documentodischeda-ricercaforestale.doc

Ri.Selv.Italia Subproject 4.1 INVENTORYING AND MONITORING OF FOREST RESOURCES AND ENVIRONMENTS Scientific responsible: Vittorio Tosi (C.R.A. – Research unit for forest monitoring and management, Trento) Final report Title: 4.1.3 – The monitoring of arboriculture for wood production Operating Unit: CRA - Poplar Research Institute, Casale M. Responsible of the operating unit: Domenico Coaloa (coaloa@populus.it) Project goal The poplar plantations, main element of arboriculture for wood production in Italy, have an important role in the economy of several Northern Regions, namely Piemonte, Lombardia, Emilia Romagna, Friuli Venezia Giulia and Veneto: in fact, about 60 % of the national poplar area is located in Piemonte and Lombardia (CGA, 2000). Forest inventories are as important for poplar cultivation as they are for silvyculture, since they provide a valuable instrument for territory planning. In this context remote sensing represents a very useful aid for forest inventories and for the frequent updates made necessary by the short rotation (10 years) that characterizes Italian poplar cultivation and the consequently frequent variations in land cover. The analysis of the first aerial photography in 1973 (Cellerino and Lapietra, 1977) and the possibility to obtain an automatic classification of Landsat satellite images in 1974 (Lapietra and Megier, 1977), showed the great potentialities of these new instruments for the estimation of wood production. However, identifying forest plantations is not easy either on low spatial resolution images, owing to the difficulty in recognizing their peculiar layout in rows, or applying conventional classification techniques to very high spatial resolution data, since the images obtained will have a characteristic “salt and pepper” effect. Recently developed approaches cluster pixels sharing a common attribute into a region, and then classify all the pixels within it in the same class. This project had two main