The STIS instrument was installed into HST in February 1997 during the Servicing Mission 2. It has almost completed checkout and is beginning its science program, and is working well. Several scientific demonstration observations were taken to illustrate some of the range of scientific uses and modes of observation of STIS.

Over the last decade a series of techniques has been developed to correct hyperspectral imaging sensed at to apparent surface reflectance. The techniques range from the empirical line method that makes use of ground target measurements to model-based methods such as the atmospheric removal program and MODTRAN that derive parameters from the data themselves to convert radiance to reflectance. Hybrid methods have been developed to augment the model calculations to provide better quality reflectance data. The model methods are computing intensive and,therefore, there is interest in developing more rapid methods to correct the data to reflectance. A parametric technique described here is in the early stages of development and could provide a breakthrough in speed of correction.

IR spectral imagers are being considered for air to ground target detection applications. IR Fourier transform spectrometers have been used to make field measurements of targets and background. Analysis of these measurements has shown the presence of target to background color and the high spectral band to band correlation of many backgrounds. Detection of low contrast targets in high thermal clutter backgrounds can therefore be improved by using IR spectral sensors as opposed to broad or single narrow band IR sensor. However, the improvement requires a high quality IR spectral sensor. In particular, the sensor must be able to achieve low noise levels. This paper establishes, parametrically, the relationships between ground resolution distance, range, aperture size, spectral bandwidth, integration time, and sensor noise level. Likely required sensor noise levels for target detection are taken from previous estimates based on Fourier transform spectrometer data. Current performance of IR detector arrays is established from the open literature. Using these inputs and making reasonable assumptions as to the values of the other parameters, the trade space between detector array size, ground coverage rate, and sensor noise level is explored.

This paper describes a methodology we have developed for wavelength band selection. This methodology combines an information theory-based criterion for selection with a genetic algorithm for searching for a near-optimal solution. We have applied this methodology to 302 material spectra in the Nonconventional Exploitation Factors database to determine the band locations for 7, 15, 30, and 60-band sets that permit the best material separation. These optical band sets wee also evaluated in terms of their utility related to anomaly/target detect in using multiband images generated from a hyperspectral digital imagery collection experiment image cube. The optimal band locations and their corresponding entropies are given in this paper. Also presented are the anomaly/target detection results obtained from using these optimal band sets.

The covered lantern project was initiated by the central MASINT Technology Coordination Office to demonstrate the tactical use of hyperspectral imagery with real time processing capability. We report on the design and use of the HYCORDER system developed for Covered Lantern that was tested in June 1995. The HYCORDER system consisted of an imaging spectrometer flying in a Pioneer Uncrewed Aeronautical Vehicle and a ground based real time analysis and visualization system. The camera was intensified allowing dawn to dusk operation. The spectral information was downlinked to the analysis system as standard analog video. The analysis system was constructed from 17 Texas Instrument C44 DSPs controlled by a 200 MHz Pentium Pro PC. A real time, parallel version of NRL's optical real-time adaptive spectral identification system algorithm was developed for this system. The system was capable of running continuously, allowing for broad area coverage. The algorithm was adaptive, accommodating changing lighting conditions and terrain. The general architecture of the algorithm will be discussed as well as results from the test.

Spectral imagery from the HYDICE instrument was analyzed for the purpose of target detection and identification. Data in the 0.4-2.5 micron wavelength range were acquired during the FOREST RADIANCE data collect. Data were analyzed from an area consisting of grassy fields and forest areas. A variety of targets were deployed in the field, with ground truth spectral measurements made. Analysis of reflectance data utilizing 'ground truth' and 'in-scene' spectra was conducted. The former suffered somewhat from inaccuracies in calibration, but training on subsets of the data allowed for relatively successful detection and classification in the remainder of the data. Spectral angle mapper and matched filter techniques were used. Both were successful in locating targets, but the later seemed to suffer more from 'false positives', though this may have been a function of thresholds set in the classification process.

