Germany, with a consumption level of around 490 million tons coal equivalent (tce) the world's fifth largest energy market, relies on importing around 60% of its energy to cover requirements. Its own energy basis is limited essentially to hard coal and lignite coal, whose domestic production accounts in all for around two thirds of the primary energy production in Germany.
In 2004 the German hard coal market had a volume of around 66 million tce, which is the largest in Western Europe. The hard coal covers about 14% of the German primary energy requirement.
The rapid reduction of domestic hard coal production over the past few decades has meant that of 153 mines in 1957 only nine have remained in operation under DSK and of 64 coking plants only one. The number of jobs in Germany hard coal mining has declined by more than half a million. The German hard coal market is therefore currently supplied to the tune of a good 40% from domestic sources. The current annual production of approximately 26 million tons of hard coal from an average depth of 1,200 metres is concentrated at seven mines in the Ruhr and one each in the Saarland and at Ibbenbüren (Figure 1). The remaining 60% of the requirement is covered by imported hard coal (mainly from Poland, South Africa and Columbia).
Once German hard coal production has been cut further to
a base level of 16 million tons per year in 2012, as has already been decided, the percentage of the supply accounted for by imported coal will increase further. This base-line mining would, however, still be inappropriate as an option for utilizing the national deposits to guard against the risk of structural changes to the world market.
The endeavour to win acceptance in society at large will remain a major challenge for mining, while at the same time the three major objectives of German energy policy will play the crucial role: economic efficiency, supply reliability and acceptable environmental impact.
For a long time the subject of economic efficiency or subsidies for domestic hard coal dominated the media. Competitiveness is of course largely determined by the situation as regards mineral resources on the world market and the resulting world market prices. But by establishing efficient working processes and using state-of-the-art technology DSK is also contributing substantially to an increase in the economic efficiency of German hard coal.
Only the past few years has the subject of supply reliability through domestic mineral resources acquired a significant status once more. Politics, industry and the general public are being made increasingly aware of the dependencies that exist in the energy and mineral resources domain by the changes in world politics, the globalization of industry and the recurrent turbulence on the international energy and mineral resources market. There is a threat of price and supply risks not only from imponderables and power positions at the suppliers of mineral resources, but increasingly also from the greater competition in demand from other countries and consumers. One result of globalization is that the supply situation in Germany today depends to a high degree on the pace of growth and the consumption of mineral resources on the part of other markets, primarily in China and South-East Asia. More than 60% of the population of Germany is again in support of promoting domestic hard coal.
Dealing with nature and the environment in a responsible fashion is a central concern of German hard coal mining. The mining industry engages in preventative environmental protection by endeavouring to achieve continuous improvement on its own responsibility – beyond statutory regulations and requirements. It gears its activities to the principle of sustainability. Accordingly economic, social and ecological development must be designed such that they meet the needs of the present generation without jeopardizing the possibilities of future ones.
In order to ensure optimum environmental protection, DSK has defined and put into effect a set of internal environmental protection guidelines. They document the notion of environmental protection from exploration and planning through to the winning, preparation and further processing of hard coal and they impose a mandatory obligation on all employees. The effects on the environment are determined and evaluated not only before new activities are taken up or new technical processes introduced, but also when measures are taken to close down operational facilities and to recycle sites and buildings.
The focal elements of environmental protection are handling mineral resources and energy in a managed fashion, protecting the atmosphere, reducing noise and air pollution, environmentally sound waste management, protecting bodies of water and soil, environmentally sound handling of hazardous substances and freight and taking timely precaution to prevent and limit environmentally relevant events.
The winning of hard coal is important for the economy as a whole and is of general interest. The forms of impact this has on the surface as a matter of course are of relevant both in terms of private law and public law. The Federal Mining Act (Bundesberggesetz) and the plans of operations embedded in it which have to be drawn up by the competent regional ("Land") authority, especially the framework operations planning procedure involving environmental impact assessment, are a major starting point when coping with the related challenges.
The aim of the operations planning procedure is to balance the interests of individuals and the common and to create transparency in the decision taken by the planning authority at the end of the procedure for all involved. A central element therefore is the participation of the general public; this means involving the municipal authorities, private individuals affected and the bodies representing public concerns in the procedures in a wide variety of ways.
