèse a été effectuée au sein du Groupe de Spectrométrie Moléculaire et Atmosphérique (GSMA) de l’Université de Reims Champagne- Ardenne. En collaboration avec l’équipe (SMPCA) au sein du laboratoire Interdisciplinaire Carnot de Bourgogne (ICB) à Dijon, l’équipe (CQP) au sien du service de Chimie Quantique et Photophysique à Bruxelles, l’Instituto de Estructura de la Materia (CSIC) à Madrid et les membres de la ligne AILES du centre de rayonnement synchrotron SOLEIL à Saint-Aubin.Cette thèse est structurée en 4 chapitres. Le premier chapitre est consacré sur les généralités de la molécule étudiée. Le second chapitre présente l’aspect théorique de la spectroscopie. Le troisième chapitre portant sur l’explication du modèle théorique qui nous avons utilisé pour effectuer l’analyse et le traitement des spectres. Dans le quatrième chapitre, nous présentons les résultats de l’analyse du spectre de la molécule de l’éthylène en différentes régions spectral.Ces résultats alimenteront des bases de données internationales telles que HITRAN (L. Rothmann) et GEISA (M. Rotger et le CNES) mais aussi peuvent servir de données modèle pour la start-up AEROVIA initiée par G. Durry, directeur de notre laboratoire. Avec ces données, nous alimenterons également la base de données européenne VAMDC. / The spectroscopy is a powerful analytical technique capable of providing Important Information physical quantities such as concentration, temperature, pressure, ... and other questions about gas.The work presented in this manuscript is devoted to analysis of high resolution infrared and Raman spectroscopy of the ethylene for atmospheric, astrophysical and planetological applications.The work described in this thesis was performed within the "Groupe de Spectrométrie Moléculaire et Atmosphérique" (GSMA) in the university of Reims Champagne-Ardenne in Reims. In national collaboration with the team (SMPCA) In collaboration in the laboratory "Interdisciplinaire Carnot de Bourgogne" (ICB) in Dijon, the team (CQP) in the service de "chimie quantique et photophysique" in Brussels. l’Instituto de Estructura de la Materia (CSIC) in Madrid and the members of the line AILES of the Synchrotron SOLEIL in Saint-Aubin.The thesis is structured into 4 main chapters. The first chapter deals with the generality of molecule studied. The second chapter represents the theoretical aspects of the spectroscopy. The third chapter dealing with the explanation of the theoretical model that we used for the analysis and processing of spectra. In the fourth chapter, we present the results of the analysis of the spectrum of the molecule of ethylene in different spectral regions.These results will feed the international databases such as HITRAN (L. Rothmann) and GEISA (Mr. Rotger and CNES), and it can also serve as a data for modeling the start-up AEROVIA initiated by G. Durry, the director of the laboratory. with these data, we can also feed the European database VAMDC.

Les satellites de télédétection sont devenus incontournables pour la société civile. En effet, les images satellites ont été exploitées avec succès pour traiter plusieurs applications, notamment la surveillance de l'environnement et de la prévention des catastrophes naturelles. Dans les dernières années, l'augmentation de la disponibilité de très haute résolution spatiale (THR) d'images de télédétection abouti à de nouvelles applications potentiellement pertinentes liées au suivi d'utilisation des sols et à la gestion environnementale. Cependant, les capteurs optiques, en raison du fait qu'ils acquièrent directement la lumière réfléchie par le soleil, ils peuvent souffrir de la présence de nuages dans le ciel et / ou d'ombres sur la terre. Il s'agit du problème des données manquantes, qui induit un problème important et crucial, en particulier dans le cas des images THR, où l'augmentation des détails géométriques induit une grande perte d'informations. Dans cette thèse, de nouvelles méthodologies de détection et de reconstruction de la région contenant des données manquantes dans les images THR sont proposées et appliquées sur les zones contaminées par la présence de nuages et / ou d'ombres. En particulier, les contributions méthodologiques proposées comprennent: i) une stratégie multirésolution d'inpainting visant à reconstruire les images contaminées par des nuages ; ii) une nouvelle combinaison d'information radiométrique et des informations de position spatiale dans deux noyaux spécifiques pour effectuer une meilleure reconstitution des régions contaminés par les nuages en adoptant une régression par méthode a vecteurs supports (RMVS) ; iii) l'exploitation de la théorie de l'échantillonnage compressé avec trois stratégies différentes (orthogonal matching pursuit, basis pursuit et une solution d'échantillonnage compressé, basé sur un algorithme génétique) pour la reconstruction d'images contaminés par des nuages; iv) une chaîne de traitement complète qui utilise une méthode à vecteurs de supports (SVM) pour la classification et la détection des zones d'ombre, puis une régression linéaire pour la reconstruction de ces zones, et enfin v) plusieurs critères d'évaluation promptes à évaluer la performance de reconstruction des zones d'ombre. Toutes ces méthodes ont été spécialement développées pour fonctionner avec des images très haute résolution. Les résultats expérimentaux menés sur des données réelles sont présentés afin de montrer et de confirmer la validité de toutes les méthodes proposées. Ils suggèrent que, malgré la complexité des problèmes, il est possible de récupérer de façon acceptable les zones manquantes masquées par les nuages ou rendues erronées les ombres.

