The development of hyperspectral remote sensing has been conducted in the middle 1980s. Since then a number of hyperspectral imagers has been developed. The spectral response of it has covered the visible, near infrared, short- wave infrared and thermal infrared regions. In some time of the technical developments, the applications in various areas have also been conducted. For mineral exploration, the difference minerals as well as the alteration zones have been analyzed and extracted. Hyperspectral imagery enabled the detailed classification of vegetation. In case of study of wetland environment, some subspecies of wetland vegetation have been distinguished. Based on the spectral characteristics of vegetation and some artificial objects the urban environment and mapping have been tested by using of the hyperspectral remote sensing integrated with high spatial resolution data. In the studies the thermal infrared multispectral remote sensing has also been developed and dealt with the geological application.

Airborne imaging spectral technology, principle means in airborne remote sensing, has been developed rapidly both in the world and in China recently. This paper describes Modular Airborne Imaging Spectrometer (MAIS), Operational Modular Airborne Imaging Spectrometer (OMAIS) and Pushbroom Hyperspectral Imagery (PHI) that have been developed or are being developed in Airborne Remote Sensing Lab of Shanghai Institute of Technical Physics, CAS.

The remote sensing of chlorophyll-a concentration of inland water is troublesome because of the numerous pollutant substances. It was found that the reflectance ratio R(700 nm)/R(675 nm) depends on not only chlorophyll-a concentration but also the scattering and absorption coefficients of suspensions and dissolved organic matter. The absorption of phycocyanin at 624 nm is a very important spectral feature for estimation of chlorophyll-a concentration as well as that of chlorophyll-a. New algorithms were developed, in which reflectance ratio R(600 nm)/R(624 nm) or R(560 nm)/R(624 nm) was used as water quality correction.

Inflight experiment of Modular Airborne Imaging Spectrometer (MAIS) and ground-based measurements using GER MARK-V spectroradiometer simultaneously with the MAIS overpass were performed during Autumn 1995 at the semiarid area of Inner Mongolia, China. Based on these measurements and MAIS image data, we designed a method for the radiometric calibration of MAIS sensor using 6S and LOWTRAN 7 codes. The results show that the uncertainty of MAIS calibration is about 8% in the visible and near infrared wavelengths (0.4 - 1.2 micrometer). To verify our calibration algorithm, the calibrated results of MAIS sensor was used to derive the ground reflectances. The accuracy of reflectance retrieval is about 8.5% in the spectral range of 0.4 to 1.2 micrometer, i.e., the uncertainty of derived near-nadir reflectances is within 0.01 - 0.05 in reflectance unit at ground reflectance between 3% and 50%. The distinguishing feature of the ground-based measurements, which will be paid special attention in this paper, is that obtaining simultaneously the reflectance factors of the calibration target, atmospheric optical depth, and water vapor abundance from the same one set of measurement data by only one suit of instruments. The analysis indicates that the method presented here is suitable to the quantitative analysis of imaging spectral data in China.

The general physical model of high etendue static imaging Fourier transform spectrometer is deduced. On the basis of this model the static imaging Fourier transform spectrometer based on the solid Sagnac interferometer is developed. It covers a spectral range of approximately 0.4 micrometer to approximately 1.0 micrometer with 100 resolvable spectral bands. The experimental results are presented.

Airborne Imaging Spectrometer data were acquired over zhangjiakou area in May 1997. To identify different types of rocks of the area by using the images, visible and near infrared reflectance spectra of main types of rocks in the area are analyzed. Based on spectral features of rocks, some spectral indexes and bands be extracted to combine different color-composite images for mapping lithology. Some keys for lithological discrimination are built according to the spectral performances of images. Geological units are identified and mapped at 1:50000 scale. Comparing the results with previous geological map, modifications have been done about the boundaries and attributes of some geological units, and some new geological contents have been added into the geological map. Our results demonstrate that imaging spectrometer images contain richer lithological information than the images acquired by other remote sensors, and can discern subtle spectral differences between different rocks which is difficult to do by using TM images or infrared photograph etc. It is proved that imaging spectrometer data are effective and efficient in mapping geological unit at larger scale.