Urban areas provide a complex material environment, both in the number of materials present, and in the spatial scale of material variation. Classification in urban environments using multispectral sensors has typically been limited to discrimination of major terrain classes due to both the limited spatial resolution of currently available sensors and to the inability to consistently discriminate between similar materials. High spectral and spatial resolution imagery, such as collected with the HYDICE sensor, provides the opportunity to develop detailed material maps for urban areas, and to perform precise material discrimination for cultural objects. Referencing a comprehensive set of material spectra, this paper describes a procedure for land cover classification which can be automated and performed with little or no a-priori knowledge of objects in the scene.

With its invariant property in reflectance based spectral signature, hyperspectral imagery has gained a prominent role in material identification and automatic target recognition since the late 1980's. Hyperspectral imagery loses its power when the targets are not directly visible to the sensor system. To conceal the targets, however, certain human actions are required which creates another type of signature like textural patterns that are unnatural in comparison to its surrounding. This paper examines how texture features can be exploited from a hyperspectral image cube for object recognition n general, and disturbed earth in particular under Susquehanna Resources and Environment's IMaG system, which integrates image processing, multisource analysis and GIS into one single environment.

Most spectrometric image analysis algorithms either require or can be augmented by estimates of target/background spectral signatures. The prediction of these spectra is complicated by the complex interplay of the target's spectra, background spectra, energy matter interaction effects, atmospheric effects and sensor response, an noise effects. Signatures can be further confused in the thermal IR by the temperature and temperature variation of targets and backgrounds. Finally, in nearly all cases, the image signature is the result of spatial mixing of target and background spectra. This paper addresses the potential for using synthetic image generation modeling tools to help in the prediction and understanding of hyperspectral signature. The DIRSIG model is discussed in terms of how it deals spectrally with target/background interactions, atmospheric propagation and sensor spectral, geometric MTF and noise effects. The DIRSIG model enables the estimation of mixed pixel 'image' spectra as they would be observed by an actual system imaging a complex 3D scene.

Thermal imagery from the spatially enhanced broadband array spectrograph system was analyzed for target detection purposes. The push-broom sensor was operated as part of the WESTERN RAINBOW experiment in October 1995. Data from 7.8- 13.4 microns were collected in 128 wavelength bands, with 128 pixels in the cross-track direction. The data set had nominal ground-resolution of better than one meter. Analysis techniques normally used in the reflective domain, with traditional imaging spectrometers, were used for the thermal data. Analysis was done in both the radiance and emissivity domains, following careful thermal calibration and atmospheric compensation. The techniques utilized were principal components, spectral angle mapper, and spectral matched filter. All wee successful, with the first two showing a success rate comparable to that found in similar experiments in the reflective domain. The principal components techniques was successful in discriminating man- made objects and disturbed earth from the desert background, much as expected. It was also successful in distinguishing between different categories of man-made objects. Of the latter two techniques, the spectral matched filter was more successful. This relatively greater success is attributed to the sensitivity of the spectral angle mapper to calibration errors, particular in the conversion from radiance to emissivity.

Improved vegetation maps are required for fire management and biodiversity assessment, from critical inputs for hydrological and biogeochemical models and represent a means for scaling-up point measurements. At scales greater than 10 meters, vegetation communities are typically mixed consisting of leaves, branches, exposed soil and shadows. To map mixed vegetation, many researchers employ spectral mixture analysis (SMA). In most SMA applications, a single set of spectra consisting of green vegetation, soil, non- photosynthetic vegetation and shade are used to 'unmix' images. However, because most scenes contain more than four components, this simple approach leads to fraction errors and may fail to differentiate many vegetation types. In this work, we apply a new approach called multiple endmember spectral mixture analysis, in which the number and types of endmembers vary per-pixel. Using this approach, hundreds of unique models are generated that account for community specific differences in plant chemistry, physical attributes and phenology. Additionally, we describe a new strategy for developing and organizing regionally specific spectral libraries. We present result from a study in the Santa Monica Mountains using AVIRIS data, in which we map grassland and chaparral communities, mapping species dominance in some cases to a high degree of accuracy.

