《遥感图像应用处理与分析》专著,由中国科学院中国遥感卫星地面站戴昌达、姜小光、唐伶俐三人撰写,是一本有助于促进我国卫星遥感应用事业进一步发展的、颇具创意的好书。该书上、中篇是作者多年从事遥感应用研究实践中感悟到的有关遥感机理和遥感图像应用处理与分析所涉及的基本原理、关键技术。下篇各章节分别介绍在一些重要应用领域开展遥感图像应用处理与分析研究的进展及其实际效果,写得很有特色。

As a multiplex system of man and nature, city is also a complex geographical entity with a high concentration of population, resources, environmental, social and economic elements. Digital City is an important application of Digital Earth. Constructing three-dimensional graphics in Digital City is one of the essential components of constructing Digital City. This thesis, starting from the aspect of technical application, synthesizes achievements in various scientific researches and briefly illustrates the technique of constructing three-dimensional graphics in Digital City. Furthermore,it puts forward some of the authors' own ideas.

This paper discusses detailed key technology of utilizing 1∶50 000 scale DLG(Digital Line Graphic) data to make topographic map of 1∶50 000 scale. Such as preprocessing of the data, the compilation of note layer on map, the compilation of graphic layer, the compilation of the data, the renewal of data. The paper gives technological processes of producing 1∶50 000 topographic map by utilizing 1∶50 000 DLG data, which offers scientific basis for spatial database transforming topographic map and visual electronic map product.

There are many papers about the relationship between satellite thermal infrared anomaly and earthquake in the world, but most of them study thermal anomalies of surface and air of focus region We could predict earthquakes not only by temperature anomalies of satellite thermal infrared imageries, but also by special abnormities appearing on satellite thermal infrared imageries The special abnormities are produced by stochastic resonance near antenna between background noise,electromagnetic wave from strong earthquake and carrier wave that transmits satellite data They include black holes and diamonds by disturbance The blackholes are black areas of satellite imagery because the standard location table is disturbed The diamonds are white enclosed diamond line on satellite imagery when isolines of temperature are showed because the standard temperature table of gray rank is disturbed They are not real temperatures of cloud layers, and they can appear on the cloud layers that the temperatures are lower or higher The reason for disturbance is because electromagnetic waves are produced by geo stress action during earthquake preparation They have very wide frequencies If the frequency and strength is about carrier frequency near receiving antenna of satellite, it can produce stochastic resonance and result in disturbance of receiving data The observation data of our receiving station in more than one and half years time show that the abnormities obviously correspond to strong earthquakes It adds one new applied field of stochastic resonance of electromagnetic wave impending prediction Through further studies, we can design one kind apparatus used for catching earthquake precursors by stochastic resonance principle.

An earthquake(M6 8) occurred on Feb 24,2003 in Jiashi and Bachu counties of Xinjiang of China This earthquake(epicenter:N39 5°and E77 2°) produced many casualties,over 2000 injured and over 200 died, as well ten thousands of buildings destroy Its occurrence shows again the importance of precise prediction Taking this event as an example, the paper raises a new technique involving 3 elements for predicting earthquake,i e ,tracking each information in element order and step by step It's process as follows:The first step is the precise judgment of seismic intensity (magnitude) According to the “Gutenberg Law”, there existed a linear relationship bet ween the different seismic magnitudes(M) and the sequence number(N),i e , “Log N = a-bM” Because the seismic actions are of self exciting characteristic, so we can base on the historic seismic information and establish a self exciting mathematical model for predicting the forth or the next time earthquake magnitud e According to the result of modeling and calculating to Xinjiang of China, we can precisely predict out that an earthquake of M5 4 will occur in Xinjiang followed by an M6 8 strong earthquake event in 2003 The second step is the precise judgment of risk region A lot of investigations show that there exists heat radiation omen before a strong earthquake,especi ally in epicentral region How to moniter this thermal omen information before earthquake for judging risk region in short term,the paper suggested to use OLR (Outgoing Longwave Radiation) information which comes from the NOAA satellite remote sensing As a result,we found that the most high value of monthly departure of OLR were present at the Jiashi epicenter region in January of 2003,and reached +21 9(W/m 2) which is not only the highest value in whole Xinjiang, but also breaks through the historic record The last step is the precise judgment of the imminent earthquake time A good many of investigations of earthquake in situ shows that the small pertubation in external factors, such as atmospheric pressure etc , can trigger the strong earthquake, especially at the imminence time We found that the pressure vari ograph (24 hours) at Jiashi epicentral region had changed +6 5(hp) on February 21,2003, i e , three days ahead of the Jiashi earthquake (M6 8) The process above mentioned is not only gradually reflecting earthquake preparat ion and evolution, but also gradually revealing it's omen features We believ e this predicting technique on line based on combining satellite remote sensing with ground information to be of far reaching importance.

