Дисертації з теми "High spatial and spectral remote sensing"

Leafy spurge is a designated noxious weed. Accurate mapping and monitoring of this species are needed to understand leafy spurge's extent and spread. Current methods are based on ground crews who survey patches. Development of an affordable technique to map and monitor leafy spurge would contribute to the control of this species. High spatial, temporal, and spectral resolution imagery was used to classify the amount of leafy spurge present with ground and aerial-based imagery. A proof of concept study was performed in 2008 using ground-based images of an area infested with leafy spurge. This proof of concept project guided the development of the methods to be used for the 2009 aerial portion of the study. Thirty-five randomly selected reference points were selected in a range area in southwest Montana. These reference points were ground surveyed to record the density of leafy spurge in a 0.5-m radius area around the reference point. Images were captured approximately 108-m from the study area and classified using random forest classification. Multiple images were collected throughout the summer in order to determine at which time period leafy spurge is most easily detected. A classification using multiple image dates was also performed to determine if a time series of images improves classification. Single date accuracies were highest late in the summer with the highest single date classification achieving 83% accuracy. The multiple date classification significantly increased overall accuracy. Several aerial images were acquired in southwest Montana over the 2009 summer. Fifty randomly selected 2-m x 2-m reference areas were surveyed for percent cover of leafy spurge as well as several other variables. Aerial images were collected at flight elevations between 300-m to 460-m. Classifications were performed using random forest classifier, and both single date and multiple date classifications were performed. Leafy spurge was most accurately detected early and late in the growing season, and significant classification accuracy increases were observed with the multiple date classification. Single date accuracies achieved 90% accuracy in early June, while multiple date classifications achieved over 96% accuracy.

Thesis (Ph. D.)--West Virginia University, 2000.
Title from document title page. Document formatted into pages; contains x, 132 p. : ill. (some col.), maps (some col.). Includes abstract. Includes bibliographical references (p. 124-132).

2006/2007
Gli incendi boschivi rappresentano uno dei maggiori problemi ambientali nella regione Mediterranea con vaste superfici colpite ogni estate. Una stima dell’impatto ambientale degli incendi (a breve e a lungo termine) richiede la raccolta di informazioni accurate post-incendio relative al tipo di incendio, all’intensità, alla rigenerazione forestale ed al ripristino della vegetazione. L’utilizzo di tecniche avanzate di telerilevamento può fornire un valido strumento per lo studio di questi fenomeni. L’importanza di queste ricerche è stata più volte sottolineata dalla Commissione Europea che si è concentrata sullo studio degli incendi boschivi ed il loro effetto sulla vegetazione attraverso lo sviluppo di adeguati metodi di stima dell’impatto e di mitigazione. Scopo di questo lavoro è la stima dell’impatto post-incendio sulla vegetazione in ambiente Mediterraneo per mezzo di immagini satellitari ad alta risoluzione, di rilievi a terra e mediante tecniche avanzate di analisi dei dati. Il lavoro ha riguardato lo sviluppo di un sistema per l’integrazione di dati telerilevati ad altissima risoluzione spaziale e spettrale. Per la stima dell’impatto a breve termine, un modello di classificazione ad oggetti è stato sviluppato utilizzando immagini Ikonos ad altissima risoluzione spaziale per cartografare il tipo di incendio, differenziando l’incendio radente dall’incendio di chioma. I risultati mostrano che la classificazione ad oggetti potrebbe essere utilizzata per distinguere con elevata accuratezza (87% di accuratezza complessiva) le due tipologie di incendio, in particolare nei boschi Mediterranei aperti. È stata inoltre valutata la capacità della classificazione ad oggetti di distinguere e cartografare tre livelli di intensità del fuoco utilizzando le immagini Ikonos e l’accuratezza del risultato è stimata all’ 83%. Per la stima dell’impatto a lungo termine, la mappatura della rigenerazione post-incendio (pino) e la ripresa della vegetazione arbustiva sono state valutate mediante tre approcci: 1) la classificazione ad oggetti di immagini ad altissima risoluzione QuickBird che ha permesso di mappare la ripresa della vegetazione e l’impatto sulla copertura a seguito dell’incendio distinguendo due livelli di intensità dell’incendio (accuratezza della classificazione 86%). 2) l’analisi statistica di dati iperspettrali rilevati in campo che ha permesso una riduzione del 97% del volume di dati e la selezione delle migliori 14 bande per discriminare l’età e le specie di pino e le 18 migliori bande per la caratterizzazione delle specie arbustive. Successivamente, i dati iperspettrali Hyperion sono stati utlizzati per mappare la rigenerazione forestale e la ripresa della vegetazione. L’accuratezza complessiva della classificazione è stata del 75.1% considerando due diverse specie di pino ed altre specie vegetali. 3) una classificazione ad oggetti che ha combinato l’analisi dei dati QuickBird ed Hyperion. Si è registrato un aumento dell’accuratezza della classificazione pari all’8.06% rispetto all’utilizzo dei soli dati Hyperion. Complessivamente, si osserva che strumenti avanzati di telerilevamento consentono di raccogliere le informazioni relative alle aree incendiate, la rigenerazione forestale e la ripresa della vegetazione in modo accurato e vantaggioso in termini di costi e tempi.
Forest fires are a major environmental problem in the Mediterranean region, where large areas are affected each summer. An assessment of the environmental impact of forest fires (in the short-term and in the long-term) requires the collection of accurate and detailed post-fire information related to fire type, fire severity, forest regeneration and vegetation recovery. Advanced tools in remote sensing provide a powerful tool for the study of this phenomenon. The importance of this work was often emphasized by the European Commission, which focused on the studying of forest fires and their effect on vegetation through the development of appropriate impact assessment and mitigation methods. The aim of this study was to assess the post-fire impact on vegetation in a Mediterranean environment by employing high quality satellite and field data and by using advanced data processing techniques. The work entailed the development of a whole system integrating very high spatial and spectral resolution remotely sensed data. For short-term impact assessment, an object-oriented model was developed using very high spatial resolution Ikonos imagery to map the type of fire, namely, canopy fire and surface fire. The results showed that object-oriented classification could be used to accurately distinguish and map areas of surface and crown fire spread (overall accuracy of 87%), especially that occurring in open Mediterranean forests. Also, the performance of object-based classification in mapping three levels of fire severity by employing high spatial resolution Ikonos imagery was evaluated, and accuracy of the obtained results was estimated to be 83%. As for long-term impact assessment, the mapping of post-fire forest regeneration (pine) and vegetation recovery (shrub) was performed by following three different approaches. First, the developed object-based classification of QuickBird (very high spatial resolution) allowed post-fire vegetation recovery and survival mapping of canopy within two different fire severity levels (86% of classification accuracy). The main effect of fire has been to create a more homogeneous landscape. Second, statistical analysis of field hyperspectral data allowed a 97% reduction in data volume and recommended 14 best narrowbands to discriminate among pine trees (age and species) and 18 bands that best characterize the different shrub species. Then, hyperspectral Hyperion was employed for mapping post-fire forest regeneration and vegetation recovery. The overall classification accuracy was found to be 75.81% when mapping two different regenerated pine species and other species of vegetation recovery. Third, an object-oriented combined analysis of QuickBird and Hyperion was investigated for the same objective. An improvement in classification accuracy of 8.06% was recorded when combining both Hyperion and QuickBird imageries than by using only the Hyperion image. Overall, it was observed that advanced tools in remote sensing provided the necessary means for gathering information about the burned areas, the regenerated forests and the recovered vegetations in a successful and a timely/cost effective manner.
XX Ciclo
1977

Native vegetation condition provides an indication of the state of vegetation health or function relative to a stated objective or benchmark. Measures of vegetation condition provide an indication of the vegetation's capacity to provide habitat for a range of species and ecosystem functions through the assessment of selected vegetation attributes. Subsets of vegetation attributes are often combined into vegetation condition indices or metrics, which are used to provide information for natural resource management. Despite their value as surrogates of biota and ecosystem function, measures of vegetation condition are rarely used to inform biodiversity assessments at scales beyond individual stands. The extension of vegetation condition information across landscapes, and approaches for achieving this, using remote sensing technologies, is a key focus of the work presented in this thesis. The aim of this research is to assess the utility of multi-spectral remotely sensed data for the recovery of stand-level attributes of native vegetation condition at landscape scales. The use of remotely sensed data for the assessment of vegetation condition attributes in fragmented landscapes is a focus of this study. The influence of a number of practical issues, such as spatial scale and ground data sampling methodology, are also explored. This study sets limitations on the use of this technology for vegetation condition assessment and also demonstrates the practical impact of data quality issues that are frequently encountered in these types of applied integrated approaches. The work presented in this thesis demonstrates that while some measures of vegetation condition, such as vegetation cover and stem density, are readily recoverable from multi-spectral remotely sensed data, others, such as hollow-bearing trees and log length, are not easily derived from this type of data. The types of information derived from remotely sensed data, such as texture measures and vegetation indices, that are useful for vegetation condition assessments of this nature are also highlighted. The utility of multi-spectral remotely sensed data for the assessment of stand-level vegetation condition attributes is highly dependent on a number of factors including the type of attribute being measured, the characteristics of the vegetation, the sensor characteristics (i.e. the spatial, spectral, temporal, and radiometric resolution), and other spatial data quality considerations, such as site homogeneity and spatial scale. A series of case studies are presented in this thesis that explores the effects of these factors. These case studies demonstrate the importance of different aspects of spatial data and how data manipulation can greatly affect the derived relationships between vegetation attributes and remotely sensed data. The work documented in this thesis provides an assessment of what can be achieved from two sources of multi-spectral imagery in terms of recovery of individual vegetation attributes from remotely sensed data. Potential surrogate measures of vegetation condition that can be derived across broad scales are identified. This information could provide a basis for the development of landscape scale multi-spectral remotely sensed based vegetation condition assessment approaches, supplementing information provided by established site-based vegetation condition assessment approaches.

