HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Related articles in JEQ Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Coops, N. C. Articles by Chisholm, L. Search for Related Content PubMed PubMed Citation Articles by Coops, N. C. Articles by Chisholm, L. Agricola Articles by Coops, N. C. Articles by Chisholm, L. Related Collections Scaling Ecosystem Management Ecosystem Restoration Plant Disease Plant and Environment Interactions

TECHNICAL REPORTS

Plant and Environment Interactions

Assessment of Crown Condition in Eucalypt Vegetation by Remotely Sensed Optical Indices

^a CSIRO Forestry and Forest Products, Private Bag 10, Clayton South, VIC 3169, Australia
^b Research & Development Division, State Forests of NSW, PO Box 100, Beecroft, NSW 2119, Australia
^c School of Geoscience, University of Wollongong, Northfields Avenue, Wollongong, NSW 2522, Australia

^* Corresponding author (nicholas.coops{at}csiro.au).

Received for publication March 4, 2003.

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Leaf and crown damage and discoloration characteristics are important variables when defining the health of eucalypt tree species and have been used as key indicators of environmental quality. These indicators can vary significantly over a few hectares, especially in mixed-species forests, making field-based environmental surveillance of crown condition an extremely expensive and logistically impractical task. Reflectance in narrow spectral wavelengths obtained from a field-based spectroradiometer and a Compact Airborne Spectrographic Imager 2 (CASI-2) were collected over eucalypt vegetation of varying condition in southeastern Australia and compared with leaf- and crown-based attributes including percent red foliage discoloration, percent leaf damage, and crown density and crown foliage condition. Of the leaf attributes sampled, percent leaf damage was well correlated with a red-green spectral index (r = 0.68, p < 0.01), and percent red discoloration was well correlated with the slope of the red-edge for selected species (r = 0.89, p < 0.001). Within-tree crown density was well correlated with the slope of the red-edge (r = 0.77, p < 0.001) and a previously published index of plant stress with crown foliage condition (r = 0.88, p < 0.01) for selected species. Despite evidence of strong interspecific variability, a set of narrow spectral wavelengths in the visible and near-infrared regions of the spectrum have been identified that will be useful in the development of forest ecosystem environmental quality indicators.

Abbreviations: ANOVA, analysis of variance • CASI-2, Compact Airborne Spectrographic Imager 2 • CI, Carter Index • CI_CASI-2, Carter Index calculated using CASI-2 wavelengths • GSI, Gamon and Surfus Index • GSI_CASI-2, Gamon and Surfus calculated using CASI-2 wavelengths • RE_ls, lower slope of the red-edge • RE_ls-CASI-2, lower slope of the red-edge calculated using CASI-2 wavelengths • RE_ts, total slope of the red-edge • RE_ts-CASI-2, total slope of the red-edge calculated using CASI-2 wavelengths

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
OVERSTOREY CONDITION of forest ecosystems is commonly used as an indicator of environmental quality in numerous countries (e.g., de Vries et al., 2000). A range of deleterious environmental processes have been shown to directly influence the physiological functioning and structure of tree crowns, including air pollution, soil acidification, and eutrophication by nitrogen (e.g., Klein and Perkins, 1988; Thimonier et al., 1994; Fredericksen et al., 1995; Dezzo et al., 1997). The resultant effects on foliar biochemical concentration and leaf age and structure can, in turn, directly influence the occurrence of phytophagous insect pests and diseases (e.g., Adams et al., 1995).

Traditional assessment of forest health and condition in Australia has been limited to ground-based visual assessment or aerial surveys coupled with aerial photographic interpretation (API) (Stone et al., 2000). While these assessments provide valuable information to forest managers they are acknowledged to be subjective, labor intensive, and often cannot reveal physiological changes that characterize early stress responses (Sampson et al., 2001; Zarco-Tejada et al., 2002).

Advances in remote sensing technology are now routinely demonstrating the possibility of developing forest canopy condition indicators based on detection of leaf pigments (Datt, 1998; Zarco-Tejada et al., 2002), and biochemicals (Smith et al., 2002), foliage biomass (Spanner et al., 1990a, 1990b; Coops et al., 1999), and structure (Lefsky et al., 2002). If a forest health surveillance system could be developed utilizing remote sensing methods this tool could help identify crown condition and health over large areas, at regular temporal intervals, providing more detailed and spatially extensive information than is currently possible (Stone et al., 2000). In addition, the strength of any health monitoring program would be greatly enhanced if it were capable of detecting early stress in forest stands (Mohammed et al., 1997), thus enabling forest managers to take a proactive course of remedial action before the stand reaches a point of nonrecovery.