In this paper, we reported some research result in applying hyperspectral remote sensing data in identification and classification of wetland plant species and associations. Hyperspectral data were acquired by Modular Airborne Imaging Spectrometer (MAIS) over Poyang Lake wetland, China. A derivative spectral matching algorithm was used in hyperspectral vegetation analysis. The field measurement spectra were as reference for derivative spectral matching. In the study area, seven wetland plant associations were identified and classified with overall average accuracy is 84.03%.

Selecting the best classification bands from hyperspectral images for particular remote sensing application is one of the most important problems in utilizing hyperspectral images. In this paper, the best classification bands selection problem is regarded as optimal feature subset selection problem and the bands in original bands set are divided into redundant and irrelevant. In order to eliminate these two type bands, a multi-level optimal classification bands selection model from hyperspectral images based on genetic algorithm and rough set theory is proposed. Through the initial two steps of the multi-level model, the dimension reduction step and the genetic algorithm based filter step, most of redundant and irrelevant bands are deleted from the original images bands set. From the machine learning perspective, the multi-level model can take both advantages of the filter and wrapper models.

Quality and interpretability of airborne imaging spectrometer data generally are degraded by two types of radiometric distortions. The first is called edge radiometric distortion or radiometric distortion in scanning direction which is related to the larger viewing field of remote sensor. During the period of data acquisition, some parameters, such as path radiance, atmospheric attenuation and the sun-object-sensor geometry, continuously change with viewing angle, which causes a severe radiometric distortion on airborne imaging spectrometer data in scanning direction. Additionally, non- Lambertin reflection of ground objects and relief of terrain strongly complicate the radiometric distortion. The second is random noise which may be caused by dark current of remote sensor, dust and other environmental factors. It is impossible to do data processing and quantitative analysis if radiometric distortions of data are not corrected properly. This paper discusses and analyses various factors which cause radiometric distortions of airborne imaging spectrometer data. Two new methods based on wavelet transform are developed to correct edge radiometric distortion and remove random noises of airborne imaging spectrometer data. Experimental results presented in the paper illustrate that proposed methods are more practical, effective and efficient for radiometric restoration of airborne imaging spectrometer data.

During December 20 - 28, 1995, an Airborne Imaging Spectrometer (AIS, 30 bands, spectral resolution 20 nm) was flown to investigate Poyang Lake area, a famous wetland, in China. The ground spectra of various materials were also collected from the area at the same time. All the data acquired from this area were used to conduct a comparative study of different calibrations for image data. In this paper, three different calibration models were intensively studied and applied in the calibration of AIS remote sensing data acquired from Poyang Lake area. The advantage and drawbacks of the three calibrate models have been analyzed and examined. The calibrated results made by authors have demonstrated that the Empirical Linear (EL) calibration model is the best model among the three model. The retrieved spectral results of soils and vegetation from the EL calibration model were correspond field spectra very well. However, authors also found the deficiency of the EL model was that the calibration sites had to selected carefully, its choice had an important effect on the quality of retrieval results.

When the input slit in the spatially modulated imaging interferometer is placed in front of or behind the focal plane of the Fourier Transform lens, distortions arise in the interferogram. The relation of the spectral resolution and the off-focus length is derived.

