| Popis: |
New technologies, sensors and techniques increase every day the human capabilities to monitoring the ecosystem and understanding the anthropogenic influence in its evolution. Particularly remote sensing, offering the possibility of covering large spatial area with a high temporal frequency, represents the ideal solution to environmental monitoring, but at the same time it is a challenge for any conventional computational resource due to the algorithms complexity and to the huge data volume to process. EOS satellites with their Moderate Resolution Imaging Spectroradiometers (MODIS) view the entire Earth's surface every days, acquiring data in 36 spectral bands. These data are used to make several products, i.e. Oceans Sea Surface Temperature Maps, essential for understanding a number of oceanographic and atmospheric phenomena. On coastal areas the SST product is unreliable requiring ad hoc processing and calibration. These requirements fit better on high performance computing and grid technologies that can provide powerful computational resources and efficient distributed data management. Several grid infrastructures have been proposed in literature for Earth observation projects (Fusco et al., 2006), however they report excellent performance but a too complex computing environment for non grid expert users. In this paper we present a grid portal for remote sensing data processing that allows scientists to access resources, execute and monitor data processing on grid technology using a simple Web browser interface. As an example of services that can be offered in this portal we implement services for producing calibrate coastal area SST maps and for combining these with standard product to obtain improved SST maps. |
