Beginning in 1984, three Earth Radiation Budget Experiment (ERBE) spacecraft were launched to measure the Earth's radiation budget components: Solar-incident, Earth-emitted, and Earth-reflected radiation. The National Aeronautics and Space Administration's Earth Radiation Budget Satellite was launched in 1984 in a low-inclination orbit, and the National Oceanic and Atmospheric Administration's NOAA-9 and NOAA-10 satellites were launched in Sun-synchronous polar orbits in 1984 and 1986, respectively. This paper describes the method used for correcting for changes in the gains and offsets used for determining the final count conversion coefficients used in processing the in-flight calibration data. We also describe a technique for the determination of WFOV shortwave offsets which proxies for nighttime offsets during months of continuous daytime only orbits near sunrise and sunset when no nighttime measurements are available. The final result covering the period of 1984 through 1991 are discussed in this paper.

The Earth Radiation Budget Experiment (ERBE) active cavity radiometers are used to measure the incoming solar, reflected-solar, and emitted longwave radiation from the Earth and its atmosphere. The radiometers are carried by the National Aeronautics and Space Administration's Earth Radiation Budget Satellite (ERBS) and the National Oceanic and Atmospheric Administration's NOAA-9 and NOAA-10 spacecraft. Four Earth-viewing nonscanning active cavity radiometers are carried by each platform. Two of the radiometers are sensitive to radiation in the spectral range from 0.2 to 50 micrometers , while the other two radiometers are sensitive to radiation in the spectral range from 0.2 to 5.0 micrometers . Each set of radiometers comes in a wide field-of-view (WFOV) and a medium field-of-view (MFOV) configuration. The cavities of the shortwave (visible) radiometers are covered with Suprasil dome to filter out the incoming longwave radiation. An absolute calibration of the radiometers was obtained by allowing them to view laboratory radiometric sources based upon the International Practical Temperature Scale of 1968 (IPTS68), and inflight stability of the radiometers is monitored by internal calibration sources.

The National Aeronautics and Space Administration's (NASA) Earth Radiation Budget Experiment (ERBE) measures the three radiant energy components (incident Solar, reflected, and emitted) of the Earth-Atmosphere System. This experiment employs four Earth-viewing nonscanning active cavity type radiometers on each the NASA Earth Radiation Budget Satellite (ERBS), and the National Oceanic and Atmospheric Administration's NOAA-9 and NOAA-10 spacecraft. The NOAA-9 and the NOAA-10 spacecraft are in Sun-synchronous polar orbits while the low-inclination orbit of the ERBS spacecraft precesses in local solar time. Two of the radiometers are total channels which measure radiation in the wavelength region of 0.2 to > 50 micrometers, while the other two radiometers are shortwave channels which measure the solar reflected energy in the spectral band of 0.2 to < 5 micrometers. A validation technique based on normalized measurements of independent nonscanning radiometers on different platforms observing the same geographical scene at the same time is described in this paper. This technique was applied to 1985, 1986, and 1987 measurements from the ERBS, NOAA-9, and NOAA-10 spacecraft to determine abrupt changes in the radiometric calibration procedure. The results for three years are presented.

Many scanning sensors produce raw outputs in which their response to the signal is superimposed on a large background. The signal changes rapidly, due to the scanned input, while the background varies more slowly, due to thermal drifts and 1/f noise. Whenever such a sensor is used as a radiometer, it is necessary to perform a differential measurement: to measure a known reference and subtract it from the raw signals, cancelling the common-mode background. Calibration is also a differential measurement: the difference between each channel's response to two known inputs is divided by the difference between these two inputs to determine the linear gain of the channel. The GOES-I Imager obtains its background subtraction references by viewing space, with radiance virtually equal to zero, during the turn- around intervals at the ends of scan lines. It used a temperature-monitored blackbody as a second reference to measure the gain. We have verified our analytical predictions by computer simulations. Gaussian and 1/f noise were generated and combined, filtered, and processed using the differential measurement algorithms. Excellent agreement was demonstrated between these simulations and the analytical model.