Beginning from the concept and development of Geographic Information System (GIS), this paper completely and systematically introduces the developing necessities, characters and key problems of military GIS. Due to very strong spatial analysis function of Geographic Information System and it′s broad application fields, GIS should be incorporated into military field to combine with patrol, antiterrorism, dealing with emergency accidents for armed police especially so that armed police can integrate, dynamically acquire and manage all kinds of informations about military and police themes. Concurrently, the enterprise-wide need to improve organizational effectiveness, efficiency, and services of Armed Police Force to protect national benefits and life safety. GIS may be used to analyze a lot of informations on regional environment to help resolving safety problems at best speed. Moreover, based on GIS, the military command automatic application system may support decision-making and bring into play its benefits adequately. Based on Arcview, we built a military GIS Demo version of Armed Police Force, which emphasies functions of network analysis, map plotting and 3D simulation. The demo system consists of six sub-function modules such as the best route design, mobile reinforcement, spatial analysis, drawing and plotting, 3D simulation and satellite image display. Hence, the demo system realized the best route design, nearest infrastructure selection, drawing and plotting of military map, 3D simulation and display of Tian′anmen Square within the second ring road in Beijing. Furthermore, the demo system can provide geographic information guarantee for the command automatic aid system on decision making and realize connection and sharing of military information. It can improve the efficiency and save human resources and materials.

This article firstly introduces the applications of LODs.Then particularly expatiates on the categoriesof LODs-based database expression and the theory of morphing between the levels of LODs that can make themodels switch smoothly.In the application aspect,the LODs technique was firstly employed to several models ofDazhuang Coal Mine,which preferably resolved the problems of visual discontinuity and image leap when ex-ploring the original models.Secondly,the VRML was employed to the LODs of network release models,whichpreferably resolved the problem of discontinuity of models when exploring in IE explorer.

The Metadata Standards of data in geo-information science have been created by many international organizations.But they only lay great stress on the public and common object.For the Geo-Object classes,which are important to the exchanging,searching and sharing web geo-information sources,have not been described in details.Therefore we need an essential metadata standard for integrating searching, exchanging, and sharing all sorts of web geo-information sources on the Internet. In this paper, considering the large amount of Geo-Object description in various sources, we present the GOMS Metadata description approach based on XML.It enables the possibility of the conversion from the specific format to the general machine-readable and machine-understandable metadata. GOMS and Extensible Markup Language annouced by W3C, standardize the Geo-Object resources description semantics and syntax on the Internet respectively, and offer an importance for the encoding exchange and reuse of Metadata.

The key technologies R&D Programme of National 9th Five-Year Plan Period have had the establishment of the sustainable development information network as the demonstration in the national sustainable development experiment regions.In this paper, the main contents of the programme are introduced.It proposes that the dynamic balance between resources, environment, economy and society can lead to the realization of the sustainable development with emphasis on regional development.Two kinds of sustainable development mode are suggested.At the same time, the models of spatial-temporal integration, simulating historically, forecasting, geocoding, developing knowledge capacity and environment and ecology carrying capacity are developed, and sustainable development information network is established.As the cases, Dafeng city sustainable development level and Zhengding County planting structure are analyzed.The programme is meaningful for the regional sustainable development and the networks information system.

