Alan R. Gillespie

A temperature and emissivity separation algorithm for Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) images

The Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) scanner on NASAs Earth Observing System (EOS)-AM1 satellite (launch scheduled for 1998) will collect five bands of thermal infrared (TIR) data with a noise equivalent temperature difference (NE/spl Delta/T) of /spl les/0.3 K to estimate surface temperatures and emissivity spectra, especially over land, where emissivities are not known in advance. Temperature/emissivity separation (TES) is difficult because there are five measurements but six unknowns. Various approaches have been used to constrain the extra degree of freedom. ASTERs TES algorithm hybridizes three established algorithms, first estimating the normalized emissivities and then calculating emissivity band ratios. An empirical relationship predicts the minimum emissivity from the spectral contrast of the ratioed values, permitting recovery of the emissivity spectrum. TES uses an iterative approach to remove reflected sky irradiance. Based on numerical simulation, TES should be able to recover temperatures within about /spl plusmn/1.5 K and emissivities within about /spl plusmn/0.015. Validation using airborne simulator images taken over playas and ponds in central Nevada demonstrates that, with proper atmospheric compensation, it is possible to meet the theoretical expectations. The main sources of uncertainty in the output temperature and emissivity images are the empirical relationship between emissivity values and spectral contrast, compensation for reflected sky irradiance, and ASTERs precision, calibration, and atmospheric compensation.

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.

Vegetation in deserts: I. A regional measure of abundance from multispectral images

Abstract A method was tested in the semiarid Owens Valley, California for measuring sparse vegetation cover using Landsat Thematic Mapper (TM) multispectral images. Although green vegetation has a characteristic reflectance spectrum in the visible and near-infrared, using conventional image-processing methods, it has been difficult to quantify vegetation cover of less than about 40%, owing to the spectral dominance of the background soils and rocks. Thus multispectral images have been of limited use in mapping variations in vegetation cover in arid and semiarid regions. In this study fractions of vegetation, soils, and shading and shadow within the smallest resolution elements (30 × 30 m pixels) of the TM images were computed by applying a mixing model based on laboratory and field reference spectra. Fractions of vegetation were calculated for each pixel in TM images taken in December 1982 and May 1985, and the results were compared with ground transects. Despite spatial variations in background soil, temporal differences in satellite instrument response, and differences in atmospheric and lighting conditions, the fractions of vegetation computed from each image gave a spatially consistent measure of the projected vegetation cover. Results were obtained for a 150-km segment of Owens Valley; they indicate that the method can facilitate mapping and monitoring sparse vegetation cover over large regions covered by satellite images.

Color enhancement of highly correlated images. I. Decorrelation and HSI contrast stretches

Abstract Conventional enhancements for the color display of multispectral images are based on independent contrast modifications or “stretches” of three input images. This approach is not effective if the image channels are highly correlated or if the image histograms are strongly bimodal or more complex. Any of several procedures that tend to “stretch” color saturation while leaving hue unchanged may better utilize the full range of colors for the display of image information. Two conceptually different enhancements are discussed: the “decorrelation stretch”, based on principal-component (PC) analysis, and the “stretch” of “hue”-“saturation”-intensity (HSI) transformed data. The PC transformation is scene-dependent, but the HSI transformation is invariant. Examples of images enhanced by conventional linear stretches, decorrelation stretch, and by stretches of HSI transformed data are compared. Schematic variation diagrams or two- and three-dimensional histograms are used to illustrate the “decorrelation stretch” method and the effect of the different enhancements.

Red River and associated faults, Yunnan Province, China: Quaternary geology, slip rates, and seismic hazard

