Effective map revision using satellite imagery

Abstract

Currently, the Royal Australian Survey Corps (RA Svy) revises its topographical mapping by having personnel traverse each map sheet. at regular intervals, seeking to identify differences between the map sheet and what is on the ground. Once change has been found, it is annotated onto the map sheet using techniques that are not much better than dead reckoning. In recent years use of commercial satellite imagery has been seen as a potential improvement over this process. This project investigates this possibility and devises a methodology not only lo monitor landscape feature change but to capture this information to revise a digital topographic database. Landsat Thematic Mapper ™ and SPOT Panchromatic satellite imagery, offering the greatest spatial and spectral resolutions presently available. were acquired for a study area centred on Collarenebri, New South Wales, over corresponding dates in the summers of 1991 and 1992. A suitable processing schema was developed, using MicroBRIAN V3 .1 image processing software, whereby these images were registered to each other. merged to improve the spectral information’s spatial resolution and then the total number of channels reduced to eliminate any redundant information. Updating of the existing 1 : 50 000 map sheet for the study area was accomplished by visually comparing the 1991 enhanced imagery to this map using a Bausch and Lombe Zoom Transfer Scope. Digital change detection was then performed on the J 991 and 1992 imagery. Landsat TM imagery was processed to identify the likelihood of temporal change occurring between these images. The merged images were then processed to capture those instances of observed feature change in more detail. This information was then applied as updates to a digital database created for the study area. Both the positional accuracy and completeness of this captured revision information were then verified in the field using a Global Positioning System as a source of accurate positional co-ordinates. Of thirty-six observed feature changes all were confirmed as having occurred. However, there were two instances where feature change had not been identified, and five occurrences where features had been mis-interpreted. Also it was observed that all changes had a positional accuracy approaching ± thirty-five metres which is marginally outside the ± thirty metres required under current map specifications. While these results confirm that commercial sa1ellite imagery can be used to effect map revision, accuracy of this information is presently limited by the spatial resolution of the satellite imagery available. However, the methodology developed was found to represent a substantial improvement over the existing RA Svy map revision technique.

Date of Award	1993
Original language	English