Thesis (Ph.D.)--George Mason University, 2009. / Vita: p. 151. Thesis director: John J. Qu. Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Earth Systems and Geoinformation Sciences. Title from PDF t.p. (viewed Oct. 11, 2009). Includes bibliographical references (p. 136-150). Also issued in print.

Thesis (M.S.) in Forest Resources--University of Maine, 2008. / Includes vita. Includes bibliographical references (leaves 140-152).

Canada has 418 million hectares of forests, representing 10% of the forested land in the world [I]. In 1997, Canada signed the Kyoto Protocol and agreed to cut greenhouse gas emissions by six percent below the 1990 level between 2008 and 2012 [2]. This agreement was ratified in December 2002. It requires Canada to report Canada's sustainable forest resources, including information about forest carbon, afforestation, reforestation, and deforestation (ARD). To fulfill this commitment, effective and accurate measuring tools are needed. One of these tools is satellite remote sensing, a cost-effective way to examine large forested areas in Canada for timely forest information. Historically, the study of forest aboveground carbon was carried out with detailed forest inventory and field sampling from temporary and permanent sample plots, which severely limited the forest area that could be studied. For regional and global scales, it is necessary to use remote sensing for aboveground carbon and ARD mapping due to time and financial constraints. Therefore, the purpose of this research is to develop, implement, and evaluate a computing system that uses multitemporal Landsat satellite images [3] to estimate the Kyoto-Protocol-related forest parameters and create geo-referenced maps, showing the spatial distribution of these parameters in a Geographic Information System (GIs). The new computing system consists of a segment-based and supervised classification engine with feature selection functionality and a Kyoto-Protocol-products estimation unit. The inclusion of the feature selection reduces the large dimensionality of the feature space of multitemporal remote Landsat data sets. Thus, more images could be added into the data sets for analysis. The implementation of the segment-based classifiers provides more accurate forest cover classifications for estimating the Kyoto Protocol products than pixel-based classifiers. It is expected that this approach will be a new addition to the current existing methodologies for supporting Canada's reporting commitments on the sustainability of the forest resources in Canada. This approach can also be used by other countries to monitor Canada's compliance with international agreements.

Thesis (M.S.)--Rochester Institute of Technology, 1993. / Typescript. P. [81] is separate folded map, placed in a mylar sheet. Includes bibliographical references (leaves 86-89).

Although remotely sensed data have been employed to assess various environmental problems, relatively few previous studies have focused on the impacts of mining. In Indonesia, mining activities have increasingly become one of major drivers of land cover change. The majority of remote sensing research projects on mining environments have exploited optical data which are frequently complicated by tmospheric disturbance, especially in tropical territories. Active remote sensors such as Synthetic Aperture Radar (SAR) are invaluable in this case. Monitoring by Independent SAR data has been limited due to single polarisation. Dual-polarised data have been employed considerably, although for some forestry applications the data were found insufficient to retrieve basic information. This Masters thesis is devoted to assess fully polarimetric SAR data for environmental monitoring of the tailings deposition zone of the PT Freeport Indonesia Grasberg mine in Papua, Indonesia. The main data were two granules of the AIRSAR datasets acquired during the PACRIM-II campaign. To support the interpretation and analysis, a scene of Landsat ETM February 2001) was used, juxtaposed with classified aerial photographs and a series of SPOT VEGETATION images. Both backscattering information and complex coherence matrices, as common representations of polarimetric data, were studied. Primary applications of this research were on degraded forest and environmental rehabilitation. Most parts of Indonesian forests have experienced abrupt changes as an impact of clear-cut deforestation. Gradual changes such as those due to fire or flooded tailings, however, are least studied. It was shown that the Cloude-Pottier polarimetric decomposition provided a convenient way to interpret various stages of forest disturbance. The result suggested that the Entropy parameter of the Cloude-Pottier decomposition could be used as a disturbance indicator. Using the fully polarimetric dataset combined with Support Vector Machine learning, the outcomes were generally acceptable. It was possible to improve classification accuracy by incorporating decomposition parameters, although it seemed insignificant. Land rehabilitation on tailings deposits has been a central concern of the government and the mining operator. Indigenous plant pioneers such as reeds (Phragmites) can naturally grow on dry tailings where soil structure is fairly well developed. To assist such efforts, a part of this research involved identification of dry tailings. On the first assessment, interpretation of surface scatterers was aided by polarimetric signatures. Apparently, longer wavelengths such as L- and P-band were overpenetrated; hence, growing reeds on dry tailings were less detectable. In this case, the use of C-band data was found fairly robust. Employing Mahalanobis statistics, the combination of HH and VV performed well on classification, having similar accuracy with quad polarimetric data. Extension on previous results was made through the Freeman-Durden decomposition. Interpretation using a three-component image of odd, even bounce and volume scattering showed that dry and wet tailings could be well distinguished. The application was benefited from unique responses of dielectric materials in the tailings deposit on SAR signals; hence it is possible to discriminate tailings with different moisture levels. However, further assessment of tailings moisture was not possible due to security reasons and access limitations at the study site. Fully polarimetric data were also employed to support rehabilitation of stressed mangrove forest on the southern coast. In this case, the Cloude-Pottier decomposition was employed along with textural parameters. Inclusion of textural properties was found invaluable for the classification using various statistical trees, and more important than decomposition parameters. It was concluded that incorporating polarimetric decompositions and textural parameters into coherence matrix leads to profound accuracy.