To prospect large or superlarge gold ore deposits, blending of remote sensing techniques and modern metallogenitic theories is one of the effective measures. The theory of metallogeny plays a director role before and during remote sensing technique applications. The remote sensing data with different platforms and different resolutions can be respectively applied to detect direct or indirect metallogenic information, and to identify the ore-controlling structure, especially, the ore-controlling structural assemblage, which, conversely, usually are the new conditions to study and to modify the metallogenic model, and to further develop the exploration model of large or superlarge ore deposits. Guidance by an academic idea of 'adjustment structure' which is the conceptual model of transverse structure, an obscured ore- controlling transverse structure has been identified on the refined TM imagery in the Hadamengou gold ore deposit, Setai Hyperspectral Geological Remote Sensing Testing Site (SHGRSTS), Wulashan mountains, Inner Mongolia, China. Meanwhile, The MAIS data has been applied to quickly identify the auriferous alteration rocks with Correspondence Analysis method and Spectral Angle Mapping (SAM) technique. The theoretical system and technical method of remote sensing strategic exploration of large or superlarge gold ore deposits have been demonstrated by the practices in the SHGRSTS.

Although characteristics of visible and near-infrared spectra (0.4 - 2.5 micrometer) of rocks are dominated by electronic processes and molecular vibration processes that is related to chemical composition and structure of rocks, external behavior of spectra of rocks is adjusted by some factors, such as environmental conditions, physical structure of surface and viewing geometry, etc. The spectra of rocks can vary with the changes of these factors to a certain extent. It is obvious that variability of spectra of ground objects would degrade the interpretability of remote sensing images, and especially degrade the effectiveness of techniques of discriminating rocks based on remote sensing spectral features and spectral indexes to a large extent. To discriminate earth surface mineralogy and lithology more effectively and efficiently by remote sensing, especially by hyperspectral remote sensing, requires a good knowledge of variability of spectra of rocks. This paper describes and analyzes variability of reflectance spectra of rocks sampled in Zhangjiakou district of Hebei Province of China, under different light incident direction and incident angle, viewing direction and viewing angle, surface forms of rocks, grades of samples, and types and extents of weathering, etc.

The bidirectionality of water light field is one of the error sources of the ocean color models that now commonly used to retrieve pigments' concentrations. As the launching of new generation ocean color sensors, e.g. SeaWiFS, or the coming MODIS, the goal of accuracy of ocean color sensing is now much higher than that of CZCS. The original minor factors, such as the bidirectionality of water-leaving radiance, become more important for retrieving algorithms. This paper is to give some results based on 3D Monte Carlo simulations and proposes some aspects that should be considered for in situ optical data collection and related remote sensor's calibration and validation activities.

The lake beach and grass moor land is a kind of typical wetlands. The area varies greatly with season in Poyang Lake region. Moreover, the field investigation of wetlands is almost impossible as geographical features and difficulties in transportation. The notes address the potential role of remote sensing in the surveying of the lake beach and grass moor land. In particular, the notes reflect the characteristics relationships between MOS-1 MESSR image and the wetlands. The application results show that MOS-1 MESSR image is effective in surveying the wetland area variation and distribution (lake, river, grass moor, mud flat, sand beach, etc.). detecting lake base shape, and analyzing eco-environment surrounded.

We had a synthetic survey and test to the Yellow River water on the Ke-Ling I ship near the mouth of the Yellow River, which includes the test to the Yellow River water spectrum reflectivity by using WFH2.5 near infrared spectrometer and the water sample collection. In the laboratory we analyzed the water sample by the photoelectric method and obtained the percentage of sample's silt diameter to the silt weight, middle-number silt diameter and silt content etc. Besides, we also obtained the spectrum reflectivity curve of the relevant water sample by the calculation.

In this paper we will first introduce the basic theory and algorithms of wavelet techniques, and then will try to use it to process the remote sensing spectral data acquired by IR Fourier transform spectrometer, we also discuss that self- adaptive wavelet network applied to classify the spectra. The experimental results show that wavelet techniques is very efficient in processing and classification of spectral data.

