The goal of this thesis aims two fundamental points. The first consists in making the development on the optimal spectral index to characterize the dynamics of a forest cover of a fir tree independently of the internal (underlying soil, topography, saturation and linearity) and external disturbing factors (BRDF and atmosphere) to covers starting from the data of simulation. The second point makes it possible to compare for the first time the potential of three spectral resolutions different (fine, average and broad) from sensors MODIS-EOS, VEGETATION-SPOT and AVHRR-NOAA to minimize the disturbing effects on the indices of vegetation in forest medium. The results obtained show that in general, the influence of the disturbing effects on the indexes of vegetation are complex. There is not only one component which dominates the met variations, but rather, of the combined influences of several factors. (Abstract shortened by UMI.)

Anticosti Island, located in the Gulf of St. Lawrence in eastern Canada, is one of the few places in the world where the Ordovician/Silurian boundary is well preserved and exposed. Its relatively undeformed shallow-water carbonate sequence of approximately 900 m in thickness is rich in fossils and is known to contain traces of hydrocarbons. The island has been for decades the subject of several geological studies, but its stratigraphic succession was never successfully mapped precisely because of its dense forest cover present over almost 95% of its vast territory. This study provides new mapping tools and techniques to support the geological representation of the island stratigraphic succession. Airborne SAR (Synthetic Aperture Radar) data acquired with the active radar system onboard of the former CCRS (Canada Centre for Remote Sensing) Convair-580 aircraft, in single and fully polarimetric modes and with different viewing geometry, were qualitatively and quantitatively evaluated by means of image interpretation and polarimetric analysis for their mapping potential over the densely forested study area. The airborne SAR data, supported with ancillary geoscience data sets and derivative topographic related products, have resulted in the availability of valuable and accurate terrain information such as topographic variations associated with the gently inclined recessive and resistant strata of the island succession. It also provided with information on the polarimetric scattering mechanism of the vegetation cover overlying the surface deposits and bedrock geology, suggesting a possible preferential distribution. With almost 50% of the Canadian territory covered by forest, radar remote sensing, as demonstrated by this study, is a cost-effective tool to produce more accurate regional structural and geological map in areas where traditional mapping campaigns failed due to the presence of an extensive vegetation cover.

The estimation of chlorophyll content is an essential biochemical parameter to track the main developmental stages and yield of cereals relevant for precision agriculture. Traditional techniques for chlorophyll content measurements are time consuming, expensive and laborious. Measurements at field level have proven to be a good alternative, but their use is limited due to extensive sampling designs and techniques. Several spectral chlorophyll indices have been developed to estimate chlorophyll content both at the leaf and canopy level using remote sensing data. A methodology of using spectral chlorophyll indices to estimate chlorophyll content from laboratory and satellite hyperspectral data was carried out in this study for wheat crops. The application of this technique under agricultural field conditions has been very limited and not rigorously validated for wheat crops. The main objective of this study is to validate the chlorophyll content estimation using spectral chlorophyll indices, and to examine the potential for chlorophyll content estimation using hyperspectral remote sensing data in the context of precision agriculture. (Abstract shortened by UMI.)

Data semantic and schematic heterogeneity is a major obstacle to the reuse and sharing of geospatial data. This research focuses on developing an ontology-based methodology to logically integrate heterogeneous geographic data in a cross-border context Three main obstacles hindering data integration are semantic, schematic, and syntactic heterogeneity. Approaches to overcome these obstacles in previous research are reviewed. Among the different approaches, an ontology-based approach is selected for horizontal geospatial data integration in the context of cross-border applications. The integration methodology includes the extraction of application schemas and application ontologies, ontology integration, the creation of a reference model (or ontologies), schema matching and integration, and the creation of usable integrated datasets. The methodology is conceptual and integrates geospatial data based on the semantic content and so is not tied to specific data formats, geometric representations, or feature locations. In order to facilitate the integration procedure, four semantic relationships are used: refer-to, semantic equivalence, semantic generalization, and semantic aggregation. A hybrid ontology approach is employed in order to facilitate the addition of new geospatial data sources to the integration process. As such, three levels of ontologies are developed and illustrated within a MS ACCESS database: application, domain, and a reference model. Furthermore, a working integration prototype is designed to facilitate the integration of geospatial data in the North American context given the semantic and schema heterogeneities in international Canadian-US geospatial datasets. The methodology and prototype provide users with the ability to freely query and retrieve data without knowledge of the heterogeneous data ontologies and schemas. This is illustrated via a case study identifying critical infrastructure around the Ambassador Bridge international border crossing. The methodology and prototype are compared and evaluated with other GDI approaches and by criteria introduced by Buccella et al. (2009). Specific challenges unique to GDI were uncovered and include geographic discrepancies, scale compatibility and temporal issues.

