Compact Airborne Spectrographic Imager Data Research Articles

LAI scale effect research based on compact airborne spectrographic imager data in the Heihe Oasis

LAI scale effect research based on compact airborne spectrographic imager data in the Heihe Oasis

Fusion of Hyperspectral CASI and Airborne LiDAR Data for Ground Object Classification through Residual Network.

Modern satellite and aerial imagery outcomes exhibit increasingly complex types of ground objects with continuous developments and changes in land resources. Single remote-sensing modality is not sufficient for the accurate and satisfactory extraction and classification of ground objects. Hyperspectral imaging has been widely used in the classification of ground objects because of its high resolution, multiple bands, and abundant spatial and spectral information. Moreover, the airborne light detection and ranging (LiDAR) point-cloud data contains unique high-precision three-dimensional (3D) spatial information, which can enrich ground object classifiers with height features that hyperspectral images do not have. Therefore, the fusion of hyperspectral image data with airborne LiDAR point-cloud data is an effective approach for ground object classification. In this paper, the effectiveness of such a fusion scheme is investigated and confirmed on an observation area in the middle parts of the Heihe River in China. By combining the characteristics of hyperspectral compact airborne spectrographic imager (CASI) data and airborne LiDAR data, we extracted a variety of features for data fusion and ground object classification. Firstly, we used the minimum noise fraction transform to reduce the dimensionality of hyperspectral CASI images. Then, spatio-spectral and textural features of these images were extracted based on the normalized vegetation index and the gray-level co-occurrence matrices. Further, canopy height features were extracted from airborne LiDAR data. Finally, a hierarchical fusion scheme was applied to the hyperspectral CASI and airborne LiDAR features, and the fused features were used to train a residual network for high-accuracy ground object classification. The experimental results showed that the overall classification accuracy was based on the proposed hierarchical-fusion multiscale dilated residual network (M-DRN), which reached an accuracy of 97.89%. This result was found to be 10.13% and 5.68% higher than those of the convolutional neural network (CNN) and the dilated residual network (DRN), respectively. Spatio-spectral and textural features of hyperspectral CASI images can complement the canopy height features of airborne LiDAR data. These complementary features can provide richer and more accurate information than individual features for ground object classification and can thus outperform features based on a single remote-sensing modality.

Estimación de variables biofísicas del pastizal en un ecosistema de dehesa a partir de espectro-radiometría de campo e imágenes hiperespectrales aeroportadas

<p>The aim of this paper is the estimation of biophysical vegetation parameters from its optical properties. The variables Fuel Moisture Content (<em>FMC</em>), Canopy Water Content (<em>CWC</em>), Leaf Area Index (<em>LAI</em>), dry matter (Cm) and AboveGround Biomass (AGB) were estimated in the laboratory from vegetation samples collected simultaneously with the acquisition of spectral data from the Compact Airborne Spectrographic Imager (CASI) sensor and the field spectroradiometer ASD FieldSpec® 3. Spectral vegetation indices found in the literature were computed from hyperspectral data. Their linear relationships with the biophysical variables measured in the field were analysed. Results show consistent relationships between analysed biophysical parameters and spectral indices, mainly those using SWIR and red-egde bands which reveal the importance of these spectral regions for the estimation of biophysical variables in herbaceous covers. Determination coefficients (<em>R<sup>2</sup></em>) above 0.91 and <em>RRMSE</em> of 21.4% have been obtained for the spectral indexes calculated whit ASD data, and 0.91 <em>R<sup>2 </sup></em>and <em>RRMSE</em> of 15.5% for the spectral indexes calculated whit CASI data.</p>

COMPARING BROAD-BAND AND RED EDGE-BASED SPECTRAL VEGETATION INDICES TO ESTIMATE NITROGEN CONCENTRATION OF CROPS USING CASI DATA