Models of hydrology and climate in alpine and other seasonally snow-covered regions require input of snow- covered area (SCA) and snow surface grain size. The spectral signature of snow depends on the snow grain size. We have shown in earlier work that to map either SCA or grain size with optical data, one must know the distribution of the other variable. Hence, we solve for SCA or grain size simultaneously with spectral mixture analysis. Alpine regions frequently exhibit large snow grain size gradients due to rugged terrain. Because the spectral signature of snow is dependent on grain size, the grain size gradients translate into spectral gradients. Snow must then be represented by a range of endmembers. To provide the range and resolution of grain size, we use modeled snow spectra of varying brain size as snow endmembers. To complete the spectral mixture library, we incorporate reference endmembers of vegetation, rock, and soil. On airborne visible/IR imaging spectrometer data of the Sierra Nevada, CA, we ran multiple mixture models. We established constraints on mixture RMS, residual, and fractions to select a subset of physically realistic models. The optimal mixture was then selected from this subset by means of the least RMS. The snow endmember fraction and grain size of the optimal mixture provide the estimates of sub-pixel SCA and surface grain size, respectively.

In recent years, spectral imaging data have been acquired with a number of airborne imaging spectrometers. Similar data will soon be collected with the NASA HyperSpectral Imager instrument from the Lewis spacecraft. The majority of users of imaging spectrometer data are interested in studying surface properties. The atmospheric absorption and scattering effects must be removed from imaging spectrometer data, so that surface reflectance spectra can be derived. Previously, we developed and updated an operational atmosphere removal algorithm, which used the Malkmus narrow band model for modeling atmospheric gaseous transmittance. The narrow band model does a reasonably good job in modeling spectra at a resolution of 10 nm or coarser. Imaging spectrometers with spectral resolutions finer than 10 nm are now available. The narrow band model is not quite suitable for modeling data collected with these spectrometers. In this paper, we describe the development of a line-by-line- based algorithm for removing atmospheric effects from imaging spectrometer data. We also discus issues related to sampling and spectral resolution.

The tremendous potential for hyperspectral imagery as a remote sensing tool has driven the development of TRW's TRWIS III hyperspectral imager. This instrument provides 384 contiguous spectral channels at 5 nm to 6.25 nm spectral resolution covering the 400 nm to 2450 nm wavelength range. The spectra of each pixel in the scene are gathered simultaneously at signal to noise ratios of several hundred to one for typical Earth scenes. Designed to fly on a wide range of aircraft and with variable frame rate, the ground resolution can be varied from approximately 50 cm to 11 m depending on the aircraft altitude and speed. Meeting critical performance requirements for image quality, co- registration of spectral samples, spectral calibration, noise, and radiometric accuracy are important to the success of the instrument. TRWIS III performance has been validated and the instrument has been radiometrically calibrated using TRW's Multispectral Test Bed. This paper discusses the characterization and calibration process and results of the measurements. An example of results from a flight at the end of 1996 is included.

The MOS instruments on the Indian satellite IRS-P3 are now working for one year in orbit. Beside the calibration of the instrument during laboratory experiments the methods and tools of the in-orbit calibration gives a reliable basis for the interpretation of the remotely sounded data from the earth atmosphere-surface system. As a part of the in-orbit quality assurance the relative internal calibration together with the absolute recalibration with the extraterrestrial sun irradiance provides data for checking the instrument parameters. The principles of the system and the results of the one year operations are discussed.

