Derek R. Peddle

Derivation and validation of Canada-wide coarse-resolution leaf area index maps using high-resolution satellite imagery and ground measurements

Leaf area index (LAI) is one of the surface parameters that has importance in climate, weather, and ecological studies, and has been routinely estimated from remote sensing measurements. Canada-wide LAI maps are now being produced using cloud-free Advanced Very High-Resolution Radiometer (AVHRR) imagery every 10 days at 1-km resolution. The archive of these products began in 1993. LAI maps at the same resolution are also being produced with images from the SPOT VEGETATION sensor. To improve the LAI algorithms and validate these products, a group of Canadian scientists acquired LAI measurements during the summer of 1998 in deciduous, conifer, and mixed forests, and in cropland. Common measurement standards using the commercial Tracing Radiation and Architecture of Canopies (TRAC) and LAI-2000 instruments were followed. Eight Landsat Thematic Mapper (TM) scenes at 30-m resolution were used to locate ground sites and to facilitate spatial scaling to 1-km pixels. In this paper, examples of Canada-wide LAI maps are presented after an assessment of their accuracy using ground measurements and the eight Landsat scenes. Methodologies for scaling from high- to coarse-resolution images that consider surface heterogeneity in terms of mixed cover types are evaluated and discussed. Using Landsat LAI images as the standard, it is shown that the accuracy of LAI values of individual AVHRR and VEGETATION pixels was in the range of 50–75%. Random and bias errors were both considerable. Bias was mostly caused by uncertainties in atmospheric correction of the Landsat images, but surface heterogeneity in terms of mixed cover types were also found to cause bias in AVHRR and SPOT VEGETATION LAI calculations. Random errors come from many sources, but pixels with mixed cover types are the main cause of random errors. As radiative signals from different vegetation types were quite different at the same LAI, accurate information about subpixel mixture of the various cover types is identified as the key to improving the accuracy of LAI estimates. D 2002 Elsevier Science Inc. All rights reserved.

Spectral Mixture Analysis and Geometric-Optical Reflectance Modeling of Boreal Forest Biophysical Structure

Biophysical structural information such as biomass, LAI, and NPP are important inputs to regional scale models of ecosystem processes and photosynthetic activity within boreal forests. However, traditional methods such as NDVI for deriving these variables from remotely sensed data have been inconsistent and unsatisfactory due to factors such as the confounding influence of background reflectance and canopy geometry on the overall pixel signal. To address this problem, we present new results which use spectral mixture analysis to determine areal fractions of sunlit canopy, sunlit background, and shadow at subpixel scales for predicting these biophysical variables. Geometric-optical reflectance models are used to estimate sunlit canopy component reflectance for input to the analysis together with field measures of background and shadow reflectance. In this article, we compare cylinder, cone, and spheroid models of canopy geometry and evaluate the importance of solar zenith angle variations in reflectance estimates for mixture fractions. These are computed from helicopter MMR radiometer data for 31 stands of black spruce along a gradient of stand densities near the southern fringe of the North American boreal forest. Component fractions are evaluated against ground data derived from dense-grid point analyses of coincident high resolution color photography and also for predicting biophysical variables. In general, the Li–Strahler spheroid model was better than the cone and cylinder models and the importance of correcting for solar zenith angle (SZA) was illustrated, with significant improvements noted for higher SZA as a result of corrections for canopy mutual shadowing. The best overall results were obtained from the shadow fraction using a spheroid model of canopy geometry at SZA 45°. Linear regression analyses showed biomass could be estimated with r2 values of 0.83 and a standard error (S.E.) of 1.7 kg/m2; LAI: r2=0.82, S.E.=0.46; and NPP: r2=0.86, S.E.=0.05 kg/m2/yr. These results were significantly higher than with NDVI for estimating biomass (r2=0.44), LAI (r2=0.60), and NPP (r2=0.56). Current and future areas of research are outlined towards improving our understanding of carbon cycling in large forested ecosystems as a function of variability in the physical climate system and environmental change.

SCS+C: a modified Sun-canopy-sensor topographic correction in forested terrain

Topographic correction based on sun-canopy-sensor (SCS) geometry is more appropriate than terrain-based corrections in forested areas since SCS preserves the geotropic nature of trees (vertical growth) regardless of terrain, view, and illumination angles. However, in some terrain orientations, SCS experiences an overcorrection problem similar to other simple photometric functions. To address this problem, we propose a new SCS+C correction that accounts for diffuse atmospheric irradiance based on the C-correction. A rigorous, comprehensive, and flexible method for independent validation based on canopy geometric optical reflectance models is also introduced as an improvement over previous validation approaches, and forms a secondary contribution of this paper. Results for a full range of slopes, aspects, and crown closures showed SCS+C provided improved corrections compared to the SCS and four other photometric approaches (cosine, C, Minnaert, statistical-empirical) for a Rocky Mountain forest setting in western Canada. It was concluded that SCS+C should be considered for topographic correction of remote sensing imagery in forested terrain.

