Invasive species detection in Hawaiian rainforests using airborne imaging spectroscopy and LiDAR

Remote sensing of invasive species is a critical component of conservation and management efforts, but reliable methods for the detection of invaders have not been widely established. In Hawaiian forests, we recently found that invasive trees often have hyperspectral signatures unique from that of native trees, but mapping based on spectral reflectance properties alone is confounded by issues of canopy senescence and mortality, intra- and inter-canopy gaps and shadowing, and terrain variability. We deployed a new hybrid airborne system combining the Carnegie Airborne Observatory (CAO) small-footprint light detection and ranging (LiDAR) system with the Airborne Visible and Infrared Imaging Spectrometer (AVIRIS) to map the three-dimensional spectral and structural properties of Hawaiian forests. The CAO-AVIRIS systems and data were fully integrated using in-flight and post-flight fusion techniques, facilitating an analysis of forest canopy properties to determine the presence and abundance of three highly invasive tree species in Hawaiian rainforests.

The LiDAR sub-system was used to model forest canopy height and top-of-canopy surfaces; these structural data allowed for automated masking of forest gaps, intra- and inter-canopy shadows, and minimum vegetation height in the AVIRIS images. The remaining sunlit canopy spectra were analyzed using spatially-constrained spectral mixture analysis. The results of the combined LiDAR-spectroscopic analysis highlighted the location and fractional abundance of each invasive tree species throughout the rainforest sites. Field validation studies demonstrated < 6.8% and < 18.6% error rates in the detection of invasive tree species at  7 m² and  2 m² minimum canopy cover thresholds. Our results show that full integration of imaging spectroscopy and LiDAR measurements provides enormous flexibility and analytical potential for studies of terrestrial ecosystems and the species contained within them.     

Forest carbon densities and uncertainties from Lidar, QuickBird, and field measurements in California

Greenhouse gas inventories and emissions reduction programs require robust methods to quantify carbon sequestration in forests. We compare forest carbon estimates from Light Detection and Ranging (Lidar) data and QuickBird high-resolution satellite images, calibrated and validated by field measurements of individual trees. We conducted the tests at two sites in California: (1) 59 km² of secondary and old-growth coast redwood (Sequoia sempervirens) forest (Garcia-Mailliard area) and (2) 58 km² of old-growth Sierra Nevada forest (North Yuba area). Regression of aboveground live tree carbon density, calculated from field measurements, against Lidar height metrics and against QuickBird-derived tree crown diameter generated equations of carbon density as a function of the remote sensing parameters. Employing Monte Carlo methods, we quantified uncertainties of forest carbon estimates from uncertainties in field measurements, remote sensing accuracy, biomass regression equations, and spatial autocorrelation. Validation of QuickBird crown diameters against field measurements of the same trees showed significant correlation (r = 0.82, P < 0.05). Comparison of stand-level Lidar height metrics with field-derived Lorey's mean height showed significant correlation (Garcia-Mailliard r = 0.94, P < 0.0001; North Yuba R = 0.89, P < 0.0001). Field measurements of five aboveground carbon pools (live trees, dead trees, shrubs, coarse woody debris, and litter) yielded aboveground carbon densities (mean ± standard error without Monte Carlo) as high as 320 ± 35 Mg ha^− 1 (old-growth coast redwood) and 510 ± 120 Mg ha^− 1 (red fir [Abies magnifica] forest), as great or greater than tropical rainforest. Lidar and QuickBird detected aboveground carbon in live trees, 70-97% of the total. Large sample sizes in the Monte Carlo analyses of remote sensing data generated low estimates of uncertainty. Lidar showed lower uncertainty and higher accuracy than QuickBird, due to high correlation of biomass to height and undercounting of trees by the crown detection algorithm. Lidar achieved uncertainties of < 1%, providing estimates of aboveground live tree carbon density (mean ± 95% confidence interval with Monte Carlo) of 82 ± 0.7 Mg ha^− 1 in Garcia-Mailliard and 140 ± 0.9 Mg ha^− 1 in North Yuba. The method that we tested, combining field measurements, Lidar, and Monte Carlo, can produce robust wall-to-wall spatial data on forest carbon.     

Endmember variability in Spectral Mixture Analysis: A review

The composite nature of remotely sensed spectral information often masks diagnostic spectral features and hampers the detailed identification and mapping of targeted constituents of the earth's surface. Spectral Mixture Analysis (SMA) is a well established and effective technique to address this mixture problem. SMA models a mixed spectrum as a linear or nonlinear combination of its constituent spectral components or spectral endmembers weighted by their subpixel fractional cover. By model inversion SMA provides subpixel endmember fractions. The lack of ability to account for temporal and spatial variability between and among endmembers has been acknowledged as a major shortcoming of conventional SMA approaches using a linear mixture model with fixed endmembers. Over the past decades numerous efforts have been made to circumvent this issue. This review paper summarizes the available methods and results of endmember variability reduction in SMA. Five basic principles to mitigate endmember variability are identified: (i) the use of multiple endmembers for each component in an iterative mixture analysis cycle, (ii) the selection of a subset of stable spectral features, (iii) the spectral weighting of bands, (iv) spectral signal transformations and (v) the use of radiative transfer models in a mixture analysis. We draw attention to the high complementarities between the different techniques and suggest that an integrated approach is necessary to effectively address endmember variability issues in SMA.     