DLR's imaging spectrometer the Modular Optoelectronic Scanner (MOS) on the Indian remote sensing satellite IRS-P3 has been orbiting since March 1996. MOS consists of two spectrometers, one narrow band spectrometer around 760 nm for retrieval of atmospheric parameters and a second one in the IVS/NIR region with an additional line camera at 1,6 micrometers . The instrument was especially designed for the remote sensing of coastal zone water and the determination and distinction of its constituents. MOS was developed and manufactured at the Institute of Space Sensor Technology (ISST) and launched in a joint effort with the Indian Space Research Organization (ISRO). The high spectral resolution of MOS offers the possibility of using the differences in spectral signatures of remote sensing objects for quantitative determination of geophysical parameters. In ISST a linear estimator to derive water constituents and aerosol optical thickness has been developed, exploiting Principal Component Inversion (PCI) of modeled top-of- atmosphere and experimental radiance data sets. The estimator results in sets of weighting coefficients for each measurement band, depending on the geophysical situations. Because of systematic misinterpretation due to non- adequateness of model and real situation the further development implies the parallel improvement of used water models and recalibration with in-situ data. The paper will present for selected test sites of the European coasts results of algorithm application. It will show the improvement of the estimated water constituents by using regional specific model parameter. Derived maps of chlorophyll like pigments, sediments and aerosol optical thickness ar presented.

Chromotomographic spectral imaging techniques offer high spatial resolution, moderate spectral resolution and high optical throughput. However, the performance of chromotomographic systems has historically been limited by the artifacts introduced by a cone of missing information. The recent successful application of principal component analysis to spectral imagery indicates that spectral imagery is inherently redundant. We have developed an iterative technique for filling in the missing cone that relies on this redundance. We demonstrate the effectiveness of our approach on measured data, and compare the results to those obtained with a scanned slit configuration.

We present results of a field demonstration of the computed tomography imaging spectrometer (CTIS). The CTIS was used to collect a sequence of image cubes of a missile in flight. This instrument is based on computed-tomography concepts and operates in the visible spectrum. Raw image data was recorded at video frame rate and an integration time of 2 msec. An iterative reconstruction of the spatial and spectral scene information from each raw image took 10 seconds. We present representative missile tracking-by- spectrum results.

The compound-tomography imaging spectrometer (CTIS) captures a scene's spatial and spectral information without any type of scanning. This capability enables a variety of applications that require spectral imaging of transient events. In this work, we demonstrate a flexible CTIS calibration techniques that allows multiple scene resolutions to be reconstructed from a single detector frame. The technique combines measurements with simulations to determine the transfer matrix of the system. Reconstructions of an experimental scene are performed to demonstrate the flexibility of the approach.

Over the past several years, Pacific Advanced Technology (PAT) has developed several hyperspectral imagers using diffractive optics as the dispersive media. This new approach has been patented and demonstrated in numerous field tests. PAT has developed hyperspectral cameras in the visible, mid-wave IR and is currently under contrast to the Air Force to develop a dual band hyperspectral lens for simultaneous spectral imaging in both the mid-wave and long- wave IR. The development of these cameras over the years have been sponsored by internal research and development, contracts from the Air Force Phillips Lab., Air Force Wright Labs Armament Division, BMDO and by the Office of Naval Research. Numerous papers have been presented in the past describing the performance of these various hyperspectral cameras. The purpose of this paper is to describe the theory behind the image multi-spectral sensing (IMSS) used in these hyperspectral cameras. IMSS utilizes a very simple optical design that enables a robust and low cost hyper-spectral imaging instrument. The IMSS is a dispersive spectrometer using a single diffractive optical element for both imaging and dispersion. The lens is tuned for a single wavelength giving maximum diffraction efficiency at that wavelength and high efficiency throughout the spectral band-pass of the camera. The diffractive optics disperse the light along the optical axis as opposed to perpendicular to the axis in conventional dispersive spectrometers. A detector array is used as the sensing medium and the spectral images are rad out electronically. POst processing is used to reduce spectral cross talk and to spatially sharpen the spectral images.

