The number of remote sensing programs and systems has increased dramatically over the last years serving the needs of geographic information systems (GIS) with high demands for geospatial data. New technologies such as coupled global positioning systems (GPS) and inertial navigation systems (INS) allow airborne sensors to produce digital data of excellent geometric accuracy and challenge standard large format aerial cameras. Multi-source remote sensing systems are creating data at higher spatial and temporal resolution than have been collected at any other time on earth. GIS technology allows the efficient storage and management of spatial datasets in digital formats. Remote sensing systems, in return, acquire current, accurate and synoptic data that can be used to update GIS databases. In combination with the appropriate data transfer and interoperability standards that are currently being developed the technology is being put in place that will eventually allow standardized exchange, processing and dissemination of geospatial information.

New digital remote sensors have become available that allow applications at resolutions and scales that were deemed impossible just a few years ago. These sensors include not only the new very high resolution remote sensing satellite programs such as IKONOS, QuickBird or EROS (Ehlers, 2002a). In addition, the advent of stereo scanners of ultra high spatial resolution offers the possibility of a complete digital airborne remote sensing processing system: data acquisition, data preprocessing, data analysis and data integration into a GIS can be managed in an integrated digital environment. These sensors do not only deliver ultra high spatial resolution of about 5 to 30 cm but also 3D digital surface models (DSM). Accurate digital surface models can also be obtained by direct recordings using laser scanning systems (LiDAR).

Satellite Systems

The advent of commercial very high resolution satellite programs has opened new application fields for space-based remote sensing. Satellite data offer, for the first time, the potential for large scale applications such as urban planning and environmental monitoring at the highest level of detail (Ehlers et al., 2003a; Möller, 2003). Spatial resolutions of 0.60-1.00 m (panchromatic) and 2.5- 4m (multispectral) have begun to challenge aerial photography. Companies such as Digital Globe or Space Imaging promise extremely fast processing. Data should be delivered within days (or even hours if downloading via internet is possible). Tiltable cameras offer short revisit periods of 2-3 days and across-track as well as along-track stereo capabilities. IKONOS II was launched in September 1999 making it the first commercial very high resolution satellite in orbit (see Table 1).

Another recent development is the use of hyperspectral technology in space. Hundreds of narrow bandwidth sensors guarantee a spectral sampling which resembles that of a spectrometer ("imaging spectrometer"). To-date, however, only experimental prototypes have been launched. The first commercial hyperspectral satellite OrbView-4 (200 spectral channels with a spatial resolution of 8 m) was launched on September 21, 2001. Unfortunately, it did not reach its orbit. The company Orbimage launched its next satellite, OrbView-3, in June 2003. OrbView-3, however, will has no hyperspectral sensor but only very high resolution panchromatic and multispectral sensors comparable to the IKONOS scanner. Table 1 presents the characteristics of four selected very high resolution satellite programs.

These new sensors offer new perspectives for digital remote sensing, but pose also great challenges, especially for automated analyses. With pixel sizes of less than 1 m, intra-class variances do no longer permit the use of standard pixel-based analysis methods (e.g., Maximum Likelihood classification). In addition, pricing policies of the companies prohibits to a large degree the purchase of more than test data sets for the nonmilitary user. It is hoped that a healthy competition between different providers will result in prices that actually promote the use of very high resolution satellite data for the civilian sector. The recently introduced digital multispectral airborne sensors will hopefully play a role in keeping the prices in an affordable range.

Digital Camera Systems

After a long period of development, we now see the emergence of operational digital camera systems which challenge aerial frame cameras. Advanced technologies such as GPS coupled navigation systems and advanced digital sensor technologies have overcome the strongest impediment of aircraft scanners: the lack of geometric stability. Public and private research has concentrated on the development of digital line array or matrix scanners that will serve as successors to the "classical" air cameras. Companies such as Leica Geosystems, Zeiss Imaging (Z/I) or Vexcel offer first commercial systems, research centers such as the German Space Center (DLR) fly their own prototypes. Such systems have to establish their market somewhere between the satellite image user seeking higher resolution and the airphoto user seeking digital input and GIS compatibility. Consequently, airborne scanner systems have to offer stereo capability and multispectral recording (Figure 1).

Two different technologies are being employed to accomplish an airborne digital recording system. Z/I and Vexcel make use of two-dimensional arrays and a set of coupled nadir-looking lenses to emulate a standard frame camera's central perspective (Ferrano and Felix, 2003; Leberl and Gruber, 2003). Leica Geosystems and the DLR employ triplet scanner technology with onedimensional line arrays arranged in fore, nadir and aft looking modes (Tempelmann et al., 2000; Hoffmann and Lehmann, 2000). The advantage of a 2D matrix camera is that all standard photogrammetric techniques can be used in a digital environment. The advantage of a stereo triplet solution is that photogrammetric preprocessing (i.e., DSM and ortho image generation) is performed before the user receives the data thus alleviating the need to run sophisticated software at the users' organization. The image data are provided in the required coordinate system and can be easily integrated into an existing GIS database. Which of the two approaches is more suitable will largely depend on the user demands and the price-performance ratios of the respective systems. Table 2 presents five selected ultra high resolution airborne digital camera systems.