In this thesis, we aim to systematically understand the relationship between cloud spatial structure and its radiation imprints, i.e., three-dimensional (3D) cloud effects, with the ultimate goal of deriving accurate radiative energy budget estimates from space, aircraft, or ground-based observations under spatially inhomogeneous conditions. By studying the full spectral information in the measured and modeled shortwave radiation fields of heterogeneous cloud scenes sampled during aircraft field experiments, we find evidence that cloud spatial structure reveals itself through spectral signatures in the associated irradiance and radiance fields in the near-ultraviolet and visible spectral range.

The spectral signature of 3D cloud effects in irradiances is apparent as a domain- wide, consistent correlation between the magnitude and spectral dependence of net horizontal photon transport. The physical mechanism of this phenomenon is molecular scattering in conjunction with cloud heterogeneity. A simple parameterization with a single parameter ϵ is developed, which holds for individual pixels and the domain as a whole. We then investigate the impact of scene parameters on the discovered correlation and find that it is upheld for a wide range of scene conditions, although the value of ϵ varies from scene to scene.

The spectral signature of 3D cloud effects in radiances manifests itself as a distinct relationship between the magnitude and spectral dependence of reflectance, which cannot be reproduced in the one-dimensional (1D) radiative transfer framework. Using the spectral signature in radiances and irradiances, it is possible to infer information on net horizontal photon transport from spectral radiance perturbations on the basis of pixel populations in sub-domains of a cloud scene.

We show that two different biases need to be considered when attempting radiative closure between measured and modeled irradiance fields below inhomogeneous cloud fields: the remote sensing bias (affecting cloud radiances and thus retrieved properties of the inhomogeneous scene) and the irradiance bias (ignoring 3D effects in the calculation of irradiance fields from imagery-based cloud retrievals). The newly established relationships between spatial and spectral structure lay the foundation for first-order corrections for these 3D biases within a 1D framework, once the correlations are explored on a more statistical basis.

Dry and wet fine earth spectral measurements were made on the Ap soil surface horizons on the Maricopa Agricultural Center by using a Barnes Modular Multiband Radiometer. Three subsets were used in the analyses 552, 101 and 11. There were three soil series, Casa Grande, Shontik and Trix, four soil mapping units, and three texture classes identified on the farm. The wet soil condition reduced the amplitude of the spectral curves over the entire spectrum range (0.45 to 2.35 μm). The spectral curves were statistically related to the soil mapping units to determine if the soil mapping units and texture classes could be separated. The wet soil condition and the smaller sample size increased the correct classification percentages for soil mapping units and texture classes. LSD tests showed there were significant differences between these groups. Simple- and Multiple-linear regression analysis were used to relate some soil physical (sand, silt and clay contents and color components) and chemical (iron oxide, organic carbon and calcium carbonate contents) to soil spectral responses in the seven bands under dry and wet conditions. There were high correlations levels among the spectral bands showing an overlap of spectral information. Generally, the red (MMR3) and near-infrared (MMR4) bands had the highest correlations with the studied soil properties under dry and wet conditions. Usually, the wet soil condition resulted in higher correlations than that for the dry soil condition over the total spectrum range. The predictive equations for sand, silt and clay and iron oxide contents were satisfactory. For organic carbon and color components, the greatest success was achieved when variation in spectral response within individual samples are smaller than that between soil mapping unit group averages. There was a poor relation between calcium carbonate and spectral response. A comparison of multi-level remotely sensed data collected by SPOT, aircraft, and ground instruments showed a strong agreement among the data sets, which correlated well to fine earth data, except for the SPOT data. Rough soil surfaces showed a reduction in reflectance altitude compared to laser level, and it appears to be directly proportional to the percent shadow in the viewing area measured by SPOT satellite and aircraft.

The recent advances in aerial- and satellite-based hyperspectral imaging sensor technologies have led to an increased availability of Earth's images with high spatial and spectral resolution, which opened the door to a large range of important applications. Hyperspectral imaging records detailed spectrum of the received light in each spatial position in the image, in which each pixel contains a highly detailed representation of the reflectance of the materials present on the ground, and a better characterization in terms of geometrical details. Since different substances exhibit different spectral signatures, the abundance of informative content conveyed in the hyperspectral images permits an improved characterization of different land coverage. Therefore, hyperspectral imaging emerged as a well-suited technology for accurate image classi fication in remote sensing. In spite of that, a signi ficantly increased complexity of the analysis introduces a series of challenges that need to be addressed on a serious note. In order to fully exploit the potential offered by these sensors, there is a need to develop accurate and effective models for spectral-spatial analysis of the recorded data. This thesis aims at presenting novel strategies for the analysis and classifi cation of hyperspectral remote sensing images, placing the focal point on the investigation on deep networks for the extraction and integration of spectral and spatial information. Deep learning has demonstrated cutting-edge performances in computer vision, particularly in object recognition and classi cation. It has also been successfully adopted in hyperspectral remote sensing domain as well. However, it is a very challenging task to fully utilize the massive potential of deep models in hyperspectral remote sensing applications since the number of training samples is limited which limits the representation capability of a deep model. Furthermore, the existing architectures of deep models need to be further investigated and modifi ed accordingly to better complement the joint use of spectral and spatial contents of hyperspectral images. In this thesis, we propose three different deep learning-based models to effectively represent spectral-spatial characteristics of hyperspectral data in the interest of classifi cation of remote sensing images. Our first proposed model focuses on integrating CRF and CNN into an end-to-end learning framework for classifying images. Our main contribution in this model is the introduction of a deep CRF in which the CRF parameters are computed using CNN and further optimized by adopting piecewise training. Furthermore, we address the problem of over fitting by employing data augmentation techniques and increased the size of the training samples for training deep networks. Our proposed 3DCNN-CRF model can be trained to fully exploit the usefulness of CRF in the context of classi fication by integrating it completely inside of a deep model. Considering that the separation of constituent materials and their abundances provide detailed analysis of the data, our second algorithm investigates the potential of using unmixing results in deep models to classify images. We extend an existing region based structure preserving non-negative matrix factorization method to estimate groups of spectral bands with the goal to capture subtle spectral-spatial distribution from the image. We subsequently use these important unmixing results as input to generate superpixels, which are further represented by kernel density estimated probability distribution function. Finally, these abundance information-guided superpixels are directly supplied into a deep model in which the inference is implicitly formulated as a recurrent neural network to perform the eventual classifi cation. Finally, we perform a detailed investigation on the possibilities of adopting generative adversarial models into hyperspectral image classifi cation. We present a GAN-based spectral-spatial method that primarily focuses on signifi cantly improving the multiclass classi cation ability of the discriminator of GAN models. In this context, we propose to adopt the triplet constraint property and extend it to build a useful feature embedding for remote sensing images for use in classi cation. Furthermore, our proposed Triplet- 3D-GAN model also includes feedback from discriminator's intermediate features to improve the quality of the generator's sample generation process.
Thesis (PhD Doctorate)
Doctor of Philosophy (PhD)
School of Info & Comm Tech
Science, Environment, Engineering and Technology
Full Text

Thesis (Ph. D.)--University of Adelaide, Dept. of Electrical and Electronic Engineering, 1994?
One computer disk in pocket inside back cover. System requirements for accompanying computer disk: Macintosh computer. Includes bibliographical references (leaves 148-160).

El análisis de imágenes ha impulsado muchos descubrimientos en la ciencia actual. Esta tesis se centra en el análisis de imágenes remotas para inspección aérea, exactamente en el problema de segmentación y clasificación de acuerdo al uso del suelo. Desde el nacimiento de los sensores hiperespectrales su uso ha sido vital para esta tarea ya que facilitan y mejoran sustancialmente el resultado. Sin embargo el uso de imágenes hiperespectrales entraña, entre otros, problemas de dimensionalidad y de interacción con los expertos. Proponemos mejoras que ayuden a paliar estos inconvenientes y hagan el problema mas eficiente.
Lately image analysis have aided many discoveries in research. This thesis focusses on the analysis of remote sensed images for aerial inspection. It tackles the problem of segmentation and classification according to land usage. In this field, the use of hyperspectral images has been the trend followed since the emergence of hyperspectral sensors. This type of images improves the performance of the task but raises some issues. Two of those issues are the dimensionality and the interaction with experts. We propose enhancements overcome them. Efficiency and economic reasons encouraged to start this work. The enhancements introduced in this work allow to tackle segmentation and classification of this type of images using less data, thus increasing the efficiency and enabling the design task specific sensors which are cheaper. Also, our enhacements allow to perform the same task with less expert collaboration which also decreases the costs and accelerates the process.

This dissertation consists of three studies dealing with particulate organic carbon (POC). The first study describes the temporal and spatial variability of particulate matter (PM) and POC, and physical processes that affect the distribution of PM and POC with synchronous remote sensing data. The purpose of the second study is to develop POC algorithms in the Gulf of Mexico based on satellite data using numerical methods and to compare POC estimates with spectral radiance. The purpose of the third study is to investigate climatological variations from the temporal and spatial POC estimates based on SeaWiFS spectral radiance and physical processes, and to determine the physical mechanisms that affect the distribution of POC in the Gulf of Mexico. For the first and second studies, hydrographic data from the Northeastern Gulf of Mexico (NEGOM) study were collected on each of 9 cruises from November 1997 to August 2000 across 11 lines. Remotely sensed data sets were obtained from NASA and NOAA using algorithms that have been developed for interpretation of ocean color data from various satellite sensors. For the third study, we use the time-series of POC estimates, sea surface temperature (SST), sea surface height anomaly (SSHA), sea surface wind (SSW), and precipitation rate (PR) that might cause climatological variability and physical processes. The distribution of surface PM and POC concentrations were affected by one or more factors such as river discharge, wind stress, stratification, and the Loop Current/Eddies. To estimate POC concentration, empirical and model-based approaches were used using regression and principal component analysis (PCA) methods. We tested simulated data for reasonable and suitable algorithms in Case 1 and Case 2 waters. Monthly mean values of POC concentrations calculated with PCA algorithms. The spatial and temporal variations of POC and physical forcing data were analyzed with the empirical orthogonal function (EOF) method. The results showed variations in the Gulf of Mexico on both annual and inter-annual time scales.