Crown- and individual-leaf-based assessment of condition are the common scales at which indicators of forest health have been developed and incorporated into forest health surveillance programs (e.g., Innes, 1993). At the crown scale there can be a reduction in crown biomass due to leaf loss (defoliation) and crown contraction (Stone et al., 2000, 2001). Assessment of leaf-scale condition usually involves estimating leaf pigment content (typically chlorophyll a + b [Coops et al., 2003; Zarco-Tejada et al., 2002], carotenoids, or anthocyanins [Sims and Gamon, 2002]), leaf fluorescence (Zarco-Tejada et al., 2002), or leaf damage attributes (Stone et al., 2000). At the leaf scale in eucalypts, for example, initial exposure to stressful processes commonly results in a decrease in leaf chlorophyll content, which is often accompanied by, or the precursor to, the visual symptom of chlorosis. A red or purple discoloration of eucalypt leaves results from the presence of anthocyanin leaf pigments. While there is no general consensus for the presence of anthocyanins (Gould et al., 2000), increased leaf anthocyanin content has been associated with exposure to several stressful abiotic and biotic processes such as cold temperatures or feeding by sap-sucking psyllids (Close et al., 2001; Stone et al., 2001). A wide range of visual symptoms can follow (Reuter and Robinson, 1997) such as a reddening of the leaf tissue, occurrence of necrotic lesions, or removal of leaf tissue. At the crown scale there can be a reduction in crown density due to leaf loss and crown contraction (Stone et al., 2000, 2001). As the decline of eucalypt crowns proceeds, there is a shift of growing shoots away from the small peripheral branches toward the larger, more central branches and trunk. This contraction, along with a greater presence of dead branches, alters both the regularity of crown outline and the degree of intracrown variation when compared with healthy crowns (Stone et al., 2000).

Physical principles have confirmed that concentrations and composition of leaf pigments directly influence the spectral response of leaves in the visual wavelengths while cellular structure and water content of leaves are the main determinants in the near- and mid-infrared response of leaves (e.g., Gaussman, 1977; Horler et al., 1983; Carter and Knapp, 2001). These principles thereby allow specific bandwidths to be selected to infer the extent of stress and damage in foliage. This has been confirmed with numerous spectral indices, acquired at very narrow spectral resolution (in the order of 5–10 nm) and known as "hyper-spectral" data, being well correlated with chlorophyll pigments (e.g., Curran et al., 1997; Jago et al., 1999; Gitelson et al., 2003), carotenoid to chlorophyll ratios (Peñuelas et al., 1995a; Zarco-Tejada et al., 1999; Gitelson et al., 2003), xanthophyll cycle pigments and related photosynthetic performance (Gamon et al., 1992; Peñuelas et al., 1995a, 1995b; Gamon and Surfus, 1999), and other measures of integrated leaf stress (Carter, 1994; Peñuelas and Filella, 1998; Merton, 1999). These advances in our understanding of leaf physiology and spectral reflectance provide a sound basis for developing foliage and crown health indicators based on remote sensing observations (e.g., Rock et al., 1988; Zarco-Tejada et al., 2000; Sampson et al., 2001).

The primary goal of an ongoing research program in Australia (Stone et al., 2000) is to develop a forest condition indicator to assess the health and vitality of Australian eucalypt forests. Like a similar project in Canada (Mohammed et al., 1997; Zarco-Tejada et al., 2000; Sampson et al., 2001, 2003), the primary aim is to develop links between physiologically based bioindicators of tree condition from field and laboratory data and then to predict and spatially extrapolate optical indices from remote sensing to assess forest health. The potential then exists to analyze the spatially explicit characterization of forest canopy condition with other environmental quality coverages using GIS software. In previous work (Coops et al., 2003), we demonstrated that a number of spectral indices derived from hand-held and airborne spectroradiometer (Compact Airborne Spectrographic Imager-2; ITRES, Ottawa, ON, Canada) data correlated well with relative leaf chlorophyll content of eucalypt foliage from a range of species. The aim of this study was to examine if the narrow spectral wavelengths can discriminate between a number of key eucalypt species and if spectral indices, based in the red-green and red-edge region of the electromagnetic spectrum, are sensitive to a range of foliar condition categories of individual leaves and crown condition of eucalypt species in southeastern Australia.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Data Collection
Olney State Forest (33°07' S, 151°22' E) is situated in southeastern Australia (Fig. 1) . The climate is characterized as maritime temperate with high summer and low winter rainfall (mean annual precipitation of approximately 1200 mm). Summer mean maximum temperature is 27°C and winter mean maximum temperature is 15°C (Stone et al., 2000).

View larger version (13K): [in this window] [in a new window]   Fig. 1. Location of the study area.

Each leaf sample was measured for leaf condition in three ways: (i) leaf area lost to insect herbivory (% damage) using the method described in Stone and Bacon (1995) (mean = 10.1%, = 6.6%, n = 22); (ii) visual estimation of percent area of red or purple discoloration on each leaf (% red) (mean = 20.08%, = 22.3%, n = 22); and (iii) visual estimation of percent necrosis on each leaf (mean = 8.5%, = 10.2%, n = 22).

In addition to the leaf-based analysis, each crown was visually assessed by a single forest health professional for two variables: foliage condition and crown density. Foliage condition rates insect damage directly visible on leaves into five classes and crown density has nine classes ranging from very dense to very sparse (Table 1).

