Effects of Delineation Errors in Forest Stand Boundaries on Estimated Area and Timber Volumes

Abstract

A 717 ha forest area dominated by Norway spruce [Picea abies (L.) Karst.] and Scots pine [Pinus sylvestris L.] was used to assess the effects of photointerpreter errors in the placement of stand boundaries on the estimated area of different groups of forest stands (strata) and other land use classes, and on the total timber volume estimate of the forest. The boundaries between thinning phase stands and clearcuts were classified into three different groups according to the length of the tree shadows hindering ground visibility in the aerial photographs. Different distributions of errors in stand boundaries were assigned to the three groups of boundaries. By means of Monte Carlo simulations of errors in each boundary, the placement of each boundary was altered according to a normally distributed random deviate. In total, 125 independent estimates of total timber volume and area of different land use classes were produced. For thinning phase forest, the difference between the mean area of the simulations and true area was –2.0% of true area. The corresponding difference for total timber volume of all stands was –2.1%. For individual simulations, the minimum and maximum biases in timber volume were –4.8% and 3.1%, respectively

Keywords:

• error assessment
• forest survey
• GIS
• Monte Carlo simulation