In recent decades, many spectral vegetation indices (SVIs) have been proposed to estimate the leaf nitrogen concentration (LNC) of crops. However, most of these indices were based on the field hyperspectral reflectance. To test whether they can be used in aerial remote platform effectively, in this work a comparison of the sensitivity between several broad-band and red edge-based SVIs to LNC is investigated over different crop types. By using data from experimental LNC values over 4 different crop types and image data acquired using the Compact Airborne Spectrographic Imager (CASI) sensor, the extensive dataset allowed us to evaluate broad-band and red edge-based SVIs. The result indicated that NDVI performed the best among the selected SVIs while red edge-based SVIs didn’t show the potential for estimating the LNC based on the CASI data due to the spectral resolution. In order to search for the optimal SVIs, the band combination algorithm has been used in this work. The best linear correlation against the experimental LNC dataset was obtained by combining the 626.20nm and 569.00nm wavebands. These wavelengths correspond to the maximal chlorophyll absorption and reflection position region, respectively, and are known to be sensitive to the physiological status of the plant. Then this linear relationship was applied to the CASI image for generating an LNC map, which can guide farmers in the accurate application of their N fertilization strategies.

COMPARING BROAD-BAND AND RED EDGE-BASED SPECTRAL VEGETATION INDICES TO ESTIMATE NITROGEN CONCENTRATION OF CROPS USING CASI DATA

Abstract. In recent decades, many spectral vegetation indices (SVIs) have been proposed to estimate the leaf nitrogen concentration (LNC) of crops. However, most of these indices were based on the field hyperspectral reflectance. To test whether they can be used in aerial remote platform effectively, in this work a comparison of the sensitivity between several broad-band and red edge-based SVIs to LNC is investigated over different crop types. By using data from experimental LNC values over 4 different crop types and image data acquired using the Compact Airborne Spectrographic Imager (CASI) sensor, the extensive dataset allowed us to evaluate broad-band and red edge-based SVIs. The result indicated that NDVI performed the best among the selected SVIs while red edge-based SVIs didn’t show the potential for estimating the LNC based on the CASI data due to the spectral resolution. In order to search for the optimal SVIs, the band combination algorithm has been used in this work. The best linear correlation against the experimental LNC dataset was obtained by combining the 626.20nm and 569.00nm wavebands. These wavelengths correspond to the maximal chlorophyll absorption and reflection position region, respectively, and are known to be sensitive to the physiological status of the plant. Then this linear relationship was applied to the CASI image for generating an LNC map, which can guide farmers in the accurate application of their N fertilization strategies.

Fusion of Airborne Discrete-Return LiDAR and Hyperspectral Data for Land Cover Classification

Accurate land cover classification information is a critical variable for many applications. This study presents a method to classify land cover using the fusion data of airborne discrete return LiDAR (Light Detection and Ranging) and CASI (Compact Airborne Spectrographic Imager) hyperspectral data. Four LiDAR-derived images (DTM, DSM, nDSM, and intensity) and CASI data (48 bands) with 1 m spatial resolution were spatially resampled to 2, 4, 8, 10, 20 and 30 m resolutions using the nearest neighbor resampling method. These data were thereafter fused using the layer stacking and principal components analysis (PCA) methods. Land cover was classified by commonly used supervised classifications in remote sensing images, i.e., the support vector machine (SVM) and maximum likelihood (MLC) classifiers. Each classifier was applied to four types of datasets (at seven different spatial resolutions): (1) the layer stacking fusion data; (2) the PCA fusion data; (3) the LiDAR data alone; and (4) the CASI data alone. In this study, the land cover category was classified into seven classes, i.e., buildings, road, water bodies, forests, grassland, cropland and barren land. A total of 56 classification results were produced, and the classification accuracies were assessed and compared. The results show that the classification accuracies produced from two fused datasets were higher than that of the single LiDAR and CASI data at all seven spatial resolutions. Moreover, we find that the layer stacking method produced higher overall classification accuracies than the PCA fusion method using both the SVM and MLC classifiers. The highest classification accuracy obtained (OA = 97.8%, kappa = 0.964) using the SVM classifier on the layer stacking fusion data at 1 m spatial resolution. Compared with the best classification results of the CASI and LiDAR data alone, the overall classification accuracies improved by 9.1% and 19.6%, respectively. Our findings also demonstrated that the SVM classifier generally performed better than the MLC when classifying multisource data; however, none of the classifiers consistently produced higher accuracies at all spatial resolutions.