La thèse propose des nouvelles techniques pour la classification et le démelange spectraldes images obtenus par télédétection iperspectrale. Les problèmes liées au données (notammenttrès grande dimensionalité, présence de mélanges des pixels) ont été considerés et destechniques innovantes pour résoudre ces problèmes. Nouvelles méthodes de classi_cationavancées basées sur l'utilisation des méthodes traditionnel de réduction des dimension etl'integration de l'information spatiale ont été développés. De plus, les méthodes de démelangespectral ont été utilisés conjointement pour ameliorer la classification obtenu avec lesméthodes traditionnel, donnant la possibilité d'obtenir aussi une amélioration de la résolutionspatial des maps de classification grace à l'utilisation de l'information à niveau sous-pixel.Les travaux ont suivi une progression logique, avec les étapes suivantes:1. Constat de base: pour améliorer la classification d'imagerie hyperspectrale, il fautconsidérer les problèmes liées au données : très grande dimensionalité, presence demélanges des pixels.2. Peut-on développer méthodes de classi_cation avancées basées sur l'utilisation des méthodestraditionnel de réduction des dimension (ICA ou autre)?3. Comment utiliser les differents types d'information contextuel typique des imagés satellitaires?4. Peut-on utiliser l'information données par les méthodes de démelange spectral pourproposer nouvelles chaines de réduction des dimension?5. Est-ce qu'on peut utiliser conjointement les méthodes de démelange spectral pour ameliorerla classification obtenu avec les méthodes traditionnel?6. Peut-on obtenir une amélioration de la résolution spatial des maps de classi_cationgrace à l'utilisation de l'information à niveau sous-pixel?Les différents méthodes proposées ont été testées sur plusieurs jeux de données réelles, montrantresultats comparable ou meilleurs de la plus part des methodes presentés dans la litterature.

As developing countries experience substantial urban growth and expansion, remotely sensed based estimates of population and demographic characteristics can provide researchers and humanitarian aid workers timely and spatially explicit information for planning and development. In this exploratory analysis, high spatial resolution satellite imagery, in combination with fine resolution census data, is used to determine the degree to which spatial features are able to identify spatial patterns of demographic variables in Accra, Ghana. Traditionally when using satellite imagery, spectral characteristics are used on a per-pixel basis to produce land cover classifications; however, in this study, a new methodology is presented that quantifies spatial characteristics of built-up areas, and directly relates them to census-derived variables. Spatial features are image metrics that analyze groups of pixels in order to describe the geometry, orientation, and patterns of objects in an image. By using spatial features, city infrastructure variations, such as roads and buildings, can be quantified and related to census-derived variables, such as living standards, housing conditions, employment and education. To test the associations between spatial patterns and demographic variables, five spatial features (line support regions, PanTex, histograms of oriented gradients, local binary patterns, and Fourier transform) were quantified and extracted from the imagery, and then correlated to census-derived variables. Findings demonstrate that, while spectral information (such as the normalized difference vegetation index) reveals many strong correlations with population density, housing density, and living standards, spatial features provide comparable correlation coefficients with density and housing characteristics. The results from this study suggest that there are relationships between spatial features derived from satellite imagery and socioeconomic characteristics of the people of Accra, Ghana.

The large extension and diversity of the Cerrado vegetative cover, the second largest biome in South America, has a strong impact on regional, and possibly global, energy, water, and carbon balances. Nevertheless, as a major farming frontier in Brazil, it is estimated that about 40% of the Cerrado land cover has already been converted into cultivated pastures, field crops, urban development, and degraded areas. Despite this aggressive pace of land conversion, there have been few investigations on the operational utilization of remote sensing data to effectively monitor and understand this biome. Within this context, and within the goals and framework of the Large Scale Biosphere-Atmosphere Experiment in Amazonia (LBA), we evaluated the usefulness of spectral vegetation indices (VIs), to effectively monitor the Cerrado, detect land conversions, and discriminate and assess the conditions of the major structural types of Cerrado vegetation. Using a full hydrologic year (1995) of AVHRR, local-area-coverage (LAC), data over the Cerrado, converted to normalized difference vegetation index (NDVI) and soil adjusted vegetation index (SAVI), we were able to spatially discriminate three major communities based on their phenologic patterns. These included savanna formations and pasture sites, forested areas, and agricultural crops. We also analyzed wet and dry season, aircraft-based radiometric data and a ground-based set of biophysical measurements, collected over the Brasilia National Park (BNP), the largest LBA core site in the Cerrado biome. Overall, we found the MODIS vegetation indices, which include a continuity NDVI and the new enhanced vegetation index (EVI), to provide better performance capabilities with improved dynamic ranges and contrasts in seasonal dynamics. Land cover discrimination was favored by the NDVI, while the EVI more strongly responded to the seasonal contrast of the vegetative cover. Thus, the synergistic use of the MODIS VI products will very likely result in an improved monitoring capability and understanding of the Cerrado biome.

The Middle Rio Grande River (MRGR) is the main source of fresh water for the state of New Mexico. Located in an arid area with scarce local water resources, this has led to extensive diversions of river water to supply the high demand from municipalities and irrigated agricultural activities. The extensive water diversions over the last few decades have affected the composition of the native riparian vegetation by decreasing the area of cottonwood and coyote willow and increasing the spread of invasive species such as Tamarisk and Russian Olives, harmful to the river system, due to their high transpiration rates, which affect the river aquatic system. The need to study the river hydrological processes and their relation with its health is important to preserve the river ecosystem. To be able to do that a detailed vegetation map was produced using a Utah State University airborne remote sensing system for 286 km of river reach. Also a groundwater model was built in ArcGIS environment which has the ability to estimate soil water potential in the root zone and above the modeled water table. The Modified Penman- Monteith empirical equation was used in the ArcGIS environment to estimate riparian vegetation ET, taking advantage of the detailed vegetation map and spatial soil water potential layers. Vegetation water use per linear river reach was estimated to help decision makers to better manage and release the amount of water that keeps a sound river ecosystem and to support agricultural activities.

The capability to monitor bushfires on a large scale from space has long been identified as an important contribution to climate and atmospheric research as well as a tool an aid in natural hazard response. Since the work by Dozier (1981), fire monitoring from space has relied on the principles he described. His method of identifying fires within a pixel significantly larger than the fire by utilizing the different responses of the 3 μm and 11 μm channels has been applied to a number of sensors. Over the last decade a lot of work has been invested to refine and validate fire detections based on this approach. So far, the application of the method proposed by Dozier (1981) reached its peak with the launch of the MODIS instrument on board the Terra satellite. In contrast to earlier sensors, MODIS was equipped with spectral channels specifically designed for the detection of fires with algorithms based on the work by Dozier (1981). These channels were designed to overcome problems experienced with other platforms, the biggest of which is the saturation of the 3 μm channel caused by big, hot fires. Since its launch, MODIS has proven itself to be a capable platform to provide worldwide fire detection at a moderate resolution of 1 km on a daily basis.It is the intention of this work to open up new opportunities in remote sensing of fires from satellites by showing capabilities and limitations in the application of other spectral channels, in particular the 2.1 μm channel of MODIS, than the ones currently used. This channel is chosen for investigation as fires are expected to emit a significant amount of energy in this bandwidth and as it is available at a native resolution of 500 m on MODIS; double the resolution of the 3 μm and 11 μm channels. The modelling of blackbodies of typical bushfire temperatures shows that a fire detection method based on the 2.1 μm channel will not be able to replace the current methods. Blackbodies of temperatures around 600 to 700 K, that are common for smoldering fires, do not emit a great amount of energy at 2.1 μm. It would be hardly possible to detect those fires by utilizing the 2.1 μm channel. The established methods based on the 3 μm and 11 μm channels are expected to work better in these cases. Blackbodies of typically flaming fires (above 800 K) however show a very high emission around 2.1 μm that should make their detection using the 2.1 μm channel possible.In order to develop a fire detection method based on the 2.1 μm channel, it is necessary to differentiate between the radiance caused by a fire of sub pixel size and the radiance of a pixel caused by the reflection of sunlight. This is attempted by using time series of past observations to model a reflectance value for a given pixel expected in absence of a fire. A fire detection algorithm exploiting the difference between the expected and observed reflectance is implemented and its detection results are compared to high resolution ASTER fire maps, the standard MODIS fire detection algorithm (MOD14) and burnt area maps. The detections of the method based on the 2.1 μm channel are found to correspond very well with the other three datasets. However, the comparison showed detections that do not align with MOD14 active fire detections but are generally aligned with burn areas. This phenomena has to be investigated in the future.

Satellite imagery such as Landsat has been in use for decades for many landscape and regional scale mapping applications, but has been too coarse for use in detailed forest inventories where stand level structural and compositional information is desired. Recently available high spatial resolution satellite imagery may be well suited to mapping fine-scale components of ecosystems, however, this remains an area of ongoing research. The first goal of this thesis was to assess the capacity of high spatial resolution satellite imagery to detect the variability in late seral coastal temperate rainforests in British Columbia, Canada. Using an object-based classifier, two hierarchical classification schemes are evaluated: a broad classification based on structural (successional) stage and a finer classification of late seral vegetation associations. The finer-scale classification also incorporates ancillary landscape positional variables (elevation and potential soil moisture) derived from Light Detection and Ranging (LiDAR) data, and the relative contribution of spectral, textural and landscape positional data for this classification is determined. Results indicate that late seral forests can be well distinguished from younger forests using QuickBird spectral and textural data. However, discrimination among late seral forest associations is challenging, especially in the absence of landscape positional variables. Classification accuracies were particularly low for rare forest associations. Given this finding, the objective of the third chapter was to explicitly examine the caveats of using high spatial resolution imagery to map rare classes. Classification accuracy is assessed in several different ways in order to examine the impact on perceived map accuracy. In addition, the effects on habitat extent and configuration resulting from post-classification implementation of a minimum mapping unit are examined. Results indicate that classification accuracies may vary considerably depending on the assessment technique used. Specifically, ignoring the presence of fine-scale heterogeneity in a classification during accuracy assessment falsely lowered the accuracy estimates. Further, post-classification smoothing had a large effect on the spatial pattern of rare classes. These findings suggest that routinely used image classification and assessment techniques can greatly impact mapping of rare classes.