Hard coal is mined largely below conurbations (the Ruhr region, the Saarland) with correspondingly complex and sensitive structure on the surface (e.g. industrial plants, power plants, infrastructure, building development), but also with numerous natural spaces containing a wealth of animals and plants that have to be protected. All these elements on the surface are affected by underground mining in the form of ground movements. The main influencing factor is subsidence – since the start of mining operations this has amounted to as much as 25 metres under urban areas and 10 metres under major shipping routes (the Rhine). These effects give rise to tensions between the interests of members of the public, those of mining companies and those of government. For some time procedures and methods have existed to regulate effects on man-made components.
The impact on the natural world has, on the other hand, only been accorded any major significance in the past few decades and this must be forecast as part of the licensing procedures for future mining operations by means of environmental impact studies.
Figure 2: Example of Prosper-Haniel Mine as a Study Area (approx. 89 sq. km, with impact area and forecast subsidence 1999 - 2019 in metres).
Since underground mining is a dynamic, not precisely predictable process with a very long running time of 15 to 20 years, the changes forecast for the ecosystem must be observed and analyzed by means of preventative, ongoing environmental monitoring in the mining operations' area of impact (Figure 2) during and after mining [Christiansen et al. 2001].
A central role here is played by the protected asset of water, with its sub-areas of ground water, flowing waters and still water – in addition to the assets of animals, plants and their habitats, and man and his living environment. The monitoring of the environmental impact of mining encompasses the following tasks to be performed at fixed intervals:
- recording of the actual state of nature
- assessment of the current condition, determination of degrees of impact
- comparison of various conditions measured/observed and forecast
- recognition, assessment and classification of time-dependent changes in the environment
- establishment of practical recommendations for the implementation of measures to avoid and reduce adverse environmental effects from mining
- performance check regarding measures taken
- communication of results
Because of conditions imposed by the authorities and, increasingly, greater sensitivity to the subject on the part of the general public, there is a need to collect more and more extensive data and to perform an ever more complex analysis.
Monitoring also places special demands on the technical side of data collection and data processing. For example, up-to-date information on the condition of the ground surface and ecosystem must be available as promptly as possible, recurrently at fixed points in time, in a quality that makes it possible to recognize individual objects, to analyze them and to store their history. To meet these rigorous requirements for the monitoring state-of-the-art technologies from the fields of photogrammetry, remote sensing and geo-information systems are used, among others. These technologies were essentially developed at DSK. The most important technologies are described below.
METHODS AND RESULTS
REMOTE SENSING METHODS TO OBTAIN DIGITAL ELEVATION MODELS (DEM):
Monitoring the influence of mining normally begins with the licensing of the planned winning operations. To document the condition at the start as a reference for future conditions and as a basis for the forecasts needed in the course of monitoring a digital elevation model (DEM) of the entire area under examination (up to 100 square kilometres per mine) is needed first of all. The precise and up-to-date data needed for this is mainly collected using procedures of analytical and digital photogrammetry and by means of supplementary terrestrial surveys. A photogrammetric aerial flight campaign is therefore performed for each mine in the year a framework plan of operations is approved. Using digital photogrammetric procedures a 5-metre geodetic point grid is automatically computed from the previously scanned aerial images. This point grid is then revised manually at digital stereo-workstations and supplemented with breaklines and formlines of artificial and topographical objects which describe the ground surface (e.g. dykes, embankments, still bodies of water) (Figure 3).
Elements not visible from the air (e.g. beds of watercourses) are supplemented by including terrestrial measurements. This means that a model of the surface is available which is highly precise in terms of positions and elevation (+/- 10 cm). This DEM is initially used to represent an elevation situation in the form of grey scale values, contour lines, contour levels, 3D views etc. (Figure 4).
Figure 3: DEM measuring elements (detail in the area of a meandering, flowing body of water, green - grid points at 5 metre intervals, brown – breaking edges, blue – water axis, background Orthophoto) .
Figure 4: DEM for the Prosper-Haniel Mine as a Study Area (contour levels from approx. 30 m above sea level in light blue up to approx. 170 m above sea level in dark brown, overlaid with still and flowing bodies of water in blue).
As a basis for ground water monitoring the ground water level is calculated by combining the DEM with the corresponding ground water model. To monitor flowing water a three-dimensional, stationed flowing water network is obtained from the DEM as the basis of, for example, longitudinal sections of bodies of water or ecological and hydrological maps (Figure 5) [Hentrich et al. 1999].