Mapping estuarine habitats using airborne hyperspectral imagery, with special focus on seagrass meadows

Estuaries and coasts are among the most productive ecosystems and constitute valuable habitats for biodiversity and ecosystem services. Amongst nearshore ecosystems, seagrass beds play a major role enhancing biodiversity and water quality. Consequently, the development of new approaches to create extensive and high-resolution habitat maps is required not only to implement conservation, restoration and management plans, but also to establish adaptation plans to face climate change impacts. This study particularly assesses the capability of hyperspectral airborne imagery acquired with Compact Airborne Spectrographic Imager (CASI) to discriminate and map estuarine habitats, with special focus on Zostera noltii seagrass meadows. To this end, 13 habitats were defined along the supralittoral, intertidal and subtidal zones of an estuary, including Z. noltii seagrass meadows. The CASI sensor was configured to acquire 25 bands in the visible and near infrared wavelengths with a ground sampling distance of 2 m. Spectral bands were selected for species discrimination based on the spectral signature of the different habitat classes. Six different band combinations were tested applying maximum likelihood classification algorithm. The most accurate classification was obtained with 10 band combination (a mean producer accuracy 92% and a mean user accuracy 94%). The classification of Z. noltii beds has been found to be restricted to moderate and high dense meadows, however a vegetation index has been defined which could be applied for mapping Z. noltii meadow cover. These results highlight the value of CASI data to discriminate and map estuarine habitats, providing key information to be used in supporting the implementation of environmental legislation, protection and conservation of coastal habitats.

Validation of the MODIS reflectance product under UK conditions

Surface reflectance obtained from remote-sensing data is the main input to almost all remote-sensing applications. The availability and special characteristics of Moderate Resolution Imaging Spectroradiometer (MODIS) products have led to their use worldwide. Validation of the MODIS reflectance product is then crucial to provid information on uncertainty in the reflectance data, and in other MODIS products and in the applied surface–atmosphere models. Compact Airborne Spectrographic Imager (CASI) and Système Pour l'Observation de la Terre (SPOT) data, collected during the Network for Calibration and Validation in Earth Observation (NCAVEO) 2006 Field Campaign, were applied to validate daily MODIS reflectance data over a site in the southern UK. The difference in the view geometry of at-nadir CASI and SPOT data and off-nadir MODIS data was dealt with using a semi-empirical bidirectional reflectance distribution function (BRDF) model. The validation results showed that for our particular study site, the absolute errors in the MODIS reflectance product were too large to allow the albedo data to be used directly in climate models. The errors were mainly related to the uncertainties in the MODIS atmospheric variables, the BRDF model, and undetected clouds and cloud shadows. More generally, the study highlights the extreme difficulty of achieving pixel-level validation of coarse spatial resolution satellite sensor data in an environment in which the atmosphere is constantly changing, and in which the landscape is characterized by high space–time heterogeneity.

Modelling leaf chlorophyll content in broadleaf and needle leaf canopies from ground, CASI, Landsat TM 5 and MERIS reflectance data

