Résultats exploitables
Harmonisation of ingestion and access to Earth observation (EO) data products
With the established concept of metadata handling along the Open GIS consortium (OGC) specifications, the integration of different kinds of products has successfully shown to be feasible. A variety of different products from optical and SAR sensors (both space-borne and air-borne), as well as derived products as e.g. line plots can be handled by the Assist demonstrator.
Also the standardised web-based way of querying the archive and accessing the archived via an OGC compliant catalogue service for web (CSW) could successfully be shown. At time of starting the implementation, however, the open source "degree" catalogue service seemed to be the only implementation of an OGC compliant CSW. In the version, which was actual when starting the development, this product showed a couple of minor bugs, which hampered integration into the Assist demonstrator. It is however expected, that these problems will be solved in future versions, and that other products implementing OGC catalogue services for web will become available. With this perspective also the access to remote catalogues will be fostered as several providers of EO data are actually in the process of adopting their previous proprietary catalogue access methods to these standards.
Integration of Mobile Communication
The integration of mobile communication has provided some major challenges to the project. The used Windows Mobile PDAs (which is de facto the only operating system for PDAs with a market share justifying the selection) does e.g. not provide a Bluetooth stack supporting the Bluetooth PAN profile needed for TCP/IP based communication over Bluetooth. As solution to this problem a series of open source Bluetooth stacks were tested, until a stack was found, which worked on the rugged PDAs procured for this project.
As lesson learned from this exercise (and considering the general enhancements in mobile communication since the design of the Assist demonstrator started), for an operational system, different means for connecting mobile field staff to central sources of information would be considered, as e.g. PDAs with an integrated WLAN interface, that connects to a set of rugged WLAN access points spanning up an easily deployable WLAN network for emergency operations.
Such a rugged access point could provide e.g. high bandwidth connectivity via DVB-RCS or other satellite services with return link capability to the ASN directly. Other technologies as e.g. WiMax or TETRA could as well be considered to provide the connection to mobile field staff, given the corresponding interfaces are integrated into the portable device.
Handling and Acceptance of Mobile Devices for In-Field Staff
The market for rugged PDAs is not that wide compared to standard PDAs. In order to cope with harsh outdoor environment conditions, a rugged version based on the Fujitsu Siemens Pocket Loox device was selected. The rugged version provides:
- shock Resistance acc. to MIL810F (i.e. 26 drops from 122 cm height onto concrete)
- water Resistance acc. to IP65 (only as long as the seal on the microphone hole is maintained, otherwise acc. to IP54)
Further a specific battery pack was used providing supply up to 12,5 hours.
In practical handling it showed, that the PDA can only be recharged, when a part of the closure is removed by means of tools, which hampers practical usage. During outdoor tests and demonstrations, it showed further, that the display reflectance hampers the view to the data.
In general the ergonomics is always a trade-off to the size of a device. Although in the design of the PDA application care has been taken to minimise the need for using the stylus, this cannot completely circumvented (e.g. to mark POIs or to enter text messages). It is expected, that with further feedback and iterations of the PDA GUI, the ergonomics for the use in field can be optimised to the users' needs.
Another topic concerning usability is the size and weight of the MCB. It is expected, that carrying the separate MCB in a back pack is acceptable to routine outdoor operations. In emergency and rescue operations it is expected, that another solution has to be found, in order not hamper the infield rescuers in their core tasks.
Use of SAR EO Data
The avalanche map as well as the snow cover map are in good agreement with validation data and are produced on a regular base using C-band satellite data. Limitations of the products are due to the observation geometry, repeat time, and the wavelength of the available radar systems. For the improvement of the presented products and the development of new products the proposed CoReH2O mission is of particular interest. With its two frequency ScanSAR and at a repeat rate of 3d and 15d it will become an important source for new cryosphere related applications.
Due to the higher frequency the system is much more sensitive to the snow structure than the current C-band satellites and the concurrent measurements at two frequencies allows to be sensitive to different structure sizes and with different penetration depths at the same time. Modeling and the sparse data available at these frequencies prove the potential.