Several alien plants are invading subtropical forest ecosystems through canopy gaps, resulting in the loss of native species biodiversity. The loss of native species in such habitats may result in reduced ecosystem functioning. The control and eradication of these invaders requires accurate mapping of the levels of invasion in canopy gaps. Our study tested (i) the utility of WorldView-2 imagery to map forest canopy gaps, and (ii) an integration of WorldView-2 data with environmental data to model the probability of occurrence of invasive Chromolaena odorata (triffid weed) in Dukuduku forest canopy gaps of KwaZulu- Natal, South Africa. Both pixel-based classification and object-based classification were explored for the delineation of forest canopy gaps. The overall classification accuracies increased by ± 12% from a spectrally resampled 4 band image similar to Landsat (74.64%) to an 8 band WorldView-2 imagery (86.90%). This indicates that the new bands of WorldView such as the red edge band can improve on the capability of common red, blue, green and near-infrared bands in delineating forest canopy gaps. The maximum likelihood classifier (MLC) in pixel-based classification yielded the overall classification accuracy of 86.90% on an 8 band WorldView-2 image, while the modified plant senescence reflectance index (mPSRI) in object-based classification yielded 93.69%. The McNemar’s test indicated that there was a statistical difference between the MLC and the mPSRI. The mPSRI is a vegetation index that incorporates the use of the red edge band, which solves a saturation problem common in sensors such as Landsat and SPOT. An integrated model (with both WorldView-2 data and environmental data) used to predict the occurrence of Chromolaena odorata in forest gaps yielded a deviance of about 42% (D2 = 0.42), compared to the model derived from environmental data only (D2 = 0.12) and WorldView-2 data only (D2 = 0.20). A D2 of 0.42 means that a model can explain about 42% of the variability of the presence/absence of Chromolaena odorata in forest gaps. The Distance to Stream and Aspect were the significant environmental variables (ρ < 0.05) which were positively correlated with presence/absence of Chromolaena in forest gaps. WorldView-2 bands such as the coastal band (λ425 nm) yellow band (λ605 nm) and the nearinfrared- 1 (λ833 nm) are positively and significantly related to the presence/absence of invasive species (ρ < 0.05). On the other hand, a significant negative correlation (ρ < 0.05) of near-infrared-2 band (λ950 nm) and the red edge normalized difference vegetation index (NDVI725) suggests that the probability of occurrence of invasive Chromolaena increases forest gaps with low vegetation density. This study highlights the importance of WorldView- 2 imagery and its application in subtropical indigenous coastal forest monitoring. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2013.