Master of Science
Department of Agronomy
William T. Schapaugh
Remote sensing technologies might serve as indirect selection tools to improve phenotyping to differentiate genotypes for yield in soybean breeding program as well as the assessment of soybean cyst nematode (SCN), Heterodera glycines. The objective of these studies were to: i) investigate potential use of spectral reflectance indices (SRIs) and canopy temperature (CT) as screening tools for soybean grain yield in an elite, segregating population; ii) determine the most appropriate growth stage(s) to measure SRI’s for predicting grain yield; and iii) estimate SCN population density among and within soybean cultivars utilizing canopy spectral reflectance and canopy temperature. Experiment 1 was conducted at four environments (three irrigated and one rain-fed) in Manhattan, KS in 2012 and 2013. Each environment evaluated 48 F4- derived lines. In experiment 2, two SCN resistant cultivars and two susceptible cultivars were grown in three SCN infested field in Northeast KS, in 2012 and 2013. Initial (Pi) and final SCN soil population (Pf) densities were obtained. Analyses of covariance (ANCOVA) revealed that the green normalized vegetation index (GNDVI) was the best predictive index for yield compared to other SRI’s and differentiated genotype performance across a range of reproductive growth stages. CT did not differentiate genotypes across environments. In experiment 2, relationships between GNDVI, reflectance at single wavelengths (675 and 810 nm) and CT with Pf were not consistent across cultivars or environments. Sudden death syndrome (SDS) may have confounded the relationships between remote sensing data and Pf. Therefore, it would be difficult to assess SCN populations using remote sensing based on these results.

The current study has been conducted in response to the growing problem of water scarcity and the need for more effective methods of irrigation water management. Remote sensing techniques have been used to match spatially and temporally distributed crop water demand to water application rates. Remote sensing approaches using Landsat imagery have been applied to estimate the components of a soil water balance model for an agricultural field by determining daily values of surface/root-zone soil moisture, evapotranspiration rates, and losses and by developing a forecasting model to generate optimal irrigation application information on a daily basis. Incompatibility of coarse resolution Landsat imagery (30m by 30m) with heterogeneities within the agricultural field and potential underestimation of field variations led the study to its main objective, which was to develop models capable of representing spatial and temporal variations within the agricultural field at a compatible resolution with farming management activities. These models support establishing real-time management of irrigation water scheduling and application. The AggieAirTM Minion autonomous aircraft is a remote sensing platform developed by the Utah Water Research Laboratory at Utah State University. It is a completely autonomous airborne platform that captures high-resolution multi-spectral images in the visual, near infrared, and thermal infrared bands at 15cm resolution. AggieAir flew over the study area on four dates in 2013 that were coincident with Landsat overflights and provided similar remotely sensed data at much finer resolution. These data, in concert with state-of-the-art supervised learning machine techniques and field measurements, have been used to model surface and root zone soil volumetric water content at 15cm resolution. The information provided by this study has the potential to give farmers greater precision in irrigation water allocation and scheduling.

The accurate estimation of forest attributes at many different spatial scales is a critical problem. Forest landowners may be interested in estimating timber volume, forest biomass, and forest structure to determine their forest's condition and value. Counties and states may be interested to learn about their forests to develop sustainable management plans and policies related to forests, wildlife, and climate change. Countries and consortiums of countries need information about their forests to set global and national targets to deal with issues of climate change and deforestation as well as to set national targets and understand the state of their forest at a given point in time.

This dissertation approaches these questions from two perspectives. The first perspective uses the process model Biome-BGC paired with inventory and remote sensing data to make inferences about a current forest state given known climate and site variables. Using a model of this type, future climate data can be used to make predictions about future forest states as well. An example of this work applied to a forest in northern California is presented. The second perspective of estimating forest attributes uses high resolution aerial imagery paired with light detection and ranging (LiDAR) remote sensing data to develop statistical estimates of forest structure. Two approaches within this perspective are presented: a pixel based approach and an object based approach. Both approaches can serve as the platform on which models (either empirical growth and yield models or process models) can be run to generate inferences about future forest state and current forest biogeochemical cycling.

The capability to monitor bushfires on a large scale from space has long been identified as an important contribution to climate and atmospheric research as well as a tool an aid in natural hazard response. Since the work by Dozier (1981), fire monitoring from space has relied on the principles he described. His method of identifying fires within a pixel significantly larger than the fire by utilizing the different responses of the 3 μm and 11 μm channels has been applied to a number of sensors. Over the last decade a lot of work has been invested to refine and validate fire detections based on this approach. So far, the application of the method proposed by Dozier (1981) reached its peak with the launch of the MODIS instrument on board the Terra satellite. In contrast to earlier sensors, MODIS was equipped with spectral channels specifically designed for the detection of fires with algorithms based on the work by Dozier (1981). These channels were designed to overcome problems experienced with other platforms, the biggest of which is the saturation of the 3 μm channel caused by big, hot fires. Since its launch, MODIS has proven itself to be a capable platform to provide worldwide fire detection at a moderate resolution of 1 km on a daily basis.
It is the intention of this work to open up new opportunities in remote sensing of fires from satellites by showing capabilities and limitations in the application of other spectral channels, in particular the 2.1 μm channel of MODIS, than the ones currently used. This channel is chosen for investigation as fires are expected to emit a significant amount of energy in this bandwidth and as it is available at a native resolution of 500 m on MODIS; double the resolution of the 3 μm and 11 μm channels. The modelling of blackbodies of typical bushfire temperatures shows that a fire detection method based on the 2.1 μm channel will not be able to replace the current methods. Blackbodies of temperatures around 600 to 700 K, that are common for smoldering fires, do not emit a great amount of energy at 2.1 μm. It would be hardly possible to detect those fires by utilizing the 2.1 μm channel. The established methods based on the 3 μm and 11 μm channels are expected to work better in these cases. Blackbodies of typically flaming fires (above 800 K) however show a very high emission around 2.1 μm that should make their detection using the 2.1 μm channel possible.
In order to develop a fire detection method based on the 2.1 μm channel, it is necessary to differentiate between the radiance caused by a fire of sub pixel size and the radiance of a pixel caused by the reflection of sunlight. This is attempted by using time series of past observations to model a reflectance value for a given pixel expected in absence of a fire. A fire detection algorithm exploiting the difference between the expected and observed reflectance is implemented and its detection results are compared to high resolution ASTER fire maps, the standard MODIS fire detection algorithm (MOD14) and burnt area maps. The detections of the method based on the 2.1 μm channel are found to correspond very well with the other three datasets. However, the comparison showed detections that do not align with MOD14 active fire detections but are generally aligned with burn areas. This phenomena has to be investigated in the future.

Vorliegende Arbeit zeigt die jüngsten Möglichkeiten hochauflösender Fernerkundung am Beispiel von Stadtbäumen in Berlin, Deutschland. Es wurden neuste methodische Ansätze eingesetzt, wie beispielsweise maschinelles Lernens und individuelle Baumdetektion. Sie erwiesen sich von großem Vorteil für die detaillierte Analyse urbaner Ökosystemdienstleistungen in einer heterogenen Umwelt. Neueste Fernerkundung von hoher zeitlicher Auflösung hat Möglichkeiten gezeigt, Veränderungen des Stadtwaldes präziser zu untersuchen. Diesbezüglich konnten Baumspezies klassifiziert werden auf Grundlage saisonaler Veränderungen, die mittels Fernerkundungsdaten aufgenommen wurden. Dies ist für den urbanen Bereich einmalig und über große Flächen noch nicht durchgeführt worden. Darüber hinaus haben diese Baumarten einzelnen Bäumen zugeordnet werden können, deren Abmessung fernerkundlich erfasst worden ist. Diese neu erzeugten Umweltinformationen einzelner Bäume können damit verbundene urbane Ökosystemdienstleistungen präzise aktualisieren. Zum Beispiel haben so Unsicherheiten in der Schätzung zur Kohlenstoffspeicherung städtischer Wälder reduziert werden können. Es ist zudem von Vorteil gewesen, den gegenwärtigen Mangel an räumlich expliziten dreidimensionalen Informationen über Stadtwälder anzusprechen. Allerdings ist die Rolle städtischen Wälder, das Treibhausgas CO2 langfristig auszugleichen, immer noch wenig untersucht. Gerade der Mangel an präzisen, konsistenten und aktuellen Details führt zu großen Unsicherheiten im Rahmen von Lebenszyklus-Analysen. Auf Grund des aktuellen Fortschritts in hochauflösender Fernerkundung könnten diese Unsicherheiten reduziert werden. Dazu werden Möglichkeiten ausgiebig kritisch bewertet und anhand einer Lebenszyklus-Analyse am Beispiel Berlin andiskutiert, inwieweit sie präzisere langfristige Prognosen zum Stadtwald als Kohlenstoffspeicher liefern.
This work shows recent options for implementing high resolution remote sensing in assessing urban trees in Berlin, Germany. State-of-the-art methodological approaches like machine learning and individual tree detection proved to be highly advantageous for analyzing details of urban ecosystem services within a heterogeneous urban environment. Recent remote sensing of high temporal resolution offers new options for more precisely addressing urban forest dynamics. This successfully shows that tree species could be identified from seasonal changes of remotely sensed imagery, though this has not yet been applied across cities. Furthermore, these tree species results could be combined with remotely sensed individual tree dimensions. This newly generated data can be suggested to update spatially explicit information on related urban ecosystem services. For example, this could reduce the uncertainties of such estimates as urban forest carbon storage, and also address the present lack of spatially explicit three-dimensional information on urban forests. However, few studies have considered the local scale of urban forests to effectively evaluate their potential long-term carbon offset. The lack of precise, consistent and up-to-date forest details is challenging within the scope of life cycle assessments. This can cause high uncertainties in urban forest carbon offset. Although, recent progress in high resolution remote sensing is promising to reduce these uncertainties. For this purpose, remote sensing options are extensively reviewed and briefly discussed using an example of life cycle assessment for Berlin, which allow more precise long-term prognoses of urban forest carbon offset.

