James R. Carr

Dust devil vortices seen by the Mars Pathfinder Camera

Discovery of dust devil vortices in Mars Pathfinder (MPF) images reveals a dust entrainment mechanism at work on Mars. Scattering of visible light by dust in the Martian atmosphere creates a pronounced haze, preventing conventional image processing from displaying dust plumes. Spectral differencing techniques have enhanced five localized dust plumes from the general haze in images acquired near midday, which we determine to be dust devils. Processing of 440 nm images highlights dust devils as distinct occultation features against the horizon. The dust devils are interpreted to be 14–79 m wide, 46–350 m tall, travel at 0.5–4.6 m/s, with dust loading of 7E-5 kg m-3, relative to the general haze of 9E-8 kg m-3, and total particulate transport of 2.2–700 kg. The vortices match predictions from terrestrial analog studies.

The semivariogram in comparison to the co-occurrence matrix for classification of image texture

Semivariogram functions are compared to co-occurrence matrices for classification of digital image texture, and accuracy is assessed using test sites. Images acquired over the following six different spectral bands are used: 1) SPOT HRV, near infrared; 2) Landsat thematic mapper (TM), visible red; 3) India Remote Sensing (IRS) LISS-II, visible green; 4) Magellan, Venus, S-band microwave; 5) shuttle imaging radar (SIR)-C, X-band microwave; 6) SIR-C, L-band microwave. The semivariogram textural measure provides a larger classification accuracy than a classifier based on a co-occurrence matrix for the microwave images and a smaller classification accuracy for the optical images.

Spectral and textural classification of single and multiple band digital images

Abstract Single and multiple band images are classified using supervised algorithms. Two programs, MXTEXT and MXMULT, are presented that use minimum-distance-to-mean or Bayesian, maximum likelihood algorithms for spectral classification (pattern recognition), further allowing classification of image texture based on the local variogram surrounding each image pixel. Classification of texture can be performed independently of the classification of spectral information, or a combined spectral/textural classification can be performed. Combining the classification of texture with that of spectral information is shown to be of particular value for single band radar images. Improved classification accuracy is demonstrated also for multiple band images when variograms and cross-variograms are used to classify texture and spectral information.

On the practice of estimating fractal dimension

Coastlines epitomize deterministic fractals and fractal (Hausdorff-Besicovitch) dimensions; a divider [compass] method can be used to calculate fractal dimensions for these features. Noise models are used to develop another notion of fractals, a stochastic one. Spectral and variogram methods are used to estimate fractal dimensions for stochastic fractals. When estimating “fractal dimension,” the objective of the analysis must be consistent with the method chosen for fractal dimension calculation. Spectal and variogram methods yield fractal dimensions which indicate the similarity of the feature under study to noise (e.g., Brownian noise). A divider measurement method yields a fractal dimension which is a measure of complexity of shape.

Application of the semivariogram textural classifier (STC) for vegetation discrimination using SIR-B data of Borneo

Abstract Classification of Shuttle Imaging Radar-B (SIR-B) data from a rainforest-covered portion of Borneo is performed using image texture. The algorithm used is the semivariogram textural classifier (STC). This is a deterministic, supervised parallelepiped type classifier which provides the option of combining textural and radiometric information. Textural information is expressed by the semivariogram function. Radiometric information is conveyed by the mean digital number (DN) value. Results of the classificaiion cmulale a previously published map obtained by visual interpretation of the same SIR-B data set.

Cokriging—a computer program

Abstract Cokriging is a process wherein several variables can be jointly estimated on the basis of intervariable and spatial structure information. Presented herein is the program COKRIG, for punctual cokriging, a program in a simple form to demonstrate the utility of cokriging. Equation solution follows a modification of a method developed for the solution of large-scale linear systems. Several example problems show that, at least for earthquake data, the inclusion of intervariable information results in a more accurate BLUE (best linear unbiased estimator).

Analysis of JERS-1 (Fuyo-1) SAR data for vegetation discrimination in northwestern Brazil using the semivariogram textural classifier (STC)

Abstract Classification of JERS-1 (Fuyo-1) SAR data from the northwestern portion of Brazil was performed using the semivariogram textural classifier (STC). This is a deterministic, supervised classifier which provides the option of combining textural and radiometric information. Textural information is expressed by the semivariogram function; radiometric information is conveyed by the mean digital number (DN) value. Results have shown that STC allows vegetation units and water bodies to be discriminated and tentatively mapped, suggesting that this is a promising approach for environmental monitoring of rainforest regions using SAR data.

COSIM: a FORTRAN IV program for coconditional simulation

Abstract Recently, a computer algorithm was presented for joint estimation of random functions; this cokriging technique demonstrated the utility and increased accuracy obtained through best linear unbiased estimation based on auto- and cross-correlation. A worthwhile extension of cokriging is coconditional simulation, a technique whereby several nonconditionally simulated random functions are conditioned using cokriging. Although coconditional simulation can be performed, one random function at a time, using kriging, this can result in an incorrect portrayal of the cross-correlation between the simulated random functions. This is particularly true if one, or several variables is sampled sparsely. Coconditional simulation based on cokriging correctly reproduces variable cross-correlation independent of variable sampling density.

Statistical self-affinity, fractal dimension, and geologic interpretation

Properties of statistical self-affinity are explored and explained. Semi-variogram analysis can be used for identifying statistical self-affine behavior. This is, however, not the only method available for such an analysis. Some error and interpretation is involved; therefore, estimating the Hurst dimension (and from this the fractal dimension) from the semi-variogram can be misleading. Simulations based on statistical self-affine properties are alternatively used to develop an empirical approach to assessing statistical self-affine behavior. Analyzing simulations using semi-variograms, and comparing these semi-variograms to those from actual data, offers an alternate and perhaps superior approach to the understanding of the statistical self-affine properties of a geologic phenomenon. This empirical approach offers a method of reverse modeling for verifying estimates of Hurst dimension from semi-variograms.

Use of indicator variograms for an enhanced spatial analysis

Flat variograms often are interpreted as representing a lack of spatial autocorrelation. Recent research in earthquake engineering shows that nearby field noise can substantially mask a prominent spatial autocorrelation and result in what appears to be a purely random spatial process. A careful selection of threshold in assigning an indicator function can yield an indicator variogram which reveals underlying spatial autocorrelation. Although this application involves use of seismic data, the results are relevant to geostatistical applications in general.