Hautamäki, S., Kilpeläinen, H., Kannisto, K., Wall, T. and Verkasalo, E. 2010. Factors Affecting the Appearance Quality and Visual Strength Grade Distributions of Scots Pine and Norway Spruce Sawn Timber in Finland and North-Western Russia.  Baltic Forestry 16 (2): 217 – 234

Logs and centre-yield sawn pieces of Scots pine (Pinus sylvestris) and Norway spruce (Picea abies) from three regions in Finland and two regions in north-western Russia were studied for the variation in and predictability of grade distributions. Log and sawn piece properties were measured and statistical analyses conducted on the differences between geographical regions, in particular. Sawn pieces were graded for appearance quality and visual strength, according to the Nordic grading rules, into NT grades and T grades (Finnish application of INSTA 142 rules), respectively. The aim was to find out how accurately it would be possible to predict the grade yields from either the log properties or sawn timber properties using binary, or multinomial regression modelling, and, in particular, whether any regional differences remained thereafter.
The type of the log (vertical position and visual grade) was the most important single factor for both species when predicting grade yield using log properties as predictors. When sawn timber properties were used as predictors, the properties related to knots were the most substantial explanatory variables. Geographical region was not the major factor in predicting grade yield in either approach, but more evident for pine than spruce.
Multinomial regression models were used to predict both appearance quality and visual strength grade distribution. They were able to predict correctly 40-50% of the NT grade and 44-59% of the T grade in individual sawn pieces. Binary regression models were used to predict visual strength grade classes in order to have only two response categories, for example, the two highest grades compared to the two lowest grades. These models were the most accurate predicting correctly 76-83% of the dichotomous grade yield. The models predicting grade yield for pine performed better than the models for spruce.
There appeared significant differences in several log and sawn timber properties between the regions. The grade yields from the Novgorod and Vologda regions were lower than in Finland, in general. For spruce, the between-region differences in log and sawn timber properties were much smaller than for pine.

Key words: Pinus sylvestris, Picea abies, sawn timber, appearance grading, visual strength grading, grade yield, Finland, Russia.