Dar A. Roberts

Classification of multispectral images based on fractions of endmembers: Application to land-cover change in the Brazilian Amazon

Abstract Four time-sequential Landsat Thematic Mapper (TM) images of an area of Amazon forest, pasture, and second growth near Manaus, Brazil were classified according to dominant ground cover, using a new technique based on fractions of spectral endmembers. A simple four-endmember model consisting of reflectance spectra of green vegetation, nonphotosynthetic vegetation, soil, and shade was applied to all four images. Fractions of endmembers were used to define seven categories, each of which consisted of one or more classes of ground cover, where class names were based on field observations. Endmember fractions varied over time for many pixels, reflecting processes operating on the ground such as felling of forest, or regrowth of vegetation in previously cleared areas. Changes in classes over time were used to establish superclasses which grouped pixels having common histories. Sources of classification error were evaluated, including system noise, endmember variability, and low spectral contrast. Field work during each of the four years showed consistently high accuracy in per-image classification. Classification accuracy in any one year was improved by considering the multiyear context. Although the method was tested in the Amazon basin, the results suggest that endmember classification may be generally useful for comparing multispectral images in space and time.

Green vegetation, nonphotosynthetic vegetation, and soils in AVIRIS data

Abstract An Airborne Visible / Infrared Imaging Spectrometer (AVIRIS) image collected over the Jasper Ridge Biological Preserve, California on 20 September 1989 was analyzed using spectral mixture analysis. The scene was calibrated to reflectance assuming a homogeneous atmosphere. The image was modeled initially as linear mixtures of the minimum number of reference endmember spectra that accounted for the maximum spectral variability. Over 98% of the spectral variation was explained by linear mixtures of three endmembers: green vegetation, shade, and soil. Additional spectral variation appeared as residuals. Nonlinear mixing was expressed as variations in the fraction of each endmember when a linear mixing model was applied to spectral subsets of the entire spectrum. After the fractions of the endmember spectra were calculated for each pixel, different types of soil were discriminated by the residual spectra. Nonphotosynthetic vegetation (NPV) (e.g., dry grass, leaf litter, and woody material), which could not be distinguished from soil when included as an endmember, was discriminated by residual spectra that contained cellulose and lignin absorptions. Distinct communities of green vegetation were distinguished by 1) nonlinear mixing effects caused by transmission and scattering by green leaves, 2) variations in a derived canopy-shade spectrum, and 3) the fraction of NPV. The results of the image analysis, supported by field observations in 1990 and 1991, indicate that the multiple bands of AVIRIS enhance discrimination of NPV from soil, and the separation of different types of green vegetation. The ability of the system to measure narrow absorption bands is one important factor; however, also important is the variation in continuum spectra expressed by the endmembers, and characteristic nonlinear mixing effects associated with green leaves.

Functional patterns in an annual grassland during an AVIRIS overflight

Abstract This study relates Airborne Visible / Infrared Imaging Spectrometer (AVIRIS) imagery to ground measurements of vegetation distribution, physiology, and productivity at Stanford Universitys Jasper Ridge Biological Preserve. Primary efforts focused on a 9-ha region of annual grassland where we completed a detailed ground-based study in conjunction with a 15 May 1991 AVIRIS overflight. Spectral mixture analysis and the normalized difference vegetation index (NDVI) calculated from AVIRIS data were used to evaluate spatial patterns of vegetation type, productivity, and potential physiological activity. Concurrent ground sampling revealed a high degree of correlation between NDVI and estimates of canopy chemistry, structure, productivity, and CO 2 flux, supporting the use of imaging spectrometry to estimate spatial and temporal trends in vegetation physiology and productivity in this relatively simple grassland ecosystem. Geostatistical analyses of both ground and AVIRIS data supported the conclusion that the AVIRIS pixel size was suitable for describing the influence of major landscape features in this grassland and that spatial detail would be lost at slightly larger pixel sizes typical of other imaging spectrometers. Analysis of fine spectral features in AVIRIS data may provide new ways of assessing physiological activity in evergreen tree and shrub communities where photosynthetic activity was not correlated with green canopy display.

Estimation of aerosol optical depth, pressure elevation, water vapor, and calculation of apparent surface reflectance from radiance measured by the airborne visible/infrared imaging spectrometer (AVIRIS) using a radiative transfer code

The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) is an imaging spectrometer that measures spatial images of the total up welling spectral radiance from 400 to 2500 nm at 10 nm spectral intervals. Quantitative research and application objectives for surface investigations require conversion of the measured radiance to surface reflectance or surface leaving radiance. To calculate apparent surface reflectance an estimation of aerosol optical depth is required for compensation of aerosol scattering and absorption across the spectral range. Determination of other atmospheric characteristics such as atmospheric water vapor and surface pressure is also required. In this paper we describe a set of algorithms to estimate aerosol optical depth, atmospheric water vapor, and surface pressure height from the AVIRIS measured radiance. Based upon these determined atmospheric parameters we described an algorithm to calculated apparent surface reflectance from the AVIRIS measured radiance using a radiative transfer code.

Exploring the correlation between Southern Africa NDVI and Pacific sea surface temperatures: Results for the 1998 maize growing season

Several studies have identified statistically significant correlations between Pacific sea surface temperature anomalies and NDVI anomalies in Southern Africa. The potential predictive value of the relationship was explored for the 1998 maize growing season. Cross-validation techniques suggested a more useful relationship for regions of wet anomaly than for regions of dry anomaly. Observed 1998 NDVI anomaly patterns were consistent with this result. Wet anomalies were observed as expected, but wide areas of expected dry anomalies exhibited average or above-average greeness.

Predicted distribution of visible and near-infrared radiant flux above and below a transmittant leaf

Abstract The effects of background reflectance, leaf size, and leaf height above the background on upward and downward radiant flux ( φ u and φ d ) from a leaf were investigated using a computer model of a horizontal, isotropically scattering leaf. This research was conducted to determine how these variables influence the light environment above, below and adjacent to a leaf. Leaf spectral properties for big-leaf maple (Acer macrophyllum) were measured in the laboratory and used in the model. Model results were reported as relative radiant flux ( φ r ), defined as a percentage of the light entered into the model. The model showed that upward relative radiant flux φ ur from a leaf was highly dependent on the reflectance of the background and the wavelength of light. The greatest variation in φ ur was observed in the near infrared (NIR). The φ ur also varied depending upon the height of the leaf above the background and the size of the leaf. Leaves were brightest when placed the farthest distance above the background. Small leaves reached maximum brightness at lower heights than larger leaves. Finally, φ ur varied spatially. Leaf edges reflected more light than the leaf center except for leaves positioned very close to the background. Additional studies using the model showed that the intensity of light within a leaf shadow varied spatially, with the greatest downward relative radiant flux φ dr , occurring directly below the center of the leaf. Furthermore, φ dr within the shadow cast by the leaf decreased as the height of the leaf above the background increased. The rate of decrease depended upon the size of the leaf. The smaller the leaf, the greater was the change in φ dr with change in leaf height. These results imply that NIR canopy reflectance, due to leaf transmittance, may be highly dependent upon the reflectance of its background. Furthermore, architecturally different canopies may show different degrees of dependence upon background reflectance in the NIR. These results extend to closed canopies, in which leaf size and spacing may vary the reflectance of the canopy. Finally, these results suggest that the amount of light scattered to the side by leaves increases the amount of NIR light measured from adjacent, unshadowed backgrounds.