Doctor of Philosophy
Department of Geography
Douglas G. Goodin
Jesse A. Poland
Kevin Price
Wheat breeders are in a race for genetic gain to secure the future nutritional needs of a growing population. Multiple barriers exist in the acceleration of crop improvement. Emerging technologies are reducing these obstacles. Advances in genotyping technologies have significantly decreased the cost of characterizing the genetic make-up of candidate breeding lines. However, this is just part of the equation. Field-based phenotyping informs a breeder’s decision as to which lines move forward in the breeding cycle. This has long been the most expensive and time-consuming, though most critical, aspect of breeding. The grand challenge remains in connecting genetic variants to observed phenotypes followed by predicting phenotypes based on the genetic composition of lines or cultivars. In this context, the current study was undertaken to investigate the utility of UAS in assessment field trials in wheat breeding programs. The major objective was to integrate remotely sensed data with geospatial analysis for high throughput phenotyping of large wheat breeding nurseries. The initial step was to develop and validate a semi-automated high-throughput phenotyping pipeline using a low-cost UAS and NIR camera, image processing, and radiometric calibration to build orthomosaic imagery and 3D models. The relationship between plot-level data (vegetation indices and height) extracted from UAS imagery and manual measurements were examined and found to have a high correlation. Data derived from UAS imagery performed as well as manual measurements while exponentially increasing the amount of data available. The high-resolution, high-temporal HTP data extracted from this pipeline offered the opportunity to develop a within season grain yield prediction model. Due to the variety in genotypes and environmental conditions, breeding trials are inherently spatial in nature and vary non-randomly across the field. This makes geographically weighted regression models a good choice as a geospatial prediction model. Finally, with the addition of georeferenced and spatial data integral in HTP and imagery, we were able to reduce the environmental effect from the data and increase the accuracy of UAS plot-level data. The models developed through this research, when combined with genotyping technologies, increase the volume, accuracy, and reliability of phenotypic data to better inform breeder selections. This increased accuracy with evaluating and predicting grain yield will help breeders to rapidly identify and advance the most promising candidate wheat varieties.

Abstract

Since 1999, Very High spatial Resolution satellite data (Ikonos-2, QuickBird and OrbView-3) represent the surface of the Earth with more detail. However, information extraction by multispectral pixel-based classification proves to have become more complex owing to the internal variability increase in the land-cover units and to the weakness of spectral resolution.

Therefore, one possibility is to consider the internal spectral variability of land-cover classes as a valuable source of spatial information that can be used as an additional clue in characterizing and identifying land cover. Moreover, the spatial resolution gap that existed between satellite images and aerial photographs has strongly decreased, and the features used in visual interpretation transposed to digital analysis (texture, morphology and context) can be used as additional information on top of spectral features for the land cover classification.

The difficulty of this approach is often to transpose the visual features to digital analysis.

To overcome this problem region-based classification could be used. Segmentation, before classification, produces regions that are more homogeneous in themselves than with nearby regions and represent discrete objects or areas in the image. Each region becomes then a unit analysis, which makes it possible to avoid much of the structural clutter and allows to measure and use a number of features on top of spectral features. These features can be the surface, the perimeter, the compactness, the degree and kind of texture. Segmentation is one of the only methods which ensures to measure the morphological features (surface, perimeter.) and the textural features on non-arbitrary neighbourhood. In the pixel-based methods, texture is calculated with mobile windows that smooth the boundaries between discrete land cover regions and create between-class texture. This between-class texture could cause an edge-effect in the classification.

In this context, our research focuses on the potential of land cover region-based classification of VHR satellite data through the study of the object extraction capacity of segmentation processes, and through the study of the relevance of region features for classifying the land-cover classes in different kinds of Belgian landscapes; always keeping in mind the parallel with the visual interpretation which remains the reference.

Firstly, the results of the assessment of four segmentation algorithms belonging to the two main segmentation categories (contour- and region-based segmentation methods) show that the contour detection methods are sensitive to local variability, which is precisely the problem that we want to overcome. Then, a pre-processing like a filter may be used, at the risk of losing a part of the information. The “region-growing” segmentation that uses the local variability in the segmentation process appears to be the best compromise for the segmentation of different kinds of landscape.

Secondly, the features calculated thanks to segmentation seem to be relevant to identify some land-cover classes in urban/sub-urban and rural areas. These relevant features are of the same type as the features selected visually, which shows that the region-based classification gets close to the visual interpretation.

The research shows the real usefulness of region-based classification in order to classify the land cover with VHR satellite data. Even in some cases where the features calculated thanks to the segmentation prove to be useless, the region-based classification has other advantages. Working with regions instead of pixels allows to avoid the salt-and-pepper effect and makes the GIS integration easier.

The research also highlights some problems that are independent from the region-based classification and are recursive in VHR satellite data, like shadows and the spatial resolution weakness for identifying some land-cover classes.

Résumé

Depuis 1999, les données satellitaires à très haute résolution spatiale (IKONOS-2, QuickBird and OrbView-3) représentent la surface de la terre avec plus de détail. Cependant, l’extraction d’information par une classification multispectrale par pixel devient plus complexe en raison de l’augmentation de la variabilité spectrale dans les unités d’occupation du sol et du manque de résolution spectrale de ces données. Cependant, une possibilité est de considérer cette variabilité spectrale comme une information spatiale utile pouvant être utilisée comme une information complémentaire dans la caractérisation de l’occupation du sol. De plus, de part la diminution de la différence de résolution spatiale qui existait entre les photographies aériennes et les images satellitaires, les caractéristiques (attributs) utilisées en interprétation visuelle transposées à l’analyse digitale (texture, morphologie and contexte) peuvent être utilisées comme information complémentaire en plus de l’information spectrale pour la classification de l’occupation du sol.

La difficulté de cette approche est la transposition des caractéristiques visuelles à l’analyse digitale. Pour résoudre ce problème la classification par région pourrait être utilisée. La segmentation, avant la classification, produit des régions qui sont plus homogène en elles-mêmes qu’avec les régions voisines et qui représentent des objets ou des aires dans l’image. Chaque région devient alors une unité d’analyse qui permet l’élimination de l’effet « poivre et sel » et permet de mesurer et d’utiliser de nombreuses caractéristiques en plus des caractéristiques spectrales. Ces caractéristiques peuvent être la surface, le périmètre, la compacité, la texture. La segmentation est une des seules méthodes qui permet le calcul des caractéristiques morphologiques (surface, périmètre, …) et des caractéristiques texturales sur un voisinage non-arbitraire. Avec les méthodes de classification par pixel, la texture est calculée avec des fenêtres mobiles qui lissent les limites entre les régions d’occupation du sol et créent une texture interclasse. Cette texture interclasse peut alors causer un effet de bord dans le résultat de la classification.

Dans ce contexte, la recherche s’est focalisée sur l’étude du potentiel de la classification par région de l’occupation du sol avec des images satellitaires à très haute résolution spatiale. Ce potentiel a été étudié par l’intermédiaire de l’étude des capacités d’extraction d’objet de la segmentation et par l’intermédiaire de l’étude de la pertinence des caractéristiques des régions pour la classification de l’occupation du sol dans différents paysages belges tant urbains que ruraux.
Doctorat en sciences agronomiques et ingénierie biologique
info:eu-repo/semantics/nonPublished

An investigation of the high spectral resolution signatures and radiative forcing of tropospheric aerosol in the thermal infrared was conducted. To do so and to support advanced modeling of optical properties, a high spectral resolution library of atmospheric aerosol optical constants was developed. This library includes new optical constants of sulfate-nitrate-ammonium aqueous solutions and the collection of a broad range of existing optical constants for aerosol components, particularly mineral optical constants. The mineral optical constants were used to model and study infrared dust optical signatures as a function of composition, size, shape and mixing state. In particular, spherical and non-spherical optical models of dust particles were examined and compared to high spectral resolution laboratory extinction measurements. Then the performance of some of the most common effective medium approximations for internal mixtures was examined by modeling the optical constants of the newly determined sulfate-nitrate-ammonium mixtures. The optical signature analysis was applied to airborne and satellite high spectral resolution thermal infrared radiance data impacted by Saharan dust events. A new technique to retrieve dust microphysical properties from the dust spectral signature was developed and compared to a standard technique. The microphysics retrieved from this new technique and from a standard technique were then used to investigate the effects of dust on radiative forcing and cooling rates in the thermal IR.

