An airborne programmable imaging spectrometer has been developed to meet the increased requirements of land and water imaging in general and for mapping of chlorophyll fluorescence in particular. This latter application gives rise to the name of the sensor which is called Fluorescence Line Lmager (FLI). An imaging spectrometer provides the opportunity to examine both narrow bandwidth features in the target spectrum and to acquire high fidelity spatial images. The FLI has also been applied to hydrographic mapping, mapping of crops, forest and rangeland as well as geobotanical exploration. Distinctive spectral features which fall near atmospheric absorption lines of water and oxygen can be readily discerned and atmospheric artifacts can be avoided in the interpretation of resulting imagery.

The Advanced Solidstate Array Spectroradiometer (ASAS) is an airborne imaging spectrometer with 30 spectral channels extending from 450 to 880 nm. A 32 x 512 element silicon Charge Injection Device (CID) array is used as the detector. The ASAS was developed at the NASA Johnson Space Center (JSC) with General Electric providing the detector package under funding from the Naval Ocean Systems Center. The instrument was transferred to NASA Goddard Space Flight Center (GSFC) after its completion and initial test flights in August 1983. Several changes and refinements have been made to the ASAS as a result of the use of this instrument for terrestrial and oceanographic remote sensing research. The most notable of these changes has been in the sensor (optics plus detector package) mounting technique. This changed the ASAS from a fixed nadir viewing instrument to a sensor capable of multiple direction observations of surface bidirectional reflectance distribution functions. This change and refinements in the ability to radiometrically and spectrally calibrate the ASAS as well as reliability improvements are presented.

The Airborne Imaging Spectrometer (AIS I) was a multispectral infrared imaging instrument utilizing a 32 x 32 element mercury-cadmium-telluride detector array. A novel optical design provided high spectral resolution and, in conjunction with the 32-element spatial resolution, made optimal use of the capabilities of the area array. The operational successes of this proof-of-concept instrument, coupled with advances in infrared detector array technology, have provided the impetus and opportunity to upgrade the instrument by replacing the sensor with a newly developed 64 x 64 detector. The original instrument configuration is reviewed, and the design considerations, limitations and trade-offs necessitated by the new detector are examined, particularly the near doubling of the field of view and the increase from 1024 to 4096 pixels per read-out frame.

The AVIRIS instrument has been designed to do high spectral resolution remote sensing of the Earth. Utilizing both silicon and indium antimonide line array detectors, AVIRIS covers the spectral region from 0.41 pm to 2.45 pm in 10-nm bands. It was designed to fly aboard NASA's U2 and ER2 aircraft, where it will simulate the performance of future spacecraft instrumentation. Flying at an altitude of 20 km, it has an instantaneous field of view (IFOV) of 20 m and views a swath over 10 km wide. With an ability to record 40 minutes of data, it can, during a single flight, capture 500 km of flight line.

The development of the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) has been completed at the Jet Propulsion Laboratory, California Institute of Technology. This paper outlines the functional requirements for the spectrometer optics subsystem, and describes the spectrometer optical design. The optical subsystem performance is shown in terms of spectral modulation transfer functions, radial energy distributions and system transmission at selected wavelengths for the four spectrometers. An outline of the spectrometer alignment is included.

The foreoptics, fiber optic system and calibration source of the Airborne Visible/ Infrared Imaging Spectrometer (AVIRIS) are described. The foreoptics, based on a modified Kennedy scanner, is coupled by optical fibers to the four spectrometers. The optical fibers allowed convenient positioning of the spectrometers in the limited space and enabled simple compensation of the scanners thermal defocus (at the -23°C operating temperature) by active control of the fiber focal plane position. A challenging requirement for the fiber optic system was the transmission of the spectral range 1.85 to 2.45 microns at .45 numerical aperture. This was solved with custom fluoride glass fibers from Verre Fluore. The on-board calibration source is also coupled to the spectrometers by the fibers and provides two radiometric levels and a reference spectrum to check the spectrometers alignment. Results on performance of the assembled subsystems are presented.

