Early detection of disease and insect infestation within crops and precise application of pesticides can help reduce potential production losses, reduce environmental risk, and reduce the cost of farming. The goal of this study was the advanced detection of early blight (Alternaria solani) in potato (Solanum tuberosum) plants using hyperspectral remote sensing data captured with a handheld spectroradiometer. Hyperspectral reflectance spectra were captured 10 times over five weeks from plants grown to the vegetative and tuber bulking growth stages. The spectra were analyzed using principal component analysis (PCA), spectral change (ratio) analysis, partial least squares (PLS), cluster analysis, and vegetative indices. PCA successfully distinguished more heavily diseased plants from healthy and minimally diseased plants using two principal components. Spectral change (ratio) analysis provided wavelengths (490-510, 640, 665-670, 690, 740-750, and 935 nm) most sensitive to early blight infection followed by ANOVA results indicating a highly significant difference (p < 0.0001) between disease rating group means. In the majority of the experiments, comparisons of diseased plants with healthy plants using Fisher’s LSD revealed more heavily diseased plants were significantly different from healthy plants. PLS analysis demonstrated the feasibility of detecting early blight infected plants, finding four optimal factors for raw spectra with the predictor variation explained ranging from 93.4% to 94.6% and the response variation explained ranging from 42.7% to 64.7%. Cluster analysis successfully distinguished healthy plants from all diseased plants except for the most mildly diseased plants, showing clustering analysis was an effective method for detection of early blight. Analysis of the reflectance spectra using the simple ratio (SR) and the normalized difference vegetative index (NDVI) was effective at differentiating all diseased plants from healthy plants, except for the most mildly diseased plants. Of the analysis methods attempted, cluster analysis and vegetative indices were the most promising. The results show the potential of hyperspectral remote sensing for the detection of early blight in potato plants.

The upcoming launch of the German hyperspectral satellite: Environmental Mapping and Analysis Program (EnMAP) will provide potential for producing improved remotely sensed maps in areas of exposed bedrock in advance of Arctic geology programs. This study investigates the usefulness of this moderate resolution (30m) sensor for predictive lithological mapping using simulated imagery to classify a map area dominated by mafic and felsic volcanics and minor sedimentary and volcaniclastic rocks in the Hope Bay Greenstone Belt of the Northwest Territories. The assessment also included the classification of high resolution and fidelity airborne (ProSpecTIR–SPECIM Dual sensor) hyperspectral imagery for comparison to understand the impact of combined lower signal-to-noise ratio (SNR), and spectral and spatial resolutions associated with EnMap. The performance of both sensors was assessed through statistical analysis of the classification results based on partial unmixing of the data as well as common geological band indices. The results obtained from these analyses were compared to a detailed published geological map of the study area. Both sensors, the airborne ProSpecTIR–SPECIM and spaceborne EnMap, provided good results however despite the simulated EnMap data’s lower resolution and SNR, the results showed it to have greater statistical accuracy and to be visually representative of the mapped geology. The results demonstrated that EnMap satellite hyperspectral technology is an effective tool for mapping lithology in the Canadian North. The discrimination of rock compositions was successful when their occurrences were spatially large and abundant; however, it was identified that spectral similarity between unit classes and spectral variability within classes are critical factors in mapping lithology. / Graduate