A focused research program has been under way for several years to discover optimally effective means for analysis of multispectral and hyperspectral data. The methods pursued are based upon fundamental principles of signal theory and signal processing. The basic approach revolves around viewing N spectral bands of data from a pixel as a single point in N dimensional space, thus, an important aspect of the work has been to discover unique aspects of higher dimensional spaces which can be exploited for their information-bearing aspects. Substantial progress on this problem has been made in the last several years, with several key algorithms having been defined. Among these are algorithms for transforms which define optimal case-specific features, and which improve the ability of the classifier to generalize. A more fundamental finding has been to understand the characteristics of high dimensional space and the significance of design samples and their use in defining the classifier. These results have been published in separate papers over the last several years. The purpose of this paper is to survey these results and to show how they relate to one another in achieving an effective overall analysis procedure for analyzing a hyperspectral image data set.

This paper presents results for the application of the orthogonal projection (OP) filter to AVIRIS hyperspectral images of the Lunar Crater Volcanic Field in Nevada. The OP filter is a special case of the simultaneous-diagonalization (SD) filter, developed to enhance a selected feature while suppressing other features and noise in an image scene. The SD filter applies to sets of images that are e spatially invariant (SI) and in which the individual features in the image scene and noise contribute linearly and additively (LA) to the recorded pixel image intensities. The OP filter uses the original LA SI image set and the signatures of the individual features to generate a new set of images in which the distinct features are separated. Applied to AVIRIS hyperspectral images, the OP filter performs spectral unmixing. The resulting filtered images are estimates of the spatial distributions, or endmembers, of the original image features, and can be used to reconstruct estimates of the original image. Since the number of individual features is much smaller than the number of images in the original set, this represents significant data compression. However, the filtered images are not perfect due to images are not perfect due to imaging system noise and imperfections in the fit of the LA SI model to the actual AVIRIS images. As a test of the accuracy of the compression method, this paper investigates the reconstruction of the original AVIRIS images from the small set of individual feature images.

Neural networks (NN) have been applied to hyperspectral image classification when traditional linear statistical classifiers have proven inadequate. The nonlinear and non- parametric properties of NN have often been cited for their apparent success. It has also been known that data preprocessing techniques such as principal component analysis (PCA) greatly improves classification accuracy. While PCA finds the axes of maximum variance in the data it does not guarantee increased separation between an arbitrary pair of classes. A transformation that is sensitive to class structure is obtained by solving the generalized eigenvalue problem of the amongst and within class covariance matrices of the data. Using this transformation, we demonstrate a case where the performance of linear statistical classifiers is comparable to that of NN classifiers for hyperspectral image classification.

Multispectral image data can be processed to 'unmix' component fractions for each pixel. Traditionally, however, these techniques do not account for the spatial spread of the signal due to the sensor point spread function (PSF). Optical PSFs typically have a width of about 0.5 to 1 pixels, which mixes spectral signatures from outside the ground-instantaneous-field-of-view of the pixel of interest. We have devised a mathematical and algorithmic framework for incorporating the sensor PSF into the unmixing problem. The approach requires solution of additional simultaneous equations from the neighboring pixels around the pixel of interest. These equations include weighting factors derived from the PSF, which is assumed to be adequately known for the specific sensor. The spatial-spectral algorithm has yielded a 28 percent error reduction for a simulated terrestrial scene when compared to a traditional unmixing method.In addition, the spatial-spectral algorithm significantly reduces the spatial width of unmixing error at class boundaries. In conclusion, the modified unmixing algorithm achieves notably improved unmixing results by including the previously ignored PSF effect on spatial- spectral mixing.