The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) instrument uses four separate focal plane assemblies consisting of line array detectors that are multiplexed to a common J-FET preamp using a FET switch multiplexing (MUX) technique. A 32-element silicon line array covers the spectral range from 0.41 to 0.70 pm. Three additional 64-element indium antimonide (InSb) line arrays cover the spectral range from 0.68 to 2.45 pm. The spectral sampling interval per detector element is nominally 9.8 nm, giving a total of 224 spectral channels. All focal planes operate at liquid nitrogen temperature and are housed in separate dewars. Electrical performance characteristics include a read noise of < 1000 e- in all channels, response and dark nonuniformity of 5% peak to peak, and quantum efficiency of ) 60%.

The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) images the ground with an instantaneous field of view (IFOV) of 1 mrad. The IFOV is scanned 30 deg from left to right to provide the cross-track dimension of the image, while the aircraft's motion provides the along-track dimension. The scanning frequency is 12 Hz, with a scan efficiency of 70%. The scan mirror has an effective diameter of 5.7 in., and its positional accuracy is a small fraction of a milliradian of the nominal position-time profile. This paper describes the design and performance of the scan drive mechanism. Trade-offs among various approaches are discussed, and the reasons given for the selection of the cam drive. The salient features of the design are presented. The method of measuring performance is described, and the performance results are given.

The AVIRIS instrument has a separate dedicated analog signal processing chain for each of its four spectrometers. The signal chains amplify low-level focal-plane line array signals (5-10 mV full-scale span) in the presence of larger multiplexing signals (- 150 mV) providing the data handling system a ten-bit digital word (for each spectrometer) each 1.3 ps. This signal chain provides automatic correction for the line array dark signal nonuniformity (which can approach the full-scale signal span).

The timing and flow of detector and ancillary data for the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) are controlled within the instrument by its digital electronics assembly. In addition to providing detector and signal chain timing, the digital electronics receives, formats, and rate-buffers digitized science data; collects and formats ancillary (calibration and engineering) data; and merges both into a single tape record. Overall AVIRIS data handling is effected by a combination of dedicated digital electronics to control instrument timing, image data flow, and data rate buffering and a microcomputer programmed to handle real-time control of instrument mechanisms and the coordinated preparation of ancillary data.

The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) has been under development at the Jet Propulsion Laboratory (JPL) for the past four years. During this same time period, a dedicated ground data-processing system has been designed and implemented to archive and process the large amounts of data expected from this instrument. This paper reviews the objectives of this ground data-processing system and presents the hardware implementation. An outline of the data flow through the system is given, and the software and incorporated algorithms developed specifically for the systematic processing of AVIRIS data are described.

The laboratory spectral and radiometric calibration of the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) used in the radiometric calibration of all AVIRIS science data collected in 1987 is described. The instrumentation and procedures used in the calibration are discussed and the calibration accuracy achieved in the laboratory as determined by measurement and calculation is compared with the calibration requirements. Instrument performance factors affecting radiometry are described. The paper concludes with a discussion of future plans.

During the summers of 1985 and 1986 a Programmable Multispectral Imager (PMI), also known as the Fluorescence Line Imager (FLI), has been used to collect airborne data over a number of forested targets in Canada and the United States. The sites were selected on the basis of suspected localized vegetation stress due to possible excess metal uptake or reported regional forest decline due to suspected acid deposition damage. This paper focuses on the characteristics of the spectral/image data available from this new sensor along with results of preliminary analysis of some of these data. Stable pixel to pixel vegetation spectral properties provide a verification of sensor calibration methods. Comparison of FLI vegetation spectra with ground-based spectra of vegetation samples show good correspondence for a variety of species studied, including spectral properties of the red edge.

Canopy lignin concentration is strongly related to annual rates of nitrogen mineralization in a series of Wisconsin forest ecosystems. High spectral resolution Airborne Imaging Spectrometer (AIS) data were acquired over these forests to investigate the potential of remotely estimating canopy lignin content. Analysis of the data using a derivative transformation and correlative techniques suggests that lignin or a closely associated cellular constituent is influential in canopy reflectance. Spatial distributions of percent canopy lignin and annual rates of nitrogen mineralization for Blackhawk Island, Wisconsin, have been estimated from a mosaic of AIS imagery.