With the development of remote sensing, new models are available continuously In order to extend the practicability of the model library, the authors introduce component object model (COM) technique COM is a software architecture that allows the components made by different software vendors to be combined into avariety of applications Remote sensing model library is composed of three parts,i e , common objects, model objects and accessorial objects Common objects include input/output procedure, solar angle calculating procedure inbi directional reflectance, and metadata about all the models Model objects include the models contributing to quantitative remote sensing applications, which comprise system models, simulant models and application models Accessorial objects include prior knowledge, measurement data and image data In the article,the executable project is validated with an instance in the end The spectrum of typical land surface observed in nadir viewing direction is simulated in pixel scale In this process, crop model, PROSPECT model and SAIL model are used to calculate the spectrum character of the pixel Crop model is used to simulate leaf area index, and PROSPECT model is to simulate reflectance and transmission of leaves The final result is calculated with SAIL model The simulated spectrum of winter wheat observed in nadir viewing direction is at 11 o'clock Apr 15, 2001 In the article, each model and each common object are designed to be components The model library based on COM technique adapts to the progress and is propitious to be expanded and modified.

Integrated storing and accessing geographical features related information is one of the most important technologies for GIS application developments. It has been a consensus of opinion in the GIS society that storing and managing mass spatial data with current business relational database management systems under the support of extended relational models is an efficient and effective technology for developing mass data and network oriented spatial database management systems. In this paper, the authors analyzed and discussed some typical spatial database management platforms including ArcSDE, Oracle Spatial, Informix Spatial DataBlade, DB2 Spatial Extender and MySQL Spatial Extensions from the view of storing models, process management schema, spatial querying and indexing, and data cache approaches. It is argued that current business spatial data engines or extension technologies still deserve much more research and improvements, especially for topology representation within relational database management systems and application schema, adaptive spatial indexing and optimization methods, and user defined complex object representation, sharing and inter-operability.

Scenarios analyses of regional land use change, influenced by various factors, have become the hot-spots of LUCC studies. Regarding system dynamics model and CLUE-S model as the modeling framework, this paper designed a Land-Use Change System Dynamics (LUCSD) model and realized the scenario analyses of baseline, ecologically-oriented and economically-oriented models in Taips County(2000～2020). The findings are concluded that land use change is a dynamic process driven by population growth, economic improvement and farming and pasturing development, and unused land in the northeastern and northwestern parts of Taips County will be the sensitive areas of land use change in the coming nearly 20 years. Based on different scenarios models, competition between land-use types led to diversity storylines in some areas. In this sense, some effective measures should be taken to hold back the irreversible but irrational conversions among land use categories to protect the eco-environment and go along with a sustainable development in Taips County.

In the domain of surveying and mapping,users present rigorous requirement for the image quality. But there are noises in the image data,hence it is possible to perform lossy compression. In this paper,an optimum technique about remote image lossy compression based on wavelete transform is presented. First,we analyse the disadvantage of data compression based on discrete cosine transform; then we adopt the method of two-layer and two-dimension wavelete transform to decompose original image; finally,according to the characteristics of remote sensing images,optimal quantizers are constructed based on LMSE principle to meet the high quantity compression. At the same time,according to the experiment,the rough relation between the compression rate and p-p SNR is acquired.

