GRID technologies for environmental monitoring using satellite data
Supported by: STCU
Main organization: SRI NASU-NSAU
Current Foreign Collaborators: George Markowsky from University of Maine
Project Manager: Kussul Natalya Nikolayevna
The overall objective of this project is the development and implementation of efficient tools for distributed computing that will provide simple and transparent way for the solution of problems of high complexity in different domains. Special attention will be paid to space images processing.
Nowadays, the solution of complex problems using satellite data in such area as environmental monitoring requires the analysis and intelligent processing of large volumes of information. This kind of information can be of different nature, e.g. Earth observation data, hydrometeorological data, geological data etc., and as a rule, this information is distributed, that is stored on different sources. That is why, the problem of integration of distributed informational and computational resources, as well as the development of intelligent methods for data processing in the framework single system, is very crucial.
To solve this problem it is needed to develop new intelligent information technologies that will enable the integration of distributed informational and computational resources and will provide the following properties, such as scalability, functioning in heterogeneous environment, transparency to users, etc. Taking into consideration high volumes of data to be processed (particularly, the size of space images), distributed nature of data sources, as well as the need of complex mathematical models construction, there is a crucial need for the development of efficient intelligent methods for distributed data processing. These methods should be based on distributed computing and efficient network interaction.
In the last years GRID technology (Global Resource and Information Database) is used to enable the solution of complex problems, particularly those arising in environmental monitoring. GRID enables the integration of different distributed data sources and efficient use of computational resources that separated from each other. Taking into consideration distributed nature of data that is used to solve the problems of environmental monitoring, as well as the need of integration of computational resources in order to provide complex data processing for decision-making, it is advisable to use GRID technology. Up to this moment this technology is on the stage of development and requires the development advanced tools that will reduce the complexity of GRID system development.
Within the project computational environment for the modeling of physical processes on the Sun, in Earth's atmosphere and soil will be developed. It will allow to integrate local networks of users' workstations with multiprocessor nodes, and will be based on open-source software such as Condor, JXTA, Alchemi.NET. Algorithms of load balancing and tasks queue control in computational network will be investigated. Methods of resources sharing in several computational systems will be developed (the creation of single computational network for the solution of physical modeling problems).
Corresponding user environment will be developed. The system and its components will be designed using up-to-date methods of system analysis in the area of information technologies with the aim to virtualize computational resources. Corresponding components for solution of physical modeling problems will be developed, and the access to the existing function libraries will be organized. RAD tools (Rapid Application Development) will be developed in order to enhance the efficiency of new computational applications development and adaptation of others to run in computational environment. Corresponding user interfaces will be developed in order to simplify the work with computational network (application execution, visualization of application running process and results of processing) and to provide transparent use of computational environment resources by experts for their needs.
Proposed system will be verified and tested on applied problems of environmental monitoring. Methods and corresponding parallel algorithms of temporal interpolation of Earth's surface images will be developed. Parallel algorithms of modeling of dynamics of basic processes occurring in multi-component soils will be developed and applied in computational network. Proposed methods of distributed data processing and parallel algorithms will be based on GRID, Web-services and intelligent methods for data processing (including neural networks). Methods for investigation of solar activity and corresponding parallel algorithms will be developed. Problem of solar activity forecasting will be investigated taking into account spatial-temporal structures of different scaling and using both space and ground-based observations. To enable solar activity forecasting, new mathematical models of non-linear phenomena concerned with discrete structure of solar plasma formation will be developed. Models of fractal structures of solar magnetized plasma that self-organizes in cluster and micro-objects will be proposed and verified. Proposed methods will be implemented in software components.
As a result of a project, new methods of physical modeling that will be based on distributed processing of large volumes of information will be developed. Obtained results will allow to virtualize computational resources by creating transparent and friendly user interface for efficient access and use of GRID system. Components, libraries and tools will be proposed in order to reduce the complexity of application development that oriented on the use in computational environment with the aim to solve different scientific problems of high complexity in different domains.