Airborne Imagery
Imagery was collected from the Compact Air-Borne Spectrographic Imager 2 (CASI-2). The CASI-2 is an air-borne push-broom imaging spectrograph that acquires imagery in the visible and near-infrared regions of the electromagnetic spectrum between 413 and 958 nm (Shepherd et al., 1995; Treitz and Howarth, 1996). To minimize bidirectional reflectance and to cover the study area, flight lines were oriented parallel to the solar azimuth (i.e., away from the sun). Imagery was flown as a series of north–south strips on 19 Oct. 1999 at 1200 h. Imagery was collected at a spatial ground resolution of 0.8 m in 10 spectral bands (Table 2), which were preselected after examination of spectra of damaged eucalypt leaves obtained from a portable spectroradiometer and the literature. Flight lines also covered pseudo invariant features (PIFs) (large sheets of different shaded material) to assist in image calibration. The CASI-2 data were converted from digital numbers to physical units of radiance (µW cm^–2 sr^–1 nm^–1). Downwelling irradiance was sampled coincidently with image acquisition from the roof of the aircraft using an incident light sensor (ILS) system that was configured into the same spectral bands as the CASI-2 sensor. Radiance was then converted to reflectance, using a model that corrected for atmospheric transmittance, path radiance, and incident radiation at the appropriate wavelengths using ambient humidity, temperature, and wind speed data collected by an automatic weather station recording at the same time as image acquisition.

Based on this result whole crown boundaries were located on large-scale hardcopies of the imagery, aerial photographs, and field maps and the boundaries manually digitized. The mean and standard deviation of the crown reflectance were then calculated for each spectral wavelength. To assess the capacity of the CASI-2 reflectance to successfully differentiate different species at the study area, an analysis of variance (ANOVA) was used to isolate species-specific effects on the acquired CASI-2 reflectances in the 10 spectral bands. Species with a single crown signature were removed from the analysis, resulting in a dataset of nine tree species groups and 65 crown signatures.

Spectral Indices
Carter (1994), Gamon and Surfus (1999), Merton (1999), and Thenkabail et al. (2000)( 2002) among others provide detailed reviews of spectral indices that have been applied to vegetation spectra. These references contain a complete derivation and description of many available indices. When reviewing indices, Thenkabail et al. (2002) states that there is no single best approach for determining the optimal number and combination of narrow wavebands required to provide best estimates of agricultural crop characteristics. We chose the following four simple reflectance indices based on the literature and our knowledge of leaf reflectance curve behavior for insect-damaged eucalypt leaves (Stone et al., 2001). The indices had to be suitable for scaling up from the spectral radiometer measurements to the CASI-2 imagery. These indices utilize key sections of the visible and near infrared electromagnetic spectrum for foliage condition studies: the lower red-edge, the total red-edge, a ratio of near-infrared to the red, and an index that compares reflectance from the red and green wavelengths. As the CASI-2 wavelength positions are at greater intervals than the field spectroradiometer, the wavelengths used in the calculation of some of the indices varied between the spectroradiometer and the CASI-2. In these cases either the nearest waveband was selected or the average of the adjoining bands was substituted (as detailed below).

Using spectral radiometers, Ahern (1988), Rock et al. (1988), Carter et al. (1996), and others have demonstrated that certain regions of the reflected electromagnetic spectrum of leaves, such as a blue shift at the inflection point between the red and near-infrared wavelengths (690–740 nm) and at the near-infrared shoulder and plateau (750–1300 nm), are sensitive to changes arising from the initial (previsual) cellular response to specific stressful agents. Red-edge indices are based not on absolute reflectance but rather on the wavelength position of the transition between low reflectance in the red region of the spectrum and high reflectance in the near infrared (Sims and Gamon, 2002). For the purpose of identifying simple measures of the red-edge reflectance, linear regression calculations based on the slope of the red-edge (red-edge versus wavelength) were used. As red-edge spectra are generally asymptotic to near-linear, simple linear regression is appropriate to model spectral trends. Curran et al. (1990) defines the wavelength intervals of the lower slope as between 690 and 710 nm and the total red-edge as between 690 and 740 nm.

The calculation for the total slope of the red-edge (RE_ts) is:

[1]

and the lower slope of the red-edge (RE_ts) is:

[2]

where ₇₁₀ = reflectance at 710 nm, ₇₄₀ = reflectance at 740 nm, ₆₉₀ = reflectance at 690 nm, and the denominator is the spectral range (in nm) between 690 and 740 nm and 690 and 710 nm, respectively.

For CASI-2, however, reflectance was not captured at 690 nm, but rather 680 and 700 nm. As there is a general linear trend in this region of vegetation spectra, we averaged values from the 680- and 700-nm bands to approximate a 690-nm reflectance. This more accurately represents the intent behind Curran's slope computations. The slope of the total red-edge, therefore, becomes:

[3]

Similarly for the lower slope, reflectance at 710 nm was computed from an average of recorded reflectance at 700 and 720 nm:

[4]

Carter (1994) measured narrow wavebands from a field-based spectroradiometer to evaluate the effectiveness of indices to predict plant stress as defined by chlorophyll content. A number of ratios of near-infrared to visible narrow wavelengths were significantly correlated with a variety of attributes of stressed and nonstressed foliage. In this research we utilized the most successful of the Carter (1994) indices (designated CI), which is shown below:

[5]

where ₆₉₅ = reflectance at 695 nm and ₇₆₀ = reflectance at 760 nm.

For CASI-2 the equation becomes:

[6]

where ₇₀₀ = reflectance at 700 nm and ₇₆₀ = reflectance at 760 nm.

