Tarmo K. Remmel

When are two landscape pattern indices significantly different

Abstract.Landscape pattern indices (LPI), which characterize various aspects of composition and configuration of categorical variables on a lattice (e.g., shape, clumping, proportion), have become increasingly popular for quantifying and characterizing various aspects of spatial patterns. Unlike in the case of spatial statistical models, when either the joint distribution of all values is characterized by a limited number of parameters, or the distribution is known for certain (usually random) cases, the distributions of LPI are not known. Therefore, comparisons of LPI or significance testing of differences among various landscapes and/or studies are uncertain. This paper scrutinizes six widely used LPI, which are computed based on categories mapped onto regular lattices. We designed a simulation using Gauss-Markov random fields to establish the empirical distributions of LPI as functions of landscape composition and configuration. We report the results for stationary binary landscapes. The confidence intervals for LPI are derived based on 1000 simulations of each given combination of parameters, and further details are evaluated for three illustrative cases. We report the distributions of the LPI along with their co-variation. Our results elucidate how proportion of cover classes and spatial autocorrelation simultaneously and significantly affect the outcome of LPI values. These results also highlight the importance and formal linkages between fully specified spatial stochastic models and spatial pattern analysis. We conclude that LPI must be compared with great care because of the drastic effects that both composition and configuration have on individual LPI values. We also stress the importance of knowing the expected range of variation about LPI values so that statistical comparisons and inferences can be made.

Open source, spatial analysis, and activity-travel behaviour research: capabilities of the aspace package

This paper reports on recent experience with the development of aspace, an Open Source (OS) library for the geographic visualization and analysis of activity-travel behaviour. The paper begins with an overview of recent progress with respect to the convergence of Open Source technology, spatial analysis, and travel behaviour research. The remainder of the paper focuses on aspace; a collection of functions that, when combined with data describing the geographical location of daily activities, can be used to visualize and describe spatial properties of individual and household activity spaces. These properties include: size, orientation, shape, and the geographical dispersion of activity locations contained within the activity space. Several planar geometries are used to transform measurable spatial properties into intuitive objects for visualizing spatial patterns of activity participation. Experiments are conducted, using data from the first wave of the 2003 Toronto Travel Activity Panel Survey, to demonstrate the potential application of aspace for basic and applied policy-based research into activity-travel behaviour. The toolkit is distributed as a downloadable ‘package’ from the Open Source R Project for Statistical Computing.

Fire mapping in a northern boreal forest: assessing AVHRR/NDVI methods of change detection

Understanding natural fire regimes is crucial to developing harvesting scenarios and conducting sustainable resource management in the boreal forest. To gain this understanding, resource professionals need efficient and cost-effective data collection methods that can operate over vast and isolated landscapes. We compared three Advanced Very High Resolution Radiometer (AVHRR)/Normalized Difference Vegetation Index (NDVI) methods of fire detection and mapping for a case study in northern Ontario, Canada, of the 1992, 1993, and 1995 fire seasons. Fire mapping accuracy was assessed by the spatial coincidence between mapped fires and ground-truthed information using a decision-tree approach and by testing the hypothesis that various calculated accuracy components were equal within an ANOVA design. Ground-truthed fire sizes and shapes were correlated with the AVHRR/NDVI-mapped areas; however, light cloud contamination increased false detection of fires due to NDVI suppression. The multiple threshold technique of change detection provided better estimates of fire areas than did single threshold methods.

Mutual information spectra for comparing categorical maps

The continual accumulation of categorical data sets, presented as nominal categories mapped onto regular grids, provides for the increased desire to compare the patterns observed between these maps. We present a measurement scheme for the comparison of categorical maps that decomposes the differences in multidimensional nested coincidence tables according to variables that record occurrence frequencies of categories (Z), at levels of spatial aggregation (Y), on specific maps (X). Sequences of conditional entropies computed according to the specific questions asked (e.g. is there coincidence between colours and locations), characterize the correspondence between the three types of variables in common units (bits) measured by mutual information. The form of these sequences, as a variable runs from coarse to fine detail, referred to as spectra, provide meaningful characterizations of the similarities/differences between categorical maps, including their spatial configuration. We introduce the information theory‐based conceptual framework and illustrate its beneficial application by comparing a series of demonstration maps.

Use of vector polygons for the accuracy assessment of pixel-based land cover maps

Identifying appropriate validation sources for large-area land cover products is a challenge, with logistical constraints frequently necessitating the use of preexisting data sources. Several issues exist when comparing polygon (vector-based) datasets to raster imagery: geolocational mismatches, differences in features or classes mapped, disparity between the scale of polygon delineation and the spatial resolution of the image, and temporal discrepancies. To evaluate the potential impact of using vector coverages to assess the accuracy of pixel-based land cover maps, five evaluation protocols are applied to test sites located in British Columbia and Newfoundland and Labrador, Canada. One protocol directly compared the land cover of the sample unit to the land cover of the forest inventory polygon within which the sample unit fell, two protocols used different regions around the sample unit to define the land cover class, and two protocols were based on homogeneity criteria that restricted the selection of sample units. For the protocols tested, the overall accuracy values ranged from 34% to 58%. Given the broad range of accuracies achieved, the results suggest that caution is needed when making spatially explicit comparisons between raster and vector datasets. When possible, the use of purpose-collected validation data is recommended for the accuracy assessment of maps derived from remotely sensed data; if preexisting vector-based data are the only option for the validation, approaches accounting for the heterogeneity of classes within a given polygon are recommended.

