This exhibit will demonstrate applications and systems software being developed under our NSF Grand Challenge grant on applying high performance computing to processing of remotely sensed image data. At the core of the exhibit is a parallel image database and geographic information system. This database system stores massive amounts of remotely sensed image data and supports flexible access to that data through queries that combine spatial and temporal constraints on relevant data sets. Using a GUI interface, users will be able to select regions in the world using the querying facility of the GIS, and also select time spans of interest to identify relevant image data. The image database system, written using the CHAOS++ library developed at Maryland to support the development of parallel object-oriented programs, also uses the Maryland Jovian parallel I/O library to organize the database on multiple disks and retrieve the relevant portions of images efficiently. The user will be able to choose from a variety of parallel application codes for processing the data, including producing one of several standard remotely sensed products (e.g., Normalized Vegetation Indices), estimating proportions of ground cover classes using a mixture modeling code, and classifying the imagery into ground cover categories using one of several classification algorithms developed at Maryland.
Equipment:
We will be bringing a workstation to SC95 and will connect to
the IBM SP2 at Maryland over a high speed network (the IWAY). Our
exhibit utilizes over 60GB of image data resident at Maryland
so cannot be practically brought physically to San Diego.
Size and Layout:
We request an exhibit area to include 1x1 blocks.
Exhibit SciNet Connections:
We will need the 10baseT Ethernet drop, but are also planning to
participate in the IWAY. We do not yet have information about what will be
required for that.
Exhibit Furnishings:
No special requests
Exhibit power connections:
No special requests
Shipping:
No special help requested