This thesis investigated the development and assessment of a simple parameterization of the multiple scattering within canopies assuming the single scattering field is known and the background beneath the canopy is completely absorbing. The parameterization is based on the concept of spectral invariants related to recollision and escape probabilities from vegetation canopies. The simplified approach is evaluated against detailed 3-D ray tracing model, PARCINOPY, as well as reference datasets from the Radiation Modelling Intercomparison Experiment On-Line Checker. Comparison with homogenous canopies simulated with PARCINOPY showed that the model's performance is best in both the solar principal and perpendicular planes at low and mid LAI levels for all solar zenith angles. The comparison to the On-line Checker datasets shows also that the model is a suitable approach to describe the multiple scattering components of physically based models. This simple parameterization is then incorporated into the Four Scale Linear Model for Anisotropie Reflectance (FLAIR) canopy radiative transfer model to enhance the description of the spectrally dependant multiple scattered radiation field of a forest canopy. The contribution of the multiply scattered radiation between the canopy and the background is also added to the parameterization of the multiple scattering component. The validation of the new version of the FLAIR model was performed using the multi-angular data sets obtained by the airborne sensor POLarization and Directionality of the Earth's Reflectances (POLDER) during the BOReal Ecosystem-Atmosphere Study (BOREAS) campaign of 1994. The results indicate that this approach is well suited to the FLAIR model. It is also demonstrated that the multiple scattering problem can be parameterized by a limited number of architectural parameters and the leaf scattering coefficient. Finally, the combined canopy-leaf PROFLAIR (PROSPECT + FLAIR) model is used to investigate the potential of simulating broadleaf forest canopy spectral reflectance. The comparison between simulated data and Hyperion reflectance data showed the ability of the PROFLAIR model to realistically simulate canopy spectral reflectance. The model was then inverted with hyperspectral Hyperion data using a look-up-table (LUT) approach to retrieve canopy leaf area index (LAI), leaf chlorophyll content (Ca+b) and canopy integrated chlorophyll content (LAI x Ca+b). The LUT was populated by simulating the mode] in forward mode using a space of realization generated based on the specific distribution of the input parameters and based on a priori information from the field. The estimated variables were then compared to ground measurements collected in the field. The results showed a reasonable performance of the PROFLAIR model to the order of performances of other well-known models. When compared to ground measurements, the model showed its ability to retrieve canopy LAI from closed forest canopy with an RMSE of 0.47 and leaf chlorophyll content with an RMSE of 4.461mug/cm2.

The capabilities of remote sensing can be utilized to map the thermal inertia of a surface. Thermal inertia is a property governing the temperature response of a medium to a heat flux at its surface and is beneficial to geologic mapping and soils stud ies. It is hypothesized that a method using only nighttime cooling will give a simple thermal inertia model requiring a minimum of input. Albedo and topographic slope and aspect data are not required. Since latent heat flux is commonly small at night the model should be applicable over surfaces of varying moisture content. The objective of this thesis is to develop a nighttime cooling model for remote sensing thermal inertia mapping. Three models (I, II, and III) are presented. They are based on solutions to the one-dimensional heat conduction equation for a semi-infinite homogeneous solid with isothermal initial temperature and time dependent boundary conditions of heat flux at the surface. Tests of the models on several soil types using ground based data indicate that all three models give meaningful relative relationships between thermal inertias and that model III often yields accurate quantitative results. For the remote sensing implementation of the model ground heat flux is determined as the residual of the energy balance of the surface. Thus, a procedure for determining net radiation using remotely sensed temperature is discussed. Also, aerodynamic heat transfer theory is used to develop a remote sensing method of estimating sensible heat flux. Corrections for the surface sublayer are necessary. Results for vegetated surfaces are expected to be unreliable. Latent heat flux is assumed to be zero or the average of several sites. Tests of these methods using ground based data give good results. An error analysis approach is used to estimate the errors resulting from a remote sensing implementation of Model III. Airborne thermal line-scan data and ground based micrometeorological observations are used to determine typical errors in the input parameters of the model. Errors in determining the energy balance components are also analyzed in detail. With good input, model III gives reasonable results (generally less than 50 percent probable error) at low thermal inertias (< 2000 J m⁻² C⁻¹ s⁻[sub ½] ). For surfaces of high thermal inertia, errors are large. The limitation of the model is not in the model itself, but in the accuracy of remotely sensed surface temperature as determined from thermal infrared line-scan surveys. For surfaces of low thermal inertia model III provides a simple thermal inertia mapping method which requires a minimum of input and is applicable over a wide variety of terrain and ground moisture conditions. The model is most suitable for the investigation of soils and may provide a useful model for planetary studies. / Land and Food Systems, Faculty of / Graduate