Both DEMs and flowing water networks are generated for the observation or forecast time sections needed by combining with subsidence calculated. These models are then used by the expert assessors to assess or forecast changes in the ground water and surface water system (e.g. changes in ground water level, waterlogged areas, changes in gradient for flowing water) (Figure 6).
In addition to "classic" photogrammetry used to generate a DEM, the increasing availability of high-resolution earth
Figure 5: Ecological stock map (yellow side strips: morphology; red/green side strips: water quality; blue centre strips: rate of flow; red centre strips: piping; red arrows: discharges; red circles: footfall damage; additional presentation of special punctiform and linear features).
Figure 6: Change in ground water level (forecast period 1993 – 2019, generated by combining DEM, ground water model and subsidence model, green – enlargement of ground water level in relation to ground elevation, beige – reduction of ground water level in relation to ground elevation).
Observation satellites and airborne digital sensors is making
remote sensing and digital image processing more interesting for future, inexpensive environmental monitoring of mining areas. In accordance with the relevant requirements regarding ground sampling distance and the geometrical accuracies, modern airborne and spaceborne sensor techniques are being tested with respect to their suitability and, if successful, they are integrated in DSK's working process.
DIGITAL AERIAL CAMERAS
In DSK research and development projects airborne sensors systems have been examined since the mid 90s with respect to their potential for DEM generation [Spreckels 2003].
Digital camera systems with area sensors now achieve a ground resolution of less than 5 cm and record 4 spectral channels, which means that real-colour and false-colour infrared images can be processed.
In December 2004 DSK undertook a flight with the area sensor camera Vexcel UltraCamD over an industrial site with an 8 cm and 10 cm ground sampling distance. The results showed a high geometric stability and – with the recording of digital pictures in 12-bit mode – it exhibited excellent possibilities for 3D stereo measurements (Figure 7).
Figure 7: Digital aerial camera (flight over an industrial site of DSK on December, 10th, 2004: extract from the digital aerial photo No. 210 – ground sampling distance 8 cm – Top: Original course of the grey scale values – Bottom: Histogram stretching of the grey scale values).
DSK will also use these camera systems in future for large-area overflights as part of the monitoring operations. The aim is to replace analytical photogrammetry by digital photogrammetry – and hence to achieve a considerable acceleration of the DEM creation by means of a completely digital workflow.
AIRBORNE LASER SCANNING AND RADAR INTERFEROMETRY
In addition non-image-making, active sensor systems, such as "airborne laser scanning" and "airborne radar interferometry", were examined as the basis for creating DEMs [Spreckels 2003]. The analyses conducted revealed that the DEMs generated form laser or radar data – with appropriate post-processing – will just satisfy DSK's accuracy requirements of +/- 10 cm. But in order to incorporate linear elements (e.g. waterway axes) a stereo image record is also invariably essential, and this is not available with laser scanner and radar systems. For this reason these active sensor systems will not be used at DSK to create DEM's.
Nevertheless airborne radar has major advantages thanks to its all-weather capability and no back-scatter (total reflection) on water surfaces. These characteristics can be exploited in the case of flood events – which mostly go alongside bad weather – in order to map quickly the flooded areas in the intensity images. The intensity images shown in Figure 8 were recorded at the beginning of 2001 in stormy weather and with heavy showers.
Figure 8: Airborne Radar Interferometry (SAR) - Intensity images from the airborne AeS1 radar system - Top: X band (1-metre ground resolution) - Bottom: P band (2.5-metre ground resolution) – Distance between grid junctions: 500 metres).
In view of the technical circumstances, radar satellite data is not suitable as a basis for high-precision DEMs. The registration of ground movements in the sub-centimetre range is possible, however, by examining the phase differences (DInSAR method). By comparing with terrestrial observation lines it was possible to confirm the potential of the DInSAR procedure for recording subsidence of up to approximately 10 cm between two satellite revisits both for urban areas (C band) [Spreckels et al. 2001] and for regions used for agriculture and forestry (L band) [Strozzi et al. 2003] (Figure 9).
At present there are no operational L band radar satellites in earth orbit. For 2006, however, it has been announced that the Japanese ALOS satellite will be launched. Its PALSAR systems works in the L band and so in future it will be possible to observe subsidence phenomena in the mostly nature-dominated mining areas on a blanket basis and with great precision from space.