In vegetation district, as a result of that the plant absorb Au and associated elements of locks and soils, the contents of Au and associated elements, pigment, water, surface temperature, and cells structure in leaves occur variations, even the plant cause abnormality of biogeochemical effect, thus it appear abnormality in the spectral information of TM image. This paper takes the Hetai gold deposit in Guangdong as researh target. The results proved that Au content of leaves in the gold deposit is as high as the background value, Au content in chlorophyll of leaves is 2 - 8 times higher than that, the contents of the chlorophyll a and b as well as the carotenoid are also higher than those of the control area, the water content and surface temperature of leaves are lower than the background value, the cells structure in leaves are deformed and broken, the spectral reflectance of the leave is 5% - 30% higher than that in the reference areas, the spectral wave shape is more shifted toward the short wavelength. According to these feature and difference, the remote sensing image processing technologies are utilized to extract the abnormal feature of biogeochemical effect, and enclose the remote sensing abnormal areas of mineral; Gold exploration was thus conducted in the surrounding areas of the Hetai bold deposits in Guangdong and two remote sensing abnormal areas of gold mineralization were rapidly evaluated in this way. The research proved that the remote sensing technology is a rapid, accurate and economical method of extracting abnormal information about the biogeochemical effect in vegetation and applying it to the mineral exploration practice, and this constitutes a new way in exploring hidden ore deposits in vegetation regions.

In this paper, an extension matrix based rule inductive learning algorithms have been presented. Exception the introduction of this rule inductive learning algorithm, we proposed a novel algorithm for discreting continuous valued attributes which is essential preprocessing step for applying symbol rule inductive algorithms to remotely sensed data analysis. Some initial results are finally given which can demonstrate the advantages of rule-based classification.

As the growth cycling of crops is usually shorter than the one of forest. The change of status of crop growth was pronounced and quickly with development stages. Therefore it is useful to know the change of the features of crop growth during growth cycling in hyperspectral remote sensing. The reflected spectra of canopies of five crops, early rice, later rice, summer maize, cotton and soybean were measured with SE590 spectrometer from June to September 1997, in Beijing. The features of first-order derivatives of canopy reflectances for each of the five crops are changed regularly with the phenological evolution. The possibilities of identification of crop types with the features of red edge are discussed.

A new method for classifying hyperspectral remote sensing data in urban area is described in this paper, that combines the edge detection and spectral analysis together. Due to the varied surface scene in city and limitation of imaging spectrometer's signal-to-noise ratio, normal classification based on pixels were not satisfied for thematic classification and mapping in urban area. This technique will improve such situation, which perform every classes with not only their similar spectral features but also the object's geometric and edge characters.

The Australian Resource Information and Environment Satellite (ARIES-1) will offer improved opportunities to map the mineralogical composition of outcropping rocks, hydrothermal alteration zones and the regolith on the Earth's surface in greater detail than before. This paper presents simulated ARIES-1 mineral mapping results for the Comstock mining district, Nevada, USA, and demonstrates how the spectrally- derived information on hydrothermal alteration has been validated through field work and supporting analytical studies. In the Comstock mining district, three types of white mica with short, medium and long Al-OH absorption wavelengths, respectively, were identified based on the simulated ARIES-1 hyperspectral data. Laboratory analytical work confirmed that the low, medium and long Al-OH absorption wavelengths correspond to low, medium and high octahedral Fe+Mg content, respectively, of the white micas. Each of the three types of white mica tends to occur in a particular alteration zone. The spatial distributions of the three types of white mica may have implications for interpreting the hydrothermal alteration processes. The white mica of short wavelength Al-OH absorption, proximal to Au-Ag mineralization, formed in acidic hydrothermal conditions both at the earliest and the last alteration stages. The white mica of medium wavelength Al-OH absorption, characteristics of the propylitic zone surrounding Au-Ag mineralization, was produced mainly in the second, low- sulfidation hydrothermal event. The white mica of long wavelength Al-OH absorption was probably formed in multiple stages, and generally occurs distal to Au-Ag mineralization.