Landsat Multispectral Scanner (MSS) data has the potential for being widely used in global change studies since it has been collected for nearly twenty years. It is crucial to the success of these studies that MSS data be as radiometrically accurate as possible. In order to assess the quality of relative radiometric correction algorithms, a study was undertaken to compare the two commonly used techniques: use of the internal calibrator, and use of scene statistics. A new technique was developed to quantify the observed differences. Results from this study indicated clear, consistent differences exist between the two commonly used relative calibration methods and suggest that relative radiometric calibration of MSS data can be performed more effectively using the scene statistics approach.

The relative degradation in time of the visible(channel 1: 0.58-0.68jim) and nearinfrared(channel 2: 0.72-1.lpm) channels of the Advanced Very High Resolution Radiometer(AVHRR), onboard the NOAA polar-orbiting operational environmental satellites(POES), has been determined, usi.ng the southeastern Libyan desert(21-23°N latitude; 2829° E longitude) as a time-invariant calibration target. A statistical procedure was used on the reflectance data for the two channels from the B3 data of the International Satellite Cloud Climatology Project(ISCCP) to obtain the degradation rates for the AVHRRs on NOAA-7, -9, and -11 spacecraft. The degradation rates per year for channels 1 and 2 are respectively: 3.6% and 4.3%(NOAA-7); 5.9% and 3.5%(NOAA-9); and 1.2% and 2.0%(NOAA-11). The use of the degradation rates thus determined, in conjunction with t1absolute" calibrations obtained from congruent aircraft and satellite measurements, in the development of correction algorithms is illustrated with the AVHRR on the NOAA-9 spacecraft.

SPOT absolute and multitemporal calibration is achieved using the on-board lamp plus regular checks over White Sands or other test sites which radiance is derived from ground measurements during the satellite overpass. In order to develop another independent method to monitor the long term evolution of the cameras sensitivity, systematic acquisitions of stable desert areas have been decided. The choice of the sites have been made based upon stability studies on images of large field of view satellites (METEOSAT and AVHRR). We now dispose of more than two years of systematic SPOT2 observations over three north Africa areas for all the modes of the two cameras, plus some SPOT1 acquisitions. For the time being, only four sites have been processed (acquired in (Chi) s mode by the HRV1 camera). The first results are promising and show: (1) a good stability of the sites (specially in (Chi) s2), (2) a good agreement with the on-board lamp, (3) a good consistency between SPOT1 and SPOT2 calibration.

The software SPACE, developed at JRC, Ispra, Italy, processes on a day-by-day basis the AVHRR images over Europe to forecast crop production which requires updated calibration of channels 1 and 2. The most suitable way to update calibration is to implement specific subroutines with, after identification and characterization, three Saharan desertic sites for inter-temporal calibration and the sunglint off-shore of Europe for an interband calibration. The paper also adresses the way to obtain absolute calibration and raises the problem of potential spectral shift of the sensor.

The calibration requirements for an earth observing imaging spectrometer system can be derived for specific applications. For the case of the detection of ocean chlorophyll a method of defining the minimum calibration requirements in relation to the required measurement accuracy is described. An end-to-end simulation including the effect of chlorophyll on ocean reflectance, the effect of the atmosphere, satellite system detection efficiency, noise and calibration uncertainty and of the inversion process is described. It is shown that for the proposed ESA NERIS instrument an inversion process based on the use of singular value decomposition methods places reduced constraints on the requirement for absolute accuracy and leads to similar requirements for relative calibration as the more conventional band ratio methods.

The in-flight calibration of the EOS Multi-angle Imaging SpectroRadiometer (MISR) will be achieved, in part, by observing deployable Spectralon panels. This material reflects light diffusely, and allows all cameras to view a near constant radiance field. This is particularly true when a panel is illuminated near the surface normal. To meet the challenging MISR calibration requirements, however, very accurate knowledge of the panel reflectance must be known for all utilized angles of illumination, and for all camera and monitoring photodiode view angles. It is believed that model predictions of the panels bidirectional reflectance distribution function (BRDF) can be used in conjunction with a measurements program to provide the required characterization. This paper describes the results of a model inversion which was conducted using measured Spectralon BRDF data at several illumination angles. Four physical parameters of the material were retrieved, and are available for use with the model to predict reflectance for any arbitrary illumination or view angle. With these data the root mean square difference between the model and the observations is currently of the order of the noise in the data, at about +/- 1%. With this success the model will now be used in a variety of future studies, including the development of a measurements test plan, the validation of these data, and the prediction of a new BRDF profile, should the material degrade in space.

