Search

Geometric optical reflectance models provide a physical linkage between image data and forest structure. We developed a 'Multiple-Forward-Mode' pseudoinversion modeling approach to produce structural lookup tables for Landsat Thematic Mapper images before and after 1993 partial harvests in New Brunswick Canada. Modeling results validated for stand density, crown radius, and stem counts enabled simple forest structural change detection. Index Terms--Canopy reflectance models, change detection, forest structure.

Digital elevation model (DEM) data were used with climate data to estimate productivity in 19 Eucalyptus plantations in Minas Gerais state, Brazil. Typically, plantation and individual stand growth and productivity estimates, such as Site Index (SI) and Mean Annual Increment (MAI), are based on field measures of height, tree diameter and age. Using a Random Forest modelling approach, SI and MAI were related to: (i) DEM-based geomorphometric variables and (ii) WorldClim historical macro-climatic measures. Three operational SI classes (high, medium and low productivity) in 180 stands were mapped with an overall accuracy of 91.6%. Medium and high productivity sites were the most accurately classified. Low productivity sites had 76.5% producer's accuracy and 92.9% user's accuracy, and were the most extensive in the study area. Such sites are considered of high importance from a plantation management perspective since additional forestry operations are likely required to address low productivity and growth. [ABSTRACT FROM AUTHOR]

In this study, we tested the effectiveness of stand age, multispectral optical imagery obtained from the Landsat 8 Operational Land Imager (OLI), synthetic aperture radar (SAR) data acquired by the Sentinel-1B satellite, and digital terrain attributes extracted from a digital elevation model (DEM), in estimating forest volume in 351 plots in a 1,498 ha Eucalyptus plantation in northern Minas Gerais state, Brazil. A Random Forest (RF) machine learning algorithm was used following the Principal Component Analysis (PCA) of various data combinations, including multispectr al and SAR texture variables and DEM-based geomorphometric derivatives. Using multispectral, SAR or DEM variables alone (i.e. Experiments (ii)-(iv)) did not provide accurate estimates of volume (RMSE (Root Mean Square Error) > 32.00 m3 ha−1) compared to predictions based on age since planting of Eucalyptus stands (Experiment (i)). However, when these datasets were individually combined with stand age (i.e. Experiments (v)-(vii)), the RF models resulted in better volume estimates than those obtained when using the individual multispectral, SAR and DEM datasets (RMSE < 28.00 m3 ha−1). Furthermore, a model that integrated the selected variables of these data with stand age (Experiment (viii)) improved volume estimation significantly (RMSE = 22.33 m3 ha−1). The large and increasing area of Eucalyptus forest plantations in Brazil and elsewhere suggests that this new approach to volume estimation has the potential to support Eucalyptus plantation monitoring and forest management practices. [ABSTRACT FROM AUTHOR]

Object-based image analysis and machine-learning classification were applied to multispectral camera array data acquired by a small rotating blade unmanned aerial vehicle (UAV) over a hardwood forest in eastern Ontario. White birch, aspen, and two species of maple were surveyed in the field. Images were segmented and the resulting objects were visually confirmed to correspond with the sampled tree crowns. Following the application of machine-learning classification using the Random Forest algorithm, an independent validation sample of 23 tree crowns was, overall, approximately 78% correct. Aspen and birch were the most distinct species; the two maples appeared to be confused with each other and with immature trees and understory shrubs. Classification accuracy, commission errors, and variable importance were interpreted to be consistent with experience documented in aerial photointerpretation selection and elimination keys for northern hardwoods. Additional tests are recommended to more fully analyse the accuracy of deciduous tree species classification using digital analysis of high spatial resolution multispectral UAV imagery. [ABSTRACT FROM AUTHOR]