ASTER Program is a joint program carried out by MITI/ERSDAC and NASA to contribute the extensive understanding of the Earth as a system through NASA's EOS Program. ASTER, the successor to the Japanese Earth Resources Satellite-1 (JERS-1) launched in 1992, is a high-resolution multi-spectral imaging sensor covering from visible to thermal infrared region. ASTER will be launched in 1999 as a one of the sensors of the NASA's EOS AM-1, the first satellite of the EOS program. Data acquisition is executed based on the schedule established using scheduler taking the priority of the acquisition request into account The algorithms for science data products are developed by ASTER Science Team composed of Japanese and U.S. scientists. ERSDAC coordinates the public solicitation of researchers to use ASTER data, called ASTER Announcement of Opportunity (ASTER AO). Any and all researchers, scientists, and research organizations are welcome to submit research proposals for non-profit and peaceful purposes to use ASTER data. They will be privileged to obtain a certain amount of ASTER data and to submit the data acquisition requests (DAR). ASTER data will be distributed to all users on a nondiscriminatory basis for peaceful purposes in consistent with national laws and regulations.

For climatic, hydrological, and snow hazard investigations, the snow covered area is one of important parameters. Moderate Resolution Image Spectrometer (MODIS) is an EOS-A facility instrument to be launched in this year. With much richer spectral and more frequent temporal measurements by MODIS, we expect that daily or weekly global snow cover information can be obtained. The objective of this study is to validate the accuracy of the MODIS snow mapping algorithm under a variety of viewing and illuminations, land cover, atmospheric and terrain conditions. We have collected a huge amount of AVIRIS image data, from which we can simulate the MODIS spectra measurements. The high resolution color infrared photo taken simultaneous as AVIRIS image data can be used to access the snow covered area as the ground truth. Through examples, we demonstrate the techniques used to access the ground truth and validation of MODIS snow mapping algorithm.

A major problem in operational land cover change detection using remotely sensed data is to separate the change signals caused by land cover changes from those due to vegetation phenology. This study provides an approach to this problem by systematically analyzing the spectral properties of major land cover change processes and the phenological profiles of different land cover types. The phenological profiles were derived from a global data set consisting of a full phenological year's data record of the 1 km monthly composites from the Advanced Very High Resolution Radiometer (AVHRR), while land cover change signals were simulated from the spectral signatures of corresponding land cover types in different seasons. A decision tree method was used to derive the decision rules that provide best separation between the change signals of land cover changes and vegetation phenology. These decision rules were referred to as land cover change trajectories. A complete set of change trajectories was developed for the globe in all seasons of a phenological year. Results from this study indicate that during most seasons of a phenological year, major land cover change processes including deforestation, denudation, revegetation, flooding, flood receding and vegetation burning, can be separated form one another and from vegetation phenology in the red-near infrared space. The derived trajectories, when validated, can serve as a theoretical basis for developing operational change detection algorithms.

During the fire-hazardous seasons 1995-1996, work was done to determine the efficiency in the use of the AVHRR/NOAA data to detect forest fires on East Siberia's territory. Promising findings of these activities were instrumental in setting up a space monitoring system for forest fires and integrating it into a common system for forest fire control. An algorithm was developed for detecting forest fires, which shows a reasonably good performance on Siberia's territory.

Generation and sample applications of an integrated set of multispectral remotely sensed products for investigations of Lake Baikal's ice cover variability are described. Different ice cover classes and unfrozen water distributions are estimated from calibrated and navigated AVHRR 1.1-km imagery of Lake Baikal for January 1994 through May 1998. The processing strategy and characteristics of the products are reviewed. The utility of this type of multiparameter data set for modeling applications and process studies is discussed.

The concept of imaging spectrometry was originated from geological communities in the early 1980s. The Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) built at NASA Jet Propulsion Laboratory has been collecting spectral imaging data in the 0.4 - 2.5 micron region from an ER-2 aircraft at 20 km for a decade. Newer generations of airborne and spaceborne imaging spectrometers have also been built. In this paper, techniques, such as spectral matching and channel ratioing for extracting information on the Earth's atmosphere and surfaces are illustrated. Examples of using the technique of imaging spectrometry for remote sensing of atmospheric water vapor, cirrus clouds, aerosols, fire, land surface properties, and ocean color are presented. The applications of imaging spectrometry to the selection and implementation of new channels on the Moderate Resolution Imaging Spectroradiometer (MODIS) for global remote sensing of atmospheric water vapor and cirrus clouds from space are also described.