Foliar chlorophyll content in forested ecosystems plays a fundamental role in plant photosynthesis and can indicate vegetation stress and disturbance. However, leaf chlorophyll retrieval is complicated as canopy reflectance in the visible and near-infrared wavelengths is affected by confounding effects not only from leaf pigment concentration but also leaf area index (LAI), canopy architecture, illumination and viewing geometry and understory vegetation. Unlike empirical indices, which are often developed at leaf-level and can be species, site and time specific, a process modelling approach can account for the variation of other variables affecting canopy reflectance; therefore providing a more accurate estimate of chlorophyll content over multiple vegetation species, time-frames and across broader spatial extents. This study used a linked canopy (4-Scale) and leaf (PROSPECT) modelling approach to investigate the ability of radiative transfer models to estimate foliar chemistry for different vegetation types (broadleaf and needle leaf) from optical remote sensing data. Coniferous and deciduous sites were selected in Ontario, Canada, representing different dominant vegetation species, including black spruce (Picea mariana), sugar maple (Acer saccharum) and trembling aspen (Populus tremuloides), and a variety of canopy closures and structures. These sites were sampled over multiple time-frames to collect ground data including leaf area index, leaf reflectance spectra (400–2500nm) and laboratory leaf chlorophyll content. Canopy reflectance data were acquired from the Compact Airborne Spectrographic Imager (CASI), Landsat 5 TM and Medium Resolution Imaging Spectrometer (MERIS). The model results show that leaf chlorophyll content derived from satellite images demonstrates a good relationship with measured leaf chlorophyll content, with validation results of R2=0.62; p<0.001 (MERIS) and R2=0.65; p<0.001 (Landsat 5 TM), and a strong linearity with negligible systematic bias. CASI data gave a regression coefficient of R2=0.41 (p<0.05) on a reduced dataset. This research provides theoretical and operational bases for the future retrieval of leaf chlorophyll content across different vegetation species, canopy structures and over broad spatial extents; crucial characteristics for inclusion in photosynthesis and carbon cycle models.

Characterising soil moisture in transport corridor environments using airborne LIDAR and CASI data

AbstractMuch of the world's transport networks are located on cutting and embankment earthworks. In the UK, many of these earthwork structures were constructed in the mid‐19th century and are susceptible to slope instability. Instability in transport corridors tends to be triggered by an increase in pore water pressure, which is directly influenced by an increase in soil moisture. This study explores the integration of high spatial resolution airborne (1 m) LIght Detection And Ranging (LIDAR) and Compact Airborne Spectrographic Imager (CASI) imagery to characterize soil moisture distribution for a transport corridor near Haltwhistle, UK. The distribution of soil moisture is estimated using these sensors through techniques that are relatively unaffected by, or make use of, the presence of vegetation, unlike other techniques using thermal, shortwave and microwave sensors. Terrain analysis calculations of potential solar radiation and a topographic wetness index were applied to a DEM interpolated from the LIDAR point data. The CASI imagery was used to map Ellenberg moisture indicator values using partial least squares regression. Individually, the remotely sensed metrics were found to have poor correlations with observed soil moisture. However, improvements could be made using an integrated model which demonstrated a correlation coefficient of 0·68. The resulting integrated model showed soil moisture content to increase on north‐facing earthworks and towards the toe of earthwork slopes. Concentrations of moisture were also predicted in cuttings where water is contributed from areas surrounding the earthworks. Copyright © 2011 John Wiley &amp; Sons, Ltd.

Airborne remote sensing of valley floor geoarchaeology using Daedalus ATM and CASI

AbstractThis paper investigates the potential of multispectral methods of airborne remote sensing for geological and archaeological prospection in river valleys, specifically the archaeologically rich, well‐documented aggregate landscape of the middle Trent Valley in the East Midlands of England. The paper reviews the systematic examination of Daedalus 1268 Airborne Thematic Mapper (ATM) and Compact Airborne Spectrographic Imager (CASI) multispectral data. Data from a single flight along the Trent Valley in 1996 were subject to a comprehensive suit of analytical techniques including the calculation of vegetation indices and multivariate analysis. The results suggest that both ATM and CASI data have great potential for detecting and characterizing archaeological and geological features. Vegetation indices relying on ratios of red to infrared reflectance were particularly effective at highlighting cropmarks, in some cases revealing crop variation not apparent in the visible spectrum. Copyright © 2008 John Wiley &amp; Sons, Ltd.