The advantages of digital cameras are widely understood: no film, no photo processing, no scanning, better radiometric quality through direct sensing, "non-aging" storage and direct integration into GIS and image processing systems. The disadvantages of digital scanners, most notable its geometric distortions and monoscopic imaging mode, no longer exist due to the stereo capabilities of the new sensors and the use of integrated INS and differential GPS technology during image acquisition.

Problems for Automated Processing of High Resolution Data

One has to point out, however, that the new sensors pose new challenges for automated interpretation. The homogenizing effects of comparably large pixel sizes are no longer valid. With these ultra/very high resolution sensors, simple pixel-based analyses are no longer applicable because of the difficulty of classifying high resolution data where each pixel is related not to the character of an object or an area as a whole, but to components of it (Blaschke and Strobl, 2001). Instead, we have image pixels that might belong to the same class but exhibit totally different reflection values. For example, a high resolution image of the class "house" can be represented by hundreds of pixels that might belong to undesired subclasses "window," "chimney," "sunlit roof," "shadowed roof," "front lawn," or "driveway" (see Figure 2).

For example, Gong et al., (1992) and Johnsson (1994) showed that spectral classification of higher resolution data does not automatically lead to more detailed classification results. Further on, using just multispectral information for classifications does not lead to accurate interpretation results because the differentiation between object classes is done not only with the help of spectral information, but also with spatial (contextual) information of the image data. For example, using only multispectral information different objects like roofs and streets might not be separated in two object-classes because they are built with the same material (Hoffmann et al., 2000). Consequently, new intelligent techniques will have to be developed that make use of GIS integration, multisensor approaches and context based interpretation schemes (see, for example, Ehlers, 2000; Schiewe, 2003). Otherwise, the last step of an all-digital image acquisition and handling process has to consist of manual on-screen digitizing.

It is therefore imperative that new techniques be developed that allow an automated processing of high resolution and multisensor images. One of the promising approaches is the use of auxiliary information in the processing steps. While the use of additional information is nothing new in the processing of remotely sensed images, the existence of geographic information systems (GIS) presents a formalized structure in which the additional information can be provided and preprocessed.

Case Study: Biotope Type Monitoring

The reason for the underlying studies were several fairway expansion projects on rivers in Northwestern Germany due to shipping requirements. Continuous environmental monitoring after the end of the expansion project had been mandated with focus on the tidally influenced riverside biotopes. Changes in composition and size of these biotopes should be documented over the long term to assess the impacts of hydraulic engineering measures.

The German Waterways and Shipping Administration was in charge to promote the development of a generally applicable automated and cost-effective method for an operational, long term monitoring process. It is well known by now that simple pixel based analyses are not applicable because of the difficulty of classifying high-resolution data where each pixel is related not to the character of an object or an area as a whole, but to components of it (Blaschke and Strobl, 2001). In cooperation with the Research Center for Geoinformatics and Remote Sensing (FZG) and the German Aerospace Center (DLR) in Berlin, an integrative monitoring concept was developed using a combination of GIS, image analysis and modeling software (Ehlers et al., 2003b). An essential improvement could be attained by using appropriately recorded digital data and applying an automated analysis process employing a rule based procedure. The selected approach seemed to guarantee the best possible accuracy compared to traditional mapping and surveying methods that are always combined with extensive fieldwork.

Study Sites

The study sites are located along the tidally influenced areas of the rivers Elbe and Weser in Northwestern Germany close to the large cities Hamburg and Bremen (see Figure 3). The choice of these sites arose from the need of highly accurate mapping as consequence of river expansion projects. In both river areas reeds and some relics of willow forests are of major interest for nature conservation.

Methodology and Results

The main objective of the projects was the development of an automated and reproducible classification process. The FZG research group developed an index based segmentation and pre-classification procedure as pragmatic data preparation approach for an automated hierarchical classification process.

The analysis takes place in several steps (see Figure 4). The first step (Level 1) is the computation of ancillary information such as texture and vegetation indices. The computation of vegetation indices is a standard procedure for satellite remote sensing applications. Indices like the NDVI and its derivatives are commonly used for separating vegetation from bare soil as well as for estimating quality and vitality of vegetation stands (Jensen, 1996).