We introduce novel robust filtering algorithms applicable to image and signal processing in the remote sensing applications. They were derived using RM-type point estimators and the restriction technique of the well-known specific for image processing KNN filter. Novel RM-KNN filters effectively remove impulsive noise while edge and fine details are preserved. The proposed filters were tested on simulated images and radar data and were provided excellent visual quality of the processed images and good quantitative quality in the MSE sense over standard median filter. Recommendations to obtain best processing results by proper selection of derived filter parameters are given in this paper. Two derived filters are suitable for impulsive noise reduction in the remote sensing image processing applications. RM-KNN filters can be used as the first stage of image enhancement following by any non-robust techniques such as Sigma-filter on the second stage.

The varying fire frequency in the Santa Monica Mountains provides unique opportunities for examining the impacts of disturbance and succession on community composition and ecosystem function at the landscape scale. Fire resets the 'successional clock' of this evergreen-dominated ecosystem and favors early successional annuals that tend to have higher photosynthetic rates and leaf transpiration rates than mature, long-lived woody perennials. Hyperspectral sensors like the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) offer promising opportunities for remotely detecting these dynamic physiological properties in changing landscapes. Specifically, we expect altered photochemical reflectance index (PRI) and water band index (WBI) values for stands in early succession following fire. PRI and WBI from Santa Monica Mountains AVIRIS imagery indeed show complex patterns varying with season and successional state, possibly revealing varying photosynthetic activity in these dynamic, fire-prone landscapes. Further work is attempting to also consider the impact of changing canopy structure and vegetation type on physiological properties detectable with hyperspectral imagery.

Satellite data was regarded as a main information source by many key projects in forestry for the past 20 years. such as the comprehensive inventory of remote sensing in "Three north" protection forest, monitoring and assessing on forest fire, evaluation Yunnan provincial tropical forest in southwest China, assessing forest ecological effect, preventing desertification, monitoring insects and innovating on current forest resources survey system, etc. Under the support of computer techniques and models, a relationship was established between remote sensing data and forest environmental factors, so they can express and estimated with quantitative methods. The satellite data being used include TM.. SPOT NOAA ERS-1 and Radarsat. This paper aims to introduce the application ofsatellite remote sensing data in Chinese forestry, especially the methods of estimating for growing stock volume, appraising the efficiency of using remote sensing data, estimating forest fuels with remote sensing, discovering early and assessment on forest fire, as well as brief introduction of remote sensing application in ecological forestry forest environment and forestry division. All ofwork mentioned in this paper is special achievements of Chinese forestry experts.

The optimum atmospheric correction and reflectance retrieval proced'ure would be based solely on the digital image and would require no in-situ field measurements during the satellite overflight, not only for operational applications but also for many research projects that use historical data sets. In this paper, The image-base atmospheric correction and reflectance retrieval approach was developed by the SPOT data about La Crau radiant calibration site and nearby research area in july, 1995. The approach was based on DOS method to calculate path radiance with the reasonable dark-object reflectance choice. The atmospheric transmittances , the downwelling atmospheric irradiance and atmospheric attenuation was infered from several atmospheric radiative transfer codes(Lowtran-7, Modtran-3, 65) with simulated atmospheres. At last, the aprroach was evaluated by comparing surface reflectances of La Crau site measured with one derived from SPOT data.

Image registration is needed in multi-spectral image fusion when images are taken from different viewpoints, or at different times, or by different optics systems. For multispectral images with global 2D distortions, such as translations, rotation, and scaling, we present a new image registration algorithm using local template matching. After preprocessing the multi-spectral images to make them fit to the task, we employ the local maximum or minimum energy points of wavelet transformation data as feature points of local multi-resolution templates. In coarse matching approach, the affine transformation to template image needed in traditional template matching method is avoid. Experimental results show that this algorithm is efficient and robust.

