Ronald J. P. Lyon

Influence of rock-soil spectral variation on the assessment of green biomass

Abstract Vegetation indices can be adversely influenced by variation in rock and soil spectral characteristics. When rocks and soils yield different vegetation index values, this is misinterpreted as changes in green biomass. This spectral influence is present to some extent in all vegetation indices. Secondly, variations in rock-soil brightness have a strong influence on ratio-based vegetation indices. Variations in rock-soil brightness have the same effect on all ratio-based vegetation indices: Vegetation is overestimated on dark backgrounds relative to bright backgrounds. Overall, it is concluded that the perpendicular vegetation index is the best available vegetation index to use in multispectral imagery of arid and semiarid regions where there is wide variation in rock and soil spectral characteristics.

Principal components analysis of multitemporal image pairs

Abstract Principal components analysis of eight-channel data sets consisting of multitemporal LANDSAT MSS image pairs often generates higher-order principal components that are related to changes in ‘brightness’ and ‘greenness’. This is the expected result of such analysis in a wide variety of biological and geological environments where the original imagery is intrinsically two dimensional; the two dimensions are ‘brightness’ and ‘greenness’; and the change in land cover between images exceeds some threshold value.

Estimation of the vegetation contribution to the 1·65/2·22 μm ratio in airborne thematic-mapper imagery of the Virginia Range, Nevada

Abstract Previous investigations have successfully used the 1·65/2·22μm ratio to map the concentration of hydroxyl-bearing minerals in imagery of sparsely vegetated terrain. However, difficulties arise in applying this ratio to semiarid regions where there is a wide variation in vegetation density. Besides mapping the mineralogic features of interest, the 1·65/2·22μm ratio also maps the density of vegetation by responding to leaf water content. A technique has been developed to estimate the vegetation contribution to this ratio based on the near-IR/red ratio. The technique has proved useful in mapping supergene alteration in the Virginia Range, Nevada

Quantitative relationships of near-surface spectra to Landsat radiometric data

Abstract The aim of our research has been to determine the quantitative relationship between the surface spectral character of a variety of geologic terrains and that sensed by the Landsat multispectral scanner. A spectral sampling and measurement program was conducted to accurately characterize the surface spectral reflectance of the Landsat resolution element and, for the first time, to establish statistically the degree of sampling required for a variety of natural terrains. Results from the study showed that for typical homogeneous and moderately heterogeneous terrains, the number of samples required to estimate the mean reflectance of a pixel is small. Only 9–20 samples are required to be within 2% reflectance at the 95% probability level. Coincident field measurements and satellite observations were used to test the equivalency and correlation of the reflectance data. Before the Landsat data could be compared with the surface measurements the satellite brightness values must be converted to absolute radiometric units, and corrected for atmospheric attenuation and scattering. A conversion method using a standard/target comparison, which indirectly compensated for atmospheric attenuation and scattering, produced a Landsat equivalent reflectance that exhibited a root-mean-square error of ± 4% reflectance, when compared with the surface measured value at 12 test sites. Although the equivalence of the surface and satellite data cannot be shown to be better than 4% reflectance, statistical study indicates that the surface and satellite data are highly correlated within defined contrast constraints. However, this correlation is present only after the satellite brightness values are corrected for between band gain differences and compensation is made for atmospheric attenuation and scattering.

A prior probability method for smoothing discriminant analysis classification maps

A statistical method is presented for smoothing discriminant analysis classification maps by including pixel-specific prior probability estimates that have been determined from the frequency of tentative class assignments in a window moving across an initial per-point classification map. The class at the center of the window is reevaluated using the data for that location and the prior probability estimates obtained from the window area. An example using Landsat spectral data demonstrates the effectiveness of the method and shows an increase in classification accuracy after smoothing.

Detection of naturally heavy-metal-poisoned areas by LANDSAT-1 digital data

Abstract Natural poisoning of soil and vegetation due to heavy metals originating from sulphide deposits in the bedrock has been demonstrated at several locations in Norway. At some of these, the features of poisoning are rather widespread, suggesting that the recognition of such naturally poisoned areas by means of satellite imagery and satellite digital tapes might be possible. An occurrence of natural Cu poisoning near Karasjok, northern Norway, was selected as a training area. LANDSAT-1 computer compatible tapes were used to prepare enhanced colour ratio images and to calculate ratios of calibrated channel reflectances. Ground reflectance measurements were also carried out using radiometers with band passes equivalent to those on the LANDSAT multispectral scanner system. Compared with the surroundings the LANDSAT data over the training area showed unique spectral characteristics. These are also indicated on the enhanced ratio images, and the digital data can be utilized in supervised discriminant analysis and unsupervised “cluster” analysis. In particular, the area of the natural Cu poisoning shows low values in the band 7/band 5 reflectance ratio. At present, a detailed, precise numerical correlation between the ground and satellite reflectance data is not possible due to differences in resolution, although spatial correlation of spectral trends can be established. It appears that features of natural heavy metal poisoning can be located with the 0.45-hectare resolution of the LANDSAT multispectral scanner system. The search for such features might therefore be used as a tool in mineral exploration. The current resolution limitation can be overcome by utilizing scanners at aircraft altitudes.