Gamon and Surfus (1999) proposed using a combination of an index of red and green wavelengths that specifically target anthocyanins. The index (GSI) is shown below and is simply the sum of the red wavelengths over the sum of the green wavelengths:

[7]

where ₇₀₀ = reflectance at 700 nm, ₆₀₀ = reflectance at 600 nm, and ₅₀₀ = reflectance at 500 nm (Gamon and Surfus, 1999).

To utilize this index using CASI-2 data we can sum CASI-2 Bands 3 to 5 (635–700 nm), which cover the red region of the spectrum as the numerator. As there is only one channel in the green region of the spectrum we simply utilize CASI-2 Band 2 (550 nm) as the denominator:

[8]

where ₇₀₀ = reflectance at 700 nm, ₆₃₅ = reflectance at 635 nm, and ₅₅₀ = reflectance at 550 nm.

All CASI-2 derived indices were analyzed using the STATISTICA statistical package (StatSoft, 1995).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Leaf Scale Assessment of Condition
Correlations among the individual-leaf-based assessments reveal that estimated percent reddish discoloration of the leaves (% red) is significantly correlated with both percent leaf area missing (% damage) (r = 0.58, p < 0.01) and estimated percent necrotic area (r = 0.56, p < 0.05). Leaf area missing and estimated necrotic area, however, are not correlated (p > 0.05) over the dataset.

As the suite of damaging insects differed between eucalypt species, there was species-specific effect associated with the extent of the various types of foliar symptoms; hence, the tree data were divided into the two dominant susceptible eucalypt species, Sydney blue gum and gray ironbark. Table 3 shows the correlation coefficient values (r) between the three leaf-scale variables and the indices derived from the spectroradiometer data.

View larger version (13K): [in this window] [in a new window]   Fig. 2. Percent leaf damage versus the Gamon and Surfus Index (GSI) for all species.

View larger version (12K): [in this window] [in a new window]   Fig. 3. Percent of leaf with red discoloration plotted against the Gamon and Surfus Index (GSI) for gray ironbark.

To investigate more closely how spectrally distinct the eucalypt crowns are, a second ANOVA was undertaken using a subset of the data for just the eucalypt species. This analysis indicates that all but three of the CASI-2 spectral bands were able to significantly discriminate the eucalypt species at the p < 0.01 level, and of these three (550, 635, and 700 nm), two bands were able to discriminate at the p < 0.05 level. The only CASI-2 band that provides no discriminatory power is CASI-2 Band 2 (550 nm). The two bands with the highest capacity to discriminate the sampled eucalypt species are 720 and 740 nm, corresponding to the mid red-edge region. This result implies that the CASI-2 imagery has the capacity to successfully discriminate different eucalypt species at the crown scale. However, as individual eucalypt species have a range of susceptibility to different herbivorous insects, these results may be confounded.

Crown-Based Assessment of Crown Density and Damage
Table 5 shows the correlation coefficient values (r) between crown density and the four indices as derived from the mean spectral responses of each crown delineated on the CASI-2 0.8-m imagery. Correlation values are provided for the full dataset of eucalypt crowns delineated on the imagery as well as individual relationships for the two dominant eucalypts species. On a general level, Table 5 indicates that moderate to highly significant relationships exist between the spectral indices and the crown-based attributes of crown density. The RE_ts-CASI-2 is well correlated with the crown density variable with a highly significant relationship across all species (r = 0.77, p < 0.001) (Fig. 4) , increasing to r = 0.89 and 0.83 for Sydney blue gum and gray ironbark, respectively (p < 0.01). Likewise, the CI_CASI-2 is moderately correlated (r = –0.75, p < 0.01) across all species with a negative correlation on an individual-species basis. By contrast, the GSI_CASI-2 is poorly correlated with crown density except for Sydney blue gum.

View larger version (13K): [in this window] [in a new window]   Fig. 4. Crown density score versus total slope of the red-edge calculated using the Compact Airborne Spectrographic Imager 2 (CASI-2) wavelengths (REts-CASI-2) for all species.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
On a leaf basis, correlations between the spectral indices and leaf damage are moderate over all the sampled species but appear highly significant for individual species. Overall, the GSI (the ratio of the sum of all the red to green reflectances) appears to provide the best correlations at the leaf scale, which may indicate that anthocyanins can be detected using this type of imagery.

For crown-scale attributes (crown density and crown foliage condition), both the RE_ts-CASI-2 and the CI_CASI-2 index have moderate to high correlations and indicate that crown-scale relationships are achievable by high spectral and spatial resolution imagery. Generally, crown density was correlated better with the spectral indices than foliage condition and overall RE_ts-CASI-2 appears to be the most correlated index.

Comparing indices derived from the leaf reflectance spectra contributes to the interpretation of relationships between reflectance data and the range of damage symptoms presented by eucalypt foliage. However, even in the small sample of reflectance indices tested, we demonstrated that the suitability of indices to scaling up from leaves to the tree canopy is variable. Canopy reflectance is not only influenced by leaf properties, but also a range of other sources of variability including atmospheric effects, shadow pattern, background composition, and instrument noise (e.g., Koch et al., 1990; Yoder and Pettigrew-Crosby, 1995; Coops et al., 2003). The potential trade-offs between the accuracy of reflectance algorithms and their robustness to scaling up can only be quantified through step-wise examination of spectral data in an integrated approach to scaling up.