In the spring of 2001 the Microwave Remote Sensing Laboratory (MIRSL) at the University of Massachusetts began the development of an advanced Multi-Frequency Radar (AMFR) system for studying clouds and precipitation. This mobile radar was designed to consist of three polarimetric Doppler subsystems operating at Ku-band (13.4 GHz), Ka-band (35.6 GHz) and W-band (94.92 GHz). This combination of frequency bands allows a measurement of a wide range of atmospheric targets ranging from weakly reflecting clouds to strong precipitation. The antenna beamwidths at each frequency were intentionally matched, ensuring consistent sampling volume. Multi-frequency radar remote sensing techniques are not widely used because few multi-frequency radars are available to the science community. One exception is the 33 GHz/95 GHz UMass Cloud Profiling Radar System (CPRS), which AMFR is intended to replace. AMFR's multi-parameter capabilities are designed for characterizing the complex microphysics of layer clouds and precipitation processes in winter storms. AMFR will also play an important role in developing algorithms and validating measurements for an upcoming generation of space-borne radars. The frequency bands selected for AMFR match those of several sensors that have been deployed or are under development. These include the Japanese Aerospace Exploration Agencies (JAXA's) Tropical Rainfall Measuring Mission (TRMM) satellite Ku-band (13 GHz) radar, the CloudSat W-band (95 GHz) radar, and the Global Precipitation Mission (GPM) satellite radars at Ku-band and Ka-band. This dissertation describes the AMFR hardware design and development. Compared to CPRS, the addition of one extra frequency band (Ku) will extend AMFR's measurement capabilities towards the larger particle sizes (precipitation). AMFR's design is based around high-power klystron amplifiers. This ensures complete coherency (CPRS uses magnetrons and coherent-on-receive technique). The partial loss in sensitivity due to lower output power of klystron amplifiers (comparing to magnetrons) is compensated by use of pulse compression (linear FM). The problem of range sidelobes (pulse compression artifacts) has been solved by using appropriate windowing functions in the receiver. Satisfactory sidelobe suppression level of 45 dB has been demonstrated in the lab. The currently best achievable range resolution of the AMFR system is 30 m (corresponds to 5 MHz receiver BW, set by the sampling rate of the Analog-to-Digital card). During the design stage, various polarization schemes have been investigated. The polarization scheme analysis showed the switching polarization scheme to be the best suited for the AMFR system. The AMFR subsystems were partially finished in the winter of 2005. Some preliminary tests were conducted in January 2006. Antenna platform was fabricated in summer 2006. The final assembly took place in the fall of 2006. Early results are presented in the dissertation. These results were helpful in revealing of certain problems in the radar system (i.e. immediate processing computer synchronization) that needed to be addressed during system development. Stratiform rain event occurred on December 18 2006 has been analyzed in detail. A number of commonly used theoretical particle size distributions is presented. Furthermore, it is shown that a fully calibrated multi-frequency radar system has capability of separating scattering and attenuation effects. This was accomplished by fitting the theoretical models into the measured data. An alternative method of estimating rain rate that relies on the dual wavelength ratios is also presented. Although not as powerful as theoretical model fitting, it has its merits for off-zenith observations. During January 2007, AMFR system participated in the C3VP experiment (Canadian CloudSat/CALIPSO Validation Project) in south Ontario, Canada. Some of the data obtained during C3VP experiment has been analyzed and presented. Analysis of these two weather events resulted in the development of the initial multi-frequency particle size distribution retrieval algorithm.