Figure 9: Differential SAR Interferometry (Picture left: ERS-1/-2 “Land use” (blue = water, green = forest, yellow = paved). Middle: Ground movements from ERS-1/-2 C-Band DInSAR, superimposed by mining areas (dT = 70 days). One colour cycle = 3 cm movement. Right: JERS L band DInSAR ground movements, approximately identical time section (dT = 88 days). One colour cycle = 12 cm movement).
SATELLITE REMOTE SENSING
OPTICAL SATELLITE IMAGERY – IKONOS
Since 2000 the optical satellite IKONOS 2 has been in earth orbit. It records with a resolution of 1 m for panchromatic data and 4 m for image data in real colour and false colour infrared. IKONOS data provides a relatively inexpensive and commercially available possibility for documenting the condition of the surface at shorter time intervals (e.g. annually).
In contrast to airborne imagery, however, there is a constant risk of cloud and shadow for IKONOS image data (Figure 10). Clouds and shadow then affect the image analysis. Nevertheless it is possible to order maximum permissible cloud cover and guaranteed cloud-free areas for the exposure.
Figure 10: IKONOS Orthophotos (false colours infrared - 1 m ground resolution – examination area of the Prosper-Haniel Mine - Left: August 2003 / cloud-free – Right: August 2003 / here the influence of clouds and shadow can clearly be seen).
IKONOS image data is geo-referenced at DSK using topographical maps or photogrammetric ground control points with a position accuracy of +/- 50 cm. Using a corresponding DEM, orthophotos of the IKONOS scenes are produced at DSK.
In the GIS there is therefore always an actual set of images available (Figure 11). In addition the IKONOS data serves as an aid in the terrestrial mapping of, among other things, fauna and flora data. One method of classifying the IKONOS data with the aim of a semi-automatic derivation of areas of change (e.g. formation of areas of water) is currently being developed.
Figure 11: IKONOS data as GIS background (3D presentation – Extract from the examination areas at the Prosper-Haniel Mine – IKONOS data with watercourse network and still waters).
HYPERSPECTRAL SENSORS - HYMAP
Within the framework of the research project MINEO [Dittmann et al. 2002], which was funded by the European Union, methods were developed for the operational use of airborne hyperspectral sensors (HyMap) for the regular observation, recording and assessment of environmental influences in mining areas, as were standards for the evaluation and provision of this remote sensing data.
Because of the underground hard coal mining conducted by DSK, only indirect environmental influences from mining can be observed. These arise in particular from subsidence-induced changes in the hydrological situation (groundwater level in relation to the surface, outlet) and the resulting effects on the vegetation.
Figure 12: Principle of Hyperspectral Data Collection.
The focus of the evaluations conducted of the remote sensing data was therefore on the development of methods and procedures which can be used to record and analyze reflection anomalies in forest stock. The HyMap sensor supplies digital pictures of the earth’s surface with 126 individual channels in the visible and infrared range. This means that each image pixel contains 126 different items of information on the reflection characteristics of the earth's surface in the various wavelength ranges, and they can be displayed in the form of a spectral cube (Figure 12)
These spectra display a different shape or different characteristics according to the material (e.g. vegetation, sand, asphalt). In addition to this high spectral resolution, HyMap is also characterized by a high spatial resolution. The flying height selected of 2200 metres resulted in a ground sampling distance of 5 * 5 metres per pixel.
The results obtained in the project make clear that it is possible to record the state of vegetation (Figure 13) and any changes in vegetation occurring. This applies both for a comparison of areas within a picture/exposure date and for the analysis of changes occurring with the use of a number of exposure/time sections.
CENTRAL GIS DATA MANAGEMENT
The remote sensing data described and the models derived there from represent only a part of the database needed for monitoring purposes. In addition there is – alongside operational data – a large quantity of ecological and hydrological data which is collected by appraisal experts during the work on the environmental impact study.
Figure 13: Presentation of vegetation stress (pines, in the area around mining-induced waterlogging – increasing stress from green to red – blue isolines of the change in water table in metres).