In this study, species-specific relationships between spectral indices and leaf and crown condition attributes may be attributed to both a species susceptibility to various damage agents and the response mechanism to declining condition. At the crown scale, gray ironbark in particular tends to respond to decline through the production of dense epicormic growth along the main stem and branches of the crown. By comparison, Sydney blue gum does not readily produce this type of pattern of growth, but rather retains sparse clumps of foliage on its upper branches while shedding lower branches. A eucalypt species that readily replaced damaged primary leaves with epicormic foliage (e.g., gray ironbark) would have a different composition of leaf ages in the crown compared with a species that retained and replaced damaged leaves less frequently (e.g., Sydney blue gum). Damage to eucalypt foliage accumulates proportionally to length of exposure to the damaging agents. Therefore, our results suggest that the ratio of red to green reflectances (Gamon and Surfus, 1999) may be sensitive to the varying proportions of leaf age cohorts present in a crown as well as foliar density. Issues such as species differences in the symptoms of affected crowns need to be considered when these types of relationships are extrapolated across the entire forest estate.

This study has highlighted that significant spectral differences can occur between eucalypt tree species. The results demonstrate that CASI-2 bands also had the capacity to discriminate between a range of eucalypt and non-eucalypt species. In the past the application of wide-spectral bandwidth imagery in Australia (such as satellite Thematic Mapper [TM] and Airborne Thematic Mapper [ATM] sensors) have shown a poor capacity to predict individual species or species types. This has been attributed to (i) the complexity of the eucalypt forest ecosystem, (ii) the effect of disturbance, both natural and man induced, as well as (iii) the structural composition of eucalypts, which generally have vertical leaves and a wide range in leaf growth from juvenile to mature on a single branch (Lees and Ritman, 1991). Using high spatial resolution imagery with similarly wide spectral bands, consistent discrimination of individual eucalypt species has been shown to be still very difficult. Even though the damage does not occur evenly over all species, the ANOVA results presented here indicate that, utilizing high spectral and spatial resolution imagery, all species could be distinguished using at least one of the CASI-2 spectral channels. Obviously, discrimination of eucalypt from rainforest species is easily accomplished with 7 of the 10 CASI-2 bands differentiating eucalypt species from rainforest (for example, gray ironbark from lilly pilly).

All but three of the CASI-2 spectral bands were able to significantly discriminate (p < 0.01) between eucalypt species with only one unable to discriminate at the p < 0.05 level. The only CASI-2 band that provides no discriminatory power is CASI-2 Band 2 (550 nm). The two bands with the highest capacity to discriminate eucalypt species are 720- and 740-nm wavelengths corresponding to the mid red-edge region.

The generation of crown spectra from the entire tree canopy appears to have produced meaningful spectral statistics for predicting crown-based indicators of condition. Accurate correspondence between field-estimated leaf and crown variables and the identification of the respective crowns on the 0.8-m CASI-2 imagery was critical to develop the relationships between CASI-2 data and health. Significant advancements in automatic tree delineation from high spatial resolution imagery have been made (Culvenor, 2002) and ongoing work is determining the effectiveness of methodologies defining operational procedures for using the technique in conjunction with forest inventory and to confirm their cost effectiveness.

The use of high spectral and spatial resolution imagery for analysis of individual crown spectra necessitates, at this stage, imagery obtained from airborne platforms. Airborne imagery geometric correction due to aircraft motion, brightness corrections due to lens and scanner characteristics, atmospheric effects, and directional variations in the forest can have important effects. Therefore, any operational program utilizing high spatial resolution imagery will require these calibration issues to be considered and addressed. In the future, it is anticipated that high spatial and high spectral resolution imagery will be available from space-based platforms. While a combination of these two capabilities may still be some way off there is significant advancement in space-borne hyperspectral sensors. Hyperspectral sensors are included on-board the new generation of satellites planned by various governments (such as the European Space Agency's Medium Resolution Imaging Spectrometer [MERIS]) and by U.S. private industry (American Society for Photogrammetry and Remote Sensing, 1995; Stoney and Hughes, 1998). The recently launched Hyperion sensor with 220 spectral bands on-board NASA's New Millennium Program's Earth Observer l (EO-l) is now providing researchers with hyperspectral images acquired from space at 30-m spatial resolution (Unger, 2001) and the moderate resolution imaging spectrometer (MODIS) with 36 channels onboard Terra offers data from 400 to 2500 nm (Thenkabail et al., 2000).

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Remotely sensed imagery is a representation of the sum of all tree and forest structural attributes, and the isolation of a particular feature is difficult (Olthof and King, 2000). The benefit of combining leaf- and crown-based estimates of forest condition allows some of these component parts to be analyzed in detail and moves toward the development of more general indicators of environmental quality. The results discussed here demonstrate that high spectral resolution remote sensing data, once correctly processed, can provide important information to ongoing forest ecosystem monitoring programs.

