HCIL has developed a prototype EOSDIS user interface in java, Sun's internet programming language. This interface is divided into two parts, Query Preview and
See this slide show for a discussion on some of the issues behind the design. The presentation of the tcl/tk prototype screens provides a walkthrough the interface.
The Query Preview provides a coarse look at a database. Datasets are selected by clicking on bars with numbers,
either on the map or with the attributes and years. Selecting
bars selects those datasets. Selections within a bar group are
"ORed"; selections between bar groups are "ANDed". The total number
of selected datasets appears at the bottom. When final selections are made, pressing the submit button at the bottom brings up the Query Refinement interface. The interface has a fixed threshold
for dataset selection. When too many datasets are selected, the bar appears
red and the submit button is deactivated.
The Query Refinement provides a more detailed look at those datasets selected in the Query Refinement. The design is not fully implemented. The interface features a starfield displaying intervals, lists of attributes, a map, and a table. The table is not implemented. The map only displays the spatial extent of datasets, and cannot be used for filtering. The interval-field zooms and pans, but does not filter. Click on an interval to highlight its attributes and spatial extent. Click on attribute values to filter intervals.
- These interfaces are for demonstation only. The datasets presented are
artificial. No real NASA EOSDIS datasets are accessed.
- A browser that supports Java is required.
- The interfaces perform best on medium-to-high performance machines with
medium-to-high resolution monitors. Your mileage may vary.
Known Problems/Open Issues
- Sometimes in the past is appeared as though the applet got "stuck" when
loading data. This generally occurs when loading one of the applets for a
second time, or going from the Query Preview to the Query Refinement. Generally, hitting "reload" on the browser fixes this problem. If not, quit the browser and load either of the two halfs of the demo first. They are listed at the end of this page.
- The Query Refinement interface can take an arbitrary number of
datasets, but the interval field becomes very large with a large
number. Techniques to reduce this size are needed.
- The table of datasets in the Query Refinement interface is not
implemented. A gif image of the table from the tcl/tk prototype is displayed.
- The interval-field on the Query Refinement interface can be panned and
zoomed. However, this interaction affects no filtering of datasets.
- Interaction with the map on the Query Refinement interface is not supported. Possible interactions are being explored.
- The display of attributes in the Query Refinement interface works since the number of attributes is small, and the number of attribute values for each attribute is small. How to effectively handle large numbers of attributes, each with a large number of valid values, is being explored.
|Query Preview Query Refinement|