Seasonal change in understory reflectance of boreal forests and influence on canopy vegetation indices

One objective of the Boreal Ecosystem-Atmospheric Study (BOREAS) is to increase our understanding of the nature of canopy spectral bidirectional reflectance in the visible/near-infrared regimes for open canopies typical of boreal forest stands. For such stands, the need to characterize the reflectance of the sunlit and shaded vegetated understory is critical. These variables are subject to temporal variability due to differences in species phenology and foliar display as well as diurnal and seasonal changes in solar illumination through a seasonally varying upper canopy foliar area. To provide for this need, a multiteam field effort was mounted to measure the nadir midday understory reflectance for the flux tower sites during 1994 BOREAS field campaigns between February and October, specifically during the winter focused field campaign (FFC-W), the spring thaw focused field campaign (FFC-T), and the three intensive field campaigns (IFC-1, IFC-2, and IFC-3) between June and September, which sample vegetation phenological change. This was accomplished by measuring at near-solar noon the sunlit and shaded nadir reflectance of the understory along a surveyed leaf area index (LAI) transect line at each flux tower site. Site-to-site comparisons of understory reflectance spectra reveal stand differences that become more significant as the season progresses. Mean midday understory reflectance spectra were observed to be remarkably consistent over the season for young jack pine stands, followed by somewhat increased variability for mature jack pine, and significant seasonal variability for black spruce stands. Derived vegetation indices for understories are generally consistent with extrapolations of previous relationships of canopy spectral vegetation indices (VIs) versus leaf area index to zero LAI. Inclusion of these “zeroLAI” understory-derived indices significantly enhance the correlation in the linear VI-LAI relationships.

Classification of SPOT HRV imagery and texture features

Abstract Spatial co-occurrence matrices were computed for a SPOT HRV multispectral image for a moderate-relief environment in eastern Canada. The texture features entropy and inverse difference moment were used with the spectral data in landcover classification, and substantive increases in accuracy were noted. These range from 10 per cent for exposed bedrock to over 40 per cent in forest and wetland classes. The average classification accuracies were increased from 511 per cent (spectral data alone) to 86.7 per cent (spectral data plus entropy measured in band 2 and inverse difference moment in band 3). Classes that are homogeneous on the ground were characterized adequately by spectral tone alone, but classes containing mixed vegetation patterns or strongly related to structure were characterized more accurately by using a mixture of spectral tone and texture.

Evidential reasoning with Landsat TM, DEM and GIS data for landcover classification in support of grizzly bear habitat mapping

Multisource data consisting of satellite imagery, topographic descriptors derived from DEMs, and GIS inventory information have been used with a detailed, field-based landcover classification scheme to support a quantitative analysis of the spatial distribution and configuration of grizzly bear ( Ursus arctos horribilis ) habitat within the Alberta Yellowhead Ecosystem study area. The map is needed to determine if bear movement and habitat use patterns are affected by changing landscape conditions and human activities. We compared a multisource Evidential Reasoning (ER) classification algorithm, capable of handling this large and diverse data set, to a more conventional maximum likelihood decision rule which could only use a subset of the available data. The ER classifier provided an acceptable level of accuracy (ranging to 85% over 21 habitat classes) for a level 3 product, compared to 71% using a maximum likelihood classifier.

Spectral texture for improved class discrimination in complex terrain

Abstract A spatial co-occurrence algorithm has been used to derive image texture from Landsat Multispectral Scanner (MSS) data to increase classification accuracy in a moderate relief, boreal environment in eastern Canada. The aim was to investigate ‘data-driven improvements’, including those available through digital elevation modelling. Overall classification accuracy using MSS data alone was 59·1 per cent when compared to a biophysical inventory of the area compiled primarily by aerial photointerpretation. This increased to 66·2 per cent with MSS plus texture and to 89·8 per cent when MSS data were analysed with geomorphometry extracted from a digital elevation model (DEM). The introduction of MSS texture resulted in statistically significant increases in individual class accuracies in classes that were also well defined using the geomorphometric and integrated data sets. This suggested that some of the additional information provided by geomorphometry was also contained in spectral texture. It was als...