	ACKNOWLEDGMENTS

 
We thank Rod Rumbachs, Simon McDonald, Grahame Price, Darren Waterson, and Alf Britton for their technical and field assistance and Bob Gittins for biometrical advice on the Olney project. We are grateful to Chris Beadle and Gina Mohammed for discussions and advice on this project. We also thank Ken Old and Phil Ryan for providing technical, administrative, and intellectual assistance to this project. We thank Ken Old, Mark Dudzinski, and Grahame Price for comments on drafts of this manuscript. Funding for this research was provided by State Forests of New South Wales, CSIRO Forestry and Forest Products, Canberra, and the Forestry and Wood Products Research and Development Corporation (FWPRDC; PN99.814), Melbourne.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION CONCLUSIONS REFERENCES

 

• Adams, M.A., P.M. Attiwill, and L.M. Wang. 1995. Effects of phosphorus supply on growth and nitrogen fractions in xylem sap and foliage of Eucalyptus regnans (F.Muell.), E. nitens (Maiden) and E. globulus (Labill.) seedlings: Implications for herbivory. Trees (Berlin) 9:324–331.
• Ahern, F.J. 1988. The effects of bark beetle stress on the foliar spectral reflectance of lodgepole pine. Int. J. Remote Sens. 9:1451–1468.
• American Society for Photogrammetry and Remote Sensing. 1995. Land satellite information in the new decade: Data notebook. ASPRS, Bethesda, MD.
• Carter, G.A. 1994. Ratios of leaf reflectances in narrow wavebands as indicators of plant stress. Int. J. Remote Sens. 15:697–703.
• Carter, G.A., W.G. Cibula, and R.L. Miller. 1996. Narrow-band reflectance imagery compared with thermal imagery for early detection of plant stress. J. Plant Physiol. 148:515–522.
• Carter, G.A., and A.K. Knapp. 2001. Leaf optical properties in higher plants: Linking spectral characteristics to stress and chlorophyll concentration. Am. J. Bot. 88:677–684.[Abstract/Free Full Text]
• Close, D.C., C.L. Beadle, P. Brown, and G.K. Holz. 2001. Cold-induced photo inhibition affects establishment of Eucalyptus nitens Deane and Maiden Maiden and Eucalyptus globulus Labill. Trees (Berlin) 15:32–41.
• Coops, N.C., H. Bi, P. Barnett, and P. Ryan. 1999. Prediction of mean and current volume increments of a Eucalypt forest using historical Landsat MSS data. J. Sust. For. 9:149–168.
• Coops, N.C., C. Stone, D.S. Culvenor, L.A. Chisholm, and R. Merton. 2003. Chlorophyll content in eucalypt vegetation at the leaf and canopy scales as derived from high resolution spectral data. Tree Physiol. 23:23–31.[ISI][Medline]
• Culvenor, D.S. 2002. TIDA: An algorithm for the delineation of tree crowns in high spatial resolution remotely sensed imagery. Comput. Geosci. 28:33–44.
• Curran, P.J., J.L. Dungan, and H.L. Gholz. 1990. Exploring the relationship between reflectance red-edge and chlorophyll content in slash pine. Tree Physiol. 7:33–48.[ISI][Medline]
• Curran, P.J., I.A. Kupiec, and G.M. Smith. 1997. Remote sensing the biochemical composition of a slash pine canopy. IEEE Trans. Geosci. Remote Sens. 35:415–420.
• Datt, B. 1998. Remote sensing of chlorophyll a, chlorophyll b, chlorophyll a+b, and total carotenoid content in eucalyptus leaves. Remote Sens. Environ. 66:111–121.
• De Vries, W., J.M. Klap, and J.W. Erisman. 2000. Effects of environmental stress on forest crown condition in Europe. Part 1: Hypotheses and approach to the study. Water Air Soil Pollut. 118:317–333.
• Dezzo, N., L. Hernandez, and H. Folster. 1997. Canopy dieback in lower montane forests of Alto Uriman, Venezuelan Guayana. Plant Ecol. 132:197–209.
• Fredericksen, T.S., B.J. Joyce, J.M. Skelly, K.C. Steiner, T.E. Kolb, K.B. Kouterick, J.E. Savage, and K.R. Snyder. 1995. Physiology, morphology, and ozone uptake of leaves of black cherry seedlings, saplings, and canopy trees. Environ. Pollut. 89:273–283.[Medline]
• Gamon, J.A., J. Peñuelas, and C.B. Field. 1992. A narrow-wave spectral index that tracks diurnal changes in photosynthetic efficiency. Remote Sens. Environ. 41:35–44.
• Gamon, J.A., and J.S. Surfus. 1999. Assessing leaf pigment content and activity with a reflectometer. New Phytol. 143:105–117.
• Gaussman, H.W. 1977. Reflectance of leaf components. Remote Sens. Environ. 6:1–9.
• Gitelson, A.A., Y. Gritz, and M.N. Merzlyak. 2003. Relationships between leaf chlorophyll content and spectral reflectance and algorithms for non-destructive chlorophyll assessment in higher plant leaves. J. Plant Physiol. 160:271–283.[ISI][Medline]