Raster to vector conversion from remotely sensed classified images is pre re quisite and kernel processing for ecological environmental mapping and its applied spatial analysis In this paper, the deficiency and limitations of procedures and their resulting land cover/land use data,associated with conversion to vector coverage database from remotely sensed classified images fore ecological environmental mapping, were explored in detail at first These procedures were qualified as ready to wear or pis aller, which applied built in function modules within some remote sensing processing and analysis systems, such as ERDAS,PCI Then a new universal approach to raster to vector conversion from remotely sensed classified images for ecological environmental mapping was delineated based on the paradigm of combining remote sensing processing & analysis systems (such as ER Mapper) with GIS (ArcGIS, for example), and its processing flowchart also presented using an pilot TM classified image in the coast zone of Fujian Province The classified images were labeled as 13 types of land cover/land use by automatic statistic classification and manually on screen digitalized in the form of ecological environmental investigation and dynamic evolution monitoring Finally, the accuracy evaluation of conversion was made through a transfer matrix between original classified images and vector to raster re converted images, and it turned out that the ratio of keeping to original land cover/land use type came up to 94 63% within the global classified image.

Digital landscape models require abstraction and simplification for realistic landscape.Tree geometricmodel and DEM are two elements in virtual forest.The tree geometric model is very important to forest GIS.Inthe process of model creation,the shape of trunk and branch have outstanding meanings.Because of theircomplexity,plant models used in computer graphics are usually created with procedural methods.A difficultproblem is the interactive control of these models:a small number of parameters is insufficient to specify plantcharacteristics,while a great number of parameters are tedious to manipulate and difficult to comprehend.Asthe shape of an actual tree is diverse,it is impossible to represent all kinds of trees in a special single curve.Toaddress this problem,the Bezier curve which is based on control points is introduced to depict the shape oftrunk and branch.And a OCX control is made to draw the Bezier curve,the result shows the manifold shape oftrunk and branch can be depicted by this method.It has many advantages such as conveniences and celeritywith good visualization effect.

The intelligent remotely sensed image understanding is a topic of general interest on geo-analysis of Remote Sensing. This paper takes a brief review on the current intelligent methods and their shortcomings, points out the necessities of change to the multi-features unit method and how to merge knowledge in the process properly.Based on the discussion above, the fundamental ideas of the multi-feature analysis of remotely sensed images and knowledge-based neural network were introduced, and the whole framework of the this study is proposed.

Geo-information mobile users service system is the technique base of Geo-information popularizing service and the platform of the commercial function of GIS industry.This passage mainly introduces the system characteristics and the frameworks of Geo-information mobile users service system,the construction of commercial GIS frame works and the commercial of the system operation.

As a result of particular advantages of Synthetic Aperture Rad ar, SAR has been applied widely in earth science and military science But, during the imaging process of SAR, these uncertainty factors, such as the environmental disturbance and the aircraft instability et al , affect the application of SAR image directly So it is very important to analyze these uncertainty factors Presently, the research of SAR uncertainty is based on the statistical method, but this method is inconvenient to unveil the mechanism of uncertainties caused and to visualize the SAR uncertainty Whereas, we first discussed the SAR system uncertainty propagation function from the view of mechanism in this paper, and then focus on discussing the visualization of uncertainties of SAR system based on this uncertainty propagation function by means of VR technique and system simulation technique Finally an example about the uncertainty visualization is given.

As the remote sensing archaeology integrates the techniques of remote sensing and field archaeology, the study mentioned in this paper ulteriorly integrates remote sensing archaeology with technique of virtual reality and the multiple subjects correlated with archaeology Sustained by the integration, the evolvement of ancient geographical environment whose features kept in some historic remains and the relationship among these remains will be studied There are such main characteristics in the study as intersection, penetration and amalgamation of physical and social sciences along with the archaeological practices in typical areas, study of the new theory and high tech of remote sensing archaeology and so on From aspect of basic research, the main contents of the study are the spectrum features of the remote sensing archaeological objects, the mechanism of remote sensing archaeology such as extracting faint information from these underground objects and making out translating mechanism of their information and theoretical system and technical criterion of remote sensing archaeology From a spect of technical development, the main symbols are these studies on earth ori ented observation under whole spectrum, whole weather, and whole time and with various platforms, on establishment of the processing and analyzing system of remote sensing archaeology information and the system will be symbolized with definitely perspective capability, multiple dimensions in time and space, multiform information resources and on digitally virtual expression of typical historic remains and the anversion of geographical environment From aspect of applications, the targets are investigation of distribution of the historic remains in the typical areas of West China, establishment of the information system of important historic remains in these typical areas, discovery of the temporally and specially distributional rules of the historic remains in these typical areas and enrichment of the civilized history of West China.

