ALBERT

Description: Spectral remote sensing has been practiced on a large scale since the launch of Landsat 1 in 1972. The limited information contained in this spectrally undersampled data set has led to the development of sophisticated statistical-inferential methods for data analysis. The results are usually limited by the availability of ground truth information. Recent technological developments have made it feasible to create narrow-band, contiguous, spectral image data sets that make possible the identification of surface cover materials based on the complete reflectance spectrum for each picture element. This capability will revolutionize the use of remote sensing data and require new deterministic image processing techniques to extract the full information content from the data. Sensors, based on the concept of imaging spectrometry and the new technology of area array infrared detectors, have been constructed and are candidates for Shuttle and space platform flights.

Description: In the present assessment of the contributions of optical earth resources remote sensing in the 0.4-15.0 micron region, attention is given to underlying principles, applications to scientific disciplines such as geology, hydrology and oceanography, the recent development history of the requisite sensors, and sensor development trends. Development status characterizations are given for thematic mapping, modular optoelectronic multispectral scanning, the telescope/CCD 'SPOT' program of France, the thermal IR multispectral scanner for mineral signature identification, airborne imaging spectrometry, and the Advanced Visible and IR Imaging Spectrometer that is nearing deployment. Technology development trends and the capabilities they portend are projected.

Description: The Shuttle Multispectral Infrared Radiometer (SMIRR) is a spectroradiometer covering the region from 0.5 to 2.5 microns in 10 channels that acquired data from spots 100 m in diameter along the subspacecraft ground track. It was flown aboard the second flight of the Space Shuttle Columbia, November 12-14, 1981. Data collected during orbit 16 over southern Egypt show that carbonate rocks, kaolinite, and possibly montmorillonite can be identified by their SMIRR spectral signatures and limited knowledge of the lithologic units present. Detailed analysis of SMIRR data for this area indicates that calcite, kaolinite, and montmorillonite rocks give rise to absorption features that result in characteristic 10 channel spectra.

Description: A method is developed for relating scene digital counts among several images of the same scene by identifying radiometric control sets with mean reflectances that are basically constant. The average digital-count values of the control sets are utilized to compute linear transforms that relate digital count values between images. Two Landsat TM images are studied by means of the technique using simulations of a wide range of atmospheric conditions. In the visible and near-IR bands the algorithm effectively adjusts the surface reflectance for the effects of relative atmospheric differences to within 1 percent. The proposed method is found to be an effective relative correction procedure that can be used when atmospheric optical-depth data and calibration coefficients are not available.

Description: Initial results of the novel remote earth sensing technique of imaging spectrometry, which is technically feasible from both spacecraft and aircraft platforms, indicate that the direct identification of surface materials on a picture-element basis is possible through proper sampling of absorption features in the reflectance spectrum. Sensors of this type are able to acquire images simultaneously in 100-200 contiguous spectral bands. Computerized data reduction and storage techniques are available for the large data sets thus generated, and novel analytic techniques are under development to maximize information content extraction.

Description: Imaging spectrometry for the remote sensing of the Earth from aircraft and satellites is discussed. Results with an aircraft instrument show that remote, direct identification of surface materials is possible. The airborne and spaceborne sensors can acquire images in 100 to 200 spectral bands. The next generation aircraft scanner (AVIRIS) is expected to be operational in 1987. Plans are underway for a Shuttle instrument (SISEX) to fly in 1991.

Description: An overview of the Seasat land experiments is presented. The potential roles for active microwave imaging systems on board satellites were reviewed with particular emphasis on the Seasat Synthetic Aperture Radar (SAR). Recommendations were made concerning the type of experiments that could most profitably be conducted over land with the Seasat SAR system capabilities available.

Description: Many algorithms for spectral analysis of imaging spectroscopy data of the Earth's surface require that the data be calibrated to surface reflectance. Calibration requires removing instrumental response, solar irradiance, atmospheric transmittance, and atmospheric scattering from the radiance detected at the sensor. Depending on the amount of support data, this can be a formidable task. This paper examines four methods of calibration: (1) a radiative transfer model from the University of Colorado (ATREM: Gao and Goetz, 1990; Gao et al., 1992), (2) a MODTRAN-based method developed at the Jet Propulsion Lab by Green et al., (1191), (3) a ground calibration using known sites as standards, and (4) a combined approach using radiative transfer methods and ground calibration. Data from the Airborne Visual and Infra-Red Imaging Spectrometer (AVIRIS) instrument were evaluated from data sets obtained over multiple years and multiple sites.

Description: Accurate wavelength calibration of imaging spectrometer data is essential if proper atmospheric transmission corrections are to be made to obtain apparent surface reflectance. Accuracies of 0.1 nm are necessary for a 10 nm-sampling instrument in order to match the slopes of the deep atmospheric water vapor features that dominate the 0.7-2.3 micrometer regions. The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) is calibrated in the laboratory to determine the wavelength position and full-width-half-maximum (FWHM) response for each of the 224 channels. The accuracies are limited by the quality of the monochromator used as a source. The accuracies vary from plus or minus to plus or minus 1.5 nm depending on the wavelength region, in general decreasing with increasing wavelength. Green et al. make corrections to the wavelength calibrations by using the known positions of 14 atmospheric absorption features throughout the 0.4-2.5 micrometer wavelength region. These features, having varying width and intensity, were matched to the MODTRAN model with a non-linear least squares fitting algorithm. A complete calibration was developed for all 224 channels by interpolation. Instrument calibration cannot be assumed to be stable to 0.1 nm over a flight season given the rigors of thermal cycling and launch and landing loads. The upcoming sensor HYDICE will require a means for in-flight spectral calibration of each scene because the calibration is both temperature and pressure sensitive. In addition, any sensor using a two-dimensional array has the potential for systematic wavelength shifts as a function of cross-track position, commonly called 'smile'. Therefore, a rapid means for calibrating complete images will be required. The following describes a method for determining instrument wavelength calibration using atmospheric absorption features that is efficient enough to be used for entire images on workstations. This study shows the effect of the surface reflectance on the calibration accuracy and the calibration history for the AVIRIS B spectrometer over the 1992 flight season.

Description: Vegetation dynamics and changes in ecological patterns were measured by remote sensing over a 10 year period (1973 to 1983) for 148,406 landscape elements, covering more than 500 sq km in a protected forested wilderness. Quantitative measurements were made possible by methods to detect ecologically meaningful landscape units; these allowed measurement of ecological transition frequencies and calculation of expected recurrence times. Measured ecological transition frequencies reveal boreal forest wilderness as spatially heterogeneous and highly dynamic, with one-sixth of the area in clearings and early successional stages, consistent with recent postulates about the spatial and temporal patterns of natural ecosystems. Differences between managed forest areas and a protected wilderness allow assessment of different management regimes.