A vicarious calibration of band 6 of the Thematic Mapper sensor carried on Landsat-5 was conducted at White Sands Missile Range (NM) in August 1992. The method utilized was based on direct measurements of surface temperature and local meteorological factors and an indirect measurement of surface emittance. Atmospheric path transmission and radiances were characterized using LOWTRAN7, and surface reflectance was derived from surface emittance. The difference between the preflight calibration and this calibration was less than 5%.

Since their launch in February 1986 and January 1990, we have, from time to time, conducted radiometric calibrations of the SPOT-1 and -2 Haute Resolution Visible (HRV) cameras at White Sands, New Mexico. We summarize the results of the calibrations, comparing the absolute calibration coefficients obtained from the White Sands data to the relative calibrations obtained from the on-board tungsten lamp.

The reflectance-based method is used to determine an absolute radiometric calibration of Landsat-5 Thematic Mapper for the solar reflective portion of the spectrum. Results are given for data collected at White Sands Missile Range in New Mexico on 1992-08-15. These results are compared to those obtained from applying a similar processing approach to data collected in 1984, 1985, 1987, and 1988.

The astronomical ultra-violet space telescope, TAUVEX, being developed in Israel by EL-OP Ltd., in conjunction with Tel Aviv University's Dept. of Astronomy and Astrophysics, has three co-aligned 20 cm diameter telescopes, each with an imaging photon-counting detector of the Wedge & Strip Anode type. The geometric and radiometric parameters of the system must be calibrated before launch in order that the image data acquired by the detector and signal processing sub-system can be converted into accurate maps of the UV sources in the sky. We describe the calibration philosophy and methodology involved in the TAUVEX system and sub-system calibration process. Also presented are the facilities and equipment specially designed and adapted for this purpose.

The next generation Wide Field Planetary Camera will include a calibration system which will enable in situ relative radiometric field calibration of the CCD detectors from the VUV to the near IR. The system features various incandescent and UV light sources which are configured to produce flat field. Preliminary tests and characterization of the system show performance to be quite good both spatially and spectrally. A description of the system, test procedures, and results will be described.

Emissive reference spheres will be used to calibrate the orbiting SPIRIT III LWIR sensor with respect to point sources. Such spheres can be made to approximate very closely the spectrum of an ideal blackbody. However, the temperature of the sphere must be modelled to an accuracy of 1.6K or better during the calibration measurements in spite of various sources of uncertainty over the sphere's near-Earth orbit. Observations of both standard stars and emissive reference spheres will provide an unprecedented opportunity to compare the results of two entirely distinct calibration methods.

The Infrared Radiance Comparator was designed for the French setting of the International Scale Temperatures (IST 90). This apparatus is devoted to high accuracy calibration of radiative sources (black bodies, tungsten ribbon lamps) for the 250 K - 2500 K temperature scale. The basic concept used for this instrument minimizes the systematic errors well known in this metrology field by measuring the critical parameters (effective wavelength of internal interference filter, linearity defects of the analog signal of the detector, polarization ratio of measured radiatives sources). Some secondary devices are used for the in-situ measurements of these critical parameters.

A novel approach to estimate the in-band noise of an EO sensor from its power spectrum is developed and experimental results using this approach are presented.

Synthetic Aperture Radar's data are today widely used in Remote Sensing applications, in order to obtain high resolution ground images. For this purpose a good estimation of Doppler parameters is required. The focusing quality of the final image will depend mainly on the update rate and accuracy of Doppler rate estimations obtained directly from sensor's raw-data. Although the autofocus algorithm normally used has not theoretical limitations, from a practical point of view it works well only under appropriate conditions. In this paper an alternative technique will be proposed, based on the Wigner distribution analysis of sensor's data; results about its application to simulated and real signals will also be reported.