Le travail présenté dans cette thèse a pour objectif d’évaluer le potentiel de modèles probabilistes multivariés appliqués sur les sous-bandes d’une décomposition en ondelettes pour la classification d’images de télédétection optiques à très haute résolution spatiale. Trois thématiques principales ont été investiguées dans ce travail : la différenciation de classes d’âge de peuplements de pins maritimes, la détection de parcelles viticoles et la détection de parcs ostréicoles. Une contribution originale concerne la proposition d’une chaîne traitement pour une classification supervisée orientée objet se basant sur des mesures de similarité adaptées au contexte de modélisation probabiliste. Celle-ci implique la création d’une base de données de patchs de texture pour l’apprentissage et l’utilisation d’une pré-segmentation de l’image à classifier. Les modèles probabilistes multivariés testés ont tout d’abord été évalués dans une procédure d’indexation d’images. Les modèles les plus performants identifiés par cette procédure ont été ensuite appliqués dans la chaîne de traitement proposée. Dans les trois thématiques explorées, les modèles multivariés ont révélé des capacités remarquables de représentation de la texture et ont permis d’obtenir une qualité de classification supérieure à celle obtenue par la méthode des matrices de co-occurrence. Ces résultats démontrent l’intérêt de la représentation multi-échelles et multi-orientations de la texture dans l’espace transformé en ondelettes et la pertinence de la modélisation multivariée des coefficients d’ondelettes issus de cette décomposition. / The prime objective of this thesis is to evaluate the potential of multivariate probabilistic models applied on wavelet subbands for the classification of very high resolution remote sensing optical data. Three main applications are investigated in this study: the differentiation of age classes of maritime pine forest stands, the detection of vineyards and the detection of oyster fields. One main contribution includes the proposal of an original supervised and object-oriented classification scheme based on similarity measurements adapted to the context of probabilistic modeling. This scheme involves the creation of a database of texture patches for the learning step and a pre-segmentation of the image to classify. The tested multivariate models were first evaluated in an image retrieval framework. The best models identified in this procedure were then applied in the proposed image processing scheme. In the three proposed thematic applications, multivariate models revealed remarkable abilities to represent the texture and reached higher classification accuracies than the method based on co-occurrence matrices. These results confirm the interest of the multi-scale and multi-orientation representation of textures through the wavelet transform, as well as the relevance of the multivariate modeling of wavelet coefficients

We used long-term observations of grassland aboveground net plant production (ANPP, 19392016), growing seasonal advanced very-high-resolution radiometer remote sensing normalized difference vegetation index (NDVI) data (1982-2016), and simulations of actual evapotranspiration (1912-2016) to evaluate the impact of Pacific Decadal Oscillation (PDO) and El Nino-Southern Oscillation (ENSO) sea surface temperature (SST) anomalies on a semiarid grassland in northeastern Colorado. Because ANPP was well correlated (R-2 = 0.58) to cumulative April to July actual evapotranspiration (iAET) and cumulative growing season NDVI (iNDVI) was well correlated to iAET and ANPP (R-2 = 0.62 [quadratic model] and 0.59, respectively), we were able to quantify interactions between the long-duration (15-30 yr) PDO temperature cycles and annual-duration ENSO SST phases on ANPP. We found that during cold-phase PDOs, mean ANPP and iNDVI were lower, and the frequency of low ANPP years (drought years) was much higher, compared to warm-phase PDO years. In addition, ANPP, iNDVI, and iAET were highly variable during the cold-phase PDOs. When NINO-3 (ENSO index) values were negative, there was a higher frequency of droughts and lower frequency of wet years regardless of the PDO phase. PDO and NINO-3 anomalies reinforced each other resulting in a high frequency of above-normal iAET (52%) and low frequency of drought (20%) when both PDO and NINO-3 values were positive and the opposite pattern when both PDO and NINO-3 values were negative (24% frequency of above normal and 48% frequency of drought). Precipitation variability and subsequent ANPP dynamics in this grassland were dampened when PDO and NINO-3 SSTs had opposing signs. Thus, primary signatures of these SSTs in this semiarid grassland are (1) increased interannual variability in ANPP during cold-phase PDOs, (2) drought with low ANPP occurring in almost half of those years with negative values of PDO and NINO-3, and (3) high precipitation and ANPP common in years with positive PDO and NINO-3 values.