Mapping an inland wetland complex using hyperspectral imagery

The goal is to determine the extent to which heterogeneous inland wetland vegetation communities and their dominant species, as well as adjacent upland vegetation types, can be mapped using 4‐m hyperspectral Compact Airborne Spectrographic Imager (CASI) data. Two classification algorithms, the maximum‐likelihood classifier (MLC) and the spectral angle mapper (SAM), are applied to CASI data acquired over an inland wetland complex located in southern Ontario, Canada. Application of the MLC algorithm to all bands of the CASI data produced better classification results than use of the SAM. Using the MLC, 10 classes were identified with an overall accuracy of 92%. This approach permitted differentiation between areas of shrub‐dominated vegetation communities, floating aquatic communities, emergent aquatics and shallow open water. In the SAM classification, 11 image‐derived spectral endmembers were generated. Wetland classes identified were shrub‐dominated wetlands, floating aquatic vegetation communities, shallow open water and moderately turbid shallow open water. Upland vegetation types were accurately mapped with both algorithms. Reasons why the SAM did not perform as well as the MLC in this complex environment are suggested. It is concluded that high‐resolution hyperspectral data can provide information needed by wetland managers about inland wetland plant communities and their dominant species.

Retrieving forest biomass through integration of CASI and LiDAR data

To increase understanding of forest carbon cycles and stocks, estimates of total and component (e.g. leaf, branch and trunk) biomass at a range of scales are desirable. Focusing on mixed species forests in central south‐east Queensland, two different approaches to the retrieval of biomass from small footprint Light Detection and Ranging (LiDAR) and Compact Airborne Spectrographic Imager (CASI) hyperspectral data were developed and compared. In the first, stems were located using a LiDAR crown openness index, and each was associated with crowns delineated and identified to species using CASI data. The component biomass for individual trees was then estimated using LiDAR‐derived height and stem diameter as input to species‐specific allometric equations. When summed to give total above‐ground biomass (AGB) and aggregated to the plot level, these estimates showed a reasonable correspondence with ground (plot‐based) estimates (r 2 = 0.56, RSE = 25.3 Mg ha−1, n = 21) given the complex forest being assessed. In the second approach, a Jackknife linear regression utilizing six LiDAR strata heights and crown cover at the plot‐scale produced more robust estimates of AGB that showed a closer correspondence with plot‐scale ground data (r 2 = 0.90, RSE = 11.8 Mg ha−1, n = 31). AGB aggregated from the tree‐level and Jackknife regression plot‐based AGB estimates (for 270 plots—each of 0.25 ha) compared well for more mature homogeneous and open forests. However, at the tree level, AGB was overestimated in taller forests dominated by trees with large spreading crowns, and underestimated AGB where an understorey with a high density of stems occurred. The study demonstrated options for quantifying component biomass and AGB through integration of LiDAR and CASI data but highlighted the requirement for methods that give improved estimation of tree density (by size class distributions) and species occurrence in complex forests.

A scale-wise model inversion method to retrieve canopy biophysical parameters from hyperspectral remote sensing data

Biophysical parameters, such as leaf area index (LAI) and leaf chlorophyll content, play an important role in precision agriculture management, forest ecology monitoring, and global change research. Although many efforts have been made, robust and accurate estimation of these parameters from remote sensing data is still a challenge. In this study, a scale-wise scheme was developed for inverting a physical surface model. With this scheme, the model parameters were gradually approximated with dynamic scales during iterations. This scale-wise inversion method was validated based on the coupled PROSPECT and SAIL model using simulated and hyperspectral Compact Airborne Spectrographic Imager (CASI) data over agricultural fields. It was also compared to the widely used Marquet–Levenberg (ML) optimization method. With the simulated data, the results showed that the scale-wise inversion method generated very accurate LAI and leaf chlorophyll content, with maximum errors of less than 0.03 and 0.10 µg/cm2, respectively. Using the same initial values, the errors for the ML method were large, and for some cases the iterations were not converged. The results also showed that the scale-wise method was not sensitive to noise in the data. With the hyperspectral CASI data, the retrieved LAI values were shown to have a strong correlation with the ground measurements, with a root mean square error (RMSE) between them of as low as 0.36 and a R2 of 0.91. The RMSE and R2 values for the ML method are 0.79 and 0.76, respectively.