Wavelet package analysis has been rapidly developed based on wavelet transform in recent years. Signal was expressed as a linear combination of a set of best bases of wavelet package,the energy of expansion coefficients are concentrated in a small number of coefficient because of the good location properties of wavelet package best bases. So it is possible to achieve high compression ratios without any loss of useful and important information of the compressed signal. Owing to the limitation of wavelet transform,the authors of this paper have researched wavelet package analysis in the signal compression and given some experimental verifications.

Spectral characteristic is one of the basic characteristics of ground objects, and is the basis of qualitative and quantitative remote sensing. Although we've developed a lot of spectral databases during the past several years, some basic problems still existed in the remote sensing applications, such as the problems of same objects but with different spectra or the opposition, are still not solved completely, which have severely obstructed the development of remote sensing application and industrialization. The main reason lies in that we haven't paid enough attention to the structure, material, and aspection of the measured objects before, and the constructed databases are short of supports of remote sensing models, so the measured data in the databases can't be extended to other cases. The new generation spectral database will be featured by the matching of ground objects' structure, spectra, knowledge library, and model library. Its aim is to solve problems confronted in the previous spectral databases. In this paper, we state the status and function of remote sensing model library in spectral knowledge library, and discuss its structure, system architecture and development technique under network environment. Its operational mechanism under the support of spectral knowledge library and metadata is also discussed. We put forward the web-based three-tiered distributed system architecture and component-based construction method of remote sensing model library. Meanwhile, we also discuss the definition of model metadata from the point of view of remote sensing model library,and put forward metadata-based management and operational mechanism of remote sensing model library. Finally, an application instance and some key codes are presented in the paper.The construction of distributed Web-based remote sensing libraries will greatly enhance the functions of spectral databases, and expand their usable ranges. The component-based design philosophy and metadata-based application and management method ensure that the model library can be updated and modified dynamically to keep its progress and practicability and enhance the reusability of codes.

Remote sensing technology is a highly efficient method applied to investigate rapid land use change.We might acquire detailed land use situation from the fused production of multi-source RS data at least cost.These conventional fusion methods such as HSI transform,Brovey transform and principal components transformcould merge two optical image data of different resolutions——a high spatial resolution panchromatic image anda multi-spectral image at low spatial resolution.However,these fusion methods required same or analogousspectral response range between the high spatial resolution panchromatic image and the low spatial resolutionmulti-spectral image.This paper brought forward an improved fusion method called LoG Filter Fusion(LFF)based on HSI transform,that could merge different kinds of optical image data with different spectral responseranges of multi-source RS data.LFF arithmetic was put forward in this paper,that is,firstly to filter on the highspatial resolution panchromatic image with LoG filter,and then to substitute the intensity component throughHSI transform to engender the LFF image by HSI invert transform.The result of analysis on these fused imagesshowed the gray variance index of LFF fusion method is higher than HSI transform,the color of LFF image isanalogous to the low spatial resolution multi-spectral image,so the LFF fusion method is more excellent at thespectrum preservation capability than HSI transform.And the classification accuracy of the LFF image is 5.68% higher than HSI image.

