Experimenting the design-based k-NN approach for mapping and estimation under forest management planning

Abstract

Estimation and mapping of forest attributes are a fundamental support for forest management planning. This study describes a practical experimentation concerning the use of design-based k-Nearest Neighbors (k-NN) approach to estimate and map selected attributes in the framework of inventories at forest management level. The study area was the Chiarino forest within the Gran Sasso and Monti della Laga National Park (central Italy). Aboveground biomass and current annual increment of tree volume were selected as the attributes of interest for the test. Field data were acquired within 28 sample plots selected by stratified random sampling. Satellite data were acquired by a Landsat 5 TM multispectral image. Attributes from field surveys and Landsat image processing were coupled by k-NN to predict the attributes of interest for each pixel of the Landsat image. Achieved results demonstrate the effectiveness of the k-NN approach for statistical estimation, that is compatible with the produced forest attribute raster maps and also proves to be characterized, in the considered study case, by a precision double than that obtained by conventional inventory based on field sample plots only

Estimating crown defoliation of Scots pine (Pinus sylvestris L.) trees using small format digital aerial images

“…However, the main advantage of non-parametric prediction is that it avoids the need for explicit models but it does presume that the total variation of the variable under focus is represented in the B-observations (Tomppo 2005). Usually, the operational use of k-NN should be preceded by the optimization of the settings for the estimator and the list of auxiliary variables, using, e.g., the "Leave One Out" cross-validation technique (Tomppo 2005, Mattioli et al 2012. The approach was to use the same approach for both types of images analyzed -all the image characteristics, t=2 and k=10.…”

K-NN FOREST: a software for the non-parametric prediction and mapping of environmental variables by thek-Nearest Neighbors algorithm

“…The software is driven through a GUI which guides the user through the different phases of the method, from input data formatting, to algorithm optimization by LOO cross validation, and to final spatial prediction of the response variable. The software K-NN FOREST was extensively tested in several research applications [Chirici et al, 2008;Chirici et al, 2010;Lasserre et al, 2011;Mattioli et al, 2012] and it is now freely distributed on-line. Private or public stakeholders involved in environmental monitoring and assessment may benefit of this software to derive spatial predictions of environmental attributes useful for quantifying the conditions and trends of environmental resources [Corona, 2010].…”

Modeling of forest above-ground biomass and evapotransportation dynamics

“…The attractiveness of the method is that it is distribution free in that it does not rely on any underlying probability distribution for estimations, but on forest conditions (Mattioli et al, 2012). When there is good representation of ground sample plots, the method has performed well for biomass estimations.…”