Forest cut-over mapping from SPOT satellite data

Abstract

Abstract Cut-over area and boundary placement accuracies from analogue SPOT Panchromatic (PLA) and Multispectral (MLA) Linear Array images were determined for forest cut-overs typical of the Canadian Prairies. Cut-overs were mapped using a Procom-2 to optically project images on to a photogrammetrically-produced planimetric base map. Overall area accuracies for the SPOT PLA and MLA images were 90-7 per cent, and 89-6 per cent, respectively. The influence of increasing accuracy with increasing cut-over size must be considered, however, when interpreting these figures. Proportional regression models were fit to improve the estimates of cut-over areas. On average, boundary placement errors were 24 m for the SPOT PLA, and 26 m for the SPOT MLA. These errors were influenced by the 25 times enlargement factor encountered in the optical projection, and the scale of the base map itself. The base map scale of 1:20 000 was a limiting factor in this study as both plotting errors and pen line widths contributed to existing map errors. Under the forest conditions evaluated, SPOT images offered an approximate five to one cost savings in data acquisition, and are more efficient to interpret and map in comparison to conventional air photos. Additional informationNotes on contributorsR. V. DAMS Formerly with Intern Technologies Ltd