This thesis introduces a model-based technique for the automatic recognition and three-dimensional reconstruction of buildings directly from a single range image or stereo processing of multiple optical views of an urban site. Initially, focus-of-attention regions that are likely to contain buildings are segmented from the scene. A perceptual grouping algorithm detects building boundaries as closed polygons in the optical image. When a digital elevation map (DEM) is the only input source available, building regions are detected through direct analysis of the elevation data. Both methods then utilize the key idea of matching a database of shape models against the DEM using a model-indexing procedure that compares orientation histograms for each parameterized model in the database to a histogram that corresponds to the DEM region. The set of models (surfaces) that most closely match the DEM region are used as the initial estimates in a robust surface fitting technique that refines the model parameters (such as orientation and peak-roof angle) of each hypothesized roof surface. The surface model that converges to the DEM with the lowest residual fit error is retained as the most likely description of the surface. The database of surface models contains a limited number of canonical shapes common to rooftops, such as planes, peaks, domes, and gables. Reconstruction of complex shapes is achieved through a composition of different parameterizations of the canonical shape models. We show how the technique can be recursively applied to a range image to segment and reconstruct buildings as well as rooftop substructure. The ability of the model-indexing technique to separate surface models under different resolutions of the parameter space and different levels of noise in the DEM is studied. The approach is evaluated on several datasets, and we demonstrate that this two-phase reconstruction approach allows robust and accurate reconstruction of a wide variety of building types. The building reconstruction process is at the heart of a general knowledge-driven system called Ascender II that incorporates contextual control of computer vision algorithms comprising a processing library. The system operates in the aerial image domain and is composed of a number of different computer vision algorithms that discriminate object classes based on evidence extracted from the available data. Algorithms are stored in evidence policies that encode contextual information about their data requirements and expected performance. Explicit knowledge about a site is stored in a Bayesian network that is used to fuse information gathered from the execution of a subset of the evidence policies on an image and forms the basis for automatic control of the library of algorithms. Based on the state of the Bayesian network and information encoded in the evidence policies, algorithms are selectively applied to the data in order to segment and recognize different object classes. Using this mechanism, the building reconstruction processes are more likely to be applied to building regions that have already been discriminated from other objects present in an urban area. Our conjecture is that this will lead to significantly better performance of the algorithms (fewer false positives, for example). The Ascender II system is evaluated on three different data sets. Acquired models are evaluated with respect to both geometric and semantic accuracy. Furthermore, the robustness of the system is analyzed with respect to incorrect and incomplete knowledge within the Bayesian network and errors within the vision algorithms. (Abstract shortened by UMI.)

This thesis describes the Turbulent Eddy Profiler (TEP), a volume-imaging, UHF radar wind profiler designed for clear-air measurements in the atmospheric boundary layer on scales comparable to grid cell sizes of Large Eddy Simulation models. TEP employs a large array of antennas--each feeding an independent receiver--to simultaneously generate multiple beams within a 25$\sp\circ$ conical volume illuminated by the transmitter. Range gating provides 30 m spatial resolution in the vertical dimension. Each volume image is updated every 2-10 s, and long data sets can be gathered to study the evolution of turbulent structure over several hours. This thesis provides a summary of the design and operational principles of the Turbulent Eddy Profiler, including an analysis of the calibration and data processing. The TEP engineering tests are described, along with an analysis of precipitation data. Finally, the atmospheric experiments in North Carolina are presented, along with analysis of clear-air data.

This dissertation investigates the use of digital and adaptive beamforming techniques for remote sensing applications. While the theoretical foundations for digital and adaptive beamforming are well established, the application of these results to remote sensing imaging radar has seen little development in the literature. Practical radar systems suffer from a variety of component errors and simple logistical issues that complicate the direct application of theoretical results. The objective of this work is to investigate the limitations of the current theory, and to demonstrate the practical application of the theory where possible. Antenna hardware is a critical component in a digital beamforming system and the first part of this thesis details the design, fabrication, and testing of the antenna hardware for the Turbulent Eddy Profiler (TEP) radar system. A corrugated pyramidal horn antenna serves as the high power transmit horn, and a printed microstrip antenna is used as an element in a ninety-one element receive array. Detailed design procedures are given for both transmit and receive antennas, and a complete set of machine drawings are included. The antennas were fully tested, and measured results are given for the transmit antenna, a single receive element, and a seven element array. The second major section of this dissertation introduces digital and adaptive array processing principles, and investigates the impact of common system errors on the capabilities of these systems. A unified treatment of system errors is presented, and individual error sources are examined in terms of their impact on important performance indexes. A new result is obtained relating the achievable null depth to the cross-correlation terms of the array's correlation matrix. The final section of this dissertation examines the practical application of digital and adaptive beamforming theory to remote sensing imaging radar. Computer simulations and experimental data are used to demonstrate the application of theoretical results to signal environments typical of remote sensing radars. Limitations on antenna performance derived from the theory are discussed, and suboptimal array processing architectures are considered. Experimental results from the FOPAIR (linear array) and TEP (planar array) systems, both developed by MIRSL, form the basis of this section.