This paper takes Poyang Lake where the famous 98's catastrophic flood happened in 1998 as an example. With available data of a during-flood SAR image and one pre-flood TM image, an approach named TIN (Triangulated Irregular Network) was used for registration, with better accuracy than other general methods. Then according to statistic analysis including interband correlation and entropy, etc. the selection of the optimal bands of TM was discussed. Finally, in order to monitor flood better and more quickly, various fusion approaches were applied and compared based on time-consuming operations. As a result, the pseudocolor composition of SAR, one near infrared TM band (TM5 or TM7) and one visible TM band (TM3 is optimal) is the best, which not only can make us easily identify flood extend area, soil area full of water content and non-flood area from each other, but also is simple and less time-consuming. It will meet the need of rapid response to flood.

Raster and vector are two major types of data format in GIS, especially at presenst, remote sensing images have been very important data resource of GIS. With the development of GIS, remote sensing data will play an even greater role in GIS. But however, remote sensing data is usually raster data, and cannot meet the requirements of spatial analysis in GIS, so a conversion of raster data into vector one is of great importance for GIS. Although there are many algorithms and softwares for vectorization of raster data, yet some problems still exist in the process of vectorizing raster data, for example, the vector images thus produced have some islands and self-intersect polygons, or there are only polylines without good topological relation. Hence this can hardly gratify the need of our work.On the basis of summarizing the existing algorithms and following the topological theory, an improved method for converting raster data to vector one is suggested in this paper. In other words, firstly it's determined by the relation of similarities or differences between four neighboring pixels(Fig.2) whether the crosses of neighboring pixels(Fig.1) belong to nodes or coordinate points, then in the process of searching nodes and coordinate points, both vertical and horizontal line segments are picked up at the same time in order to realize the vectorization of raster data more quickly and more efficiently. In addition, by use of the methods such as automatically breaking and widening grid size, some common problems in the course of vectorization of raster data such as self-intersected polygon and island are successfully solved. The authors programme a programme in Visual C⁺⁺ to realize the algorithm in this paper, and contrast its result with other softerwares or algorithms. The algorithm has some better virtues:1)it is easy and convenient to practice; and 2) it has perfect topological relation. Therefore, the improved method is of great benefit to retrieve useful information from remote sensing images, and to promote the integration of Remote Sensing and Geography Information System.

Sharing of geographic information is important because the more it is shared, the more it is used, and the greater becomes society's ability to evaluate and address the wide range of pressing problems to which such information may be applied. To facilitate sharing, the Geographic Information System (GIS) user and research communities must to establish a structured set of standards for information concerning objects or phenomena that are directly or indirectly associated with a location relative to the Earth. These standards may specify the methods, tools and services for both technical and institutional aspects of collecting, structuring, analyzing, accessing, presenting, disseminating, integrating, maintaining spatial data and transforming such data in digital/electronic form between different users, systems and locations. In this paper, the newest development status of geographic information standards is introduced at International Organization for Standardization (ISO), Open GIS Consortium (OGC) and the European Committee for Standards(Comité Européen de Normalisation, CEN). The organizaion structure, Working items of geographic information standards are presented at ISO/TC 211, OGC and CEN/TC 287 and the strategies for developing geographic information standards in China are proposed.

Evapotranspiration is an important parameter for water resource management Compared to the traditional ET computation methods, the ET computation method based on Remote Sensing has the characteristics of quickness, precision, large scale and visualization mapping SEBAL (Surface Energy Balance Algorithms for Land) is a model which is used to compute evapotranspiration in large scale region based on Remote Sensing images which have the visible, near infrared and infrared bands SEBAL model is based on surface energy balance equation which includes net radiance flux, soil heat flux and sensible heat flux and latent heat flux All these three heat fluxes are computed in sequence, then the latent heat flux which provides energy for evapotranspiration can be computed, finally instantaneous ET and daily ET can be obtained In this paper SEBAL model was used to compute the evapotranspiration for the Yellow River Delta where water resource is facing an urgent situation, in the end the evapotranspiration characteristics of Yellow River Delta are analyzed This study is helpful for better management of water resource of Yellow River Delta.