Seven flightlines of Airborne Imaging Spectrometer (AIS) data were analyzed for an area of hydrothermally altered rocks in the northern Grapevine Mountains, Nevada and California. Atmospheric and solar irradiance effects were removed from the raw data by first normalizing the data using an equal-area normalization and then dividing each spectrum (pixel) in an AIS flightline by an average spectrum calculated from all of the flightlines in a given mission. These procedures produced "Internal Average Relative (IAR) Reflectance" spectra that closely resemble laboratory and field reflectance spectra. Once the data were reduced to IAR reflectance, then removal of a continuum from the data was used to put all of the absorption bands on a common reference plane. The strongest absorption feature in each spectrum was analyzed by automatically calculating band position, band depth, and band width, and mapping these parameters into a hue, saturation, intensity (HSI) coded image for pixels with absorption features deeper than a selected cutoff depth. The color coded image allowed rapid identification of mineral groups based upon band position (hue). Band width was shown as variation in color saturation while image intensity allowed a subjective estimate of the band depth to be made. Examination of individual spectra and comparison to spectra of laboratory standards allowed positive identification of sericite (muscovite), montmorillonite, calcite, and dolomite.

Preliminary data acquired by the Fluorescence Line Imager (FLI) of the Canadian Department of Fisheries and Oceans over coastal and inland water scenes is analyzed with a focus on the need for imaging spectrometry in water remote sensing applications. Various examples are given wherein the information contribution of spectral images is advantageous if not essential towards solving the remote sensing inversion problem.

Recent advances in remote sensing have enabled scientists to collect image data in literally hundreds of spectral channels simul-taneously, from the near ultraviolet through the short-wavelength infrared, using imaging spectrometers. In many cases this data is of sufficient resolution to provide a direct surface materials identification. Yet the volume and complexity of the data produced requires new algorithms and approaches beyond those traditionally used for multispectral image analysis, including algorithms for fast image segmentation, spectral identification and mixture analysis. This paper describes a software system specifically designed to provide the scientist with the tools necessary for exploratory analysis of imaging spectrometer data using only modest computa-tional resources.

A field experiment and its results involving AIS-2 data for Rogers Lake, CA are described. The radiometry and spectral calibration of the instrument are critically examined in light of laboratory and field measurements. Three methods of compensating for the atmosphere in the search for ground reflectance are compared. We find, preliminarily, that the laboratory-determined responsivities are 30 to 50% less than expected for conditions of the flight for both short-and long-wavelength observations. The spectral sampling interval is 20 to 30 nm. The combined system-atmosphere-surface signal-to-noise ratio, as indexed by the mean response divided by the standard deviation for selected areas, lies between 40 and 110, depending upon how scene averages are taken, and is 30% less for flight conditions than for the laboratory. Atmospheric and surface variations may contribute to this difference. It is not possible to isolate instrument performance from the present data. As for methods of data reduction, the so-called scene average or log-residual method fails to recover any feature present in the surface reflectance, probably because of the extreme homogeneity of the scene. The empirical line method returns predicted surface reflectances that are systematically high but within a few percent of actual observed values using either calibrated or uncalibrated data. LOWTRAN-6, acting as an approximate theoretical model of the atmosphere for these exercises, predicts reflectance values 30 to 50% below the measured ones, based on the lower than expected radiances under solar illumination given by the instrument. This emphasizes the importance of accurate radiometric calibration in the study of surface or atmospheric properties.

Spaceborne imaging spectrometers such as the High Resolution Imaging Spectrometer (HIRIS) are presently under development. Major trade-offs in weight, cost, and performance are related to the spatial, spectral, and radiometric resolution. In this paper, we have investigated the relationship between spectral and radiometric resolution by applying maximum likelihood classification and binary vector analysis to a set of model data. The model data consist of Gaussian spectral absorption features to which various levels of random noise have been applied. The results show that if the rapid, binary vector analysis techniques are to be used, feature depth-to-noise ratios of 10 or greater and Nyquist sampling are required to achieve acceptable classification errors. This translates into a NEAp of 0.5% for a 5% feature depth.

After engineering flights aboard the NASA U-2 research aircraft in the winter of 1986-87 and spring of 1987, extensive data collection across the United States was begun with the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) in the summer of 1987 in support of a NASA data evaluation and technology assessment program. This paper presents some of the first results obtained from AVIRIS. Examples of spectral imagery acquired over Mountain View and Mono Lake, California, and the Cuprite Mining District in western Nevada are presented. Sensor performance and data quality are discussed, and in the final section of the paper, plans for the future are described.