Les prairies représentent une source importante de biodiversité dans les paysages agricoles qu’il est important de surveiller. Les satellites de nouvelle génération tels que Sentinel-2 offrent de nouvelles opportunités pour le suivi des prairies grâce à leurs hautes résolutions spatiale et temporelle combinées. Cependant, le nouveau type de données fourni par ces satellites implique des problèmes liés au big data et à la grande dimension des données en raison du nombre croissant de pixels à traiter et du nombre élevé de variables spectro-temporelles. Cette thèse explore le potentiel des satellites de nouvelle génération pour le suivi de la biodiversité et des facteurs qui influencent la biodiversité dans les prairies semi-naturelles. Des outils adaptés à l’analyse statistique des prairies à partir de séries temporelles d’images satellites (STIS) denses à haute résolution spatiale sont proposés. Tout d’abord, nous montrons que la réponse spectrotemporelle des prairies est caractérisée par sa variabilité au sein des prairies et parmi les prairies. Puis, pour les analyses statistiques, les prairies sont modélisées à l’échelle de l’objet pour être cohérent avec les modèles écologiques qui représentent les prairies à l’échelle de la parcelle. Nous proposons de modéliser la distribution des pixels dans une prairie par une loi gaussienne. A partir de cette modélisation, des mesures de similarité entre deux lois gaussiennes robustes à la grande dimension sont développées pour la classification des prairies en utilisant des STIS denses: High-Dimensional Kullback-Leibler Divergence et -Gaussian Mean Kernel. Cette dernière est plus performante que les méthodes conventionnelles utilisées avec les machines à vecteur de support (SVM) pour la classification du mode de gestion et de l’âge des prairies. Enfin, des indicateurs de biodiversité des prairies issus de STIS denses sont proposés à travers des mesures d’hétérogénéité spectro-temporelle dérivées du clustering non supervisé des prairies. Leur corrélation avec l’indice de Shannon est significative mais faible. Les résultats suggèrent que les variations spectro-temporelles mesurées à partir de STIS à 10 mètres de résolution spatiale et qui couvrent la période où ont lieu les pratiques agricoles sont plus liées à l’intensité des pratiques qu’à la diversité en espèces. Ainsi, bien que les propriétés spatiales et temporelles de Sentinel-2 semblent limitées pour estimer directement la diversité en espèces des prairies, ce satellite devrait permettre le suivi continu des facteurs influençant la biodiversité dans les prairies. Dans cette thèse, nous avons proposé des méthodes qui prennent en compte l’hétérogénéité au sein des prairies et qui permettent l’utilisation de toute l’information spectrale et temporelle fournie par les satellites de nouvelle génération
Grasslands are a significant source of biodiversity in farmed landscapes that is important to monitor. New generation satellites such as Sentinel-2 offer new opportunities for grassland’s monitoring thanks to their combined high spatial and temporal resolutions. Conversely, the new type of data provided by these sensors involves big data and high dimensional issues because of the increasing number of pixels to process and the large number of spectro-temporal variables. This thesis explores the potential of the new generation satellites to monitor biodiversity and factors that influence biodiversity in semi-natural grasslands. Tools suitable for the statistical analysis of grasslands using dense satellite image time series (SITS) with high spatial resolution are provided. First, we show that the spectro-temporal response of grasslands is characterized by its variability within and among the grasslands. Then, for the statistical analysis, grasslands are modeled at the object level to be consistent with ecological models that represent grasslands at the field scale. We propose to model the distribution of pixels in a grassland by a Gaussian distribution. Following this modeling, similarity measures between two Gaussian distributions robust to the high dimension are developed for the lassification of grasslands using dense SITS: the High-Dimensional Kullback-Leibler Divergence and the -Gaussian Mean Kernel. The latter outperforms conventional methods used with Support Vector Machines for the classification of grasslands according to their management practices and to their age. Finally, indicators of grassland biodiversity issued from dense SITS are proposed through spectro-temporal heterogeneity measures derived from the unsupervised clustering of grasslands. Their correlation with the Shannon index is significant but low. The results suggest that the spectro-temporal variations measured from SITS at a spatial resolution of 10 meters covering the period when the practices occur are more related to the intensity of management practices than to the species diversity. Therefore, although the spatial and spectral properties of Sentinel-2 seem limited to assess the species diversity in grasslands directly, this satellite should make possible the continuous monitoring of factors influencing biodiversity in grasslands. In this thesis, we provided methods that account for the heterogeneity within grasslands and enable the use of all the spectral and temporal information provided by new generation satellites

The impacts of floods on buildings in urban areas are increasing due to the intensification of extreme weather events, unplanned or uncontrolled settlements and the rising vulnerability of assets. There are some approaches available for assessing the flood damage to buildings and critical infrastructure. To this point, however, it is extremely difficult to adapt these methods widely, due to the lack of high resolution classification and characterisation approaches for built structures. To overcome this obstacle, this work presents: first, a conceptual framework for understanding the physical flood vulnerability and the physical flood susceptibility of buildings, second, a methodological framework for the combination of methods and tools for a large-scale and high-resolution analysis and third, the testing of the methodology in three pilot sites with different development conditions. The conceptual framework narrows down an understanding of flood vulnerability, physical flood vulnerability and physical flood susceptibility and its relation to social and economic vulnerabilities. It describes the key features causing the physical flood susceptibility of buildings as a component of the vulnerability. The methodological framework comprises three modules: (i) methods for setting up a building topology, (ii) methods for assessing the susceptibility of representative buildings of each building type and (iii) the integration of the two modules with technological tools. The first module on the building typology is based on a classification of remote sensing data and GIS analysis involving seven building parameters, which appeared to be relevant for a classification of buildings regarding potential flood impacts. The outcome is a building taxonomic approach. A subsequent identification of representative buildings is based on statistical analyses and membership functions. The second module on the building susceptibility for representative buildings bears on the derivation of depth-physical impact functions. It relates the principal building components, including their heights, dimensions and materials, to the damage from different water levels. The material’s susceptibility is estimated based on international studies on the resistance of building materials and a fuzzy expert analysis. Then depth-physical impact functions are calculated referring to the principal components of the buildings which can be affected by different water levels. Hereby, depth-physical impact functions are seen as a means for the interrelation between the water level and the physical impacts. The third module provides the tools for implementing the methodology. This tool compresses the architecture for feeding the required data on the buildings with their relations to the building typology and the building-type specific depth-physical impact function supporting the automatic process. The methodology is tested in three flood plains pilot sites: (i) in the settlement of the Barrio Sur in Magangué and (ii) in the settlement of La Peña in Cicuco located on the flood plain of Magdalena River, Colombia and (iii) in a settlement of the city of Dresden, located on the Elbe River, Germany. The testing of the methodology covers the description of data availability and accuracy, the steps for deriving the depth-physical impact functions of representative buildings and the final display of the spatial distribution of the physical flood susceptibility. The discussion analyses what are the contributions of this work evaluating the findings of the methodology’s testing with the dissertation goals. The conclusions of the work show the contributions and limitations of the research in terms of methodological and empirical advancements and the general applicability in flood risk management
In vielen Städten nehmen die Auswirkungen von Hochwasser auf Gebäude aufgrund immer extremerer Wetterereignisse, unkontrollierbarer Siedlungsbauten und der steigenden Vulnerabilität von Besitztümern stetig zu. Es existieren zwar bereits Ansätze zur Beurteilung von Wasserschäden an Gebäuden und Infrastrukturknotenpunkten. Doch ist es bisher schwierig, diese Methoden großräumig anzuwenden, da es an einer präzisen Klassifizierung und Charakterisierung von Gebäuden und anderen baulichen Anlagen fehlt. Zu diesem Zweck sollen in dieser Arbeit erstens ein Konzept für ein genaueres Verständnis der physischen Vulnerabilität von Gebäuden gegenüber Hochwasser dargelegt, zweitens ein methodisches Verfahren zur Kombination der bestehenden Methoden und Hilfsmittel mit dem Ziel einer großräumigen und hochauflösenden Analyse erarbeitet und drittens diese Methode an drei Pilotstandorten mit unterschiedlichem Ausbauzustand erprobt werden. Die Rahmenbedingungen des Konzepts grenzen die Begriffe der Vulnerabilität, der physischen Vulnerabilität und der physischen Anfälligkeit gegenüber Hochwasser ein und erörtern deren Beziehung zur sozialen und ökonomischen Vulnerabilität. Es werden die Merkmale der physischen Anfälligkeit von Gebäuden gegenüber Hochwasser als Bestandteil der Vulnerabilität definiert. Das methodische Verfahren umfasst drei Module: (i) Methoden zur Erstellung einer Gebäudetypologie, (ii) Methoden zur Bewertung der Anfälligkeit repräsentativer Gebäude jedes Gebäudetyps und (iii) die Kombination der beiden Module mit Hilfe technologischer Hilfsmittel. Das erste Modul zur Gebäudetypologie basiert auf der Klassifizierung von Fernerkundungsdaten und GIS-Analysen anhand von sieben Gebäudeparametern, die sich für die Klassifizierung von Gebäuden bezüglich ihres Risikopotenzials bei Hochwasser als wichtig erweisen. Daraus ergibt sich ein Ansatz zur Gebäudeklassifizierung. Die anschließende Ermittlung repräsentativer Gebäude beruht auf statistischen Analysen und Zugehörigkeitsfunktionen. Das zweite Modul zur Anfälligkeit repräsentativer Gebäude beruht auf der Ableitung von Funktion von Wasserstand und physischer Einwirkung. Es setzt die relevanten Gebäudemerkmale, darunter Höhe, Maße und Materialien, in Beziehung zum erwartbaren Schaden bei unterschiedlichen Wasserständen. Die Materialanfälligkeit wird aufgrund internationaler Studien zur Festigkeit von Baustoffen sowie durch Anwendung eines Fuzzy-Logic-Expertensystems eingeschätzt. Anschließend werden Wasserstand-Schaden-Funktionen unter Einbeziehung der Hauptgebäudekomponenten berechnet, die durch unterschiedliche Wasserstände in Mitleidenschaft gezogen werden können. Funktion von Wasserstand und physischer Einwirkung dienen hier dazu, den jeweiligen Wasserstand und die physischen Auswirkung in Beziehung zueinander zu setzen. Das dritte Modul stellt die zur Umsetzung der Methoden notwendigen Hilfsmittel vor. Zur Unterstützung des automatisierten Verfahrens dienen Hilfsmittel, die die Gebäudetypologie mit der Funktion von Wasserstand und physischer Einwirkung für Gebäude in Hochwassergebieten kombinieren. Die Methoden wurden anschließend in drei hochwassergefährdeten Pilotstandorten getestet: (i) in den Siedlungsgebieten von Barrio Sur in Magangué und (ii) von La Pena in Cicuco, zwei Überschwemmungsgebiete des Magdalenas in Kolumbien, und (iii) im Stadtgebiet von Dresden, das an der Elbe liegt. Das Testverfahren umfasst die Beschreibung der Datenverfügbarkeit und genauigkeit, die einzelnen Schritte zur Analyse der. Funktion von Wasserstand und physischer Einwirkung repräsentativer Gebäude sowie die Darstellung der räumlichen Verteilung der physischen Anfälligkeit für Hochwasser. In der Diskussion wird der Beitrag dieser Arbeit zur Beurteilung der Erkenntnisse der getesteten Methoden anhand der Ziele dieser Dissertation analysiert. Die Folgerungen beleuchten abschließend die Fortschritte und auch Grenzen der Forschung hinsichtlich methodischer und empirischer Entwicklungen sowie deren allgemeine Anwendbarkeit im Bereich des Hochwasserschutzes
El impacto de las inundaciones sobre los edificios en zonas urbanas es cada vez mayor debido a la intensificación de los fenómenos meteorológicos extremos, asentamientos no controlados o no planificados y su creciente vulnerabilidad. Hay métodos disponibles para evaluar los daños por inundación en edificios e infraestructuras críticas. Sin embargo, es muy difícil implementar estos métodos sistemáticamente en grandes áreas debido a la falta de clasificación y caracterización de estructuras construidas en resoluciones detalladas. Para superar este obstáculo, este trabajo se enfoca, en primer lugar, en desarrollar un marco conceptual para comprender la vulnerabilidad y susceptibilidad física de edificios por inudaciones, en segundo lugar, en desarrollar un marco metodológico para la combinación de los métodos y herramientas para una análisis de alta resolución y en tercer lugar, la prueba de la metodología en tres sitios experimentales, con distintas condiciones de desarrollo. El marco conceptual se enfoca en comprender la vulnerabilidad y susceptibility de las edificaciones frente a inundaciones, y su relación con la vulnerabilidad social y económica. En él se describen las principales características físicas de la susceptibilidad de edificicaiones como un componente de la vulnerabilidad. El marco metodológico consta de tres módulos: (i) métodos para la derivación de topología de construcciones, (ii) métodos para evaluar la susceptibilidad de edificios representativos y (iii) la integración de los dos módulos a través herramientas tecnológicas. El primer módulo de topología de construcciones se basa en una clasificación de datos de sensoramiento rémoto y procesamiento SIG para la extracción de siete parámetros de las edficaciones. Este módulo parece ser aplicable para una clasificación de los edificios en relación con los posibles impactos de las inundaciones. El resultado es una taxonomía de las edificaciones y una posterior identificación de edificios representativos que se basa en análisis estadísticos y funciones de pertenencia. El segundo módulo consiste en el análisis de susceptibilidad de las construcciones representativas a través de funciones de profundidad del impacto físico. Las cuales relacionan los principales componentes de la construcción, incluyendo sus alturas, dimensiones y materiales con los impactos físicos a diferentes niveles de agua. La susceptibilidad del material se calcula con base a estudios internacionales sobre la resistencia de los materiales y un análisis a través de sistemas expertos difusos. Aquí, las funciones de profundidad de impacto físico son considerados como un medio para la interrelación entre el nivel del agua y los impactos físicos. El tercer módulo proporciona las herramientas necesarias para la aplicación de la metodología. Estas herramientas tecnológicas consisten en la arquitectura para la alimentación de los datos relacionados a la tipología de construcciones con las funciones de profundidad del impacto físico apoyado en procesos automáticos. La metodología es probada en tres sitios piloto: (i) en el Barrio Sur en Magangué y (ii) en la barrio de La Peña en Cicuco situado en la llanura inundable del Río Magdalena, Colombia y (iii) en barrio Kleinzschachwitz de la ciudad de Dresden, situado a orillas del río Elba, en Alemania. Las pruebas de la metodología abarca la descripción de la disponibilidad de los datos y la precisión, los pasos a seguir para obtener las funciones profundidad de impacto físico de edificios representativos y la presentación final de la distribución espacial de la susceptibilidad física frente inundaciones El discusión analiza las aportaciones de este trabajo y evalua los resultados de la metodología con relación a los objetivos. Las conclusiones del trabajo, muestran los aportes y limitaciones de la investigación en términos de avances metodológicos y empíricos y la aplicabilidad general de gestión del riesgo de inundaciones

