The results are presented of five in-flight absolute radiometric calibrations, made in the period July 1984 to November 1985, at White Sands, New Mexico, of the solar reflective bands of the Landsat-5 Thematic Mapper (TM) . The 23 bandcalibrations made on the five dates show a ± 2.8% RMS variation from the mean as a percentage of the mean.

A technique for radiometric normalization of images of the same area acquired on different dates is discussed. The technique accounts for illumination and atmospheric variations between the scenes. Statistical analysis of segmented scene elements is used to generate band by band transformation functions. Examples are presented where Thematic Mapper subscenes (512 x 512 pixels) from 1982 and 1984 are normalized for each reflective band. Analysis of the transformed images indicates that the residual radiometric error after the transform is of the order of one reflectance value. Examples of the utility of the technique for change detection, environmental assessment, and as a general preprocessing technique for scene classification and analysis are discussed.

Raw TM data of a single scene of southern England without PCD or MSCD have been transformed to British National Grid coordinates using 101 GCPs. For a whole scene in one transformation, third order is the minimum acceptable for raw data of this state. The standard deviation of the residuals was reduced to sub-pixel level using third or fourth order polynomial transformations, but the 90% level is about 40 metres or 1.3 pixels, allowing accurate planimetric mapping to scales of 1/100 000 and smaller. System corrected data have been shown to be marginally better, and under the most favourable conditions both raw and corrected data may be marginally suitable for mapping at 1/50 000.

The Canada Centre for Remote Sensing (CCRS) has been using the scene-statistics approach, originally implemented in 19751 for destriping LANDSAT MSS data, for the routine correction of LANDSAT Thematic Mapper (TM) reflective band data. The 0.8% within-scene radiometric accuracy specified for Canadian production systems has been achieved for all but a small percentage of requested scenes2. A significant functionality of the processing methodology includes the correction for dark signal (DC) level shifts as a function of both detector and scan line number, by utilizing background reference level measurements available during the calibration period of each mirror scanning sequence. Additionally, forward and reverse mirror scans are treated independently. The effect of extended, bright targets such as snow, ice or clouds, is to cause a lowering of the background level by as much as 4 digital numbers (DN), with recovery times equivalent to approximately 2000 pixels. Hence, in the vicinity of the target itself, localized banding in phase with forward and reverse mirror sweeps is evident. A two-fold impact to the CCRS processing system occurs when the bright targets are at the edge of the scene, since the background reference level measured during the calibration period does not represent the offset for the imaging period of the sequence. The effects are, firstly, a more pronounced banding in those scan lines which image the bright target, and secondly, a forward/reverse banding superimposed over the entire scene, causing these scenes to fail the accuracy criterion. An automated procedure has therefore been adopted to detect when the forward/reverse banding would be outside specifications using the standard techniques. In such cases, forward and reverse mirror sweeps are then corrected with identical gain and offset correction parameters and the DC level shift corrections are applied.

The radiometric performance of the Landsat Thematic Mapper instrument has been demonstrated to be within target specifications although additional system corrections can be modeled and corrected for. This paper characterizes two radiometric error sources, the so-called "scan-correlated shift" and "memory effect" from measurements performed in the shutter obscuration region. The variability of the reference zero radiance background level is described as a function of minor-frame position, detector, scan and scene.

A multitemporal TM data set (scene 191/31/2, Rome) has been analyzed in order to determine the best image chip characteristics for use in a ground control point (GCP) library. We have attempted to identify the optimum selection of image features, bands, chip size, and season of the year for the reference image. The value of correlation coefficient between chips (small subimages centred at the GCP) of the same feature, but acquired at different times, is taken as a component to measure chip quality. Also, the high contrast in a chip is taken to indicate better quality of the GCP. The "Optimum Chip Factor" (OCF) is a measure of these two aspects of GCP quality. ro determine the optimum bands, chip size and features, the OCF was computed in different spectral bands, over different window sizes and for different features. The best season was found by taking each acquisition date as a reference image and comparing the number of successfully relocated GCPs in all other images. The results can be taken as a guideline for the use of GCPs in a GCP library. We found that the best band is band 5, the best features are features with land/water boundaries, and the best chip size is 15 pixels square. To determine the best season of the year, some further investigations will be necessary.