Global modeling of gross primary production (GPP) is a critical component of climate change research. On local scales, GPP can be assessed from measuring CO₂ exchange above the plant canopy using tower-based eddy covariance (EC) systems. The limited footprint inherent to this method however, restricts observations to relatively few discrete areas making continuous predictions of global CO₂ fluxes difficult. Recently, the advent of high resolution optical remote sensing devices has offered new possibilities to address some of the scaling issues related to GPP using remote sensing. One key component for inferring GPP spectrally is the efficiency (ε) with which plants can use absorbed photosynthetically active radiation to produce biomass. While recent years have seen progress in measuring ε using the photochemical reflectance index (PRI), little is known about the temporal and spatial requirements for up-scaling these findings continuously throughout the landscape. Satellite observations of canopy reflectance are subject to view and illumination effects induced by the bi-directional reflectance distribution function(BRDF) which can confound the desired PRI signal. Further uncertainties include dependencies of PRI on canopy structure, understorey, species composition and leaf pigment concentration. The objective of this research was to investigate the effects of these factors on PRI to facilitate the modeling of GPP in a continuous fashion. Canopy spectra were sampled over a one-year period using an automated tower-based, multi-angular spectroradiometer platform (AMSPEC), designed to sample high spectral resolution data. The wide range of illumination and viewing geometries seen by the instrument permitted comprehensive modeling of the BRDF. Isolation of physiologically induced changes in PRI yielded a high correlation (r²=0.82, p<0.05) to EC-measured ε, thereby demonstrating the capability of PRI to model ε throughout the year. The results were extrapolated to the landscape scale using airborne laser-scanning (light detection and ranging, LiDAR) and high correlations were found between remotely-sensed and EC-measured GPP (r²>0.79, p<0.05). Permanently established tower-based canopy reflectance measurements are helpful for ongoing research aimed at up-scaling ε to landscape and global scales and facilitate a better understanding of physiological cycles of vegetation and serve as a calibration tool for broader band satellite observations. / Forestry, Faculty of / Graduate