L’intérêt de la télédétection appliquée aux volcans actifs et potentiellement dangereux a été démontré depuis longtemps dans la mesure où cette technique a participé à l’amélioration de la compréhension des processus éruptifs et des aléas volcaniques, amélioration qui permet une réduction des risques volcaniques. Nous avons entrepris plusieurs études volcanologiques reposant sur l’usage d’images de moyenne et haute résolution spatiale, qu’elles soient optiques (IKONOS, Pléiades, GeoEye, Quickbird and SPOT5), radar (ALOS-PALSAR) ou bien thermiques (ASTER et MODIS «hot spot»). Associées à l’analyse de MNTs et de photographies aériennes acquises par un drone, ces études ont consisté à appliquer des techniques de télédétection sur le Semeru et le Merapi, deux des volcans composites les plus actifs et les plus densément peuplés de l’ile de Java en Indonésie. Cette recherche fondée sur la télédétection a permis de mettre en évidence des structures géologiques et tectoniques, d’identifier, de classer et de cartographier des dépôts éruptifs sur les deux volcans et a servi à améliorer l’évaluation des risques à la suite des grandes éruptions de 2002-2003 au Semeru et de 2010 au Merapi. Nous avons également initié une étude afin de comprendre les interactions entre l’activité éruptive et le contexte sismo-tectonique régional en utilisant l’analyse des données MODIS avec la méthode MODVOLC. Nous avons remis à jour la carte géologique du volcan Semeru en y associant des données issues de l’interprétation d’images HSR récentes, des photographies aériennes, l’analyse de MNTs et des observations de terrain, notamment dans le réseau hydrograhique qui convoie des lahars. Nous avons décrit l’histoire éruptive postérieure à 2001 au Semeru en incluant la grande éruption à l’origine des écoulements pyroclastiques (EPs) en 2002-2003 et les éruptions effusives de 2012-2014, qui constituent un phénomène rarement observé sur ce volcan. Le Semeru a produit un volume de 2.5 ± 0.5 106m3 de coulées de lave provenant du cratère sommital entre 2010 et 2014, ce qui peut annoncer, pour la première fois depuis 1967 ou 1941, une modification profonde du style éruptif de ce volcan. Au moment de terminer cette thèse, le dome-coulée situé dans le cratère Jonggring-Seloko continue à croître et les coulées de lave dépassent 2 km de longueur dans la cicatrice majeure en pente raide sur le flanc SE ; leurs fronts pourraient s’effondrer et produire des EPs dont le volume moyen pourrait excéder les valeurs de 3 à 6.5 million de m3 mesurées sur la période 1967-2007. Les écoulements futurs pourront déborder des parois de la cicatrice vers l’aval et se propager vers les vallées des flancs est et sud-ouest. L’épisode éruptif du 26 octobre au 23 novembre 2010 s’est avéré l’événement majeur de l’activité du Merapi depuis 1872. Notre interprétation des images HSR démontre qu’à l’issue des éruptions explosives, le sommet du Merapi a perdu un volume de 10 x 106m3 et la gorge de Gendol orientée SSE a été élargie jusqu’à mesurer 1.3 x 0.3 x 0.2 km. Le nouveau cratère élargi et profond inclut le dome post-2010, qui a été fracturé en 2013, tandis que ses parois verticales instables peuvent être fragilisées par les explosions mineures de 2013 et 2014. Nous avons identifié et cartographié les dépôts pyroclastiques et de lahar de 2010 en appliquant plusieurs méthodes de classification aux images optiques HSR et aux données polarisées de Radar à Synthèse d’Ouverture (RSO). Les résultats démontrent la capacité de l’imagerie satellitaire HSR à capturer l’extension et les impacts de dépôts immédiatement après une grande éruption et avant tout remaniement. Cette technique met en exergue l’utilité de l’imagerie haute résolution et des données radar pour les volcans en activité persistante dont l’accès est souvent rendu impossible. (...)
Remote sensing has long been recognized as a tool for analysis at active and hazardous volcanoes because it can augment our understanding of the processes that underlie volcanic activity so as enable us to apply this understanding to volcanic risk reduction. This thesis presents a volcanological study using High-Spatial Resolution optical images (IKONOS, Pléiades, GeoEye, Quickbird and SPOT5 satellites), radar data (ALOS-PALSAR sensor) and thermal (ASTER satellite and MODIS hot spot) images. In association with DEMs and low-altitude aerial photographs, remote sensing techniques have been applied for tracing the evolution of activity at Semeru and Merapi, two of the most active and densely populated volcanoes in Java, Indonesia. This remotely sensing-based study has unraveled structures, geological features and erupted deposits of both volcanoes and has improved the existing hazard assessment after their most recent eruptions. The thesis also presents the first advance towards deciphering possible interactions between regional tectonic earthquakes and renewed stages of eruptive activity of Merapi and Semeru volcanoes based on the analysis of volcanic hotspots detected by the MODVOLC technique. The geological map of Semeru is updated, including additional data derived from the interpretation of the most recent satellite images, aerial photographs, DEM analysis and fieldwork. The post-2001 eruptive activity at Semeru, including the large PDC-forming eruption in 2002-2003 and uncommon lava flow eruptions in 2010-2014 are investigated. The fact that Semeru has produced several lava flows from the central summit vent between 2010 and 2014 may herald a profound change in eruption style for the first time since at least 1967. At the time of writing, a dome-fed coulée in the Jonggring-Seloko crater continues to grow and lava flows are extending to distances of >2 km down Semeru's SE-scar; their fronts may collapse and produce large-volume pyroclastic density currents (PDCs), perhaps exceeding the average (1967-2007) volume range of 3 to 6.5 million m3. Future dome-collapse PDCs may travel farther down the main SE scar and can spill over its lowermost rims towards the southwest and eastward radiating drainage network. The 26 October-23 November 2010 eruption was the Merapi’s largest event since 1872 (it attained VEI=4). The interpretation of HSR images shows that due to the explosive eruptions, the summit area lost about 10 x 106m3 and the SSE-trending Gendol Breach enlarged to reach 1.3 x 0.3 x 0.2 km in size. The new, enlarged and deep summit crater including the 2010 lava dome is extremely unstable having been weakened by the post-2010 explosive events. This instability is a result of the steep Gendol Breach below the mouth of the crater and the steep and unstable crater walls. The 2010 Merapi pyroclastic and lahar deposits have been identified by applying several classification methods to HSR optical images and dual-polarization synthetic aperture radar (SAR) data. The results show the ability of remotely sensed data to capture the extent and impacts of pristine deposits shortly after emplacement and before any reworking, and highlight the purpose of using high-spatial resolution imagery and SAR data on persistently active volcanoes where access for field survey is often impossible. The 2010 tephra and PDC deposits covered ca. 26 km2 in two catchments of Gendol and Opak Rivers on Merapi’s south flank, i.e. 60-75% of the total PDC deposit area and a total bulk volume of 45 x 106m3. The tephra-fall deposit covered an area of ca. 1300 km2 with a volume range of 18-21 x 106m3. Volumes of these deposits were estimated using the areas determined from remote sensing data and deposit thickness measured in the field. (...)
Penginderaan jauh telah lama dikenal sebagai suatu alat untuk analisis di gunungapi aktif dan berbahaya karena dapat meningkatkan pemahaman kita tentang proses yang mendasari aktivitas gunung berapi sehingga memungkinkan kita untuk menerapkan pemahaman ini dalam pengurangan risiko erupsi gunungapi. Disertasi ini menyajikan studi vulkanologi menggunakan citra satelit optik resolusi tinggi (IKONOS, Pléiades, GeoEye, Quickbird dan SPOT5), data radar (ALOS-PALSAR sensor) dan citra termal (satelit ASTER dan hotspot MODIS). Dalam kaitannya dengan DEM dan foto udara, teknik penginderaan jauh telah diterapkan untuk melihat evolusi aktivitas di Semeru dan Merapi, dua gunung berapi yang paling aktif dengan kepadatan penduduk yang tinggi terletak di Pulau Jawa, Indonesia. Studi berbasis penginderaan jauh ini telah mengkaji struktur, fitur geologi dan material erupsi dari kedua gunungapi tersebut dan telah mempertajam penilaian bahaya yang ada setelah erupsi terkini. Disertasi ini juga menyajikan kemajuan awal dalam menafsirkan kemungkinan interaksi antara gempa tektonik regional dan aktivitas gunungapi Merapi dan Semeru berdasarkan analisis hotspot vulkanik yang terdeteksi oleh MODVOLC. Peta geologi Semeru telah diperbaharui dengan memasukkan data tambahan yang berasal dari interpretasi citra satelit terbaru, foto udara, analisis DEM dan data lapangan. Aktivitas erupsi pasca-2001 di Semeru, termasuk erupsi dengan aliran pirokastik (Pyroclastic Density Current/PDC) besar pada tahun 2002-2003 dan erupsi tidak biasa dengan aliran lava pada 2010-2014, telah dikaji. Fakta bahwa Semeru telah menghasilkan beberapa aliran lava dari kawah di puncak antara tahun 2010 dan 2014, mengindikasikan perubahan besar dalam gaya erupsi untuk pertama kalinya setidaknya sejak 1967. Pada saat penulisan disertasi ini, sebuah kubah lava (Coulée) di kawah Jonggring- Seloko terus tumbuj dan aliran lava yang memanjang hingga jarak >2 km arah tenggara Semeru; ujung lava kemungkinan dapat runtuh dan menghasilkan aliran piroklastik yang mungkin melebihi volume rata-rata (tahun 1967 hingga 2007) dalam kisaran 3-6.5 juta m3. Aliran piroklastik yang akan datang mungkin mengalir sepanjang gawir utama ke arah tenggara dan dapat menyebar melampaui lereng paling bawah ke arah barat daya dan ke arah timur menyebar ke jaringan drainase. Erupsi yang terjadi pada 26 Oktober-23 November 2010 adalah erupsi terbesar Merapi (mencapai VEI 4) sejak 1872. Interpretasi citra resolusi tinggi menunjukkan bahwa daerah puncak kehilangan batuannya sekitar 10 juta m3 akibat erupsi eksplosif. Erupsi juga memperbesar “Gendol Breach” dengan orientasi tenggara menjadi berukuran 1.3x0.3x0.2 km. Kawah puncak yang baru, diperbesar dan dalam, termasuk juga kubah lava tahun 2010 sangat tidak stabil dan telah diperlemah oleh beberapa erupsi eksplosif pasca-2010. Ketidakstabilan ini diakibatkan oleh curamnya Gendol Breach di bawah mulut kawah dan kondisi dinding kawah yang curam dan tidak stabil. Deposit piroklastik dan lahar diidentifikasi dengan menerapkan beberapa metode klasifikasi terhadap citra optik resolusi tinggi dan data dual-polarisasi Synthetic Aperture Radar (SAR). Hasilnya menunjukkan kemampuan data penginderaan jauh untuk merekam jangkauan dan dampak dari deposit murni sesaat setelah pengendapan dan sebelum proses erosi, serta menyoroti tujuan penggunaan citra resolusi tinggi dan data SAR di gunungapi sangat aktif dengan akses untuk survei lapangan sering kali tidak memungkinkan. Endapan tephra dan PDC menutupi area sekitar 26 km2 di dua DAS, Kali Gendol dan Opak, di sisi selatan Merapi, atau 60-75% dari total luas endapan PDC, dan total volume 45 juta m3. Deposit tephra jatuh menutupi area seluas sekitar 1.300 km2 dengan volume 18-21 juta m3. Volume endapan vulkanik ini diestimasi menggunakan informasi luas yang ditentukan dari data penginderaan jauh dan ketebalan yang diukur di lapangan. (...)