Spectroscopic classification of tropical forest species using radiative transfer modeling

Leaf spectroscopy may be useful for tropical species discrimination, but few studies have provided an understanding of the spectral separability of species or how leaf spectroscopy scales to the canopy level relevant to mapping. Here we report on a study to classify humid tropical forest canopy species using field-measured leaf optical properties with leaf and canopy radiative transfer models. The experimental dataset included 188 canopy species collected in humid tropical forests of Hawaii. The leaf optical model PROSPECT-5 was used to simulate the leaf spectra of each species, which was used to train a classifier based on Linear Discriminant Analysis, and a canopy radiative transfer model 4SAIL2 to scale leaf measurements to the canopy level. The relationship linking classification accuracy at the leaf level to biodiversity showed an asymptotic trend reaching a maximum error of 47% when applied to the entire 188 species experimental dataset, and 56% when a simulated dataset showing amplified within-species spectral variability was used, suggesting uniqueness of the spectral signature for a significant proportion of species under study. The maximum error in canopy-level species classification was higher than leaf-level classification: 55% when canopy structure was held constant, and 64% with varying and unknown canopy structure. However, when classifying fewer species at a time, errors dropped considerably; for example, 20 species can be classified to 82-88% accuracy. These results highlight the potential of imaging spectroscopy to provide species discrimination in high-diversity, humid tropical forests.     

Dry season mapping of savanna forage quality, using the hyperspectral Carnegie Airborne Observatory sensor

Forage quality within an African savanna depends upon limiting nutrients (nitrogen and phosphorus) and nutrients that constrain the intake rates (non-digestible fibre) of herbivores. These forage quality nutrients are particularly crucial in the dry season when concentrations of limiting nutrients decline and non-digestible fibres increase. Using artificial neural networks we test the ability of a new imaging spectrometer (CAO Alpha sensor), both alone and in combination with ancillary data, to map quantities of grass forage nutrients in the early dry season within an African savanna. Respectively 65%, 57% and 41%, of the variance in fibre, phosphorus and nitrogen concentrations were explained. We found that all grass forage nutrients show response to fire and soil. Principal component analysis, not only reduced image dimensionality, but was a useful method for removing cross-track illumination effects in the CAO imagery. To further improve the mapping of forage nutrients in the dry season we suggest that spectra within the shortwave infrared (SWIR) region, or additional relevant ancillary data, are required.     

Equivalent water thickness in savanna ecosystems: MODIS estimates based on ground and EO-1 Hyperion data

The Brazilian savanna biome, known locally as the Cerrado, with an area of about 2 million km² and marked by a conspicuous seasonality, comprises a vertically structured mosaic of ecosystem types, ranging from grassland to tropical dry forests. The Cerrado is a major agricultural frontier in Brazil, with nearly 50% of its original vegetative cover already converted to pastures and crop fields. Such large-scale conversion has severely affected regional runoff, river discharge and the atmosphere water transfer from soil reservoirs through vegetation. In this study, we used multitemporal Earth Observing-1 (EO-1) Hyperion hyperspectral imagery to derive canopy water content (validated by ground truth measurements), whose estimates were regionally extrapolated, over the entire Cerrado biome, based on the normalized difference vegetation index (NDVI) Moderate Resolution Imaging Spectroradiometer (MODIS) product MOD13Q1. MODIS-based canopy-level equivalent water thickness (EWT_C) values were significantly distinct for each of the major anthropogenic and natural Cerrado land-cover types, at both the beginning and end of the dry season, and were correlated with land surface temperatures (LSTs). This method provides reasonable estimates of precipitable canopy water. Potential applications of EWT_C estimates based on moderate resolution imagery include early fire warnings and validation and constraining of regional hydrological models.     

Spectral unmixing of vegetation,soil and dry carbon cover in arid regions: Comparing multispectral and hyperspectral observations

Remote measurements of the fractional cover of photosynthetic vegetation (PV), non-photosynthetic vegetation (NPV) and bare soil are critical to understanding climate and land-use controls over the functional properties of arid and semi-arid ecosystems. Spectral mixture analysis is a method employed to estimate PV, NPV and bare soil extent from multispectral and hyperspectral imagery. To date, no studies have systematically compared multispectral and hyperspectral sampling schemes for quantifying PV, NPV and bare soil covers using spectral mixture models. We tested the accuracy and precision of spectral mixture analysis in arid shrubland and grassland sites of the Chihuahuan Desert, New Mexico, USA using the NASA Airborne Visible and Infrared Imaging Spectrometer (AVIRIS). A general, probabilistic spectral mixture model, Auto-MCU, was developed that allows for automated sub-pixel cover analysis using any number or combination of optical wavelength samples. The model was tested with five different hyperspectral sampling schemes available from the AVIRIS data as well as with data convolved to Landsat TM, Terra MODIS, and Terra ASTER optical channels. Full-range (0.4-2.5 w m) sampling strategies using the most common hyperspectral or multispectral channels consistently over-estimated bare soil extent and under-estimated PV cover in our shrubland and grassland sites. This was due to bright soil reflectance relative to PV reflectance in visible, near-IR, and shortwave-IR channels. However, by utilizing the shortwave-IR2 region (SWIR2; 2.0-2.3 w m) with a procedure that normalizes all reflectance values to 2.03 w m, the sub-pixel fractional covers of PV, NPV and bare soil constituents were accurately estimated. AVIRIS is one of the few sensors that can provide the spectral coverage and signal-to-noise ratio in the SWIR2 to carry out this particular analysis. ASTER, with its 5-channel SWIR2 sampling, provides some means for isolating bare soil fractional cover within image pixels, but additional studies are needed to verify the results.