Validation of SILVAH for tolerant hardwoods in Ontario
- Download PDF (1319631)
- This publication is available only online.
Forest Research Information Paper. Issue 128. 1995. pp. 1-28
SILVAH, a stand growth simulator commonly used in the northeastern United States, has been evaluated by comparing predicted and actual growth of tolerant hardwoods in southern Ontario. The data came from 139 stands, unmanaged or managed (thinned), even-aged or uneven-aged. The data were used to test the accuracy of diameter distribution, basal area, mean quadratic DBH, number of trees per hectare, and stand volume predictions for periods from 5 to 20 years. The accuracy of short-term basal area projections, expressed by modelling efficiency, ranged from 62% to 94%. The reliability of stand density projections was especially low (modelling efficiency from 15% to 88%). Despite underpredicting the number of trees in small DBH classes, SILVAH provided reliable predictions of DBH distribution, especially for unevenaged, unmanaged stands. Basal area of stands was underestimated by 2% to 6% on the average for 5-year projections (except for low productivity sites). During the same projection period, the error of predicted individual stand basal area ranged from -22% to 25%. Prediction bias increased with longer projection periods. The trend of underestimating stand density and overestimating mean quadratic DBH resulted from low predicted numbers of small DBH trees. Consequently, the basal area of pole and small sawlog timber classes was underestimated (on the average 1% to 28% in 5 years). A few factors contributed to the poor performance of the model: 1) lack of ingrowth data or lack of an automatic ingrowth submodel; 2) incompatibility of model requirements and data sets related to the range of tree sizes in the input list; and 3) inability of the model to account for differences in site productivity.
Bankowski, Jacek; Dey, Daniel C.; Jim, Rice; Boysen, Eric; Miller, Roj. 1995. Validation of SILVAH for tolerant hardwoods in Ontario. Forest Research Information Paper. Issue 128. 1995. pp. 1-28