An operational quality control system has been implemented at the European Space Agency Landsat stations, Earthnet Programme Office. This automatically checks the characteristics of every single TM product before distribution to users. A record of the results of quality control is kept in order to trace back any problem identified and define statistically the performance of the sensors. Examples of the behaviour of some of these parameters are presented,and the benefits of such an approach are discussed.

The pre-flight and on-board in-flight absolute radiometric calibrations of the two HRV cameras on SPOT-1 are briefly described. The results of the in-flight calibrations at White Sands, New Mexico are presented in detail and compared to the other methods. The ratio of calibration coefficients for the two SPOT-1 HRVs was obtained from a histogram match of a scene, imaged simultaneously by the two cameras. This was compared to the ratio of the calibration coefficients from the White Sands results. The ratios are in good agreement, < 3% in three bands, the exception is multispectral band 2 which shows a 7% difference. The histogram result and a comparison with spectral radiances measured from an altitude of 3000 m indicate that, at the time and under the conditions that the calibrations were madelthe specification for a ± 10% RMS absolute calibration uncertainty for the SPOT-1 HRVs had been met. The White Sands and histogram ratio results were used to update the calibration of the on-board suncalibrator system for the HRVs.

The radiometric resolution of SPOT imagery is limited by two independant phenomena : -the instrument noise generated along the image columns by the CCD arrays detectors, the electronics and the quantization process -the equalization defaults due to the fact that each CCD array detector has a different sensitivity and dark current This effect is taken care of by image radiometric preprocessing at the ground segment level ; the residual errors and imperfections may be considered as an additional noise generated along the images lines. During SPOT post launch performance assessment, the noise along columns is estimated by recording the signal delivered by an onboard calibration lamp on each detector over a several seconds period. The determination of the so-called "relative calibration coefficients" which permits the equalization preprocessing is made by analyzing the signal delivered by the calibration lamp The calibration device is designed so as to project the light of the reference lamp onto the detectors through the whole telescope optics. The characteristics of this device are measured on ground and controlled after launch by testing several technological parameters. An independant estimation of these calibration coefficients can be made by imaging areas of high and uniform reflectance such as large and flat snowy surfaces. This paper presents the results of these various investigations.

This paper presents the method used to check the viewing geometry of the SPOT satellite after the launch. Moreover, the results obtained during the first part of the post launch assessment period arc given for the various parameters which specify the geometric image quality. The relevant estimation methods are also described, and the specifications are recalled.

The SPOT MTF is tested during the post-launch performance assessment using a photointerpretative method. To this end an image catalog is prepared before launch. This catalog is based on aerial high resolution photographs of 20 cities in the Southern part of France in the panchromatic band. Each photograph is subsequently digitized and sampled at a 10 meter resolution, and for each original photograph a series of 9 images with various and known MTF is obtained through parametric convolution. SPOT MTF is then estimated from a visual comparison between SPOT images and images from the reference catalog. Moreover an algorithm of evaluation of high spatial frequency content is applied to a set of suitable SPOT images, when the two HRV instruments are in parallel viewing, to allow direct comparison between the two instruments MTF in the panchromatic band. This paper presents a detailed description of these methods and their results.

In order to take advantage of all the capabilities of SPOT system, we developed new methods for mosaicking SPOT images and increasing the ground resolution of multispectral data. The mosaicking method uses the capability of acquiring data with HRV instruments in twin mode of operation (overlap of three kilometers between the two data strips) to obtain an image size of approximately 117 by 110 kilometers. To reduce the processing time, we do not use a general mosaicking method, but a simplified one which takes into account the radiometric and geometric performances of the SPOT satellite. One of the major SPOT advantages is to provide data with a 10 meters resolution. In order to take fully advantage of this capability, we designed a process which can produce multispectral SPOT images with a 10 meters resolution. To achieve this goal, we had to design special geometric corrections which can insure a perfect superimposability of the Panchromatic and multispectral data. An other exposed problem is linked with the mixing method to be used, in order to insure a good radiometric result. These products have been developped on an operational way, in order to be sold in a near future by SPOT IMAGE.