Modelling geometric and misregistration error in airborne sensor data to enhance change detection

One of the major goals of remote sensing is to carry out monitoring programmes such as land‐cover change detection. However, the accuracy of such change detection activities can be limited by several factors. A key variable that can limit the accuracy of change detection is the misregistration error between the images used. Although the impacts of misregistration on change detection have been considered in various studies, a single global value for misregistration has typically been applied across the whole scene. The effect of misregistration, however, varies spatially and its effects on change detection could be more accurately predicted and ultimately removed if this spatial variation in error were modelled. The current study aimed to develop a model that described the spatial variation of misregistration for airborne image data. As misregistration is a function of geometric error, the geometric errors associated with the airborne data were modelled and this geometric error model was used to derive a model of misregistration. The impacts of various navigational variables on the accuracy of the automated geocorrection of compact airborne spectrographic imager (CASI) data were evaluated. A significant relationship was found between geometric error and angular acceleration (adjusted r 2 = 0.651; p = 0.017). The relationship between geometric error and angular acceleration together with a model of orthometric errors was used to derive an error model that described the spatial variation in geometric errors associated with the automated geocorrection of the CASI data. This model gave a probabilistic description of the spatial variation in geometric error. From the geometric error model, a model of misregistration between CASI images from two times was derived. This model was tested using data from an urban test site and a significant correlation, at 95% confidence, was found between predicted and measured misregistration. The models derived could be used in change detection, potentially reducing the impact of geometric errors and so misregistration in airborne sensor data, which is a major limitation in the use of remote sensing for environmental monitoring.

Integration of airborne CASI and gamma ray data for mine site characteristaion

Current airborne remotely sensed data can contribute to the characterisation of mine sites for monitoring and rehabilitation assessment. The spatial and spectral resolution required from airborne data depends on the issues to be addressed. For this paper we combine Compact Airborne Spectrographic Imager data with ground‐based reflectance measurements and gamma ray survey data to characterise both minerals indicative of acid mine drainage processes and the radiological nature of a rehabilitated uranium mine site.

The delineation of tree crowns in Australian mixed species forests using hyperspectral Compact Airborne Spectrographic Imager (CASI) data

In mixed-species forests of complex structure, the delineation of tree crowns is problematic because of their varying dimensions and reflectance characteristics, the existence of several layers of canopy (including understorey), and shadowing within and between crowns. To overcome this problem, an algorithm for delineating tree crowns has been developed using eCognition Expert and hyperspectral Compact Airborne Spectrographic Imager (CASI-2) data acquired over a forested landscape near Injune, central east Queensland, Australia. The algorithm has six components: 1) the differentiation of forest, non-forest and understorey; 2) initial segmentation of the forest area and allocation of segments (objects) to larger objects associated with forest spectral types (FSTs); 3) initial identification of object maxima as seeds within these larger objects and their expansion to the edges of crowns or clusters of crowns; 4) subsequent classification-based separation of the resulting objects into crown or cluster classes; 5) further iterative splitting of the cluster classes to delineate more crowns; and 6) identification and subsequent merging of oversplit objects into crowns or clusters. In forests with a high density of individuals (e.g., regrowth), objects associated with tree clusters rather than crowns are delineated and local maxima counted to approximate density. With reference to field data, the delineation process provided accuracies > ∼70% (range 48–88%) for individuals or clusters of trees of the same species with diameter at breast height (DBH) exceeding 10 cm (senescent and dead trees excluded), with lower accuracies associated with dense stands containing several canopy layers, as many trees were obscured from the view of the CASI sensor. Although developed using 1-m spatial resolution CASI data acquired over Australian forests, the algorithm has application elsewhere and is currently being considered for integration into the Definiens product portfolio for use by the wider community.