Increasing importance has been attached over the past few years to the centralized provision of this geo-data. DSK has therefore set up a central geo-database (GDZB) on the basis of ESRI GIS products (ArcGIS, ArcSDE, ArcIMS) and the Oracle database system. It facilitates both direct access by means of GIS clients and use of the data via the Intranet. For this purpose special map services and a metadata browser are provided in the Intranet. The metadata browser is used for navigation purposes and as a communication instrument between the user and the central data.
The map services (Figure 14) provide the users with a wide variety of thematic data/maps and basic GIS functionalities for their analysis. The centralized provision of space-related data offers an optimization and improvement of the information situation with respect to up-to-dateness, quality, simplification and cost. It also facilitates multiple uses by a number of users. Geo-data is thus made easily usable for a large number of tasks related to spatial considerations.
Figure 14: Example of Intranet Map Service of GDZB.
The blanket-coverage, detailed, multitemporal database needed as a documentation and analytical basis in monitoring in the areas of ground movements, morphology of the surface, ecology and hydrology is also integrated in the centralized GIS data stock of the GDZB. A special feature here is the consideration of time as a fourth dimension in the database, thus making it possible to depict dynamic changes in nature during the course of the monitoring period in the form of versions. The DSK Environmental Monitoring Information System (DSK-UMIS) is therefore being developed - as an extension of the GDZB and the ArcGIS functionalities. This will create a user-friendly system which will enable one to:
- store, change and analyze consistently all space-related and time-related data arising in DSK monitoring,
- conduct GIS operations via a user-friendly interface,
- support evaluation of data collected by remote sensing,
- generate on the basis of the data stored information for various user groups, taking account of the corresponding security aspects, and also to pass this information on (as hardcopy, on CD, via Intranet and via Internet).
The concrete specialist requirements for DSK-UMIS have been determined with the help of an object-oriented requirements analysis. The major requirements for GIS arise from the storage of the data collected and from the spatial, temporal and spatial-temporal analyses of the objects.
Integration of time in the GIS as an additional dimension is absolutely essential [Roosmann et al. 2003]. Basically it makes it possible to conduct spatial, thematic, temporal and spatial-temporal analyses. In order to perform these analyses the functionality of ArcGIS is extended to include new, specialist methods with integrated user interfaces (e.g. for spatial-temporal selections and analyses, object monitoring, the processing of terrain models and watercourse networks).
Figure 15: Selection of Objects by Time Indications.
One basis for more extensive evaluations and more complex analyses is provided by selective enquiries over a certain spatial and temporal expanse, such as: Select all objects within a certain region at a certain point in time (Figure 15).
This four-dimensional monitoring database is not only to be capable of being processed by GIS experts with a correspondingly extended ArcGIS, but is also to be made available – like other GDZB data – by an Intranet service to all DSK employees who are interested and are authorized. For this purpose a special map server is being developed on the basis of ArcIMS, which offers in addition selection possibilities for certain versions and themes, and which provides corresponding thematic maps with possibilities for analysis (Figure 16).
Figure 16: Monitoring Intranet Map Service (rear window: Selection possibility for versions and themes – front window: Presentation of the thematic map and analysis possibilities via integrated toolbox).
DISCUSSION AND CONCLUSIONS
The effects on the natural environment due to ground movements during the underground mining of hard coal will be forecast for future mining operations within the framework of licensing procedures. By preventative and continuous environmental monitoring in the area impacted by mining, the relevant changes in the balance of nature during and after mining operations must be observed and analyzed.
Because of conditions imposed by the authorities and, increasingly, greater sensitivity to the subject on the part of the general public, there is a need to collect more and more extensive data and to perform a ever more complex analysis. The monitoring imposes special requirements on the recording, keeping, processing, analysis and output of dynamic, complex, spatially related objects, taking account of time as the fourth dimension. To meet these rigorous requirements for monitoring, the techniques used include the following:
The morphology of the surface in the form of digital elevation models (DEM) is a major component of the monitoring database. For the creation of this DEM the only feasible approach is to record the surface by photogrammetric methods because of the special requirements regarding up-to-dateness, accuracy and economic efficiency. Both the classic evaluation methods of analytical photogrammetry and those of digital photogrammetry are used. Laser and radar data cannot be taken as a basis for special reasons. Digital cameras and digital stereo workstations will increasingly replace classic photogrammetry in the next few years. The result is a very precise DEM with integrated three-dimensional watercourse network.