In the field of airborne remote sensing those computations actually have only rarely been used, mostly due to the limited availability of an appropriate database. Scanned aerial photos have generally proven to be inappropriate for these methods due to their heterogeneous radiometric properties and their evident inconsistencies even within single scenes. Even with the first models of operational digital scanners for airborne applications computations of vegetation indices were limited. For example, the HRSC-A offered an inappropriate spectral resolution (i.e., no true red band) for vegetation mapping tasks. The new HRSC-AX sensor displays better suited multispectral bands for terrestrial applications and higher radiometric resolution (see Table 1). The excellent quality of the HRSC-AX data enables the computation vegetation indices. However, the applicability of each one has to be examined thoroughly because its spectral bands and scales differ from those of standard satellite systems.

During our research, numerous indices have been tested for the HRSC image data. For the HRSC-A, best results were obtained with a combination of the near infrared, the panchromatic and a calculated virtual red band (Ehlers et al., 2003a). For the HRSC-AX, standard NDVI methods seem to be sufficient. It has to be noted that the use of vegetation indices combined with a hierarchical classification procedure improves the classification process with respect to speed and accuracy.

The vegetation indices are used in the next step (Level 2) to identify and separate four coarse classes (non-vegetation/ sparse vegetation; vegetation, water, shadow) (see Figure 4). With the additional incorporation of height information from the DSM, the vegetation class is further divided into high vegetation (e.g., trees) and low vegetation (e.g., shrubs, grass). This can be achieved by incorporating all data in an integrated GIS/image processing environment using the appropriate GIS or image analysis procedures. Thus, biotope types that do not show a difference in their multispectral reflectance characteristics but are of different height can easily be separated. This "separation of information" step, therefore, permits the detail and accuracy of the classification to be improved. The resulting segments are already pre-classified into the semantic layers "non-vegetation/sparse vegetation," "water," "shadows," "low vegetation" and "high vegetation." For each segment and each object class a minimum size is defined. Smaller segments are eliminated using standard GIS operations.

In the next step (Level 3), the separated layers (for an example see Figure 5) were treated with appropriate classification algorithms (i.e., isodata clustering for the non-vegetation/sparse vegetation layer and supervised classification for the herbaceous vegetation layers; see Figure 4). With this approach, the level of detail in the biotope type classification could be significantly improved. It was possible to identify more than 20 different classes.

Finally, a GIS-based postprocessing is involved to produce the final classification result (Level 4). GIS operations such as majority filtering, logical overlay, definition of neighborhood relations and minimum area functions were used to estimate appropriate classes for shadow areas and to combine the individual information layers. The final output was a GIS layer with 21 biotope types for the study sites. The differences between the final classification result and maps created by field work and photointerpretation are presented in Figure 6 for a small test site located on an island in the Elbe River. The richness of detail of the classification results corresponds well with the structures in the original image. The visual interpretation result already shows the generalization that was performed by the human operator. The older reference map shows the subset mapped with only a few polygons whereas the new classification result consists of more and better fitting polygons, described by over 1400 vertices. Even single trees, shrubs or open forest areas smaller than 100 square meters can be detected over large areas. Preliminary accuracy checks indicate that the procedure yields an average overall accuracy of better than 85%. Classes of particular ecological interest (e.g., ruderal vegetation, scirpus, willow trees) could be mapped with an accuracy exceeding 95%. First results also indicate changes in the distribution of these ecological vulnerable species between 1999 and 2002 (see Figure 7).

Conclusions

The potential of high resolution digital airborne scanner data with excellent geometric fidelity was investigated in a pilot project. It could be shown that digital image data in combination with an integrated GIS/image processing environment allowed the development of an automated classification procedure for detailed and accurate biotope mapping. This automated hierarchical technique facilitates the documentation of dynamic processes in environmental monitoring projects. The original images as well as the classification results can easily be integrated in a GIS environment. This allows operational analysis and measuring of changes over time. Of particular importance proved the integration with the digital surface model that is simultaneously produced from the HRSC sensors. With this, it was possible to differentiate between high and low vegetation despite their similarity in spectral reflectance. Using GIS operators such as majority filtering and rule-based overlay techniques, shadows could be eliminated and individual classification layers be combined. The hierarchical classification procedure could be formalized and stored in a flow-chart environment. Next steps include the extension of the monitoring areas to the tidal North sea coast and the inclusion of laser scanning technology for more accurate DSM generation.

About the Authors

Dr. Manfred Ehlers is Professor of GIS and Remote Sensing at the University of Vechta, Germany.

Monika Göhler and Ronald Janowsky are Research Associates with the Institute for Environmental Sciences (IUW) and the Research Center for Geoinformatics and Remote Sensing (FZG) at the University of Vechta, Germany.

References

http://www.gis.leica-geosystems.com/products/files/ADS40_Software_Brochure.pdf (last visited 02/03).