Variability of seasonal CASI image data products and potential application for management zone delineation for precision agriculture

The delineation of management zones is an important step to implementing site-specific crop management practices. Remote sensing is a cost-effective way to acquire information needed for delineating management zones, since it has been successfully used for mapping soil properties and monitoring crop growth conditions. Remotely sensed hyperspectral data are particularly effective in deriving crop biophysical parameters in agricultural fields; therefore, the potential of hyperspectral data to contribute to management zone delineation needs to be assessed. In this study, the spatial variability of soil and crops in two agricultural fields was studied using seasonal compact airborne spectrographic imager (CASI) hyperspectral images. Different spectral features including soil brightness and colouration indices, principal components of soil reflectance data, and crop descriptors (leaf area index (LAI) and leaf chlorophyll content) were derived from CASI data and used to partition the fields into homogeneous zones using the fuzzy k means unsupervised classification method. The reduction of variances of soil electrical conductivity, LAI, leaf chlorophyll content, and yield was inspected to determine the appropriate number of zones for each field. The zones obtained were interpreted according to the soil survey map and field practices. Analysis of variance (ANOVA) was conducted to examine the effectiveness of the delineation. The study shows that the spatial patterns of the resulting soil zones faithfully represent the soil classes described by the soil survey maps, and the spatial patterns of the resulting crop classes discriminated the different crop growth conditions well. These results show that hyperspectral data provide important information on field variability for management zone delineation in precision agriculture.

Artificial neural networks to predict corn yield from Compact Airborne Spectrographic Imager data

In the light of recent advances in spectral imaging technology, highly flexible modeling methods must be developed to estimate various soil and crop parameters for precision farming from airborne hyperspectral imagery. The potential of artificial neural networks (ANNs) for the development of in-season yield mapping and forecasting systems was examined. Hyperspectral images of corn ( Zea mays L.) plots in eastern Canada, subjected to different fertilization rates and various weed management protocols, were acquired by a compact airborne spectral imager. Statistical and ANN approaches along with various vegetation indices were used to develop yield prediction models. Principal component analysis was used to reduce the number of input variables. Greater prediction accuracy (about 20% validation RMSE) was obtained with an ANN model than with either of the three conventional empirical models based on normalized difference vegetation index, simple ratio, or photochemical reflectance index. No clear difference was observed between ANNs and stepwise multiple linear regression models. Although the high potential usefulness of ANNs was confirmed, particularly in the creation of yield maps, further investigations are needed before their application at the field scale can be generalized.

Retrieval of the canopy leaf area index in the BOREAS flux tower sites using linear spectral mixture analysis

This paper reports on the use of linear spectral mixture analysis for the retrieval of canopy leaf area index (LAI) in three flux tower sites in the Boreal Ecosystem-Atmosphere Study (BOREAS) southern study area: Old Black Spruce, Old Jack Pine, and Young Jack Pine (SOBS, SOJP, and SYJP). The data used were obtained by the Compact Airborne Spectrographic Imager (CASI) with a spatial resolution of 2 m in the winter of 1994. The convex geometry method was used to select the endmembers: sunlit crown, sunlit snow, and shadow. Along transects for these flux tower sites, the fraction of sunlit snow was found to have a higher correlation with the field-measured canopy LAI than the fraction of sunlit crown or the fraction of shadow. An empirical equation was obtained to describe the relation between canopy LAI and the fraction of sunlit snow. There is a strong correlation between the estimated LAI and the field-measured LAI along transects (with R 2 values of 0.54, 0.71, and 0.60 obtained for the SOBS, SYJP, and SOJP sites, respectively). The estimated LAI for the whole tower site is consistent with that obtained by the inversion of a canopy model in our previous study where values of 0.94, 0.92, and 0.63 were obtained for R 2 for the SOBS, SYJP and SOJP sites, respectively. The CASI 2-m summer data over the SOBS site was also employed to investigate the possibility of deriving canopy LAI from the summer data using linear mixture analysis. At a spatial resolution of 10 m, the correlation between the field-measured LAI and the estimated LAI along transects is small at R 2 less than 0.3, while R 2 increases to 0.6 at a spatial resolution of 30 m. The difficulty in canopy LAI retrieval from the summer data at a spatial resolution of 10 m is likely due to the variation of the understory reflectance across the scene, although spatial misregistration of the CASI data used may also be a possible contributing factor.