The objectives of dissertation were to examine vegetation and water indices from AVIRIS and MODIS data for monitoring semiarid and upland vegetation communities related with moisture condition and their spatial and temporal dependencies in estimating evapotranspiration (ET). The performance of various water indices, including the normalized difference water index (NDWI) and land surface water index (LSWI), with the chlorophyll-based vegetation indices (VIs), the normalized difference vegetation index (NDVI) and enhanced vegetation index (EVI) was evaluated in 1) investigating sensitivity of vegetation and land surface moisture condition 2) finding optimal indices in detecting seasonal variations in vegetation water status at the landscape level, and 3) their spatial and temporal scale dependency on estimating ET. The analyses were accomplished through field radiometric measurement, airborne-based and satellite data processing accompanied with water flux data.The results of these studies showed vegetation and landscape moisture condition could be identified in VI - WI scatter-plot. LSWI (2100) showed the biggest sensitivity to variation of vegetation and background soil moisture condition as well. Multi-temporal MODIS data analysis was able to show water use characteristic of riparian vegetation and upland vegetation. Results showed water use characteristics of riparian vegetation are relatively insensitive to summer monsoon pulse, while upland vegetation is highly tied to summer monsoon rain. The relationship between water flux measurement from eddy covariance tower and satellite data has shown that MODIS derived EVI and LSWI (2100) have similar merit to estimate ET rate, but better correlation was observed from the relationship between MODIS EVI and ET.Pixel aggregation results using fine resolution AVIRIS data showed moderate resolution spatial scale 250m or 500m, best predicted ET rates over all study areas. Surface fluxes temporally aggregated to weekly or biweekly intervals showed the strongest ET versus EVI relationships. ET measured at flux towers can be scaled over heterogeneous vegetation associations by simple statistical methods that use meteorological data and flux tower data as ground input, and using the MODIS Enhanced Vegetation Index (EVI) as the only source of remote sensing data.

The debris-flow initiation variable of clay mineralogy is examined for Holocene age debris-flow deposits across the Colorado Plateau. A Kolmogorov-Smirnov two-sample test between 25 debris-flow producing shale units and 23 shale units rated as not producing debris-flows found a highly significant difference between shale unit kaolinite-illite and montmorillonite clay content. Debris-flow producers tend to have abundant kaolinite and illite (61.5% of clays) and small amounts of montmorillonite (10.4%). Clay sample soluble cation (Na, Ca, K, and Mg) content could not be used to accurately divide the data set into debris-flow producers and debris-flow non-producers by either cluster analysis or a Kolmogorov-Smirnov two-sample test.AVIRIS hyperspectral data reveal that debris-flow deposits, colluvium, and some shale units in Cataract Canyon, Utah display the double-absorption feature characteristic of kaolinite at 2.2 µm. Lab-based reflection spectra and semi-quantitative x-ray diffraction results show that Cataract Canyon debris-flow matrix clays are dominated by kaolinite and illite and lacking in montmorillonite. A surface material map showing the spectral stratigraphy of the study area was created from AVIRIS data classified using an artificial neural network and compares favorably to existing geologic data for Cataract Canyon. A debris-flow initiation potential map created from a GIS-based analysis of surface materials, slope steepness, slope aspect, and fault maps shows the greatest debris-flow initiation potential in the study area to coincide with outcrops of the Moenkopi Formation on steep (>20%), southwest-facing slopes. Small areas of extreme debris-flow initiation potential are located where kaolinite and illite clay-rich colluvial wedges are located on southwest-facing walls of Colorado River tributary canyons. The surface materials map shows formations clearly when they remain relatively consistent in composition and exposure throughout the study area, such as the White Rim Sandstone and most clay-rich members of the Moenkopi Formation. The debris-flow producing Organ Rock Shale and Halgaito Formation were shown inconsistently on the surface materials map, likely as a result of compositional variations in the study area. The results of this study provides evidence that hyperspectral imagery classified using an ANN can be successfully used to map the spectral stratigraphy of a sparsely vegetated area such as Cataract Canyon.

Hyperspectral remote sensing image data, such as that recorded by AVIRIS with 224 spectral bands, provides rich information on ground cover types. However, it presents new problems in machine assisted interpretation, mainly in long processing times and the difficulties of class training due to the low ratio of number of training samples to the number of bands. This thesis investigates feasible and efficient feature reduction and image classification techniques which are appropriate for hyperspectral image data. The study is reported in three parts. The first concerns a deterministic approach for hyperspectral data interpretation. Multigroup and multiple threshold spectral coding procedures, and associated techniques for spectral matching and classification, are proposed and tested. By coding on subgroups of bands using one or three thresholds, spectral searching and matching becomes simple, fast and free of the need for radiometric correction. Modifications of existing statistical techniques are proposed in the second part of the investigation A block-based maximum likelihood classification technique is developed. Several subgroups are formed from the complete set of spectral bands in the data, based on the properties of global correlation among the bands. Subgroups which are poorly correlated with each other are treated independently using conventional maximum likelihood classification. Experimental results demonstrate that, when using appropriate subgroup sizes, the new method provides a compromise among classification accuracy, processing time and available training pixels. Furthermore, a segmented, and possibly multi-layer, principal components transformation is proposed as a possible feature reduction technique prior to classification, and for effective colour display. The transformation is performed efficiently on each of the highly correlated subgroups of bands independently. Selected features from each transformed subgroup can be then transformed again to achieve a satisfactory data reduction ratio and to generate the three most significant components for colour display. Classification accuracy is improved and high quality colour image display is achieved in experiments using two AVIRIS data sets.