• Gould, K.S., K.R. Markham, R.H. Smith, and J.J. Goris. 2000. Functional role of anthocyanins in the leaves of Quintinia serrata A. Cunn. J. Exp. Bot. 51:1107–1115.
• Greaves, B.L., and R.D. Spencer. 1993. An evaluation of spectroradiometry and multispectral scanning for differentiating forest communities. Aust. For. 56:68–79.
• Horler, D.H.H., K. Dockray, and J. Barber. 1983. The red edge of plant reflectance. Int. J. Remote Sens. 4:273–288.
• Howard, J.A. 1966. Spectral energy relations of isobilateral leaves. Aust. J. Biol. Sci. 19:757–766.
• Innes, J.L. 1993. Forest health: Its assessment and status. CABI Publ., Wallingford, UK.
• Jago, R., M.E. Cutler, and P.J. Curran. 1999. Estimating canopy chlorophyll concentration from field and airborne spectra. Remote Sens. Environ. 68:217–224.
• Jurskis, V., and J. Turner. 2003. Eucalypt dieback in eastern Australia: A simple model. Aust. For. 65:87–98.
• Koch, B., U. Ammer, T. Schneider, and H. Wittmeier. 1990. Spectroradiometer measurements in the laboratory and in the field to analyse the influence of different damage symptoms on the reflection spectra of forest trees. Int. J. Remote Sens. 11:1145–1163.
• Klein, R.M., and T.D. Perkins. 1988. Primary and secondary causes and consequences of contemporary forest decline. Bot. Rev. 54:2–43.
• Leckie, D.G., X. Yuan, D.P. Ostaff, H. Piene, and D.A. Maclean. 1992. Analysis of high resolution multispectral MEIS imagery for spruce budworm damage assessment on a single tree basis. Remote Sens. Environ. 40:125–136.
• Lees, B.G., and K. Ritman. 1991. Decision-tree and rule induction approach to integration of remotely sensed and GIS data in mapping vegetation in disturbed or hilly environments. Environ. Manage. 15:823–831.
• Lefsky, M.A., W.B. Cohen, G.G. Parker, and D.J. Harding. 2002. Lidar remote sensing for ecosystem studies. Bioscience 52:19–30.
• Li, X., and A.H. Strahler. 1985. Geometric-optical modelling of a conifer forest canopy. IEEE Trans. Geosci. Remote Sens. GE-23: 705–721.
• Merton, R.N. 1999. Multi-temporal analysis of community scale vegetation stress with imaging spectroscopy. Ph.D. diss. Dep. of Geogr., Univ. of Auckland, New Zealand.
• Mohammed, G.H., P.H. Sampson, S.J. Colombo, T.L. Noland, and J.R. Miller. 1997. Bioindicators of forest sustainability: Development of a forest condition rating system for Ontario. For. Res. Info. Paper 137. Ontario For. Res. Inst., Ottawa, ON.
• Olthof, I., and D.J. King. 2000. Development of a forest health index using multispectral airborne digital camera imagery. Can. J. Remote Sens. 26:166–176.
• O'Neill, A.L., S. Hardy, and S.I. Fraser. and K.R. McCloy. 1990. Leaf morphology and the spectral reflectance of some Australian plant species. p. 1096–1103. In Proc. of the 5th Australasian Remote Sensing Conf., Perth, Australia. Australasian Photogrammetry and Remote Sensing Soc., Perth.
• Peñuelas, J., F. Baret, and I. Filella. 1995a. Semi-empirical indices to assess carotenoids/chlorophyll a ratio from leaf spectral reflectance. Photosynthetica 31:221–230.[ISI]
• Peñuelas, J., and I. Filella. 1998. Visual and near-infrared reflectance techniques for diagnosing plant physiological status. Trends Plant Sci. 3:151–156.
• Peñuelas, J., I. Filella, P. Lloret, F. Muñoz, and M. Vilajeliu. 1995b. Reflectance assessment of mite effects on apple trees. Int. J. Remote Sens. 16:2727–2733.
• Reuter, D.J., and J.B. Robinson. 1997. Plant analysis: An interpretation manual. CSIRO Publ., Melbourne, Australia.
• Rock, B.N., T. Hoshizaki, and J.R. Miller. 1988. Comparison of in situ and airborne spectral measurements of the blue shift associated with forest decline. Remote Sens. Environ. 24:109–127.
• Sampson, P.H., G.H. Mohammed, P.J. Zarco-Tejada, J.R. Miller, T.L. Noland, D. Irving, P.M. Treitz, S.J. Colombo, and J. Freemantle. 2001. The bio-indicators of forest condition project: A physiological, remote sensing approach. For. Chron. 76:941–952.
• Sampson, P.H., P.J. Zarco-Tejada, G.H. Mohammed, J.R. Miller, and T.L. Noland. 2003. Hyperspectral remote sensing of forest condition: Estimating chlorophyll content in tolerant hardwoods. For. Sci. (Bethesda, MD) 49:381–391.
• Shepherd, P.R., J.R. Freemantle, S. McArdle, and J.R. Miller. 1995. A comparison of different operational reflectance generation methods applied to airborne CASI imagery. p. 268–273. In Proc. of the 17th Canadian Symp. on Remote Sensing, Saskatoon, SK. Vol. 1. Canadian Soc. of Remote Sensing, Ottawa, ON.
