| The remote sensing observations provide us an opportunity to have better understanding of the Earth system as a whole and then to have better predictions for global climate and environmental changes which have significant socioeconomic impacts to our societies. The large volumes of collected or to be collected future remote sensing data give a serious challenge to the on-line Earth science data information system design. The related design issues include data management, data access, data selection, data ordering, data distribution, and knowledge discovery from the data.; This dissertation addresses these issues through our design approaches in two on-line Earth science data information systems: (1) the Virtual Domain Application Data Center (VDADC) system, and (2) a more advanced distributed system, the Seasonal to Interannual Earth Science Information Partner (SIESIP) project. Our approach consists of: (1) a federated system architecture for dealing with the diversity in both user communities and archived data sets; (2) a Web-based interactive user interface for easy data access; (3) a pyramid data model for data management; (4) a content-based browsing approach for data selection; and (5) a spectrum of data analysis and knowledge discovery tools for data browsing and information extraction.; In addition, several Earth science application areas have been explored in this dissertation work. The first area of application is investigating the interannual vegetation variability in the United States to explore a significant relationship between one interannual vegetation variability signal and the El Nino/Southern Oscillation (ENSO) index. Here, we also find possible signatures of long-term vegetation responses due to human-induced effects. This work may be indicating potentially wide economic and ecological applications. The second application is global and regional correlation studies between vegetation and other geophysical parameters. The third application area investigated here is ENSO cause and effect hypothesis study through a machine learning and inference approach. |