Several advanced imaging spectrometers will be deployed on the Space Station Polar Platforms as a part of the Earth Observing System (Eos) program. Two of these, the Moderate-Resolution Imaging Spectrometer (MODIS) and High-Resolution Imaging Spectrometer (HIRIS), will be provided as facility instruments, and currently are under conceptual study at the Jet Propulsion Laboratory and the Goddard Space Flight Center. Other imaging spectrometer concepts, including proposals for a thermal infrared imaging spectrometer, are expected in response to the Eos Announcement of Opportunity scheduled for release in January of 1988.

The concept of the imaging spectrometer is becoming established as a major new thrust in remote sensing of the Earth. For several years, JPL has operated the Airborne Imaging Spectrometer on a NASA C-130; this instrument has demonstrated the direct identification of surface materials using imaging spectrometry. An advanced aircraft instrument, the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS), began operation on a NASA U-2 in 1987. The Shuttle Imaging Spectrometer Experiment (SISEX) was conceived as the next step in the sequence, and would provide a relatively inexpensive demonstration of the concept in Earth orbit. This paper will describe the design and development status of SISEX, and the status of the enabling technology.

The High Resolution Imaging Spectrometer, related data system, orbit, and mission operations are described. The pushbroom instrument simultaneously images the terrestrial surface in 192 spectral bands from 0.4 to 2.5 microns. The swath width is 30 kilometers and spatial resolution is 30 meters. It is planned to be launched with the Earth Observing System aboard the Space Station Polar Platform in 1995. Array detectors allow concurrent integration of the signals at 192,000 detector elements.

The Near-Infrared Mapping Spectrometer (NIMS) is one of the science instruments in the Galileo mission, which will explore Jupiter and its satellites in the mid-1990's. The NIMS experiment will map geological units on the surfaces of the Jovian satellites and characterize their mineral content; and, for the atmosphere of Jupiter, investigate cloud properties and the spatial and temporal variability of molecular abundances. The optics are gold-coated reflective and consist of a telescope and a grating spectrometer. The balance of the instrument includes a 17-detector (silicon and indium antimonide) focal plane array, a tuning fork chopper, microprocessor-controlled electronics, and a passive radiative cooler. A wobbling secondary mirror in the telescope provides 20 pixels in one dimension of spatial scanning in a pushbroom mode, with 0.5 mr x 0.5 mr instantaneous field of view. The spectral range is 0.7 - 5.2μm; resolution is 0.025μm. NIMS is the first infrared experiment to combine both spatial and spectral mapping capability in one instrument.

Infrared mapping spectrometry, a new remote sensing tool in which a scene is imaged simultaneously in hun-dreds of wavelengths, will be used on several approved planetary missions. A second-generation visible and infrared mapping spectrometer (VIMS) has been selected for both the Mars Observer and Comet Rendezvous Asteroid Flyby (CRAF) missions. The modular VIMS design can be adapted easily to the differing characteristics of several planetary missions planned through the end of the century. VIMS is a scanning spectrometer with a focal plane consisting of linear arrays of visible and infrared detectors, cooled by a radiative cooler. The foreoptics may be tailored to different missions, according to their field-of-view and resolution requirements. A wide-angle scan is implemented for Mars Observer, using a full-aperture scan mirror. A narrow-angle scan is achieved for the CRAF mission, using a scanning secondary mirror within a Cassegrain foreoptic. A significant on-board data processing capability has been designed to provide software flexibility, thus allowing for varying mission objectives and highly variable telecommunication data rates.

We describe the development of an imaging photon counting detector using microchannel plates and a wedge-and-strip readout system. Several detectors of this type have already been fabricated ranging from open face detectors for the Extreme Ultraviolet (EUV) (<1000 Å) to sealed tubes sensitive to Far Ultraviolet (FUV) (1000-2000 Å) radiation. Recently, we have developed a similar sensor to operate in the visible (400-800 nm) spectral region using a S20 photocathode. This device is compact and its construction is ruggedized and suitable for airborne and spaceborne applications. We will describe the results of our evaluations of the visible sensor in terms of their position resolution, gain, pulse height, quantum efficiency, non-linearity and dark count characteristics. Imaging applications of this sensor with a Fabry Perot spectrometer will be presented.

A low-cost, high-performance sensor design for low-earth-orbit ocean-color remote sensing is presented with system tradeoffs and performance estimates.