Pointek, J. (November 1995)
NASA's Earth Observing System Data and Information System (EOSDIS) seeks to manage large volumes of complex data from satellite and field management programs, allowing users to query and search for relevant subsets of data. The challenge in developing an appropriate user interface for this system lies in meaningfully representing these data to a diverse user community in a compact, easy-to-use manner. The dynamic query approach lends itself well to this problem. Dynamic queries involve range searches on multi-key data sets. Users directly manipulate slider or button widgets to specify ranges for each key value. Within a dynamic query interface, changes to query fields are reflected in a visual display of the data almost instantaneously, usually within 100 ms. To reduce lengthy network delays due to the transfer of enormous volumes of data, we plan to implement a two-level query architecture. The initial level will allow the user to narrow the query selection on a coarser scope, while the second refined query level will allow users to locally view and manipulate the previously selected data subset. Because of the need for immediate feedback and display, efficient query computation must be supported using suitable data structures and search algorithms. Previous work in dynamic queries has resulted in favorable performance for smaller data sets (less than 5000) using simple data structures. However, these data structures are poorly suited for large, complex data sets such as those present in EOSDIS, prompting the need for development of new multidimensional data structures and algorithms.