Reflectance processing of remote sensing spectroradiometer data

Abstract Spectral reflectance is the ratio of incident-to-reflected radiant flux measured from an object or area over specified wavelengths. Unlike radiance and irradiance values, reflectance is an inherent property of an object and is independent of time, location, illumination intensity, atmospheric conditions and weather. Although reflectance is a key unit of measure in remote sensing, it is not measured directly and instead must be derived. Accordingly, the conversion of field and laboratory measurements of spectral radiance into reflectance values is a frequent requirement with ground data in support of airborne and satellite remote sensing applications in the environmental and earth sciences. In this paper, laboratory and computer methods for processing field spectroradiometer measurements of spectral radiance into calibrated absolute reflectance values are described. Target radiance measures are obtained under direct and diffuse illumination using a portable field spectroradiometer, with irradiance spectra captured by near simultaneous acquisition of reflected radiation from a reference panel. The approach for converting raw target and panel radiance spectra to calibrated reflectance involves five major processing stages: (i) panel calibration, (ii) solar zenith angle computations, (iii) spectral and angular interpolation, (iv) computation of reflectance, and (v) automated batch mode execution of stages (ii)–(iv) for processing large data volumes. Equipment, methods, and computer programs for achieving these stages are described. Example forestry ground spectra acquired in the Boreal Ecosystem Atmosphere Study (BOREAS) are presented to illustrate raw field measurements and final reflectance products. These methods would also be useful in other applications such as agriculture, water resources, oceanic studies, rangeland management, and geological exploration and mineral identification.

Ground and remote estimation of leaf area index in Rocky Mountain forest stands, Kananaskis, Alberta

Leaf area index (LAI) is an important measure of canopy structure that is related to biomass, carbon and energy exchange, and is an important input to ecological and climate change models. LAI can be estimated using algorithms applied to airborne and satellite imagery, with ground-based measurements of LAI being required for calibration and validation. A variety of methods exist for ground-based and remote estimation of LAI, and this can lead to confusion and uncertainty regarding selection of methods, experimental design, and instrumentation. As a contribution towards clarifying these protocols, this paper investigated and compared three optical methods and an allometric technique for ground-based estimation of LAI, and these were related to remote LAI estimates derived from the compact airborne spectrographic imager (casi) using three vegetation indices (normalized difference, weighted difference, and soil-adjusted vegetation indices, or NDVI, WDVI, and SAVI, respectively) and subpixel-scale spectral mixture analysis (SMA). The study was conducted in the Kananaskis region of Alberta in the Canadian Rocky Mountains and considered four species compositions within a montane ecological subregion: lodgepole pine, white spruce, composite deciduous (aspen and balsam poplar), and mixedwood (mixture of deciduous and lodgepole pine or white spruce). LAI data were obtained in the field using a LI-COR, Inc. LAI-2000 instrument, a tracing radiation and architecture of canopies (TRAC) system, an integrated (LAI-2000 and TRAC) method, and an allometric technique that used the ratio of sapwood basal area to leaf area. A subsample of plots was assessed with hemispherical photographs and LAI-2000 data from which similar effective leaf area index (eLAI) values were derived for two of the four species analyzed. The results highlight the importance of ensuring that samples represent the range of stand structures and canopy architecture inherent in the species group being assessed. Foliage clumping was observed to be similar in both coniferous and deciduous species and an important element to measure. LAI estimates were influenced by the field methods used to estimate LAI, species and their canopy architecture, and the form of the vegetation index or subpixel-scale mixing derived from the casi image. Of the three vegetation indices, the SAVI was the statistically strongest predictor of LAI for mixedwood species, but all were poor LAI estimators for lodgepole pine and deciduous species. The subpixel-scale scene fractions from SMA provided the best prediction of LAI for white spruce compared with the three vegetation indices. The result for white spruce provides an encouraging basis for further investigation of SMA as a sampling tool to scale from field to high-resolution airborne and satellite imagery for local to landscape-level biophysical estimation.

Classification of permafrost active layer depth from remotely sensed and topographic evidence

Abstract The remote detection of permafrost (perennially frozen ground) has important implications to environmental resource development, engineering studies, natural hazard prediction, and climate change research. In this study, we present results from two experiments into the classification of permafrost active layer depth within the zone of discontinuous permafrost in northern Canada. A new software system (MERCURY⊕) based on evidential reasoning was implemented to permit the integrated classification of multisource data consisting of landcover, terrain aspect, and equivalent latitude (potential insolation), each of which possessed different formats, data types, or statistical properties that could not be handled by conventional classification algorithms available to this study. In the first experiment, four active layer depth classes were classified using ground based measurements of the three variables with an accuracy of 83 % compared to in situ soil probe determination of permafrost active layer depth at over 500 field sites. This confirmed the environmental significance of the variables selected, and provided a baseline result to which a remote sensing classification could be compared. In the second experiment, evidence for each input variable was obtained from image processing of digital SPOT imagery and a photogrammetric digital elevation model, and used to classify active layer depth with an accuracy of 79%. These results suggest the classification of evidence from remotely sensed measures of spectral response and topography may provide suitable indicators of permafrost active layer depth.