Desert areas are good candidates for the assessment of multitemporal, multiband or multiangular calibration of optical satellite sensors. This paper describes a selection procedure of desert zones in North Africa and Saudi Arabia, of size 100 X 100 km, using a criterion of spatial uniformity in a series of METEOSAT-4 visible data. Twenty such zones are selected with a spatial uniformity better than 3% in a multitemporal series of cloud free images. The temporal stability of the spatially averaged reflectance of each selected zone is investigated at seasonal and hourly time scales with same series of images. It is found that the temporal variations, of typical peak-to-peak amplitude 8 - 15% in relative value, are mostly controlled by directional effects, with residual rms variations, not accounted for by directional effects, of the order of 1 to 2 % in relative value.

As a part of the Earth Radiation Budget Experiment (ERBE), the Earth Radiation Budget Satellite (ERBS) has continuously acquired active-cavity radiometric measurements of the Earth's outgoing radiative flux since October 1984. We have analyzed the daily average of daytime and nighttime total and shortwave flux measurements from the ERBS, at satellite altitude, for the period from January 1985 through December 1991. We found the annual cycle of total and shortwave flux measurements to be highly repeatable. We averaged the ERBS data for the years 1985 through 1989, and used these as a mean for comparison of 1990 and 1991 data. We observed (in daytime) some differences in both the total and shortwave flux regions. Similar trends were not apparent in the derived daytime longwave data (i.e., total- sw). There is evidence of generally increasing outgoing flux (and hence a warming trend) over the years 1985 through 1991.

Thermal Infrared Multispectral Scanner (TIMS) imagery and Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) imagery are registered to each other for seven scenes. Cloud area, as determined by applying the 3-band ratio method to AVIRIS imagery, is used to determine the optimum temperature threshold in the TIMS imagery. For this set of scenes, a threshold of 2 degree(s) to 3 degree(s) below the background temperature provides for the most accurate cloud pixel identification. Five to 8% differences in cloud area are found when comparing cloud pixel identification between AVIRIS and TIMS. Some of the differences are due to misregistration; however, at least half are due to differences in the mensuration process. It is demonstrated that cloud edges have a distribution of temperature thresholds, indicating the difficulty of locating cloud edges with a single temperature threshold. It is found that cloud edges occupy nearly half of the entire range of scene temperatures, significantly overlapping the distribution of background temperatures.

In field studies involving radiometry the irradiance variation is an important factor affecting measurements. There are tradeoffs between methods for obtaining the reflectance of a target. A widely spread method of field radiometry is the one where an instrument measures a reference panel and then the target. In some cases, for one measurement of reference, many target measurements may be taken, and between the target and the reference measurements irradiance may change. This change can dramatically affect the detected signal, and consequently the bidirectional reflectance factor. This paper reports an experiment conducted to study the influence of irradiance variations on the determination of the bidirectional reflectance factor of a target. For a clear day the bidirectional reflectance factor was measured at different time-interval rates. Irradiance was measured at a one second rate. The time spent on three different reflectance measurements method was evaluated at first. Results showed that for a unfavorable situation, in which a substitution method was tested, the error introduced with a 180 seconds interval between target and reference measurements is of the same magnitude as the nominal error of the radiometer intercalibration. For an unfavorable day, in which there were clouds, error as great as 16% can be introduced in the BRF measurements.

Effective radiometric models in IR and millimeter waves ranges were developed for remote measurements of a sea surface in the real time. Real time processing of IR-radiometer signals was accelerated by decreasing of integration intervals in the space of angles of sight. A 3D model was developed in the millimeter wave range for a 2-scale rough sea surface. The suggested models yield the wind velocity (IR-model), wave spectrum and direction of the sea surface waves (millimeter-range model).

The simulation of imagery that will be produced by a new sensor is an important element in the design process. To properly account for a new sensor's spatial and spectral degradation, it is necessary to supply as input to the simulation a relatively high spatial and spectral representation of the type of scene to be imaged. The performance of the sensor under design is then evaluated by comparison of the simulated degraded image with the input scene and/or by comparison with simulated images from existing sensors. In this paper we demonstrate how realistic scene with high spatial and spectral resolution can be synthesized from color aerial photography and AVIRIS hyperspectral imagery. Image processing techniques for fusion of aerial photography with AVIRIS imagery that take into account the topography of the scene are described. Sample composite images for a site near Cuprite, Nevada, are presented along with evidence that spectral signatures are preserved after applying the synthesis procedure.

