The need for the community of earth science to process and analyze a huge amount of data from diverse distributed sources is of paramount importance. Due to the wide variety of data formats and structures of existing earth science data, scientists have to spend a significant amount of effort to convert the data from their original heterogeneous formats and structures into usable information: such as writing specialized computer front-end programs ¨C one for each kind of datasets before using these data to conduct any scientific analyses

Another challenge that earth scientists often face is the difficulty in effectively exploring and visualizing a huge amount of data, especially the multi-dimensional data, which could then significantly deteriorate researchers' efforts from gaining meaningful knowledge out of the large amounts of data under investigation. However, most data publishers do not yet provide these services at present. On the other hand, it is difficult and very time-consuming for earth scientists to develop advanced data analyses and visualization applications by themselves alone.

In order to facilitate earth scientists to use, share and visualize data more efficiently and effectively, so that they can spend less time as computer programmers and therefore focus on researches in their areas, we have developed an integrated data management, retrieval and visualization system for hydrological studies. Our system covers many heterogeneous data sources with different data structures and formats, and is able to manage, retrieve, share, analyze, and visualize large data volumes easily. The system is designed with full flexibility, extensibility, scalability, uniformity, transparency and heterogeneity. XML based metadata mechanism is the foundation of our data management system. The dynamically generated query GUI (Graphical User Interface) in our system makes it easy and convenient for scientists to access and retrieve heterogeneous datasets. Scientific data visualization methods are developed to display the huge amounts of data graphically to help researchers have better understanding of the data and gain valuable insights of the datasets under investigation. Without knowing any information of the physical storage location, content, structure and format of each dataset instance, and without programming a single line of codes, scientists can now query heterogeneous data easily, and view and understand the retrieved data in analytical and graphical ways.