The 900-km-long right-slip Red River fault of southernmost China and northern Vietnam is a profound structural discontinuity that is mechanically associated with the collision of the Indian and Eurasian plates. Although history records no large earthquakes resulting from slippage along at least the principal segment of the fault in China, youthful landforms and disruptions of young sedimentary rocks indicate that it has generated large earthquakes during the Pleistocene and Holocene epochs. The historic quiescence thus must be regarded as being indicative of a current seismic gap, although the recurrence interval between major earthquakes is evidently much longer than for many other major active fault systems. That recent displacement has been primarily right lateral is indicated by consistently displaced drainages, ranging in offset from 9 m to 6 km, and the freshness of the smallest and most recent offsets implies repeated Holocene movements. Although physiographic features typical of active faulting such as scarps and drainage diversions are present throughout, the general absence of sag ponds reflects both the high rate of dissection of the fault by the Red River and its tributaries and the lower degree of activity as compared to highly active faults such as the San Andreas fault of California. In its middle 170 km, the fault zone is made up of two branches. The range-front branch demarcates the northeastern base of the Ailao Mountains and, at least locally, has an appreciable component of dip slip. The mid-valley branch, in large part previously unrecognized, traverses principally deeply dissected Cenozoic valley fill northeast of the range-front fault and has undergone almost pure lateral slip. Lateral postfill offsets along the range-front branch diminish toward the southeast, whereas those along the mid-valley branch diminish northwestward; the net effect is that the total postfill offset across both branches is almost uniform. The Red River and its major tributaries appear to have experienced about 5.5 km of right slip since the beginning of a major episode of incision that continues to the present day. Restoration of this offset provides a remarkable alignment of most large tributaries as well as removing a major kink in the course of the Red River itself. Using maximum credible rates of incision, we estimate an average fault-slip rate of 2 to perhaps 5 mm/yr. At this long-term rate of slip, the smallest offsets observed along the fault (9 m) would occur no more frequently than every 1,800 to 4,500 yr on the average. This is consistent with the historical record of fault dormancy for the past 300 yr. North of the Red River fault, there is a large seismically active region laced with numerous faults of north and northwesterly trends. Several of these faults display clear and even spectacular evidence of youthful normal faulting, and some appear to have left-lateral components as well. These faults, as well as the Red River fault itself, are accommodating regional east-west crustal extension and north-south shortening.

Color enhancement of highly correlated images. II - Channel ratio and 'chromaticity' transformation techniques

Abstract Conventional enhancements for the color display of multispectral images are usually based on independent contrast modifications (“stretches”) of three image channels. This approach generally does not produce colorful pictures if the image channels are strongly correlated. Procedures that selectively emphasize the weakly correlated component of the image data may better utilize the full range of colors. In Part I of this series of articles, two such methods were discussed that exaggerated color saturation without greatly modifying hue. These methods utilized principal-component analysis and HSI transformation of the image data. Herein we discuss two other methods, based on ratioing of data from different image channels. In the first approach, images of the ratioed data are “stretched” and assigned primary colors for display as color pictures. An alternative ratio method uses the difference, normalized to the intensity, instead of the ratio itself. Hues in pictures thus enhanced bear no resemblance to those in the unenhanced picture: hence, interpretation may be difficult. In the second approach, the intensity for each of three image channels is ratioed, pixel by pixel, to the sum of the intensities. Thus the data are transformed to image-domain “chromaticity” coordinates. These data are next “stretched” and then multiplied by the summed intensities to return them to the original coordinates for display. This technique is similar to those of Part I in that the hues need not be greatly altered.

Topographic normalization of Landsat TM images of forest based on subpixel sun-canopy-sensor geometry

Abstract Because trees are geotropic (perpendicular to the geoid), topography has no control over the Sun–crown geometry. What topography does influence is the relative positioning of trees and thus the amount of shadowing cast by them within the canopy. As satellite sensors in general measure the collective radiance of many trees inside their instantaneous field of view, the overall canopy brightness at the pixel scale is strongly controlled by canopy shadowing and hence by the topography. The removal of or compensation for topographic effects on forest images should be based on the normalization of mutual shadowing at the subpixel scale, rather than on the normalization of Sun–terrain–sensor geometry at the pixel scale. The Sun–canopy–sensor (SCS) topographic correction model was developed to characterize and hence correct the topographic effects on forest images. Testing with simulated image data showed the SCS model to be accurate (root-mean-squared residual error

Spectral mixture analysis of multispectral thermal infrared images

Remote spectral measurements of light reflected or emitted from terrestrial scenes is commonly integrated over areas sufficiently large that the surface comprises more than one component. Techniques have been developed to analyze multispectral or imaging spectrometer data in terms of a wide range of mixtures of a limited number of components. Spectral mixture analysis has been used primarily for visible and near-infrared images, but it may also be applied to thermal infrared data. Two approaches are reviewed: binary mixing and a more general treatment for isothermal mixtures of a greater number of components.