L’exploration spatiale effectuée par les missions orbitales et in situ a permis de mettre en évidence la grande diversité d’objets planétaires rencontrés dans le Système Solaire. Les propriétés de leurs surfaces et notamment leurs compositions sont d’excellents témoins des différents processus physiques endogènes et exogènes ayant façonné ces différents corps depuis leur formation jusqu’à aujourd’hui. Ma thèse a pour contexte l’exploration spatiale multi-échelle des surfaces de deux types d’objets que sont les astéroïdes de type-C et Mars. Les astéroïdes de type C sont des corps très primitifs et ainsi reconnus pour contraindre les premiers stades d’évolution du Système Solaire. Ils peuvent notamment apporter de nombreux indices concernant la présence de phases altérées et de matière organique lors des phases primordiales du Système Solaire. Quant à Mars, sa surface se révèle fascinante en particulier grâce à la grande variété des minéraux détectés par les différentes sondes spatiales. Ces minéraux sont traceurs de nombreux processus physiques qui ont prévalu à la surface permettant ainsi de mieux comprendre les interactions entre les différentes enveloppes que sont la structure interne, la surface, l’atmosphère et l’environnement spatial de Mars. Dans ce travail de thèse je me suis intéressée à l’étude de la minéralogie des surfaces obtenue par imagerie hyperspectrale, d’une part à l’échelle macroscopique avec l’instrument OMEGA/Mars Express dans le but de quantifier les abondances des minéraux traceurs des roches ignées sur Mars et, d’autre part, à l’échelle microscopique avec le microscope MicrOmega, en préparant les futures investigations de la mission Hayabusa-2 à destination de l’astéroïde de type-C Ryugu grâce à l’étalonnage de l’instrument et la caractérisation d’échantillons en laboratoire.Concernant Mars, les résultats majeurs sont les suivants. Un nouveau produit basé sur le jeu complet de données de l’instrument OMEGA dans le proche infra-rouge a été construit, combinant toutes les observations adaptées à l’étude globale des minéraux. Ce cube 3D global de réflectance de Mars a été utilisé pour produire de nouvelles cartes de détections et obtenir un niveau supplémentaire d’analyse en comparaison aux études passées. Un modèle de transfert radiatif a alors été appliqué à tous les spectres des zones présentant des signatures de minéraux mafiques dans le but de quantifier les abondances de ces minéraux traceurs de l’activité magmatique et volcanique. Les cartes globales de pyroxènes, olivine et plagioclase présentées dans cette thèse représentent les premières cartes de minéralogie modale de Mars à une résolution de ~1.5 km/px. Une méthode de classification a mis en lumière la présence de plusieurs classes minéralogiques variées révélant ainsi une hétérogénéité de la surface à différentes échelles. La composition chimique a ensuite été calculée et comparée avec les mesures orbitales et in situ.Dans le cadre de la mission Hayabusa-2, j’ai exploité les données d’étalonnage du microscope hyperspectral proche infra-rouge MicrOmega développé à l’IAS. La réduction et l’analyse des données a permis la construction de la fonction de transfert 4D (position sur le champ de vue, longueur d’onde et température d’opération) de l’instrument dans la gamme complète des valeurs de ces paramètres fonctionnels. Les performances instrumentales concernant l’aspect détection ont été également validées. Cet étalonnage a aussi mis en évidence l’importance de bien définir les opérations en amont de façon à maximiser le rapport signal sur bruit en fonction des paramètres fonctionnels et ainsi d’interpréter au mieux les données scientifiques qui seront acquises une fois au sol de l’astéroïde. Enfin, j’ai également participé à différentes campagnes de mesures d’échantillons naturels avec MicrOmega révélant la capacité de cet instrument à caractériser des échantillons à l’échelle microscopique (10s µm/px). / Space exploration carried out through orbital and in situ missions enables us to highlight the great diversity of objects found in the Solar System. The properties of planetary surfaces and especially their compositions are excellent witnesses of the various endogenous and exogenous physical processes that have shaped these different bodies from their formation to the present day. My thesis is based on the multi-scale spatial exploration of the surfaces of two types of objects, the C-type asteroids and Mars. C-type asteroids are very primitive bodies and are thus recognized as excellent candidates to constraint the early stages of evolution of the Solar System. In particular, they should provide numerous insights concerning the presence of altered phases and organic matter during the primordial phases of the Solar System. As for Mars, its surface is fascinating especially thanks to the wide variety of minerals detected by various space probes. These minerals are tracers of many physical processes that have prevailed on the surface, allowing us to better understand the interactions between the different envelopes that are the internal structure, the surface, the atmosphere and the space environment of Mars. In the work I carried out during my thesis, I focused on the study of the surface mineralogy obtained by hyperspectral imagery, at a macroscopic scale with the OMEGA/Mars Express instrument for quantifying the abundance of minerals tracing the magmatic and volcanic activities on Mars and, at the microscopic scale, with the microscope MicrOmega, focusing on the calibration of the instrument and the characterization of samples in laboratory, within the framework of the future investigations of the Hayabusa-2 mission to the C-type asteroid Ryugu.Regarding Mars, the major results are the following. A new product based on the entire OMEGA instrument dataset acquired in the near infrared has been constructed, combining all the observations adapted to the study of mineral distribution at the global scale. This global 3D reflectance cube of Mars was used to produce new mineral maps providing a further step of analysis compared to past studies. A radiative transfer model was then applied to all spectra presenting mafic mineral signatures in order to quantify the abundances of minerals tracing the magmatic and volcanic activities. The global maps of pyroxenes, olivine and plagioclase presented in this thesis represent the first maps of modal mineralogy of Mars at a resolution of ~ 1.5 km/px. A classification method has highlighted the presence of several distinct mineralogical classes revealing a certain heterogeneity of the surface at different scales. The chemical composition was then calculated and compared with the orbital and in situ measurements.As part of the preparation of the Hayabusa-2 mission, I exploited the calibration data of the hyperspectral imaging microscope MicrOmega. The data reduction and analysis allowed us to derive the 4D transfer function (position on the field of view, wavelength and operating temperature) of the instrument in the full range of values of its functional parameters. The instrumental performances regarding to the identification were also validated: detection of absorption bands of the order of 1% in reflectance with an accuracy of 5 nm on the position of the band and a quantification of the overall reflectance level of the order of 20%. This calibration also highlighted the fact to carefully prepare in advance the operations so as to maximize the signal-to-noise ratio as a function of the functional and environmental parameters and, thus to interpret as much as possible the scientific data that will be acquired once on the asteroid surface. Finally, I also participated in various campaigns of measurements of natural samples with MicrOmega revealing the ability of this instrument to characterize samples at a microscopic scale (10s μm/px).

