From sample to spatial dataset - Regionalization of Forest Monitoring Data in Saxony

Enormous damages on woodland in Saxonies mountain regions occur a in the 1970 and 1980s as an effect of air pollution. Due to this an intensive research of damage causes was preceded and a monitoring established. From its inception the monitoring evolved and the used methods extensively harmonised within Europe. Since 1991 exists a sample for the yearly estimation of crown defoliation and the periodical (1992 and 2007) measurement of soil condition also in Saxony. The sample has 280 sample points in a 4 kilometre grid.

Geostatistics makes it possible

Stepwise multiple linear regression analysis enables the prediction of soil chemical attributes (base saturation, pH, CEC, Mg-stocks and C-stocks) in Saxony. The role of spatial variance and dependence of measurements and model residuals were analyzed using geostatistical methods. By using 2D-variogram surfaces anisotropic features could be described. The area-related transfer of sample measurements, the regionalization, reveals chemical soil properties for the forested area of Saxony.

A high intensity of the evaluation was focused on using landscape morphology as a predictor for the soil chemical status. Besides topographic variables (e.g. elevation, sine- and cosine-transformation of aspect, slope, profile and plan curvature, topographic wetness index, stream power index, etc.) and digital information to soil type classification, forest site survey indices and stand characteristics from forest inventory data were the auxiliary variables that provided indirect information about soil chemical attributes due to their availability for the whole forested area of Saxony.

Identifying regions with serious soil status and appropriate treatment

The soil chemical data show plausible and valuable correlation to landscape characteristics. The multiple regression models account for 79 % - 85 % (CEC; figure 2) and 61 % - 75 % (bases saturation) of the variance presented in the sample of Saxony. The study results show that it is possible to regionalize data of the forest soil monitoring (BZE) on a spatial scale relevant for forestry practice and forest management. The cards expose the serious nutrient loss in woodland soils over a widespread area. Especially in the Oremountains, where enormous sulphur and nitrogen immissions occur during decades, exists a lack between the actual and recommended nutrient status.

To discuss and evaluate landscape related influencing factors and their interactions constellations of hypothetical events could be modelled on the basis of the regression equations. Results are the influence of different tree species on soil chemical properties [figure 3] and the model scenarios to effects of forest liming [figure 4]. Hiding the age effects on crown condition it is possible to show soil, orientation and slope effects and the resultant exposure to pollutant emissions more clearly [figure 5].