Photogrammetry consists in taking a photo from the same part of the ground with the same camera but from two different places and then use this pair of pictures for stereoplotting Stereoplotting consists in placing the stereopair of photos into the two projectors of an optical and mechanical stereoplotter with the same interior orientation as the camera used Interior orientation of the camera is measured by calibration. Then relative and absolute orientation of two projectors are done. This mean orientate the two projectors into the same relative and absolute position angles as were the camera during the shot. This is done taking at least three control points for hight and two control points in X and Y to match will the map manuscript plane. Nowaydays optico-mechanical plotters which have some limitations and are very long and difficult to operate are being replaced by analytical equipment which give : - better accuracy due to - simplier construction - digital corrections of parameters such as earth curvature, films deformation, atmospheric refraction lens distorsion. - faster operation typically 15' for model orientation instead of one or one and half hour. - larger field of application, no limitation of : - focal lens of camera, - convergence.

The potential value of the broader spectral coverage and higher resolution of digitally processed Thematic Mapper data to geologic exploration is enormous. Already there are reported successes in improved geologic mapping and the detection of spectral anomalies associated with mineralization and hydrocarbon microseepage.

Detailed ground work has been undertaken to study the correlations among Landsat TM spectral data, lithologic units, geobotanical forest associations, and geomorphic site parameters in the deciduous forests of the Ridge and Valley Province, Pennsylvania. There is a strong correlation between lithologic unit and forest association; this correlation together with a strong geomorphic component is seen in the Landsat TM data.

Spring and summer acquired Thematic Mapper data of a spruce forest growing in southeastern Austria show the normalised differences ND1 and ND3, ratio R41, band differences BD1 and BD3, and the greenness index TMG employing Thematic Mapper bands 1, 3, and 4 best discriminate copper-lead-zinc stress in spruce trees of thirty Thematic Mapper single bands and transformations examined. Stress detection is slightly better in the summer versus the spring scene data.

Satellites can provide cost-effective remotely sensed images. The usefulness of these images is increasing, as the sensors improve with each new satellite. Furthermore, satellite-acquired imagery is in digital form, suggesting the possibility of automating remote sensing tasks such as mapping. To date however, satellite imagery has only yielded two-dimensional planimetric information. With stereo pairs of satellite imagery, the capability for generating the third dimension, height, exists as well. The French SPOT Satellite (Chevrel, Courtois and Weill, 1981) for example, can image high-resolution stereo pairs. Depths are generated from such stereo pairs by stereo matching, normally the task of the human stereo vision system. This paper describes a computer system which automates the process of stereo matching. With this system, the digital equivalent of a contour map (a digital terrain model [DTM]) can be generated automatically directly from digital satellite images. Digital terrain models are quite useful, serving every purpose that a contour map does, and others as well. For example, terrain-dependent parameters such as volumes can be computed easily from DTMs, and DTMs are used in the production of orthophotos. Unfortunately, generating a DTM manually requires many hours. This cost has motivated many attempts to automatically correlate stereo images, without much success. Recent computational vision research has reported some progress in this area (Barnard and Fischler, 1982). This paper describes work that extends these results and applies them to digital satellite images. The result is a system for generating DTMs with significant capability, demonstrated with results from both simulated SPOT images and real Landsat 5 Thematic Mapper (TM) images. DTMs with height accuracies better than 60 metres have been obtained from Landsat images. Before these results are presented, the problem is discussed in detail, and the algorithm which has been developed is described.

Optical remote sensing of coastal and estuarine areas concerns observations of the bottom and watercolumn features. Though the spectral characteristics of the Landsat and. SPOT sensors do not allow detailed determination of concentrations of specific components suspended and dissolved in the seawater, it can provide the data on the depth and bottom type. Modelling of the radiative transfer in the watermass gives evaluation of the upwelled radiance signals detected by the remote sensors in the specific bands. Typical watercolumn and bottom spectral signatures are used as input parameters in a two-flow model. Algorithms are developed relating the optical signals tot the features to be remotely mapped. An optimal bottom depth algorithm should be linear, highly sensitve to the depth variations and insensitive to the watercolumn and bottom composition variations. Analogously, an optimal bottom classification algorithm should be sensitive only to the bottom type variations. Generally, algorithms employing ratios and/or differences between the outputs of the sensor's spectral channels are most favourable due to their capability of canceling out some unwanted surface and atmospheric effects. From the modelling several algorithms are proposed and their (in)sensitivity to the features of interest tested. The results demonstrate the applicability of the Landsat and SPOT sensors for mapping of the bottom depth and type down to 3-20 m, dependently on the watercolumn and bottom composition.