Three-dimensional canopy relectance models provide a physical-structural basis to satellite image analysis, representing a potentially more robust, objective and accurate approach for obtaining forest cover type and structural information with minimal ground truth data. The Geometric Optical Mutual Shadowing (GOMS) canopy relectance model was run in multiple-forward-mode (MFM) using digital multispectral IKONOS satellite imagery to estimate tree height and stand volume over 100m2 homogeneous forest plots in mountainous terrain, Kananaskis, Alberta. Height was computed within 2.7m for trembling aspen and 1.8m fr lodgepole pine, with basal area estimated within 0.05m2. Stand volume, estimated as the product of mean tree height and basal area, had an absolute mean difference from field measurements of 0.85m3/100m2 and 0.61m3/100m2 for aspen and pine, respectively. / xiii, 143 leaves : ill. (some col.) ; 29 cm.

The multiple-forward-mode (MFM) inversion procedure is a set of methods for indirect canopy relectance model inversion using look-up tables (LUT). This thesis refines the MFM technique with regard to: 1) model parameterization for the MFM canopy reflectance model executions and 2) methods for limiting or describing multiple solutions. Forest stand structure estimates from the inversion were evaluated using 40 field validation sites in the Canadian Rocky Mountains. Estimates of horizontal and vertical crown radius were within 0.5m and 0.9m RMSE for both conifer and deciduous species. Density estimates were within 590 stems/ha RMSE for conifer and 310 stems/ha RMSE for deciduous. The most effective inversion method used a variable spectral domain with constrained, fine increment LUTs. A biomass estimation method was also developed using empirical relationships with crown area. Biomass density estimates using the MFM method were similar to estimates produced using other multispectral analysis methods (RMSE=50t/ha). / xvi, 156 leaves : ill. (some col.), maps ; 29 cm.

The mapping of forest age is important for effective forest inventory as age is indicative of a number of plant physiological processes. Field survey techniques have traditionally been used to collect forest inventory data, but these methods are costly and time-consuming. Remote sensing offers an alternative which is time-effective and cost-effective and can cover large areas. The aim of this research was to assess the capabilities of multispectral and hyperspectral remotely sensed image data and the statistical method, random forest, for Pinus patula age prediction. The first section of this study used spatial and spectral data derived from multispectral QuickBird imagery to predict forest age. Five co-occurrence texture measures (variance, contrast, correlation, homogeneity, and dissimilarity) were calculated on QuickBird panchromatic imagery (0.6 m spatial resolution) using 12 moving window sizes. The spectral data was extracted from visible and near infrared (NIR) QuickBird imagery (2.4 m spatial resolution). Using the random forest ensemble, various methods of combining the spectral and texture variables were evaluated. The best model was achieved using backward variable selection which aims to find the fewest number of input bands while maintaining the highest predictive accuracy. Only five of the original 64 variables were used in the final model (R2 = 0.68). The second part of this study examined the utility of the random forest ensemble and AISA Eagle hyperspectral image data to predict P. patula age. Random forest was used to determine the optimal subset of hyperspectral bands that could predict P. patula age. Two sequential variable selection methods were tested: forward and backward variable selection. Although both methods resulted in the same root mean square error (3.097), the backward variable selection method was unable to significantly reduce the large hyperspectral dataset and selected 206 variables for the model. The forward variable selection method successfully reduced the large dataset to only nine optimal bands while maintaining the highest predictive accuracy from the hyperspectral dataset (R2 = 0.6). Overall, we concluded that (i) remotely sensed data can produce accurate models for P. patula age prediction, (ii) random forest is an effective tool for the combination of spectral and spatial multispectral data, (iii) random forest is an effective tool for variable selection of a high dimensional hyperspectral dataset, and (iv), although random forest has mainly been used as a classifier, it is also a very effective tool for prediction. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2010.

Thesis (M.S.)--Mississippi State University. Department of Forestry. / Title from title screen. Includes bibliographical references.