Rig. Morrison, THE USE OF REMOTE-SENSING TO EVALUATE SHOREBIRD HABITATS AND POPULATIONS ON PRINCE-CHARLES-ISLAND, FOXE BASIN, CANADA, Arctic, 50(1), 1997, pp. 55-75
Landsat-5 Thematic Mapper imagery was used to produce a 17-habitat cla
ssification of Prince Charles Island, Fore Basin, Northwest Territorie
s, through a combination of supervised and unsupervised approaches. Br
eeding shorebirds and habitats were surveyed at 35 study plots in July
1989. Habitat-specific breeding densities calculated from these obser
vations were used to estimate total populations of breeding shorebirds
on the island based on areas of habitat derived from the classified i
mage. Breeding densities were further modelled in two ways: first, to
adjust for distance from the coast, where regression analyses found a
significant relationship between distance and density, and second, to
include only those pixels of areas considered suitable for breeding, u
sing results of a proximity analysis to determine habitat associations
between known breeding locations (pixels) and other habitats. Six spe
cies of shorebirds were found breeding on Prince Charles Island, with
a combined population (after modelling) estimated at 294000 pairs. Com
parison of breeding densities and estimated populations of shorebirds
with those recorded at other arctic locations indicated that Prince Ch
arles Island supports highly significant numbers of shorebirds, especi
ally white-rumped sandpipers and red phalaropes. Comparison of referen
ce areas of known habitat with those on the classified image indicated
classification accuracy averaged over 90%. Remote sensing appears to
offer a reliable method for assessing habitats and regional breeding p
opulations of birds in at least some areas, providing that classificat
ion methods are carried out in a carefully controlled manner. Use of t
he method over broad areas of the Arctic would require considerable wo
rk to recalibrate imagery for different geographic regions.