On April 20, 2010 a tragic explosion aboard the Deepwater Horizon (DWH) drilling rig marked the beginning of one of the worst environmental disasters in history. For 87 days oil and gas were released into the Gulf of Mexico. In August 2010, anomalous phytoplankton activity was identified in the Northeastern Gulf of Mexico, using the Fluorescence Line Height (FLH) ocean color product. The FLH anomaly was bound by approximately 30-28 degrees North and 90 and 86 degrees West and there was a suggestion that this anomaly may have occurred due to the presence of oil. This study was designed to examine alternative explanations and to determine what influence the Mississippi River and the freshwater diversions, employed in the response efforts, may have had on the development of the FLH anomaly. The combination of the anomalously high flow rate in the Mississippi River observed in June-August 2010, the use of freshwater diversions, and three severe storms increased the flow of water through the adjoining marshes. We propose that these conditions reduced the residence time of water and nutrients on the wetlands, and likely mobilized nutrients leading to increased fresh water and nutrients being discharge to the coasts around the Mississippi Delta. Salinity contour maps created from data collected by ships operating in the Northeastern Gulf of Mexico showed that the 31 isohaline was upwards of 250km east of the Mississippi River Birds Foot Delta in August 2010. The American Seas (AmSeas) numerical circulation model was used to examine the dispersal and distribution of water parcels from the Mississippi River and freshwater diversions. Two virtual particle seeding locations were used to trace particles to obtain a measure of the percentage of particles entering a Region of Interest (ROI) located in the center of the FLH anomaly, i.e. 150 km east of the Mississippi Delta. All environmental data examined suggest that the eastward dispersal of the Mississippi River water including that derived from freshwater diversions and storm activity contributed to the development of FLH anomaly in August 2010. Chapter two examines the spectral characteristics of water and oil collected by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). Several peaks in the spectral features of the total radiance of surface oil between 1907nm and 2400nm appear to be absent for water. An algorithm (Spectral Line Height) was created to measure the height of the peak at 2142nm relative to a baseline between 2013nm and 2390nm. A normalized difference technique developed by the USGS was used as a validation tool. Preliminary results of the SLH technique appear to compare favorably with the results derived using the USGS technique. The SLH technique worked in areas that did not show sunglint or shallow bottom features. Sunglint areas would require additional correction to remove the effect of specular reflection. The SLH technique shows promise but will require validation to develop into an operational remote sensing method.

Ce mémoire de maîtrise présente une nouvelle approche non supervisée pour détecter et segmenter les régions urbaines dans les images hyperspectrales. La méthode proposée n ́ecessite trois étapes. Tout d’abord, afin de réduire le coût calculatoire de notre algorithme, une image couleur du contenu spectral est estimée. A cette fin, une étape de réduction de dimensionalité non-linéaire, basée sur deux critères complémentaires mais contradictoires de bonne visualisation; à savoir la précision et le contraste, est réalisée pour l’affichage couleur de chaque image hyperspectrale. Ensuite, pour discriminer les régions urbaines des régions non urbaines, la seconde étape consiste à extraire quelques caractéristiques discriminantes (et complémentaires) sur cette image hyperspectrale couleur. A cette fin, nous avons extrait une série de paramètres discriminants pour décrire les caractéristiques d’une zone urbaine, principalement composée d’objets manufacturés de formes simples g ́eométriques et régulières. Nous avons utilisé des caractéristiques texturales basées sur les niveaux de gris, la magnitude du gradient ou des paramètres issus de la matrice de co-occurrence combinés avec des caractéristiques structurelles basées sur l’orientation locale du gradient de l’image et la détection locale de segments de droites. Afin de réduire encore la complexité de calcul de notre approche et éviter le problème de la ”malédiction de la dimensionnalité” quand on décide de regrouper des données de dimensions élevées, nous avons décidé de classifier individuellement, dans la dernière étape, chaque caractéristique texturale ou structurelle avec une simple procédure de K-moyennes et ensuite de combiner ces segmentations grossières, obtenues à faible coût, avec un modèle efficace de fusion de cartes de segmentations. Les expérimentations données dans ce rapport montrent que cette stratégie est efficace visuellement et se compare favorablement aux autres méthodes de détection et segmentation de zones urbaines à partir d’images hyperspectrales. / This master’s thesis presents a new approach to urban area detection and segmentation in hyperspectral images. The proposed method relies on a three-step procedure. First, in order to decrease the computational complexity, an informative three-colour composite image, minimizing as much as possible the loss of information of the spectral content, is computed. To this end, a non-linear dimensionality reduction step, based on two complementary but contradictory criteria of good visualization, namely accuracy and contrast, is achieved for the colour display of each hyperspectral image. In order to discriminate between urban and non-urban areas, the second step consists of extracting some complementary and discriminant features on the resulting (three-band) colour hyperspectral image. To attain this goal, we have extracted a set of features relevant to the description of different aspects of urban areas, which are mainly composed of man-made objects with regular or simple geometrical shapes. We have used simple textural features based on grey-levels, gradient magnitude or grey-level co-occurence matrix statistical parameters combined with structural features based on gradient orientation, and straight segment detection. In order to also reduce the computational complexity and to avoid the so-called “curse of dimensionality” when clustering high-dimensional data, we decided, in the final third step, to classify each individual feature (by a simple K-means clustering procedure) and to combine these multiple low-cost and rough image segmentation results with an efficient fusion model of segmentation maps. The experiments reported in this report demonstrate that the proposed segmentation method is efficient in terms of visual evaluation and performs well compared to existing and automatic detection and segmentation methods of urban areas from hyperspectral images.