Given the scale and rate of mangrove loss globally, it is increasingly important to map and monitor mangrove forest health in a timely fashion. This study aims to identify the conditions of mangroves in a coastal lagoon south of the city of Mazatlán, Mexico, using proximal hyperspectral remote sensing techniques. The dominant mangrove species in this area includes the red (Rhizophora mangle), the black (Avicennia germinans) and the white (Laguncularia racemosa) mangrove. Moreover, large patches of poor condition black and red mangrove and healthy dwarf black mangrove are commonly found. Mangrove leaves were collected from this forest representing all of the aforementioned species and conditions. The leaves were then transported to a laboratory for spectral measurements using an ASD FieldSpec® 3 JR spectroradiometer (Analytical Spectral Devices, Inc., USA). R2 plot, principal components analysis and stepwise discriminant analyses were then used to select wavebands deemed most appropriate for further mangrove classification. Specifically, the wavebands at 520, 560, 650, 710, 760, 2100 and 2230 nm were selected, which correspond to chlorophyll absorption, red edge, starch, cellulose, nitrogen and protein regions of the spectrum. The classification and validation indicate that these wavebands are capable of identifying mangrove species and mangrove conditions common to this degraded forest with an overall accuracy and Khat coefficient higher than 90% and 0.9, respectively. Although lower in accuracy, the classifications of the stressed (poor condition and dwarf) mangroves were found to be satisfactory with accuracies higher than 80%. The results of this study indicate that it could be possible to apply laboratory hyperspectral data for classifying mangroves, not only at the species level, but also according to their health conditions.

Removing the effects of the atmosphere from remote sensing data requires accurate knowledge of the physical properties of the atmosphere during the time of measurement. There is a nonlinear relationship that maps atmospheric composition to emitted spectra, but it cannot be easily inverted. The time evolution of atmospheric composition is approximately Markovian, and can be estimated using hyperspectral measurements of the atmosphere with particle filters. The difficulties associated with particle filtering high-dimension data can be mitigated by incorporating future measurement data with the proposal density.

In this dissertation, dimensionality reduction for hyperspectral remote sensing imagery is investigated to alleviate practical application difficulties caused by high data dimension. Band selection and band clustering are applied for this purpose. Based on availability of object prior information, supervised, semi-supervised, and unsupervised techniques are proposed. To take advantage of modern computational architecture, parallel implementations on cluster and graphics processing units (GPU) are developed. The impact of dimensionality reduction on the following data analysis is also evaluated. Specific contributions are as below. 1. A similarity-based unsupervised band selection algorithm is developed to select distinctive and informative bands, which outperforms other existing unsupervised band selection approaches in the literature. 2. An efficient supervised band selection method based on minimum estimated abundance covariance is developed, which outperforms other frequently-used metrics. This new method does not need to conduct classification during band selection process or examine original bands/band combinations as do traditional approaches. 3. An efficient semi-supervised band clustering method is proposed, which uses class signatures to conduct band partition. Compared to traditional unsupervised clustering, computational complexity is significantly reduced. 4. Parallel GPU implementations with computational cost saving strategies for the developed algorithms are designed to facilitate onboard processing. 5. As an application example, band selection results are used for urban land cover classification. With a few selected bands, classification accuracy can be greatly improved, compared to the one using all the original bands or those from other frequently-used dimensionality reduction methods.