Timing and significance of Late-glacial and Holocene cirque glaciation in the Sierra Nevada, California

Mapping and radiocarbon dates of cirque moraines in the Sierra Nevada demonstrate that the last significant pre-Little Ice Age glacier advance in the range, the Recess Peak, resulted from snowline lowering roughly twice that of the Matthes (Little Ice Age) advance, and that the Recess Peak advance is late Pleistocene in age. We mapped Recess Peak and Matthes deposits in 64 cirques along a profile of the main Sierran crest that spans the north-south limits of ‘Neoglacial’ deposits in the range. Equilibrium-line altitudes for the reconstructed Recess Peak glaciers vary greatly but coherently with those of the Matthes advance. The variability of both sets of deposits reflects strong topographic influences on snow accumulation and ablation patterns in their deep cirques. Tephrochronology and radiocarbon dates from lake-sediment cores provide limits on the timing of the two advances. Previous work documenting the absence of a young, regionally extensive tephra on Matthes moraines in the central Sierra demonstrates that they formed after ∼700 14C years BP (∼650 cal. years BP). The age of the Recess Peak advance has been less certain; we therefore collected and dated sediment cores from lakes dammed behind terminal moraines correlating to the Recess Peak advance in four widely separated drainages along the Sierran crest (north to south): South Fork American River, Lee Vining Creek, Middle Fork San Joaquin River, and Bishop Creek. Twenty-three high-precision AMS radiocarbon dates on gyttja, peat, and macrofossils from the cores are internally consistent and demonstrate that the Recess Peak advance, previously thought to be of late Holocene age (∼2500 years BP), ended before 11,190±70 14C years BP (∼13,100±85 cal. years BP). Recess Peak is therefore late Pleistocene in age and probably predates the North Atlantic Younger Dryas climatic reversal. The absence of any glacial deposits on the bedrock between the Recess Peak and Matthes deposits indicates that: (1) any advance related to the Younger Dryas event in central California was smaller than the Matthes advance; (2) the Matthes advance was the most extensive, and possibly the only, Neoglacial event in the range; and (3) climate in the Sierra between ∼13,000 cal. years BP and 650 cal. years BP was apparently too warm and/or dry to support glaciers larger than those of the Little Ice Age. Other mapping indicates that the Recess Peak is the first significant glacier advance after retreat of Tioga (local late-Wisconsin maximum) glaciers. These results suggest a regionally variable climate in western North America during the Younger Dryas event, because glaciers appear to have expanded in the Canadian Rockies at that time. The new Recess Peak age limits, combined with other dated lake cores, indicate that the Sierra was essentially deglaciated by 14,000–15,000 cal. years BP (∼12,000–13,000 14C years BP), substantially earlier than previously estimated. This finding indicates that current production rates of some in situ cosmogenic nuclides, calibrated on an assumed deglaciation of the range at 11,000 cal. years BP (∼10,000 14C years BP), may be systematically too high by as much as 20%.

Simple Models For Complex Natural Surfaces: A Strategy For The Hyperspectral Era Of Remote Sensing

A two-step strategy for analyzing multispectral images is described. In the first step, the analyst decomposes the signal from each pixel (as expressed by the radiance or reflectance values in each channel) into components that are contributed by spectrally distinct materials on the ground, and those that are due to atmospheric effects, instrumental effects, and other factors, such as illumination. In the second step, the isolated signals from the materials on the ground are selectively edited, and recombined to form various unit maps that are interpretable within the framework of field units. The approach has been tested on multispectral images of a variety of natural land surfaces ranging from hyperarid deserts to tropical rain forests. Data were analyzed from Landsat MSS (multispectral scanner) and TM (Thematic Mapper), the airborne NS001 TM simulator, Viking Lander and Orbiter, AIS, and AVRIS (Airborne Visible and Infrared Imaging Spectrometer).