Prototype Implementation of a Research Environment for Advanced Landsat Monitoring (REALM)

Principal Investigator: Dr. Goward

The application of Landsat-7 ETM+ data to global-scale Earth science questions depends on harnessing advanced computing technologies to automate processing. To this end, we are implementing a prototype computer system (Research Environment for Advanced Landsat Monitoring - REALM), that will preprocess, store, and allow users to query over 1 Terabyte of Landsat data. Preprocessing modules include systematic radiometric and geometric correction, cloud masking, cloud shadow masking, image-to- image coregistration, and atmospheric correction. Data are passed through these preprocessing modules and stored on a 1.2 Tbyte tape robot. User queries will be submitted via the World Wide Web using a procedural query language, and processed on a parallel cluster of Pentium workstations. Queries may either be simple image retrieval and compositing activities, or complex classification and change-detection tasks. When completed, REALM will address specific science questions, including the calculation of Pan-Amazonian deforestation rates, and land-cover classification for the Chesapeake watershed.

For more information, please see the REALM project page at http://realm.umd.edu/.