Simultaneous use of geological, geophysical, and LANDSAT digital data in uranium exploration☆

In southern Libya, in an 850-km2 area of uranium-bearing crystalline rocks, which had been mapped geologically in 1974 and over which concurrent low-altitude gamma-ray spectrometry (total count) and magnetics were also flown, a fourteen man-month pilot study was made using multivariate statistical methods of merged datasets from quantified geological, geophysical, and LANDSAT data. Key to the analysis were, 1. 1. The use of a new grid by which to resample the datasets, oriented with the specific azimuth of a LANDSAT-2 pass, with spacings defined by the average pixel sizes of that coverage. 2. 2. Scaled, 1:1 aspect-ratio computer printouts, which were used as the base maps for the grid and which were compared to semi-controlled photomosaics, geological, and geophysical maps, all at 1:50,000 scale. 3. 3. A “defocused” LANDSAT cell, of almost 480×458 m, achieved by averaging 8 pixels (West—East) by 6 pixels (North—South). In being still clamped to the geometry of LANDSAT, this facilitated the extraction of the radiance datasets from the satellite coverage. 4. 4. Thirty-one parameters were extracted from the 1:50,000 scale geological maps, and two geophysical parameters obtained by resampling the contoured geophysical maps. 5. 5. Appreciating that LANDSAT data are also quantified geophysical matricies and that from them 6 nonredundant ratios could be obtained. A total of ten LANDSAT variables were thereby extracted. 6. 6. Multivariate analysis of 43 variables for the 3364 cells was performed. Transformations of the variables to approximate the normal distribution were necessary to effectively apply the discriminant function analysis. The results included contoured probability maps of uranium prediction, with analysis of the role and the relative contribution of each of the 43 parameters. By far, the key parameters for the geological set were contact relationships (lengths/cell) and for LANDSAT were the average brightness of channel 6 and ratio 7/4. Adding the LANDSAT data significantly increased the definition and spatial coherence of the anomalies, retaining some of the key contact lengths but adding more information from the geological maps (rock percentages/cell, etc.) with which LANDSAT (here) shows high agreement. The extraction of the multivariate signatures (geological and geophysical and LANDSAT for this study area, would allow extension to adjoining areas, as the geological framework of this test site extends an appreciable (500 km) distance to the east and to the west, as well as south into the N. Chad uranium belts. Automation of some of the data extraction and data-handling steps has been completed recently. Considerable further development of the methodology is underway, particularly with respect to the statistical treatment of the data, as well as the continued parametric analysis of this excellent data set, especially the use of linear regression against the radiometric data.

A relation between landsat digital numbers, surface reflectance, and the cosine of the solar zenith angle

Abstract A method for estimating the reflectance of ground sites from satellite radiance data is proposed and tested. The method uses the known ground reflectance from several sites and satellite data gathered over a wide range of solar zenith angles. The method was tested on each of 10 different Landsat images using 10 small sites in the Walker Lake, Nevada area. Plots of raw Landsat digital numbers (DNs) versus the cosine of the solar zenith angle (cos Z) for the the test areas are linear, and the average correlation coefficients of the data for Landsat bands 4, 5, 6, and 7 are 0.94, 0.93, 0.94, and 0.94, respectively. Ground reflectance values for the 10 sites are proportional to the slope of the DN versus cos Z relation at each site. The slope of the DN versus cos Z relation for seven additional sites in Nevada and California were used to estimate the ground reflectances of those sites. The estimates for nearby sites are in error by an average of 1.2% and more distant sites are in error by 5.1%. The method can successfully estimate the reflectance of sites outside the original scene, but extrapolation of the reflectance estimation equations to other areas may violate assumptions of atmospheric homogeneity.

The multiband approach to geological mapping from orbiting satellites - Is it redundant or vital

Abstract Before we plan to use the multiband data returned from the twoERTS scanners (MSS and RBV) it is in order to assess their possible utility to geologists . Are we hoping for “where is it” or “what is it” images? The author feels strongly that only the “where is it” question can be answered with todays very limited knowledge of field spectra from rocks and soils. The spectral quality of data from these two scanners will represent a totally new view, and there is no field experience with comparable equipment for training. Airborne use of parallel instruments with optical characteristics (filters, detectors, etc.) designed as closely as possible to those in ERTS series is vital and can help to bridge the gap. Without this we will be merely using very costly redundancy for a job any one band of imagery could do.

Mid-Infrared Remote Sensing Systems and Their Application to Lithologic Mapping

Mid-infrared remote lithologic mapping by emittance and by reflectance are assessed in laboratory experiments. The emittance spectra of various rocks and minerals, measured in the 8-13, ¿m atmospheric transmission window, are compared with reflectance data measured in the range of 9.2-11.2 Am using a line-tuned CO2 laser. We conclude that the reflectance data are more useful for lithologic discrimination than the passive emittance data. An experimental laser suitable for terrain mapping from a low-flying aircraft is described. The low-pressure longitudinal discharge CO2 laser has a rotating mirror to scan the diffraction grating and generates 90 bursts of pulses per second. Each 1-ms burst contains 92 pulses at 92 CO2 laser wavelengths. The mean output power is 12 W and the average pulse power is 370 W. With that power, and using incoherent detection, a signal-to-noise ratio of better than 100: 1 should be obtained from terrain with an albedo of 0.01 at a height of 500 m.