Multispectral and hyperspectral image sets contain large amounts of data which are difficult to exploit by manual means because they are comprised of multiple bands of image data that are not easily visualized or assessed.Non-literal imagery exploitation refers to a process that exploits non- spatial information by focusing on individual pixel signatures that span the spectral range of the sensor. GTE has developed a system that utilizes evolutionary computing techniques as a potential aid to imagery analysts to perform automatic object detection, recognition and materials identification on multispectral and hyperspectral imagery. The system employs sophisticated signature preprocessing and a unique combination of non-parametric search algorithms guided by a model based cost function to achieve rapid convergence and pattern recognition. The system is scaleable and is capable of discriminating decoys from real objects, identifying pertinent materials that comprise a specific object of interest and estimating the percentage of materials present within a pixel of interest.

This paper discusses recently developed algorithms for the classification of pixels in hyperspectral images, used in conjunction with a library of hyperspectral hemispherical reflectance data measured in the laboratory and partitioned into usable classes of materials. The algorithms are based upon functions of the principal components of the class covariances and the corresponding null spaces, and the underlying measures used in the classification statistics are similar to Mahalanobis distances. The algorithms can be used as stand-alone processing or combined with spatial and temporal algorithms n a higher level system of hyperspectral image processing. The nature of the classification algorithms and the database will be discussed, with particular attention being paid to issues specific to this approach. The basic performance of the classifier algorithms will be demonstrated using modified laboratory data. The applicability of orthogonal subspace projection methods to problems inherent in remote sensing using hyperspectral invisible and IR data will be emphasized, while specifically dealing with the compensation for inaccuracies in necessary estimates of atmospheric attenuation and target temperature. Preliminary results of classification of field collected hyperspectral data will also be presented, and ongoing and future work in hyperspectral classification described.

The objective is to use spectral image data to determine the identity of materials present in the image. In order to accomplish material identification, material-specific spectral signatures need to be developed, as do algorithms that exploit those signatures. The performance of the algorithms and signatures will be tested against real data collected at multiple altitudes and locations, calibrated to reflectance using different techniques. Spectral signatures will be derived for the targets that were deployed at various test using the HYDICE sensor. These signatures are used to detect and locate those targets in the data sets. The false alarm performance of the algorithms and signatures provides evidence for the uniqueness of those signatures. Similar false alarm performance between the calibration techniques demonstrates the ability to create robust spectral signatures without using a priori information on any material in the image for calibration. This has implications for a)hyperspectral sensor that have pixels with a large GSD and do not facilitate the use of calibration panels, b) collection scenarios where calibration panels cannot be deployed, and c) data exploitation.

We have investigated a novel IR microcalorimetric spectroscopy technique that can be used to detect the presence of trace amounts of target molecules. The chemical detection is accomplished by obtaining the IR photothermal spectra of molecules absorbed on the surface of an uncooled thermal detector. Traditional gravimetric based chemical detectors require highly selective coatings to achieve chemical specificity. In contrast, IR microcalorimetric based detection requires only moderately specific coatings since the specificity is a consequence of the photothermal spectrum. We have obtained IR photothermal spectra for trace concentrations of chemical analytes including diisopropyl methylphosphonate (DIMP), 2-mercaptoethanol and trinitrotoluene (TNT) over the wavelength region 2.5 to 14.5 micrometers . We found that in the wavelength region 2.5 to 14.5 micrometers DIMP exhibits two strong photothermal peaks. The photothermal spectra of 2-mercaptoethanol and TNT exhibit a number of peaks in the wavelength region 2.5 to 14.5 micrometers and the photothermal peaks for 2-mercaptoethanol are in excellent agreement with IR absorption peaks present in its IR spectrum. The photothermal response of chemical detectors based on microcalorimetric spectroscopy has been found to vary reproducibly and sensitively as a consequence of adsorption of small number of molecules on a detector surface followed by photon irradiation and can be used for improved chemical characterization.