This paper presents results from the application of model based techniques for an efficient correction to the topographically induced radiometric influences on the remotely sensed imagery. Thus, in a first step, relief induced geometric distortions of optical imagery must be removed, taking terrain elevation into account. In a second step the radiometry of the image is considered. Synthetic images are generated based on the Digital Elevation Models, and sun and satellite position at the time of acquisition of the image. The synthetically derived images model the image formation process for the direct lighted and shaded areas, using direct, indirect, and diffuse illumination. The resemblance of the synthesized image to reality is evaluated for a mountainous alpine region covered by snow. In a last step the data derived from the synthetic image are used for radiometric correction of the effects of the topography.

The development of reliable techniques for mission planning and mission rehearsal require the availability of a large number of images from a variety of sensors so that the system operators and pilots can adequately plan and rehearse the missions under realistic scenarios. Conversely, the development of databases for evaluating the performance of algorithms developed to perform automatic target recognition or terminal homing require a significant number of images, but fewer and covering less area than the imagery required for mission training and mission rehearsal. However, the target recognition and terminal homing images must be of significantly higher quality to avoid introducing artifacts into the imagery which would make the algorithms perform better than would be the case against real-world imagery. This paper will define a general conunon database (COB) structure which can be used to generate imagery for a variety of sensors. The sensors for which general structures will be defined include thermal infrared (TIR), synthetic aperture radar (SAR), millimeter wave (MMW) radar and laser radar (LADAR). These general structures have been used to convert color infrared (CIR) imagery to TIR and SAR for a mission rehearsal application. This structure development is initiated by defining the three main considerations of application area, speed of processing and image quality. These three considerations are highly interrelated and impose constraints on the types of algorithm architectures which are feasible for developing the required database. The top-level structure for the COB is composed of both dynamic and stationary files. The stationary files are fixed for each scenario and the dynamic files may change under the control of the event control module. The main components of the dynamic files, stationary files and event control module will be defined with specific examples presented for each of these modules. The COB structure is developed in a modular nature so that additional information and new sensors concepts can be easily added as system and mission requirements change. In addition, the issues of validation of the CDB will be addressed as it applies to the expected usage of the COB.

The need for low-cost, high quality and application-targeted spaceborne sensing is rapidly gaining in prominence. In response to this need, Spar Aerospace Limited is a major contributor to PROGERT, a Quebec-Government funded project to enhance remote sensing services to the provincial forestry user community. The computer simulation of advanced electro-optical concepts is being performed at Spar in PROGERT and applied to the development of a new satellite imaging instrument. As part of the overall design strategy, Spar is developing a model, based on forest physical parameters and standard sensor sub-systems. The purpose of this model is to explore system-level trade-offs in the early design phases thus reducing costly hardware development. In particular, the use of this model allows optimal specification of data acquisition parameters within the constraints of mission and end-user requirements.

The Digital Imaging and Remote Sensing laboratory's Image Generation model combines computer aided design, ray tracing techniques, radiometric principles, and thermodynamic models to create synthetic imagery. The model emphasizes rigorous radiometric solutions that account for spectral reflectance effects, angular emissivities, atmospheric transmission and upwelled and downwelled sky radiance. This paper describes enhancements to the radiometric portion of the code that permits inclusion of variations with azimuth of downwelling sky radiance, solution of the radiometric propagation models using specific radiosonde data including adjustments for the time of day, and the incorporation of background effects from objects adjacent to the target. Simulated scenes are presented that show how these enhancements produce imagery that more closely match observed phenomena. In particular, the importance of properly modeled sky radiance is shown both for low altitude oblique imagery where the sky is directly observed and for near nadir imagery where reflected sky radiance is important.

An essential part of remote sensor development is constructing accurate electro-optic models which simulate the system and forecast its radiometric performance prior to deployment. Although based upon established radiative transfer equations, these models are often tailored to a particular sensor, implemented as a complex spreadsheet or a batch program written in a high-level language. Unfortunately, a complex spreadsheet can be difficult to manage, and a batch program is laborious to modify and rarely extensible to other sensors. Furthermore, both of these approaches fail to visually represent the system they are modeling. This paper describes WinRAD, a Microsoft Windows -hosted application which, when completed, will provide both scientist and engineer with graphical tools to interactively design electro-optic models the overcome these shortcomings.

