The mapping of stands of Parana pine (Araucaria angustifolia (Bert.) O. Ktze) in the forest of south-west Parana State (Brazil) using computer-aided analysis of Landsat MSS data
This study examines the value of Landsat data for mapping stands of Parana Pine (Araucaria angustifolia (Bert.) 0. Ktze) in the natural forests of southern Parana State, Brazil. This species is economically the most important forest tree in southern Brazil and estimation of its reserves and the rate of exploitation are important. Two forest areas were selected for detailed studies. For one area (Quedas do Iguacu) Landsat data in both computer compatible tape (CCT) and transparency format was used. No aerial photographs were available but an existing forest map was used as ground truth. For the second area (Mangueirinha) Landsat CCTs and aerial photographs were available. From the latter a forest cover type map was produced and used as ground truth. Additionally some field checking was undertaken. For the Quedas do Iguacu area visual qualitative temporal analyses were carried out on the products generated from transparencies of six different Landsat scenes. On these Parana Pine stands could be recognized in the MSS bands 6 and 7 and in colour composites generated from the MSS bands 4, 5 and 7. For a selected subarea a supervised computer classification using CCT data was tested success-fully for mapping both mature Parana Pine stands and reforestation areas. A supervised classification using transparencies of MSS bands 5 and 7 that were scanned and digitized by microdensitometer successfully mapped Parana Pine stands but failed to discriminate reforestation areas. For the Mangueirinha area CCTs of Landsat imagery acquired in spring and in winter were used for mapping Parana Pine stands. Both visual and computer classification methods were employed. For the latter, a small area of Parana Pine stands with two different crown densities (as ascertained from study of aerial photographs) was selected. Different combinations of MSS bands were tested using supervised and unsupervised classifications. The results showed that using a supervised classification combinations of MSS bands 5 and 7 were most effective, but that only dense stands of Parana Pine could be mapped accurately. The spring imagery yielded better discrimination than that acquired in winter.