Tree genera classification with geometric features from high-density airborne LiDAR

Categorical recognition of a trees genus is known to be valuable information for the effective management of forest inventories. In this paper, we present a method for learning a discriminative model using Random Forests to classify individual trees into three genera: pine, poplar, and maple. We believe that both internal and external geometric characteristics of the tree crown are related to tree form and therefore useful in classifying trees to the genus level. Our approach involves the extraction of both internal and external geometric features from a LiDAR point cloud as we believe that geometric features provide important information about the organization of the points inside the tree crown along with overall tree shape and form. We developed 24 geometric features and then reduced the number of features to increase efficiency. These geometric characteristics, computed for 160 sampled trees from eight field sites, were classified using Random Forests and achieved an 88.3% average accuracy rate by using 25% (40 trees) of the data for training.

Tracking Desertification in California Using Remote Sensing: A Sand Dune Encroachment Approach

Most remote sensing studies in deserts focus solely on vegetation monitoring to assess the extent of desertification. However, the application of sand dune encroachment into such studies would greatly improve the accuracy in the prediction criteria of risk-prone areas. This study applies the latter methodology for tracking desertification using sand dunes in the Kelso Dunes (in Newberry-Baker, CA, USA). The approach involves the comparison of spectral characteristics of the dunes in Landsat Thematic Mapper (TM) images over a 24-year period (1982, 1988, 1994, 2000, and 2006). During this 24-year period, two El Nino events occurred (1983 and 1993); it was concluded that despite the shift in predominant winds, the short-term variation in wind direction did not make a noticeable change in dune formation, but greatly influences vegetation cover. Therefore, relying solely on vegetation monitoring to assess desertification can lead to overestimations in prediction analysis. Results from this study indicate that the Kelso Dunes are experiencing an encroachment rate of approximately 5.9 m3/m/yr over the 24-year period. While quantifying the Kelso Dunes or any natural dynamic system is subject to uncertainties, the encroachment rate approach reflects the highly heterogeneous nature of the sand dunes (in regards to spectral variability in brightness) at Kelso Dunes and serves as an exemplar for future research.

Categorical, class-focused map patterns: characterization and comparison

We present a rigorous and simple approach for the comparison of binary landscapes by class-focused metric values that complements the ease of computing these metrics for landscape ecology research. First, we assess whether a class-focused pattern metric value could have emerged due to random chance. Second, we compare two landscapes and assess whether class-focused pattern metrics computed for each landscape are significantly different or not. Our frameworks are based on conditional autoregressive simulations to derive empirical distributions for each metric where composition and configuration parameters are controlled. Our method permits the computation of probabilities that an observed metric value is either greater than or less than a given level of expectation. We also provide means for situating any landscape on a selected pattern metric-surface defined by parameters of composition and configuration. These surfaces illustrate which parameter would be most easily adjusted to effect a desired change in a selected class-focused pattern metric’s value. Implementation is fully within the R statistical computing environment.

Hybrid Ensemble Classification of Tree Genera Using Airborne LiDAR Data

Abstract: This paper presents a hybrid ensemble method that is comprised of a sequential and a parallel architecture for the classification of tree genus using LiDAR (Light Detection and Ranging) data. The two classifiers use different sets of features: (1) features derived from geometric information, and (2) features derived from vertical profiles using Random Forests as the base classifier. This classification result is also compared with that obtained by replacing the base classifier by LDA (Linear Discriminant Analysis), kNN (k Nearest Neighbor) and SVM (Support Vector Machine). The uniqueness of this research is in the development, implementation and application of three main ideas: (1) the hybrid ensemble method, which aims to improve classification accuracy, (2) a pseudo-margin criterion for assessing the quality of predictions and (3) an automatic feature reduction method using results drawn from Random Forests. An additional point-density analysis is performed to study the influence of decreased point density on classification accuracy results. By using Random Forests as the base classifier, the average classification accuracies for the geometric classifier and vertical profile classifier are 88.0% and 88.8%, respectively,

Mapping tree genera using discrete LiDAR and geometric tree metrics

Los mapas de generos de arboles son utiles para el inventario forestal, planificacion urbana y el mantenimiento de la infraestructura de lineas de transmision. Se presenta un estudio de caso de uso de datos LiDAR de alta densidad para el mapeo de generos de arboles a lo largo del derecho de paso (ROW) de un corredor de linea de transmision. El objetivo de la investigacion fue identificar arboles individuales que mostraban o poseian una amenaza potencial a la infraestructura de la linea de transmision. Mediante el uso de mapas tridimensionales de LiDAR se derivaron metricas de arboles que estan relacionadas con la geometria de estos (formas del arbol). Por ejemplo, la direccion del crecimiento dominante de los arboles es util para identificar arboles que crecen inclinados hacia las lineas de transmision. Tambien se derivaron otras metricas geometricas que son utiles para determinar los generos de los arboles, tales como altura, forma de la copa, tamano y estructura de ramas. El area de estudio se ubico al norte de Thessalon, Ontario, Canada, a lo largo de los principales corredores de ROW y en los bosques aledanos. Los atributos geometricos usados para la clasificacion de los generos fueron categorizados en cinco amplias clases: 1) lineas, 2) agrupamiento, 3) volumenes, 4) amortiguamiento en 3D de puntos, y 5) forma general del arbol que provee parametros como una entrada para el clasificador forestal aleatorio.