This paper reviews current and future signature modeling activities at KRC and TACOM. PRISM (Physically Reasonable Infrared Signature Model) and its associated modeling tools are discussed along with the implementation of the physical principles that will evolve into the SuperCode. By continuing the current efforts with PRISM and then forming a SuperCode Research Consortium to implement additional advanced features, a universal code will be available to the modeling community.

Geographic background models constitute an important component in scene simulations as they provide one component of clutter observed through IR sensors. Accurate modeling of the spatially-varying features and signatures of the background presents a significant challenge to the developer of a scene simulation; especially for tactical applications. The approach employed within the Georgia Tech Research Institute is outlined in this paper. Emphasis is placed on the methodology for creating the spatial structure of the background, the underlying signature prediction model, and the relationship between the geographic features and the signature model.

An ocean surface model for synthetic IR/visible imaging applications has been developed at GTRI based upon the Pierson-Moskowitz wave spectrum. The model calculates a 2D grid of height values describing a given snapshot of the sea surface. This surface is a function of wind speed and direction as well as elapsed time into the simulation. The time parameter permits the animation of the sea surface during a simulated movie sequence. Sea signatures are calculated using a combination of models and tools: the GTRI IR signature code GTSIG is used to predict sea temperatures; LOWTRAN7 is used to construct tables of sky radiances; the Fresnel equations are used to construct tables of sea reflectance. These signature components are combined during image rendering with a ray-tracing approach the provides the total radiance (emitted and reflected) from the ocean surface arriving at each sensor image pixel. This ocean model has been integrated into a complete image rendering system called GTRENDER, and is available for use in GTRI applications such as the GTSIMS family of missile simulations.

Digital simulations produced by the CREATION scene generation program will be used to demonstrate effects of state of the art and proposed future sensor characteristics on the output images of IR imaging systems. Necessary conditions for the validity of the digital simulations will be illustrated in the first part of the presentation. Tradeoffs between specific FLIRs will be discussed and some suggestions for preprocessing steps of future focal planes follow. Important tradeoffs between sensor characteristics are stressed. The fidelity of CREATION does not only apply to readily available vehicle geometry; the rigorously 3D background features of the latest version of CREATION have higher resolution than the state of the art thermal signatures of targets; these additional details may be important for objective comparisons of sensor degradations.

PIPADS is a vertically integrated parallel image processing and display system designed for interactive visualization and processing of large geoscientific data sets arising from remote sensing or computational modelling. Applications include perspective views of digital terrain models and other image surfaces, interactive image warping, volume visualization and seismic data processing. The PIPADS system consists of a layered object oriented software framework resident on a Unix workstation host and a high-performance modular hardware sub-system. The design was driven by the requirements of interactive visualization and image processing using a class of highly parallel and regular image-space algorithms, known as scan-line algorithms. The software control structure allows both asynchronous and synchronous communications from the host application to the compute/display server and allows tuning for minimal latency between the user-interaction and the resulting change in data view.

Sources of remotely sensed atmospheric data have expanded rapidly within the last several years, including ground-based Doppler weather radars and several satellite-based, earth- observing systems that sense and record large-scale weather conditions. Although these systems are capable of producing voluminous amounts of data, tools for visualizing, analyzing, and applying the data for operational uses requiring 4D display have not kept up. This paper presents a discussion of a software system known as WEST (Weather Environment Simulation Technology). WEST was developed through internal funding at Southwest Research Institute, for the visualization and simulator integration of very large atmospheric data sets containing fine-grained weather features. The approach implemented within WEST allows the user to visually observe, 'fly through', and interact with recorded atmospheric data at real-time speeds using commercial off-the-shelf hardware. This new capability is designed to benefit such diverse applications as flight training, weather analysis, systems design, and mission planning.

It is our belief that new modes of research and new tools are required to handle the massive amount of diverse data that is to be stored, organized, accessed, distributed, visualized, and analyzed in this decade. The fundamental innovation required is the integration of three automation technologies: viz. knowledge-based expert systems, science visualization and science data management. This integration is based on a concept called the DataHub. In order to prove the concept, a series of software prototypes are being implemented. On the basis of used comments DataHub is continually changing. The current changes in philosophy and design are described here, including two major changes made to the Motif/X Windows DataHub interface. Additionally, because of the drawbacks of the current implementation in label or file recognition, an expert system is being investigated. Finally, short and long term design and implementation issues are discussed.

