John A. Richards

Segmented principal components transformation for efficient hyperspectral remote-sensing image display and classification

A segmented, and possibly multistage, principal components transformation (PCT) is proposed for efficient hyperspectral remote-sensing image classification and display. The scheme requires, initially, partitioning the complete set of bands into several highly correlated subgroups. After separate transformation of each subgroup, the single-band separabilities are used as a guide to carry out feature selection. The selected features can then be transformed again to achieve a satisfactory data reduction ratio and generate the three most significant components for color display. The scheme reduces the computational load significantly for feature extraction, compared with the conventional PCT. A reduced number of features will also accelerate the maximum likelihood classification process significantly, and the process will not suffer the limitations encountered by trying to use the full set of hyperspectral data when training samples are limited. Encouraging results have been obtained in terms of classification accuracy, speed, and quality of color image display using two airborne visible/infrared imaging spectrometer (AVIRIS) data sets.

Probabilistic and Evidential Approaches for Multisource Data Analysis

Two methods for combining the information contents from multiple sources of remote-sensing image data and spatial data in general are described. One is a probabilistic scheme that employs a global membership function (similar to a joint posterior probability) that is derived from all available data sources. The other is an evidential calculus based upon Dempsters orthogonal sum combination rule. A feature of both methods is that uncertainty regarding data analysis can be incorporated into the process. Both schemes are evaluated in terms of their general applicability and certain equivalences are noted. Moreover, both are shown to perform well on mixed multispectral data.

Thematic mapping from multitemporal image data using the principal components transformation

Abstract The principal components transformation is used to highlight regions of localized change evident in satellite multispectral imagery associated with bushfire damage and with vegetation regrowth following fire burns. In line with previous studies by other investigators it is the higher order components that are seen to lead to change enhancement. These components are classified by unsupervised techniques to yield thematic maps on which change classes are recorded. In this manner, confusion of class signatures between dynamic and static cover types is avoided. In the present case this relates to confusion between fire burn regions and water edge mixed pixels.

L-Band Radar Backscatter Modeling of Forest Stands

An L-band HH radar backscatter model of a coniferous forest stand is described and compared with SIR-B L-band image data of the Mount Shasta region of northern California. Being based upon an identification and implementation of the expected major components of forest backscattering, the model is simple in form and thus fast computationally, making possible extensive simulations of forest stands. A particularly important component in the model relates to representing the specular reflections expected from tree trunks to the ground and then back to the sensor. These are strong returns and are seen to be necessary to explain both the forest measurements made by the authors and the observations of others. Although the experimental data is limited in quantity and quality, agreement between available experimental and simulated values of forest backscatter is better than the residual uncertainty and relative calibration error of the experimental data, provided the model and experiment are matched initially at one set of parameter values.

An explanation of enhanced radar backscattering from flooded forests

Abstract A simpie structural backscatter model for a forest stand, suitable for use with L-band HH polarized radar imagery, is used to explain the increased level of backscattering observed from flooded forests. Measurements made of relative levels of backscatter from SIR-B image data of a flooded Australian forest are consistent with an interpretation based upon scattering mechanisms involving both the tree components and the understorey or forest floor. The change in Fresnel power reflection coefficient of the ground with flooding is advanced as the cause of the enhancement in backscattered power levels.

Efficient maximum likelihood classification for imaging spectrometer data sets

A simplified maximum likelihood classification technique for handling remotely sensed image data is proposed which reduces, significantly, the processing time associated with traditional maximum likelihood classification when applied to imaging spectrometer data, and copes with the training of geographically small classes. Several wavelength subgroups are formed from the complete set of spectral bands in the data, based on properties of the global correlation among the bands. Discriminant values are computed for each subgroup separately and the sum of discriminants is used for pixel labeling. Several subgrouping methods are investigated and the results show that a compromise among classification accuracy, processing time, and available training pixels can be achieved by using appropriate subgroup sizes. >

The effect of changing environmental conditions on microwave signatures of forest ecosystems: preliminary results of the March 1988 Alaskan aircraft SAR experiment

Abstract In preparation for the first European Space Agency (ESA) Remote Sensing(ERS-I) mission,a series of multitemporal, multifrequency, multipolarization aircraft synthetic aperture radar (SAR) data sets were acquired over the Bonanza Creek Experimental Forest near Fairbanks, Alaska in March, 1988. P-, L- and C-band data were acquired with the NASA/JPL Airborne SAR on five differentdays over a period of two weeks. The airborne data were augmented with intensiveground calibration data as well as detailed, simultaneous in situ measurements of the geometric, dielectric and moisture properties of the snow and forest canopy. During the time period over which the SAR data were collected, the environmental conditions changed significantly; temperatures ranged from unseasonably warm (I to 9°C) to well below freezing (-8 to - 15°C), and the moisture content of the snow and trees changed from a liquid to a frozenstate. The SAR data clearly indicate the radar return is sensitive to these changing environmental fa...

Analysis of remotely sensed data: the formative decades and the future

Developments in the field of image understanding in remote sensing over the past four decades are reviewed, with an emphasis, initially, on the contributions of David Landgrebe and his colleagues at the Laboratory for Applications of Remote Sensing, Purdue University. The differences in approach required for multispectral, hyperspectral and radar image data are emphasised, culminating with a commentary on methods commonly adopted for multisource image analysis. The treatment concludes by examining the requirements of an operational multisource thematic mapping process, in which it is suggested that the most practical approach is to analyze each data type separately, by techniques optimized to that datas characteristics, and then to fuse at the label level.

Remote Sensing with Imaging Radar

This book treats the technology of radar imaging for remote sensing applications in a manner suited to the mathematical background of most earth scientists. It assumes no prior knowledge of radar on the part of the reader; instead it commences with a development of the essential concepts of radar before progressing through to a detailed coverage of contemporary ideas such as polarimetry and interferometry. Because the technology of radar imaging is potentially complex the first chapter provides a framework against which the rest of the book is set. Together, the first four chapters present the technical foundations for remote sensing with imaging radar. Scattering concepts are then covered so that the reader develops the knowledge necessary for interpreting radar data, itself the topic of a later chapter which draws together the current thinking in the analysis of radar imagery. The treatment is based on the assumption that the radars of interest are, in general, multi-polarised. Polarisation synthesis and polarised interferometric SAR are among the topics covered, as are tomography and the various forms of interferometry. A full chapter is given to bistatic radar, which is now emerging as an imaging technology with enormous potential and flexibility in remote sensing. The book concludes with a summary of passive microwave imaging. A set of appendices is included that provide supplementary material, among which is an overview of the rather complicated process of image formation with synthetic aperture radar, and summaries of some of the mathematical procedures important for a full appreciation of radar as a remote sensing technology.

Knowledge-based techniques for multi-source classification

Abstract The value of utilizing multiple data sources for classifying images has long been recognized in remote sensing. However, any attempts to do so have faced enormous problems primarily due to the inadequacy of traditional single source analytical techniques. This paper demonstrates the feasability of using knowledge-based procedures to provide a new scheme for incorporating several sources in the classification process. The two schemes presented (based on numerical and qualitative reasoning) are computationally efficient and have high classification accuracies.