With the advent of very high spatial resolution (VHR) satellite, spatial details within the image scene have increased considerably. This led to the development of object-based image analysis (OBIA) for the analysis of VHR satellite images. Image segmentation is the fundamental step for OBIA. However, a large number of techniques exist for RS image segmentation. To identify the best ones for VHR imagery, a comprehensive literature review on image segmentation is performed. Based on that review, it is found that the multiresolution segmentation, as implemented in the commercial software eCognition, is the most widely-used technique and has been successfully applied for wide variety of VHR images. However, the multiresolution segmentation suffers from the parameter estimation problem. Therefore, this study proposes a solution to the problem of the parameter estimation for improving its efficiency in VHR image segmentation. The solution aims to identify the optimal parameters, which correspond to optimal segmentation. The solution to the parameter estimation is drawn from the Equations related to the merging of any two adjacent objects in multiresolution segmentation. The solution utilizes spectral, shape, size, and neighbourhood relationships for a supervised solution. In order to justify the results of the solution, a global segmentation accuracy evaluation technique is also proposed. The solution performs excellently with the VHR images of different sensors, scenes, and land cover classes. In order to justify the applicability of solution to a real life problem, a building detection application based on multiresolution segmentation from the estimated parameters, is carried out. The accuracy of the building detection is found nearly to be eighty percent. Finally, it can be concluded that the proposed solution is fast, easy to implement and effective for the intended applications.

Resumo: Na sociedade atual, discussões relacionadas à água potável tem ocupado um espaço importante, principalmente no meio científico, onde, através de pesquisas voltadas à disponibilidade e qualidade das águas é possível preparar diagnósticos e apontar soluções para planejadores e tomadores de decisões. Nesse contexto, o objetivo geral do trabalho foi desenvolver um método para inferência de variáveis limnológicas que indicam a qualidade da água e estejam associadas à sua característica espectral, em um reservatório de uso múltiplo e avaliar sua correlação com dados espectrais tomados "in situ" e extraídos de imagens orbitais de satélites de alta resolução espacial. Para tanto, uma imagem multiespectral do satélite Ikonos II foi adquirida, quase simultaneamente a coleta de dados limnológicos e espectrais "in situ", em pontos amostrados adequadamente no corpo d'água, e posicionados com GPS. Devido à heterogeneidade das condições do tempo no levantamento de campo, uma nova abordagem amostral foi necessária, que se deu pela divisão da amostra em quatro conjuntos, quais foram: conjunto 1 (céu aberto e vento fraco), conjunto 2 (céu aberto e vento de médio a forte), conjunto 3 (céu nublado e vento fraco) e conjunto 4 (céu nublado e vento de médio a forte)... (Resumo completo, clicar acesso eletrônico abaixo)
Abstract: In current society, drinkable water has been the subject of innumerable debates, mainly in scientific groups, in which, through researches focused on the availability and water quality, it is possible to prepare diagnoses and point out solutions to planners and decision makers. In this context, the general aim of the research was to develop a method for the inference of physical limnological variables that indicate the quality of the water and that are associated to its spectral characteristic, in a multiple use reservoir and evaluate its correlation to spectral data collected "in situ" and extracted from orbital images of high definition space sattelites. In order to achieve that, a multispectral image of the satellite Ikonos II was acquired, almost simultaneously to the gathering of limnological and spectral data "in situ", in points sampled adequately in the water surveyed, and positioned by means of GPS. Due to the heterogeneous weather conditions when taking the ground samples, a new sampling approach was necessary, and it occurred with the division of the sample in four settings, which were: setting 1 (clear sky and mildly windy), setting 2 (clear sky and windy), setting 3 (overcast sky and mildly windy) and setting 4 (overcast sky and windy)... (Complete abstract click electronic access below)
Orientador: Maria de Lourdes Bueno Trindade Galo
Coorientador: Edivaldo Domingues Velin
Banca: Cláudio Clemente Faria Barbosa
Banca: Waterloo Pereira Filho
Banca: Renata Ribeiro de Araújo
Banca: Vilma Mayumi Tachibana
Doutor

Lorsque la resolution spatiale devient trop fine, l'utilisation de l'analyse radiometrique pour l'extraction d'informations des images en teledetection est insuffisante. Aussi, pour pallier a ce probleme, on presente une nouvelle classification combinant une analyse spatiale a l'analyse radiometrique