Multisource Inventory Methods

A practical two-stage procedure for determination of stand parameters of entire forest stands by means of fi eld plots and airborne laser scanner data has been developed and tested. In the fi rst stage, geo-referenced fi eld sample plots are related to the corresponding laser data, and relationships (regressions) between fi eld data and various canopy height and canopy density metrics that can be derived from laser data are estimated. In the second stage, the entire forest area in question is divided into a regular grid with cell size equal to the size of the fi eld sample plots. Stand parameters of each grid cell are predicted by means of the estimated regression models and the laser data that correspond to each cell. Stand parameters of entire stands are fi nally computed from the individual cell predictions. The procedure has been tested in a forest area of 1000 ha. 173 systematically distributed sample plots in very young (h < 10 m) to mature forest were used to regress dominant height, mean height, mean diameter, stem number, basal area, and volume against the laser data. The derived models were used to predict corresponding parameters in 73 subjectively selected test stands. Verifi cation of stand estimates revealed a negligible bias for most parameters. The precision was promising: the RMSEs of the predictions were 0.61–1.17 m, 0.57–1.33 m, 1.32–1.61 cm, 128–400 ha–1 (16.9–22.2%), 2.33–2.54 m2 ha–1 (8.6–11.7%) and 18.3–31.9 m3 ha–1 (11.4–14.2%) for mean height, dominant height, mean diameter, stem number per hectare, basal area, and volume, respectively (Næsset 2002, Næsset & Bjerknes 2001). The proposed procedure was applied in a practical stand-level inventory in an area of 6000 ha during summer 2001.