Migratory bird habitat monitoring through remote sensing

Abstract

Unsupervised classification of Landsat-TM data was employed to identify habitats important for migratory birds in Costa Rica. The overall habitat classification accuracy was 70 per cent (Kappa correction). Mature forest could be identified with high accuracy (93 per cent) but Landsat-TM classification accuracy for major successional stages was low. Habitat availability and conversion rates from 1976 to 1986 were derived from multidate Landsat imagery supplemented with interpretation of historical air photos to document the specific types of habitat change. The major trend in habitat conversion between 1976 and 1984 was forest clearing followed by establishment of permanent pasture. Some of the pasture land was converted to perennial tree crops by 1986. The implication of habitat modification on groups and species of migrant land birds are discussed.