How to detect desert trees using corona images: Discovering historical ecological data

Abstract

Lack of appropriate historical data has seriously impeded research into the extent and effects of deforestation in arid lands, but the declassification of corona satellite images has now made it possible to study long-term changes in arboreal vegetation. The potential and limitations of such images for long-term vegetation studies are scrutinized in the light of this research. High-resolution images (ca. 2.7 m) from 1965 are compared to field data (2003) of individually mapped wadi trees ( Acacia tortilis and Balanites aegyptiaca) from 20 different sites in the Eastern Desert of Egypt. Of trees mapped in 2003 (canopy area (CA) >6 m 2) 70% was detected in the imagery. A spatial classification shows that between 9% and 55% of the population was concealed by landscape elements that reduce image contrast. The study indicates that 97% of the population mapped in 2003 was already present in 1965 and that trees grow slowly and are older than previously assumed. Pollarded trees were detected in the imagery, and the resultant reduction in CA may lead to misinterpretations in change analyses. corona images offer a spatial and temporal dimension for ecological information which other data sources cannot furnish at a comparable cost, coverage, resolution or accessibility.