Keywords: Image registration, Template matching, Algorithm, Wavelet transformation, Multispectral imag

This paper discusses procedures and techniques involved in a research project on present land use investigation by using Landsat data and sampling technology. which has a special emphasis on present land type stratification. sampling procedures fit for the application of satellite remote sensing. sample design. database management. statistical summary and extrapolation. statistical analysis and precision evaluation. A sampling system fit for investigation and monitoring of agricultural land use type. A framework for stratified sampling was achieved through the detennination of parameters. like sample size, sampling rate. sample number and location. The methodology was extended to be used in the •whole Sichuan province in the scale of I : 100,000. which is classified as 7 strata or 25 sub-strata. has I 18 samples of IOKmX 10km. and 2.00/t in tenns of sampling rate. Project results suggest that the method is feasible. operational. repeatable and effective both in time and cost. The accuracy is acceptable for macro-scale survey in moderate or small mapping scale, especially with well stratified strata and extensive incorporation of remote sensing and GIS. The methodology is detailed in this paper.

The paper presents a method of automatic extraction of multiangular data from multiangular remotely sensed sequential images. In the procedure of extraction, the key problem is multiangular remote sensing registration. The non-linear resampling technology is used to correct systematic geometric distortion caused by wide-angle imaging. We present a method to extract enough ground control points (GCPs) for multiangular image matching. The extracted GCPs are distributed uniformly. One of the problems in multiangular image registration is brightness variation due to differences in viewing angle, bi directional reflectance variation and atmospheric scattering. An advanced weighted template matching method is presented to remove the differences and the best matching result has been achieved. For the images with local non-linear geometric distortion, surface-fitting function is used as mapping function to improve the registration accuracy. Multiangular data is from directional information of the same target in multi-images. We designed a schedule to register multiangular images sequentially by using interleaved-angular image. We use POLDER airborne simulated multiangular images, which were provided by CNES of France , Aerial data by IRSA, CAS. And have obtained differential angle registered data of targets. The experiment result shows that the method we presented has optional accuracy in practical use for multiangular data analysis.

The paper describes an on-going project on the potential study of Landsat TM for the monitoring of wetland resources with a concern ofpeat deposit. The method is to mcoiporate remotely-sensed data and ground-based GIS data to improve the accuracy of automated interpretation. The main systems used in the project are ER MAPPER (UNiX VERSION) and ARC/iNFO. The study area is Ruergai plateau in the southwest of China. Preliminary results show the following improvements. (1) The outlines of the swamp are clearly drawn out by incorporating geomorphical data and the accurncy is reasonable. (2) The estimation of peat deposit could be unproved with the DEM data. (3) The whole procedure can be easily and automatically repeated when new data are available.

Topographic effects are the main obstacles to further analysis of satellite spectral data in mountainous area, especially to quantitative remote sensing. To obtain the true reflectance ofthe land surface, we must remove the topographic effects first. A physical model for rugged terrain is developed in this paper. Both atmospheric and topographic effects are considered in the model. 3 illumination sources are expressed analytically which include: 1) direct solar irradiance; 2) diffuse sky irradiance; 3) reflected irradiance from the adjacent terrain. Based on a quick searching algorithm for local horizon. the most complex part in the model—the reflected radiation can be calculated throughout the whole image in an economic computation time. In stead of using the field measured data or standard atmospheric condition obtained from the commercial software, we invert the atmospheric parameters from the image itself based on stochastic programming theory, a priori knowledge is used in the inversion process. To test the model, a Landsat TM-scene is matched to a digital elevation model(DEM) which has a resolution of Im for elevation. The true reflectance map is obtained from the model. It is found that most ofthe topographic effects are removed in the map.