Various remote sensing procedures are used for documentation purposes – and also as a basis for analysis. Use is made here of both airborne and spaceborne data/methods:
• colour/CIR images (spring overflight) to obtain digital elevation models (see above) and orthophotos,
• aircraft radar data (SAR) to detect areas of water,
• hyperspectral data (airborne sensor HyMap) to classify vegetation and analyze changes in vegetation,
• high-resolution satellite data (IKONOS) to preserve evidence and as the basis for ecological mapping and analysis,
• satellite radar data (DInSAR) to detect ground movements.
The monitoring requires as a documentation and analytical basis a blanket-coverage, detailed and multitemporal database in the areas of ground movements, morphology of the surface, ecology and hydrology. This data and the necessary analytical methods are provided by geo-information systems (GIS). A special feature of the DSK monitoring database is the consideration of time as a fourth dimension in the database, thus making it possible to depict dynamic changes in nature. The DSK Environmental Monitoring Information System (DSK-UMIS) is being developed for this purpose.
The far-sighted and modern approach of the intensive and long-term monitoring of the impact of mining on the environment – using the state-of-the-art procedures described – results in a reliable and constantly up-to-date information basis for corporate and government decisions, both in terms of supply reliability and of protection of the public and the environment – and hence ultimately to greater safety and satisfaction for the general public and companies.
1. CHRISTIANSEN, G.; FISCHER, C.; KELSCHEBACH, M.; VOSEN, P.: Monitoring of environmental changes caused by hard coal mining in the Ruhr district of Germany. In: Mander, ـ., e.a. (ed.): Conference Proceedings of the IALE European Conference 2001. Publicationes Instituti Geographici Universitatis Tartuensis, Bd. 92, pp. 270 – 274, Tartu, 2001.
2. DITTMANN, C.; VOSEN, P.; BRUNN, A.; ET AL.: MINEO (central Europe) environmental test site in Germany Contamination/impact mapping and modelling – Final Report. August 2002. Project funded by the European Community under the ”Information Society Technology“ Programme, IST-1999-10337.
3. HENTRICH, S.; VOSEN, P.: Kohle und Wasser: Konzept und Methodik zur GIS-gestützten Prognose, Analyse und ـberwachung des Einflusses des untertägigen Steinkohleabbaus auf Oberflächengewässer. In: Proceedings of the 14th ESRI European User Conference, Munich, Germany, 15-17 Nov. 1999.
4. ROOSMANN, R.; BUSCH, W.; GORCZYK, J.; MAUERSBERGER, F.; NICKEL, S., VOSEN, P.: Geographic Information Systems support for monitoring environmental impacts caused by deep hard coal mining. In: A. Gnauck, R. Heinrich (ed.): The information society and enlargement of the European Union. 17th International Conference Informatics for Environmental Protection, Cottbus 2003. Part 2: Applications, Workshops, Posters. pp 730-738, Metropolis Verlag 2003.
5. SPRECKELS, V.; WEGMـLLER, U.; STROZZI, T.; MUSIEDLAK, J.; WICHLACZ, H.-C.: Nutzung von InSAR-Daten zur großflächigen Erfassung von topographischen Veränderungen über Abbaubereichen der Deutschen Steinkohle AG (DSK). In: Tagungsband des Deutschen Markscheider Vereins (DMV) 2001, Trier, Germany, 26-28 Sept. 2001, pp. 49 - 70.
6. SPRECKELS, V.; WEGMـLLER, U.; STROZZI, T.; MUSIEDLAK, J.; WICHLACZ, H.-J.: Detection and Observation of Underground Hard Coal Mining-Induced Surface Deformation with Differential SAR Interferometry. In: Proceedings of ISPRS Joint Workshop “High Resolution Mapping from Space 2001”, Hanover, Germany, 19-21 Sept. 2001. On CD-ROM.
7. SPRECKELS, V.: Subsidence Determination by Different Sensor Types using Standardized Transformation Parameters. In: Proceedings of ISPRS Joint Workshop "High Resolution Mapping from Space 2003", Hanover, Germany, 6–8 October 2003. On CD-ROM.
8. STROZZI T.; WEGMـLLER, U.; WERNER, C.; WIESMANN, A.; SPRECKELS, V.: JERS SAR interferometry for land subsidence monitoring. In: IEEE Transactions on Geoscience and Remote Sensing, Vol. 41, No. 7, July 2003, pp. 1702-1708.