Lars Wilhelmsson

Models for Predicting Wood Properties in Stems of Picea abies and Pinus sylvestris in Sweden

To study and model the variation of wood properties, sample trees were selected from 42 Norway spruce and 20 Scots pine stands covering a wide variation in climatic and site conditions, stand maturation and tree sizes. Plot and tree measurements were followed by sampling wood from different heights in each sample tree and laboratory measurements of wood properties. Mixed linear and non-linear prediction models were developed using diameters, number of annual rings and climatic indices as explanatory variables. The variation in spruce properties explained by these variables was: basic density 50%, latewood content 52%, juvenile wood diameter 85%, heartwood diameter 94% and bark thickness 76%. The corresponding values for pine were 59, 54, 79, 92 and 85%. Random among-tree variance was an important contributor to the remaining variation for density and latewood. In general, only a minor part of the random variation was related to variance between stands. Predictions derived from the models for density and juvenile wood in both species, and heartwood in pine showed good agreement when validated with data sets from two other studies. The resulting models may be used for predicting wood properties in forest planning and in bucking computers in harvesters, provided that the essential information is available.

Genetic parameters of growth and wood quality traits in Picea abies

Genetic parameters were estimated for wood and growth traits in two 19-yr-old clonal trials and a 40-yr-old full-sib progeny trial of Norway spruce [Picea abies (L.) Karst.]. In the clonal trials high (>0.4) broad-sense heritabilities were found for wood density traits, lignin content, number of internal cracks, growth traits, spiral grain and number of resin canals. Moderate (0.2–0.4) heritabilities were found for tracheid lumen diameter and cell wall thickness, microfibril angle and tracheid length, while low heritabilities (<0.2) were found for pulp yield, fibre strength, wood stiffness and wood colour. Lignin content and pulp yield showed low genetic variation, whereas the genotypic coefficient of variation for most other traits ranged between 5 and 15%. Most traits showed low levels of genotype by environment interaction. Among the wood properties, latewood proportion, earlywood density and ring density showed significant, adverse correlations with volume in both clonal trials.

Two models for predicting the number of annual rings in cross-sections of tree stems

Abstract The number of annual rings is an important parameter in many models for predicting stem wood and fibre properties. To develop applicable tools for characterization of properties, two models (1 and 2) for predicting the number of annual rings in cross-sections from arbitrary heights in stems were evaluated and adapted to fit trees of Norway spruce and Scots pine in Sweden. Model (1) was a new model based on a diameter measurements (over bark) at an arbitrary stem cross-section, related to diameter at breast height (over bark) and tree age at breast height (1.3 m) as explanatory variables. To establish a reference and possible alternative for application it was compared with a previously published model (2) aimed at predicting the number of annual rings at log ends, based on tree height, log end height, tree age and stem taper. Both models were fitted using data relating to 248 stems from 42 stands of Norway spruce and 116 stems from 20 stands of Scots pine. Tree age at breast height varied from 16 to 207 years for spruce, and from 18 to 129 years for pine. Both models showed great flexibility when the parameters were fitted to measurements for each individual tree. Standard prediction errors for cross-sections were estimated. For spruce, the error was 5.9 rings for model (1) and 5.8 rings for model (2). For pine, the error was 7.8 rings for model (1) and 8.4 rings for model (2). The models provide comparable results and are both useful for operative predictions of the number of annual rings. Model (1) may have advantages as it operates with tree age and two diameter measurements only (cross-section of interest and breast height) and therefore is simple to use and, for example, implement into a harvester measurement system.

A model approach to include wood properties in log sorting and transportation planning

ABSTRACT There is a trend that sawmills are more focused on particular valuable products in their production. This has led to an increased demand for sawlogs that are better adapted to the target products and production efficiency. Depending on the product being produced there are different log properties which are better adapted for certain products than others. Sawmills can require hard constraints on log properties such as length, diameter, internode length and sound knots. Some properties are not required but are desired as they make the production more efficient or increase the frequencies of preferred products. In these cases, we include an added value corresponding to what the industry is willing to pay for improved adaptation of the raw material. To achieve this, we propose an optimization model that integrates logging operations (bucking and forwarding) at harvest areas, transportation planning and flexible description of demand at sawmills. High flexibility for sorting at harvest areas may require additional piles of different properties to be generated. Instead of using a large number of special assortments, we allow many sorting alternatives depending on the requirements used at the industries. The transportation planning decides on the flows between harvest areas and sawmills while considering demand and supply. Even if many potential piles are used in the planning model, only a few may be used in practice. We present computational results based on 16 synthesized geographically distributed harvest areas, each representing all regional variation of mature sample trees from the Swedish National Forest Inventory and a number of sawmills.