On the analysis of cubic smoothing spline-based stem curve prediction for forest harvesters

In the cut-to-length (CTL) harvesting system the telling, delimbing, and bucking processes take place at the harvesting site. The optimal Cutting points along the stem can be determined if the whole stem Curve is known. In practice, however, it is not economically feasible to measure the whole stem first before crosscutting. and hence the first cutting decisions are usually made when only a short part of the stem is known. Predictions are used to determine the cutting pattern to compensate for the unknown part of the stem. In this paper our interest focuses on stem Curve prediction in a harvesting situation and we Study it modified version of a cubic smoothing spline-based prediction method devised by Nummi and Mottonen (T. Nummi and J. Mottonen. 2004. J. Appl. Stat. 31: 105-114). The method's performance was assessed in five different final felling stands of spruce and pine, collected by harvesters in southern Finland. The results for the spline approach are very promising and show the superiority of the method over the linear mixed-model-based approach of Liski and Nummi (E. Liski and T. Nummi. 1995. Scand. J. Stat. 22: 255-269) and also over the approach based on the variable-exponent taper equation of Kozak (A. Kozak. 1988. Can. J. For. Res. 18: 1363-1368).