Landsat MSS data (and formerly ERTS) have initialized an euphoric boom to obtain excellent and specific information for the use in regional planning and environmental tasks. Nevertheless, in many cases those expectations have been proven to be overemphasized if one was faced with the concrete problems in this field of application. With some of the first results of a pilot project in southern Bavaria (F.R.G.) using TM-data, we are now able to support the expectation of providing very detailed and geometric accurate planning base data. An object-oriented presentation of spectral features for micro-climatic investigations, hydrological sealing, land use, and indicators in the vitality of vegetation canopies enables us to state a high potential for much more differentiated land information analysis. In this context, it should be pointed not only to the high-resolution capabilities but especially to the properties in the nir and swir channels of TM.

New and important data in such areas of image science as sensors (and their applications in remote sensing), robotics, CAD, holography, etc..., not to mention important developments in our understanding of the human visual system suggest that much can be gained from the careful examination of color vision. Inspired by some of the latest advances in these areas, we attempt to use and optimize some of the principles of color vision for the processing of remote sensing imagery. We also discuss briefly the implications of our results for future applications.

The availability of high quality spectral data from the current suite of earth observation satellite systems offers significant improvements in our ability to survey and monitor food and fiber production on both a local and global basis. Current research results indicate that Landsat Thematic Mapper (TM) data when used in either digital or analog formats achieve higher land cover classification accuracies than Multispectral Scanner (MSS) data using either comparable or improved spectral bands and spatial resolution. A review of these quantitative results are presented for both natural and cultivated vegetation.

The missions of the German Modular Optoelectronic Multispectral Scanner MOMS aboard two flights of the United States Space Transportation System STS demonstrated the feasibility of a novel concept with regard to both technical and scientific objectives. On account of the successful missions a cooperation was agreed between the German Federal Minister for Research and Technology and NASA for comparing MOMS observations with the more familiar operational Landsat-Thematic Mapper data over selected testsites as a means of obtaining some relative measure of performance. This paper summarizes the results obtained and presents the MOMS-02, a further experimental representative in the MOMS program aimaing at the realzation of an operational system for the mid-nineties.

The Modular Optoelectronic Multispectral Scanner (MOMS), a CCD type sensor, was flown on two Shuttle missions, STS-7 and STS-11, for experimental purpose. Two MOMS scenes from STS-11 mission were given to the Earthnet Programme Office, European Space Agency for image quality analysis. However, the Thematic Mapper (TM) data from Landsat 5 which represent the state-of-the-art of current remote sensing technology are the only available auxiliary data. Based on some primitive analyses using the TM data as reference, we can conclude as follows: 1) The registration accuracy between the two MOMS spectral bands is good. However, the algorithm for the radiometric calibration of MOMS detectors must be improved to remove the striping pattern caused by the inappropriate adjustment of radiometric values. 2) The projected ground pixel size of 20 x 20m was degraded by some unknown factors and the actual resolution of the MOMS data is much larger than the projected 20m ground size. For the time being, the geoscientific applications of the MOMS data are rather limited.

This paper describes EOSAT's (The Earth Observation Satellite Company), plans for operating the Landsat System in the Landsat-6,7 era, (1985 to 1995). The paper will provide background information relative to EOSAT, and an overview of the planned configurations of both the space segment (Spacecraft and Enhanced Thematic Mapper), and the ground segment (Space-craft Operations Center, Data Capture Site, EOSAT Processing Center) for Landsats 6 and 7. The planned enhancements to the Landsat-6/7 Thematic Mappers will be described in some detail including the implementation of a Panchromatic Band of detectors providing 15m spatial resolution on both ETM sensors, and the possible inclusion on the Landsat-7 ETM of as many as five bands of thermal detectors with 120/60 meter spatial resolution.