The article reviews more than 20 years out of 80 years of application and research work of remote sensing technology (RS), geographic information system (GIS) and global positioning system on soil erosion, and briefly introduces and comments on the technical approaches during the process of erosion investigation at different stages. It aims at making soil conservation monitoring workers understand “3S” application in soil erosion investigation process, and have a clear direction for future application study. The application of RS, GIS and GPS to soil erosion fast investigation is a beneficial attempt for the applied of Hi tech to production reality, and also is a frontier for the applied study of “3S” in water and soil conservation. Soil erosion investigation is a quite complicated task. This article, based on the experience of summary on erosion remote sensing investigation at different periods of time, aiming at the existing technical problems in the third remote sensing investigation on soil erosion, carries out an analysis and puts forward corresponding suggestions so as to consult with colleagues and to provide a reference for decision making.

The Yellow River is well-known for its highest sediment contents of all rivers in the world. Due to the huge sediment to the coast, the Yellow River Delta is in a state of continuous dynamic change. The most recent shift occurred in 1976, and a new sub-delta is being created since then. This paper describes the method and technology for monitoring the coastline change of the Yellow River Delta using remote sensing. The multi-temporal Landsat and other satellite images were used as the monitoring data source. The images were registered to the high precision GIS database, and they were matched each other and matched with the database. All the images have the same coordinate system. A mean high water (MHW) model was developed for extracting coastlines from remote sensing images. 20 coastlines were extracted from different years. The result shows that a new territory of 330km2 was formed in the present estuary area from 1976 to 2000. At the same time, a territory of 137km2 was eroded in the Diaokou estuary area to the north of the Yellow River Delta. In conclusion, it is the substantial foundation for the water and sedimentation of the Yellow River to shape the delta. That is, it is the essential factor leading to coastline evolution.

Based on multi-temporal remote sensing images of Jinjiang city in Fujian province acquired in 1988 and 2001,together with auxiliary data and geo-knowledge，we establish the formation characteristics and the classification system of the eco-environment in Jinjiang city. By interpreting the Thematic Map of ecoenviron-ment in this area and analyzing the current situation of Jinjiang city’s land use,we then get two ecoenvironmental maps in two different periods. Combined with the analysis of the quantity of eco-environment,characteristics of their distribution and the rule of their dynamic changes,we establish the spatial and temporal analysis model for measuring the rate of ecoenvironmental changes. In conclusion,the paper points out that the changes of two eco-environments in this area are vigorous,and the settlements of urban and rural eco-environment have increased greatly,while the area of eco-environment in platform dryland decreases drastically. And the obviously changing city space could be divided into three areas: the northeast developing area,the southwest developing area and the central newly developed area. In the northeast and southwest developing areas,their development trends radiate along the transportation lines. But the central newly developed area has changed as the hub shapes outwardly.

Three land use classification maps of Xiamen in SE China in 1989,1995,and 2000 with their changeconversion matrixes were obtained using the multi-stage classification and the change detection(post-classifica-tion comparison)techniques.The result clearly reveals the land use changes in Xiamen during the 11-yearstudy period with rapid economic development and urbanization process.The land use has changed significant-ly,especially in the main land use types such as the built-up land,the arable land,and the forested land.Alarge amount of arable lands are replaced by the built-up lands due to urban expansion,resulting in not onlythe decrease of quantity but also reduction in quality of the arable lands.Analysis revealed that the main driv-ing forces causing these changes in Xiamen include the progress in industrialization and urbanization,the popu-lation growth,and the topographic environment.The study finally concludes that dynamic monitoring of land useinter-annual change can rapidly and efficiently identify the characteristics and the inherent rules of land usechange of the city by using multi-temporal and muhi-spectral resolution remote sensing images,with a combi-nation of geographic information system(GIS)and global position system(GPS)technique,and the result canserve as a reference for decision-making and urban planning of the local government.