Comparison between parametric and non-parametric methods for the spazialization of forest standing volume by integrating field measures, remote sensing data and ancillary data

• Main Authors: Bertini R, Chirici G, Corona P, Travaglini D
• Format: Article
• Language: Italian
• Published: Italian Society of Silviculture and Forest Ecology (SISEF) 2007-01-01
• Series: Forest@
• Subjects: Forest management; Forest mapping; Spazialization; Multiregressive method; K-nearest neighbors method
• Online Access: http://www.sisef.it/forest@/showPaper.php?action=html&issue=11&msid=439&lang=en

The use of remotely sensed data for forest inventory and monitoring of natural resources is ever increasing. Distinctively, remotely sensed data, integrated with ancillary data, can be exploited for the spazialization of biophysical attributes measured by forest inventories or management plans. Such applications are based on the relationships between the considered attributes and the spectral information measured by multispectral satellite images. Operative applications are commonly based on parametric or, more frequently, non-parametric approaches. The final aim of the present contribution is the spazialization of forest standing volume of various tree species in a study site in northern Italy by parametric (multiregressive) and non-parametric algorithms (<i>k</i>-Nearest Neighbors). The project is based on field data measured in productive forest stands dominated by <i>Abies alba</i> Mill. and/or <i>Picea abies</i> L. in the Provincia Autonoma di Trento (eastern Alpine Region of Italy). Remotely sensed images were acquired by the Landsat 7 ETM+ sensor while ancillary information is given by the altitude obtained from DEM and the site fertility from the GIS of the management plans. The contribution compares spazialization performance of several operative configurations of the tested methods in order to provide guidelines for the operative application of such techniques on vast areas. The study results emphasize the higher suitability of the tested non-parametric methods.