The increment of wood is a central factor in the economical and ecological control of a forest enterprise. The development of models to determine this is therefore a matter that has been of concern to forestry research for a long time. The first form of models of this type were yield tables, which determine the quality of a site and thus its increment potential through the stand continual growth. The commonest tables in use today were constructed in the middle of the 20th century. Today, they provide an inadequate illustration of current increment process. And yet, yield tables are still very important in practice. The objective of the work was therefore to find out how high the discrepancy between the actual increment in forest stands and increment predictions from yield tables is and to determine the factors that influence this discrepancy. The influencing factors are to be used to construct models to correct yield table predictions. Increment data of a high level of statistical precision were obtained from the first repeated measurement of continuous forest inventory in 25 stations of the Bavarian State Forestry Service and a few private forest stations. They serve as references for the increment in the last 15 years. Their percentage deviation from the corresponding yield table prediction was the dependent variable for various forms of regression models, which were formulated in each case for the species spruce, pine, fir, European larch, beech and oak. The independent variables for the regression models were stand structure data and information from site mapping. The digital elevation model was used to assign the height above sea level, aspect and slope to each random sample point. Data from the forest health survey provided information on the needle/leaf loss of a stand. Random sample points of forest soil inventory provided data on annual nitrogen deposition. The survey grid for forest health survey and forest soil inventory did not cover the same area as the inventory grid for continuous forest inventory. Consequently, a geostatistical method was used to compress the dot-type information from forest health survey and forest soil inventory, such that assignment to the sampling points for the continuous forest inventory was possible. The simple form of the regression model incorporated stand age, site index, stocking degree and height above sea level as independent variables. Because of the low volume and the ease of availability of the required information, these models make it easy to correct yield table predictions. Based on the species, the differentiated form of the model contains additional factors such as proportion of species, slope and nitrogen deposition. For site models, site information were incorporated into the model instead of site index. To be able to account for regional site differences in increment, random sample points were divided into 7 groups of comparable climatic conditions by using a cluster analysis. These climatic groups were incorporated into the site models as Z-variables from a contrast coding. Also, in order to incorporate local site information from site mapping, the nominal site figures had to be adapted to a metric scale level in a previous stage. The climatic group models, which were the fourth model form, were separately parameterised for each tree species and climatic group. Major factors that explained the differences between the measured increment and the yield tables were found in the stand parameters such as age, stocking degree and site index. In particular increment well above the yield table level can be expected in old and dense stands with poor site indices. In comparison with yield table, the positive derivation in increment decreases as the height above sea level increases. The nitrogen saturation of many forest soils is probably the reason why the increment in regions with low annual nitrogen deposition is comparatively higher. The site-describing variables in the site models show only a minor influence on volume increment difference from the yield table. The models show that on the same site, site indices in younger stands must be better than in older stands. The models were validated using data from the continuous forest inventory within the same inventory period, which were not incorporated in the modelling process. A clear improvement in the prediction precision was found for all four model forms. The differences between the individual model forms were minimal. Generally, it can be said that the average deviation of the measured increment at the random sample point from the corrected yield table value was reduced when compared with original yield table prediction. The spread of the deviations was smaller and its distribution was more symmetrical. The suitability of the explanation models as tools for predicting increment will only be justified after a second inventory period with different weather conditions and when growth condition may have further changed.
«
The increment of wood is a central factor in the economical and ecological control of a forest enterprise. The development of models to determine this is therefore a matter that has been of concern to forestry research for a long time. The first form of models of this type were yield tables, which determine the quality of a site and thus its increment potential through the stand continual growth. The commonest tables in use today were constructed in the middle of the 20th century. Today, they pr...
»