No / The purpose of the research was to examine the practical feasibility of developing a hyperspectral camera from a Zeiss fundus camera and to illustrate its use in imaging diabetic retinopathy and glaucoma patients. The original light source of the camera was replaced with an external lamp filtered by a fast tunable liquid-crystal filter. The filtered light was then brought into the camera through an optical fiber. The original film camera was replaced by a digital camera. Images were obtained in normals and patients (primary open angle glaucoma, diabetic retinopathy) recruited at the Manchester Royal Eye Hospital. A series of eight images were captured across 495- to 720-nm wavelengths, and recording time was less than 1.6s. The light level at the cornea was below the ANSI limits, and patients judged the measurement to be very comfortable. Images were of high quality and were used to generate a pixel-to-pixel oxygenation map of the optic nerve head. Frame alignment is necessary for frame-to-frame comparison but can be achieved through simple methods. We have developed a hyperspectral camera with high spatial and spectral resolution across the whole visible spectrum that can be adapted from a standard fundus camera. The hyperspectral technique allows wavelength-specific visualization of retinal lesions that may be subvisible using a white light source camera. This hyperspectral technique may facilitate localization of retinal and disc pathology and consequently facilitate the diagnosis and management of retinal disease.

<p>  </p> <p>The Leafspec portable imaging device had provided a reliable low-cost hyperspectral image acquisition solution to a wide range of users. The Leafspec implemented a enclosed imaging environment and a build-in halogen light source, which eliminated the influence of ambient light. However, a uniform light source was hard to achieve due to dimension and power restrains. White Reference, in many cases refer to an image of a uniform white material such as a Teflon board. White Referencing is a widely used calibration technique in the effort of minimizing noises in an image, including ones that are induced by the light source. However, an abnormal spatial distribution was found remained in images collected by Leafspec even after performing White Referencing technique. In hypothesis, the huge spectral difference between Teflon board and leaves caused this issue. Following this assumption, this article proposed to use a uniform section of soybean leaf to create a Green Reference for calibrating hyperspectral images. In our experiment, 20 green reference samples were collected by imaging a 15 𝑚𝑚 × 15 𝑚𝑚 most uniform section from soybean leaf along a 100 𝑚𝑚 imaging window with a 5 𝑚𝑚 increment. </p>

The improved spectral resolution of modern hyperspectral sensors provides a means for discriminating subtly different classes of on ground materials in remotely sensed images. However, in order to obtain statistically reliable classification results, the number of necessary training samples can increase exponentially as the number of spectral bands increases. Obtaining the necessary number of training signals for these high-dimensional datasets may not be feasible. The problem can be overcome by preprocessing the data to reduce the dimensionality and thus reduce the number of required training samples. In this thesis, three dimensionality reduction methods, all based on parametric projection pursuits, are investigated. These methods are the Sequential Parametric Projection Pursuits (SPPP), Parallel Parametric Projection Pursuits (PPPP), and Projection Pursuits Best Band Selection (PPBBS). The methods are applied to very high spectral resolution data to transform the hyperspectral data to a lower-dimension subspace. Feature extractors and classifiers are then applied to the lower-dimensional data to obtain target detection accuracies. The three projection pursuit methods are compared to each other, as well as to the case of using no dimensionality reduction preprocessing. When applied to hyperspectral data in a precision agriculture application, discriminating sicklepod and cocklebur weeds, the results showed that the SPPP method was optimum in terms of accuracy, resulting in a classification accuracy of >95% when using a nearest mean, maximum likelihood, or nearest neighbor classifier. The PPPP method encountered optimization problems when the hyperspectral dimensionality was very high, e.g. in the thousands. The PPBBS method resulted in high classification accuracies, >95%, when the maximum likelihood classifier was utilized; however, this method resulted in lower accuracies when the nearest mean or nearest neighbor classifiers were used. When using no projection pursuit preprocessing, the classification accuracies ranged between ~50% and 95%; however, for this case the accuracies greatly depended on the type of classifier being utilized.