The surface plasmon tunable filter (SPTF) is a new technology invented at the Jet Propulsion Laboratory. When white light is incident on a metal/air/metal structure, under appropriate conditions, surface plasmon waves are excited at one metal/air material interface. Those photons in the surface plasmon resonance wavelength range will be converted into the energy of free electrons in the metal, then coupled into the other metal film which re-radiates light at the identical resonant wavelength. This surface plasmon resonance depends on the dielectric constant of the metal and the thickness of the air gap. When the thickness of the air gap changes, the surface plasmon resonance spectrum shifts from one wavelength to another, and the device functions as a tunable filter. The SPTF is a light weight, low power device, which can be integrated with a solid state image sensor to form a spectrometer-on-a-chip. Theoretical calculation has shown that this image spectrometer can also work in IR range up to at least 10 micrometers .

This paper describes a study in progress. Its objective is to investigate the effects of spectral and spatial resolution, spectral coverage, signal-to-noise (SNR), and terrestrial background environments on hyperspectral exploitation performance and to determine their tradeoffs. The trade space and the methodologies used to perform nonliteral exploitation functions such as anomaly/target detection, material identification, and abundance estimation are described. Of the three exploitation functions, only summary anomaly/target detection results are presented in this paper. They are presented as a function of 4 spatial resolutions, 4 spectral resolutions, 2 spectral regions, 2 SNR levels, and 2 backgrounds. Anomaly/target detection performance comparison is made in terms of detection success rate when the false alarm rate is held constant for all image cubes. Within the study's trade space,results so far indicate that spatial resolution is the most important parameter for anomaly/target detection, followed by SNR and spectral resolution. Subsequent work on the effects of the same sensor design parameters on material identification is near completion and results will be documented in the very near future.

The ability to differentiate man-made materials from natural materials depends upon exploiting recognizable differences in their respective spectral response. Of particular interest is the question of whether a given materials signature derived from a laboratory or field measurement can be used as a training vector for discrimination or identification in a given setting. Through the application of a matched filter, we can quantify the performance of training vectors which have been transferred in this way as well as identify which spectral regions are most diagnostic in a given situation.

A wide variety of applications of imaging spectrometry have been demonstrated using data from aircraft systems. Based on this experience we have developed requirements for a satellite imaging spectrometer system to best characterize the littoral environment, for scientific and environmental studies and to meet Naval needs. This paper describes the process for determining those requirements and the resulting hyperspectral remote sensing technology (HRST) program. The HRST spacecraft has a coastal ocean imaging spectrometer with adequate spectral and spatial resolution and high signal to noise to provide long term monitoring and real- time characterization of the coastal environment. It includes on-board processing for rapid data analysis and data compression, a large volume recorder, and high speed downlink to handle the required large volumes of data. This is a joint program with an industrial partner,and their commercial remote sensing requirements are included in the system design.

A computational prototyping environment has been developed for the chemical imaging sensor, an IR imaging passive standoff detection device based on the rapid-scan Fourier transform interferometer. The environment uses commercially available software (AVS) which features a visual programing interface, extensive visualization capabilities, interactive steering, and an ability to distribute computations over a heterogenous group of computers. Physical models developed to simulate device behavior are incorporated into AVS modules and data-flow networks. The main distinctions of the models are the use of a ray tracing algorithm in the simulations of interference in the optical system and the application of the radiometric model of MODTRAN for realistic simulations of atmospheric transmittance and radiance. The simulations predict abilities of the device to measure interferograms and spectra as well as spectral images of clouds containing harmful chemicals.

Imaging spectrometers are increasingly use to measure the spectral radiance of Earth surface from aircraft and satellites. The design of optics for imaging spectrometers, covering the spectral range from visible through short-wave IR, can be simplified using the Dyson and Offner concentric relay design forms. Problems in integration of dispersing elements into these concentric relays are addressed. It is shown that use of dispersing prisms with curved faces can provide good performance in designs that are simple and compact.