Subject of the paper is the investigation of the information content of high dimensional multispectral remote-sensing measurements in the VIS-NIR region for ocean-atmosphere problems. The final goal of such measurements is the separation of atmospheric influence and the retrieval of detailed information of water constituents. Primary questions appearing during interpretation process are: how many independent parameters can be found from the measurements? what is the physical sense of these parameters? what is the accuracy of the parameters? One has to take into account that the properties of the measuring device, like channel position, bandwidth, number of channels and measurement accuracy have a great influence on the interpretation. One possible method to get answers to the above questions is the Principle Component Analysis (PCA). A problem in PCA is the physical interpretation of the mathematically obtained results - Eigenvalue, Eigenvector and Principle Components. Because the results of PCA interpretation depend on the statistical properties of the measurement data, they must be mapped back to the absolute measurement quantities (radiances). To get a physical interpretation of the PCA results a detailed investigation with a simulated data set using a simplified (but nonlinear) model was realized (atmosphere after Gordon, Sturm, water reflectances after Sathyendranath, Morel, Prieur). It will be presented a concept, how in-situ measurements can be involved into interpretation model with PCA.

Geometrical registration of two images is nowadays a current and an important step in remote sensing before any further processing and interpretation of the data. Geometrical registration of images with a different ground resolution is useful for a better comprehension of dynamical processes (i.e. deforestation, desertification), as well as for extrapolating models of interpretation obtained on small regions to large areas. We present in this paper a procedure for a fully automatic registration of remotely sensed data based on the multiresolution decomposition of the images with the use of the wavelet transform. We have successfully applied this technique to register images having the same ground resolution, (SPOT HRV and LANDSAT MSS), and different ground resolutions (SPOT HRV with LANDSAT MSS, LANDSAT TM with SPOT HRV).

A new edge detection algorithm has been developed and implemented that has both good speed and accuracy properties. The accuracy of the approach is derived from scale adaptation through anisotropic diffusion. The speed of the filter is based on recursive filtering. Specifically, the algorithm is implemented through a decomposition of the recursive filter into a convolution of two independent half space filters. This further improves the speed of the recursive filter and facilitates scale adaptation. The resulting algorithm is 1 - 2 orders of magnitude faster than comparable Gaussian based frequency domain and spatial domain methods. The new filter is omni-directional and super-elongated. It is also contour following, has computational complexity which is independent of scale, and has no truncation noise. The algorithm has been implemented and successfully applied to SPOT XS and Landsat MSS and TM imagery as one component of a region based image segmentation scheme.

This paper describes two examples of the use of volumetric data visualization techniques for the exploration of remotely sensed hyper-dimensional data. The first example presents an Airborne Visible IR Imaging Spectrometer hyperspectral data set, which is organized as a stack of 2D slices, as a 3D cube. The second example extends the investigation by using volume-rendering algorithms to extract features from a multi-temporal Advanced Very High Resolution Radiometer data set.

The ATLAS (Airborne Terrestrial Applications Sensor) is a 15-channel multispectral scanning imager, currently under development for NASA's Commercial Remote Sensing Program Office. The sensor package utilizes a rotating linescan mirror, and a modified Dall-Kirkham telescope with a 7.5-inch clear aperture and 2.0 mrad ifov. Scan rates are adjustable from 6 - 50 rev/sec, with a total of approximately 73 degree(s). Three spectrometers with grating dispersive elements are used to provide wide spectral coverage. Blackbody sources (hi/lo) and a modified integrating sphere source, built into the scanhead, are employed to provide in-flight radiometric calibration data for quantitative inference of ground scene temperatures and radiance values. In this paper, the overall design of the ATLAS scanner system is reviewed. Results of spectral response, NETD, NER, and MTF calibration measurements are presented for each channel. Additionally, some initial ATLAS flight test data and analysis are described, including SNR, uniformity of TIR data over water, overall image quality and other results.