Ce mémoire de maîtrise présente une nouvelle approche non supervisée pour détecter et segmenter les régions urbaines dans les images hyperspectrales. La méthode proposée n ́ecessite trois étapes. Tout d’abord, afin de réduire le coût calculatoire de notre algorithme, une image couleur du contenu spectral est estimée. A cette fin, une étape de réduction de dimensionalité non-linéaire, basée sur deux critères complémentaires mais contradictoires de bonne visualisation; à savoir la précision et le contraste, est réalisée pour l’affichage couleur de chaque image hyperspectrale. Ensuite, pour discriminer les régions urbaines des régions non urbaines, la seconde étape consiste à extraire quelques caractéristiques discriminantes (et complémentaires) sur cette image hyperspectrale couleur. A cette fin, nous avons extrait une série de paramètres discriminants pour décrire les caractéristiques d’une zone urbaine, principalement composée d’objets manufacturés de formes simples g ́eométriques et régulières. Nous avons utilisé des caractéristiques texturales basées sur les niveaux de gris, la magnitude du gradient ou des paramètres issus de la matrice de co-occurrence combinés avec des caractéristiques structurelles basées sur l’orientation locale du gradient de l’image et la détection locale de segments de droites. Afin de réduire encore la complexité de calcul de notre approche et éviter le problème de la ”malédiction de la dimensionnalité” quand on décide de regrouper des données de dimensions élevées, nous avons décidé de classifier individuellement, dans la dernière étape, chaque caractéristique texturale ou structurelle avec une simple procédure de K-moyennes et ensuite de combiner ces segmentations grossières, obtenues à faible coût, avec un modèle efficace de fusion de cartes de segmentations. Les expérimentations données dans ce rapport montrent que cette stratégie est efficace visuellement et se compare favorablement aux autres méthodes de détection et segmentation de zones urbaines à partir d’images hyperspectrales. / This master’s thesis presents a new approach to urban area detection and segmentation in hyperspectral images. The proposed method relies on a three-step procedure. First, in order to decrease the computational complexity, an informative three-colour composite image, minimizing as much as possible the loss of information of the spectral content, is computed. To this end, a non-linear dimensionality reduction step, based on two complementary but contradictory criteria of good visualization, namely accuracy and contrast, is achieved for the colour display of each hyperspectral image. In order to discriminate between urban and non-urban areas, the second step consists of extracting some complementary and discriminant features on the resulting (three-band) colour hyperspectral image. To attain this goal, we have extracted a set of features relevant to the description of different aspects of urban areas, which are mainly composed of man-made objects with regular or simple geometrical shapes. We have used simple textural features based on grey-levels, gradient magnitude or grey-level co-occurence matrix statistical parameters combined with structural features based on gradient orientation, and straight segment detection. In order to also reduce the computational complexity and to avoid the so-called “curse of dimensionality” when clustering high-dimensional data, we decided, in the final third step, to classify each individual feature (by a simple K-means clustering procedure) and to combine these multiple low-cost and rough image segmentation results with an efficient fusion model of segmentation maps. The experiments reported in this report demonstrate that the proposed segmentation method is efficient in terms of visual evaluation and performs well compared to existing and automatic detection and segmentation methods of urban areas from hyperspectral images.