Circumboreal Canadian bogs and fens distinguished by differences in soils, hydrology, vegetation and morphological features were classified using combinations of Radarsat-2 synthetic aperture radar (SAR) quad-polarization data and Landsat-8 Operational Land Imager (OLI) spectral response patterns. Separate classifications were conducted using a traditional pixel-based maximum likelihood classifer and a machine learning algorithm following an object-based image analysis (OBIA). This study focused on two wetland classes with extensive coverage in the area (bog and fen). In the pixel-based maximum likelihood classification, accuracy increased from approximately 69% user’s accuracy and 79% producer’s accuracy using Radarsat-2 SAR data alone to approximately 80% user’s accuracy and 87% producer’s accuracy using Landsat-8 OLI data alone. Use of the Radarsat-2 SAR and Landsat-8 OLI data following principal components analysis (PCA) data fusion did not result in higher pixel-based maximum likelihood classification accuracy. In the object-based machine learning classification, higher bog and fen class accuracies were obtained when using Radarsat-2 and Landsat OLI data individually compared to the equivalent pixel-based classification. Subsequently, a PCA-data fusion product outperformed the individual bands of the Radarsat-2 and Landsat-8 imagery in object-based classification. Greater than 90% producer’s accuracy was obtained. The margin of error (MOE) was less than 5% in all classifications reported here. Further research will examine alternative data fusion techniques and the addition of Radarsat-2 SAR interferometric digital elevation model (DEM)-based geomorphometrics in object-based classification of different morphological types of bogs and fens. [ABSTRACT FROM PUBLISHER]

Forest change features related to resource exploration and extraction are important in the Alberta Oil Sands Region (AOSR), where, for example, 2486 oil and gas well sites were established in a 5000 km2area on three leases in the period 1984–2011. A newly established well site is typically readily identified visually in Landsat multispectral and high spatial resolution imagery, but poses an automated detection and mapping challenge over larger areas and long time periods relative to other major disturbance features. In this study, Landsat time series image composites from the national Composite-2-Change (C2C) change detection protocol were used in a comparison to randomly sampled, independently-generated well site reference data. The highest accuracy reported was approximately 83%, with relatively low errors of omission (13%) and high errors of commission (up to 37%). Future research will incorporate well site disturbance object characteristics in this type of regionally-sensitive forest change analysis. [ABSTRACT FROM PUBLISHER]

The goal of this research was to classify four wetland types in the Hudson Bay Lowlands in northern Canada using Radarsat-2 quad-polarization and Landsat-8 satellite sensor data and geomorphometric variables extracted from an airborne lidar digital elevation model. Segmentation was followed by object-based image classification implemented with a Random Forest machine learning algorithm. The classification accuracy was determined to be approximately 91 percent. This is a significant improvement over the accuracy that was obtained using the Radarsat-2 (80 percent) or Landsat-8 sensor data alone (84 percent). Variable importance (VI) was measured for geomorphometric measures related to the gravity-, wind- and solar-fields, which were developed to explain eco-hydrological differences and increase the separability of wetland classes. Further research will consider additional geomorphometric and spectral response variables that are useful in more detailed boreal wetland classifications and analysis of wetland characteristics over time. [ABSTRACT FROM AUTHOR]

Standardized protocols for forest change detection and classification based on Landsat time series data are becoming more common for use in characterizing multi-decadal history or trends in forest dynamics. One such protocol, referred to as Composite-2-Change (C2C), is a highly automated process developed in Canada that is applicable across extensive forest regions and includes change detection and typing based on Best-Available-Pixel image compositing, spectral trend analysis of breakpoints, object-based segmentation, and Random Forest classification. The aim of this article is to assess the classification accuracy of the Composite-2-Change (C2C) protocol in an eastern Canadian boreal forest environment in northern Ontario. Results demonstrated that the Landsat-derived change detection and attribution approach was approximately 90% accurate for stand-replacing forest change (fire, harvesting, roads), and approximately 75% for four non-stand replacing forest changes caused by spruce budworm and forest tent caterpillar defoliation, wetland and forest flooding caused by localized hydrological variations, and one class of multiple/other disturbances. The C2C protocol approach offers unique independent data layers for modelling that can be used to relate and inform on a range of substantive and subtle changes, which in turn can be labelled and tracked, offering otherwise unavailable information on forest dynamics over large areas. [ABSTRACT FROM PUBLISHER]

The goal of this research was to classify four wetland types in the Hudson Bay Lowlands in northern Canada using Radarsat-2 quad-polarization and Landsat-8 satellite sensor data and geomorphometric variables extracted from an airborne lidar digital elevation model. Segmentation was followed by object-based image classification implemented with a Random Forest machine learning algorithm. The classification accuracy was determined to be approximately 91 percent. This is a significant improvement over the accuracy that was obtained using the Radarsat-2 (80 percent) or Landsat-8 sensor data alone (84 percent). Variable importance (VI) was measured for geomorphometric measures related to the gravity-, wind-and solar-fields, which were developed to explain eco-hydrological differences and increase the separability of wetland classes. Further research will consider additional geomorphometric and spectral response variables that are useful in more detailed boreal wetland classifications and analysis of wetland characteristics over time. [ABSTRACT FROM AUTHOR]