A lack of concurrent reference data of snow at 9.6 GHz and 17.2 GHz with proper ground information was identified. The lack of reference data is a major constraint for model development and validation as well as for application development. It was stressed that the characterisation of the snow cover needs to be reproducible and feasible in the field. Unfortunately the widely used snow characterisation is not well suited to map the relevant structural parameters for the scattering. A promising description of the snow structure is by using the correlation length.
The two applications developed in the scope of the Assist project work with current SAR systems and are quite robust. However for both products limitations remain. Avalanche mapping is difficult on slopes because of layover and foreshortening. Furthermore the degree of visibility of the avalanche depends also on the avalanche type. Dense or wet snow gives a stronger change in signal than dry snow avalanches. The snow cover map is very reliable for the isothermal state of the snow pack. During the period when the snow thaws during the day and freezes again in the night, the time of acquisition is important.
The methods to support landslide survey and landslide monitoring are based on differential SAR interferometry and interferometric point target analysis. Both technologies proved to be well suited. The strong topography in the area under investigation is challenging and needs to be addressed. Both landslide survey as landslide monitoring benefit from the availability of precise height information. In both cases it was possible to receive a more robust processing and product quality through the use of a height dependent atmospheric model. Results received are in good agreement with existing information.
Use of Optical EO for landslide susceptibility analysis
The landslide susceptibility analysis using univariate statistical models is a complex and sensitive task. The selection of appropriate input parameters and representative training data sets are crucial for the successful application of any model. The resulting quality of the applied functional models is directly dependant on the quality of the inputs in terms of spatial resolution and classification accuracy. In order to fulfil these requirements it was a major task to look for the best data sources available. The use of QuickBird data was an excellent choice in a series of possible spaceborne remote sensing data due to its very high spatial resolution. The classification of QuickBird data proved to deliver statistically accurate landcover results, which later have been integrated as variables in the susceptibility analysis. Additionally, a 'pseudo-stereo' image was generated from the QuickBird satellite image which has shown to be very useful for the visual interpretation of landslides in a time- and cost-saving manner.
The development of automatic and semi-automatic tools for a fast and cost-efficient derivation of these variables has to be considered as a step further. It can be stated that the surplus of the above described investigations can be used for summer risks as well as for winter risks. The extent of forest, the occurrence of gaps as well as the curvature in the terrain are, for example, also important information for avalanche analysis. In this context it has to be emphasized that the delineation of the upper forest border is very important in mountainous terrain and a key element in the generation of the forest mask. Until recently the delineation of the upper forest border was a cumbersome work based on field work and/or aerial photo interpretation. In this investigation an algorithm was developed to derive it from QuickBird data automatically on a very high statistical accuracy, and this can be considered as an important step for alpine related investigations.
Another excellent data source was the incorporation of LiDAR data from a flight campaign initialised during the course of the project by the local government of Tyrol. It could be demonstrated, that the different qualities of the used DTMs are clearly discernable in the final susceptibility maps. Although the difference in the overall statistics is only moderate, it is recommended to use a DTM with the finest resolution available, in order to pinpoint the hazardous spots in detail.
The applied statistical model for landslide susceptibility could not yet be evaluated in real situations, since no recent landslides within the study area have occurred during the lifetime of the project. Up to now the only possible way for its evaluation was to focus on the approach encompassing a model-development area and a model-evaluation area. This test showed that the model is not unrestrictedly transferable. However, the results computed from the "weights of evidence model" showed a realistic distribution of the classes according to the experience from the field campaign and the knowledge on prior hazardous events.
It can be concluded, that the anticipated goals of third scenario are successfully realized on a high technical level, yielding excellent results from new remote sensing data sources. The transfer of the derived models to other regions was tested, but could not be performed successfully during this project. Reasons can be found in the general sensitivity of the models and in the regional specifications. More investigations are needed in the context of model development, which was not part of this project. Future research needs in this field can be seen in the development of tools for automatic extraction of buildings, which is important in vulnerability and risk analysis.
Recherche de données OpenAIRE...
Une erreur s’est produite lors de la recherche de données OpenAIRE
Aucun résultat disponible