Doctorat en Sciences / info:eu-repo/semantics/nonPublished

Building information extraction and reconstruction from satellite images is an essential task for many applications related to 3D city modeling, planning, disaster management, navigation, and decision-making. Building information can be obtained and interpreted from several data, like terrestrial measurements, airplane surveys, and space-borne imagery. However, the latter acquisition method outperforms the others in terms of cost and worldwide coverage: Space-borne platforms can provide imagery of remote places, which are inaccessible to other missions, at any time. Because the manual interpretation of high-resolution satellite image is tedious and time consuming, its automatic analysis continues to be an intense field of research. At times however, it is difficult to understand complex scenes with dense placement of buildings, where parts of buildings may be occluded by vegetation or other surrounding constructions, making their extraction or reconstruction even more difficult. Incorporation of several data sources representing different modalities may facilitate the problem. The goal of this dissertation is to integrate multiple high-resolution remote sensing data sources for automatic satellite imagery interpretation with emphasis on building information extraction and refinement, which challenges are addressed in the following: Building footprint extraction from Very High-Resolution (VHR) satellite images is an important but highly challenging task, due to the large diversity of building appearances and relatively low spatial resolution of satellite data compared to airborne data. Many algorithms are built on spectral-based or appearance-based criteria from single or fused data sources, to perform the building footprint extraction. The input features for these algorithms are usually manually extracted, which limits their accuracy. Based on the advantages of recently developed Fully Convolutional Networks (FCNs), i.e., the automatic extraction of relevant features and dense classification of images, an end-to-end framework is proposed which effectively combines the spectral and height information from red, green, and blue (RGB), pan-chromatic (PAN), and normalized Digital Surface Model (nDSM) image data and automatically generates a full resolution binary building mask. The proposed architecture consists of three parallel networks merged at a late stage, which helps in propagating fine detailed information from earlier layers to higher levels, in order to produce an output with high-quality building outlines. The performance of the model is examined on new unseen data to demonstrate its generalization capacity. The availability of detailed Digital Surface Models (DSMs) generated by dense matching and representing the elevation surface of the Earth can improve the analysis and interpretation of complex urban scenarios. The generation of DSMs from VHR optical stereo satellite imagery leads to high-resolution DSMs which often suffer from mismatches, missing values, or blunders, resulting in coarse building shape representation. To overcome these problems, a methodology based on conditional Generative Adversarial Network (cGAN) is developed for generating a good-quality Level of Detail (LoD) 2 like DSM with enhanced 3D object shapes directly from the low-quality photogrammetric half-meter resolution satellite DSM input. Various deep learning applications benefit from multi-task learning with multiple regression and classification objectives by taking advantage of the similarities between individual tasks. Therefore, an observation of such influences for important remote sensing applications such as realistic elevation model generation and roof type classification from stereo half-meter resolution satellite DSMs, is demonstrated in this work. Recently published deep learning architectures for both tasks are investigated and a new end-to-end cGAN-based network is developed, which combines different models that provide the best results for their individual tasks. To benefit from information provided by multiple data sources, a different cGAN-based work-flow is proposed where the generative part consists of two encoders and a common decoder which blends the intensity and height information within one network for the DSM refinement task. The inputs to the introduced network are single-channel photogrammetric DSMs with continuous values and pan-chromatic half-meter resolution satellite images. Information fusion from different modalities helps in propagating fine details, completes inaccurate or missing 3D information about building forms, and improves the building boundaries, making them more rectilinear. Lastly, additional comparison between the proposed methodologies for DSM enhancements is made to discuss and verify the most beneficial work-flow and applicability of the resulting DSMs for different remote sensing approaches.