• Sims, D.A., and J.A. Gamon. 2002. Relationship between leaf pigment content and spectral reflectance across a wide range of species, leaf structures and developmental stages. Remote Sens. Environ. 81:337–354.
• Smith, M.L., S.V. Ollinger, M.E. Martin, J.D. Aber, R.A. Hallett, and C.L. Goodale. 2002. Direct estimation of aboveground forest productivity through hyperspectral remote sensing of canopy nitrogen. Ecol. Appl. 12:1286–1302.
• Spanner, M.A., L.L. Pierce, D.L. Peterson, and S.W. Running. 1990a. Remote sensing of temperate coniferous forest leaf area index: The influence of canopy closure, understorey vegetation and background reflectance. Int. J. Remote Sens. 11:95–111.
• Spanner, M.A., L.L. Pierce, S.W. Running, and D.L. Peterson. 1990b. The seasonality of AVHRR data of temperate coniferous forests: Relationships with leaf area index. Remote Sens. Environ. 33:97–112.
• StatSoft. 1995. STATISTICA for Windows. StatSoft, Tulsa, OK.
• Stone, C. 1996. The role of psyllids Hemiptera: Psyllidae and bell miners Manorina melanophrys in canopy dieback of Sydney blue gum Eucalyptus saligna Sm. Aust. J. Ecol. 21:450–458.
• Stone, C. 1999. Assessment and monitoring of decline and dieback of forest eucalypts in relation to ecological sustainable forest management: A review with a case study. Aust. For. 62:55–63.
• Stone, C., and P.E. Bacon. 1995. Influence of insect herbivory on the decline of black box Eucalyptus largiflorens. Aust. J. Bot. 43:555–564.
• Stone, C., L. Chisholm, and N.C. Coops. 2001. Spectral reflectance characteristics of eucalypt foliage damaged by insects. Aust. J. Bot. 49:687–698.
• Stone, C., N.C. Coops, and D. Culvenor. 2000. Conceptual development of a Eucalypt Canopy Condition Index using high resolution spatial and spectral remote sensing imagery. J. Sust. For. 11:23–45.
• Stoney, W.E., and J.R. Hughes. 1998. A new space race is on! GIS World 11:44–46.
• Thenkabail, P.S., R.B. Smith, and E. De Pauw. 2000. Hyperspectral vegetation indices and their relationships with agricultural crop characteristics. Remote Sens. Environ. 71:158–182.
• Thenkabail, P.S., R.B. Smith, and E. De Pauw. 2002. Evaluation of narrowband and broadband vegetation indices for determining optimal hyperspectral wavebands for agricultural crop characterization. Photogramm. Eng. Remote Sens. 68:607–621.
• Thimonier, A., J.L. Dupouey, F. Bost, and M. Becker. 1994. Simultaneous eutrophication and acidification of a forest in north-east France. New Phytol. 126:533–541.
• Treitz, P., and P. Howarth. 1996. Remote sensing forest ecosystem characterization: A review. NODA/NFP Tech. Rep. TR-12. Nat. Resour. Canada and Canadian For. Serv., Ottawa, ON, Canada.
• Unger, S. 2001. Overview of EO-1. The first 120 days. Proc. IGARSS 2001, Sydney, Australia. 9–13 July 2001. IEEE Soc., Sydney.
• Yoder, B.J., and R.E. Pettigrew-Crosby. 1995. Predicting nitrogen and chlorophyll content and concentrations from reflectance spectra (400–2500 nm) at leaf and canopy scales. Remote Sens. Environ. 53:199–211.
• Zarco-Tejada, P.J., J.R. Miller, G.H. Mohammed, and T.L. Noland. 2000. Chlorophyll fluorescence effects on vegetation apparent reflectance: I. Leaf-level measurements and model simulation. Remote Sens. Environ. 74:582–595.
• Zarco-Tejada, P.J., J.R. Miller, G.H. Mohammed, T.L. Noland, and P.H. Sampson. 1999. Canopy optical indices from infinite reflectance and canopy reflectance models for forest condition monitoring: Applied to hyperspectral CASI Data. Proc. IEEE 1999 Int. Geoscience and Remote Sensing Symp., Hamburg, Germany. 28 June–2 July 1999. IEEE Soc., Hamburg.
• Zarco-Tejada, P.J., J.R. Miller, G.H. Mohammed, T.L. Noland, and P.H. Sampson. 2002. Vegetation stress detection through Chlorophyll a + b estimation and fluorescence effects on hyperspectral imagery. J. Environ. Qual. 31:1433–1441.[Abstract/Free Full Text]

Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2004 33: 799-804. [Full Text]  

This article has been cited by other articles:

				P. Aplin Remote sensing: ecology Progress in Physical Geography, March 1, 2005; 29(1): 104 - 113. [PDF]

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Related articles in JEQ Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (8) Citing Articles via Google Scholar Google Scholar Articles by Coops, N. C. Articles by Chisholm, L. Search for Related Content PubMed PubMed Citation Articles by Coops, N. C. Articles by Chisholm, L. Agricola Articles by Coops, N. C. Articles by Chisholm, L. Related Collections Scaling Ecosystem Management Ecosystem Restoration Plant Disease Plant and Environment Interactions