Abstract
The objective of this study is to demonstrate that Synthetic Aperture Radar (SAR) imagery adds information for Earth Survey applications which is not available from LANDSAT imagery alone, and that this information can be extracted by application of machine processing techniques. If optimum use is to be made of SAR imagery then its areas of impact need to be identified. SAR imagery was acquired of portions of the Delmarva Peninsula on August 8, 1973, by a United States Air Force RF-4 aircraft based at Shaw Air Force Base, in South Carolina. LANDSAT imagery (0% of cloud cover) was acquired on the same date. The SAR was a Goodyear Aerospace NA/APQlOZA (X-Band). A portion of the SAR data film was optically correlated and a section of the imagery near Wallops was selected on the basis of available (though incomplete) ground truth. The SAR data was registered with the LANDSAT imagery of the same scene and these registered scenes were displayed and/or analyzed on several facilities to identify the impact of the SAR channel. Eleven control points were used to register with a LANDSAT geometrically corrected scene. After distortion correction, the mean error along the track was 0.38 pixels with a standard deviation of 0.694 pixels. The registered data was entered into the LARS library and LARSYS was used for classification via a remote terminal to Purdue/ LARS. SEPARABILITY demonstrated preference for SAR data by several classes. Color composites were made both at Purdue/LARS with their image display and at Goddard Space Flight Center with their Image 100. The color composites also demonstrated several areas of value for the SAR data.
Date of this Version
1976