ERCIM News No.45 - April 2001 [contents]
Grid for Remote Sensing
by Giovanni Aloisio
The continuous monitoring of the earths surface by remote sensing space systems produces huge quantities of raw data that are stored in the ground station segments of the National Space Agencies. These data must be processed to produce useful images and then delivered to the final user. Remote sensing imaging based both on traditional sensors and on more sophisticated radar technologies (SAR-Synthetic Aperture Radars) can provide a potential source of information for users working in many fields, such as archaeology, geology, drawing of maps, ecology and others.
Several Earth Observation Systems (EOS) have been developed by the National Space Agencies and web technologies are usually exploited to simplify the information access. An Earth Observation System (EOS) is made up of a sensor mounted on a satellite or a space shuttle that gathers images of the Earths surface, of a distributed database in which these images are stored and of heterogeneous distributed computing resources which process and distribute them via web to the final users (see Figure 1).
Figure1: Earth Observation System.
Computing resources are needed to transform the raw data gathered by the sensor in an actual image. This preliminary step of processing is what is called pre-processing to distinguish it from the post-processing that may be needed to extract knowledge from the image. The realime pre-processing of SAR raw data requires the use of high performance computing resources.
The traditional static approaches normally used in the construction of EOS systems allow a user to request a pre-processed image of a place (if the required image is found in the distributed archive!) but they lack the intelligence needed to start a specific post-processing in response to high level user requests. On the contrary, a dynamic EOS (DEOS) should be able to manage high level requests such as show me the corn fields in the surroundings of that place or give me the image of that place and find the computer that can process that image in the fastest way and with the lower cost (see Figure 2).
Figure 2: Dynamic Earth Observation System.
To satisfy these requests, a DEOS should have flexible management and control of the distributed resources. Further challenges for a DEOS are represented by the integration of information coming from different sources, eg, the integration of maps and images of the Earths surface with any information associated with the geographic location; the integration of the functions of a friendly web browser with those of GIS and related technologies. A DEOS must also provide secure access to the distributed information and the steering of remote applications.
A Dynamic Earth Observation System can be implemented using Grid technologies. Grids can be thought of as a mixture of software and hardware infrastructure whose aim is to provide a coherent, unified view of geographically spread computing resources, smart instruments and distributed data archives. This seamless integration will enable users to share their resources and to build new classes of applications based on resource pooling and selection.
SARA-Digital Puglia (Synthetic Aperture Radar Atlas-Digital Puglia Project) is a Grid-based Dynamic Earth Observation System. The system is under development at the ISUFI HPC Laboratory of the University of Lecce, Italy, in a joint project with the Center for Advanced Computing Research of the California Institute of Technology and the Italian Space Agency.
The first prototype implements an active digital library of SAR images of the Puglia region of Italy. The purpose of SARA/Digital Puglia is to allow a user to choose from a web browser a geographic area of interest and to specify, using high level requests, the information to be extracted from the selected images. High level user requests usually require not only the retrieval of the selected image but also further post-processing on it. This must be transparently activated on some of the remote machines belonging to the Computational Grid.
A grid portal, the Grid Resource Broker (GRB) has been also implemented, to allow authorised users to create and handle computational grids on the fly. GRB was designed to be a very general grid tool providing locationransparent and secure access to Globus services using a simple web-based graphical user interface. Using the GRB, users do not need to write specialized code nor to rewrite their existing legacy codes. We plan to add other features, and to enhance the basic functionality provided, implementing several scheduling algorithms. We have also derived a library that can be used to build grid enabled applications from GRB. A demonstration of the GRB capabilities was given in the SuperComputing 2000 conference, held in Dallas, Texas. Another demonstration will be given in the context of the NPACI All Hands Meeting 2001, San Diego, California.
Giovanni Aloisio - University of Lecce
Tel: +39 30 832 320221