As a first step toward the development of a new remote sensing technique that we call 'holographic correlation spectroscopy,' we demonstrate that diffractive optics can be used to synthesize the IR spectra of real compounds. In particular, we have designed, fabricated, and characterized a diffractive element that successfully reproduces the major features of the spectrum of gaseous HF in the region between 3600 cm^-1 and 4300 cm^-1. The reflection-mode diffractive optic consists of 4096 lines, each 4.5 micrometers wide, at 16 discrete depths relative to the substrate, and was fabricated on a silicon wafer using anisotropic reactive ion-beam etching in a four-mask-level process. We envision the use diffractive elements of this type to replace the cumbersome reference cells of conventional correlation spectroscopy and thereby enable a new class of compact and versatile correlation spectrometers.

This paper describes the theory of design, operation, and testing of a tunable MWIR Fabry-Perot interferometer operating in low orders. This device is called the agile bandpass tunable filter (ABTF) due to the fact that the spectral bandwidth can be changed by a large factor by changing the order. In first order the system can be tuned over the entire 3.5-5 micrometers spectral region with only a single order sorting filter. We provide a short introduction to tunable filters an then briefly discuss the requirements that low order operation places on the Fabry-Perot dielectric mirrors. Operation in low orders forces one to abandon the classical Fabry-Perot approximation that the mirrors are negligibly thin compared to the plate separation. Rather, one must now account for the phase properties of the dielectric stack mirrors as they produce phase effects comparable to the plate separation. We next address the issue of control of the Fabry-Perot. This is accomplished through a closed-loop system using capacitive sensor on the Fabry-Perot flats to measure the separation of the plates. Additionally we describe how the ABTF is characterized using a FTIR to measure the bandpass shape and position, and we show some examples of measurements made with the ABTF used as a hyperspectral imaging system with a 256 X 256 HgCdTe camera. We conclude with a discussion of potential applications and future work.

Birefringent, nematic liquid crystals (LC) have been laminated between the substrates of several Fabry-Perot etalons. The application of an electric field allows the effective index of refraction of the LC to be varied. A polymer alignment layer is used to align the crystals perpendicular to the optical axis of the Fabry-Perot etalon. An oscillating electric field is used to rotate the crystal around the optical axis of the etalon, effectively changing the index of refraction. This change in index is used to tune the Fabry-Perot etalon in a manner similar to traditional pressure and mechanical tuning systems. However, the approach described here has the advantage of producing a solid-state etalon that is tunable without needing a bulky pressure system or environmentally sensitive piezo-electric stacks. A two etalon spectrometer consisting of two Fabry- Perot etalons coupled to a CID detector has been developed. A suppression etalon with a gap of 10 micrometers , and a LC wit a refractive index of 1.63 are used in conjunction with a high resolution etalon to produce an instrument ideal for observing the atomic spectra of hot, light neutral species and the molecular bands in the atmosphere. Several other etalons have been constructed to further develop this technology. Clear apertures greater than 2 inches have been achieved, and a hybrid spacer technique has been developed to allow for etalons with spacings of up to 1 cm. Fabry- Perot partial reflective coatings capable of operation from the visible to the NIR will also be discussed.

At the Denver meeting, last year, we presented the High Etendue Imaging Fourier Transform Spectrometer, (HEIFTS), theory and optical design. This device uses a new 'image plane interferometer' geometry to produce 'autocorrelation function modulation' in the image plane of a 2D imaging array, such that the phase offset of the modulation varies linearly across the image. As a 2D image is pushbroomed across the imaging, array, the record of an individual scene pixel is recorded for each autocorrelation phase offset. The 3D array of this data is processed to yield an 'autocorrelation function' data cube, which is Fourier transformed to yield a 'wavenumber' hyperspectral data curve. A phase I device has been demonstrated in the laboratory and initial results are presented. The significant increase in signal to noise ratio, which the HEIFTS optical design promises over conventional hyperspectral imaging schemes, has been simulated, and results will be discussed. A Phase II system is being prepared for initial field deployment, and will be described.