Zoning of forest health conditions based on a set of soil, topographic and vegetation parameters

Saxony, a federal state in the east of Germany, includes regions with a history of high deposition rates of total acidity and conspicuous impairments of the health status of forests and forest soils. Both the reduction of SO2 immissions and extensive forest liming campaigns have led to a visible regeneration of forests, especially in the Saxon low mountain ranges. There is a strong need for maps with landscape-related information about the forest health status as a basis for an ecological underpinning of forest management practices. The aim of the analyses presented was to derive a zoning of forest health conditions in Saxony on the basis of soil, topographic and vegetation parameters.The upscaling approach of the present paper is based on multiple linear regression analyses coupled with geostatistics using a two-stage procedure with global and regional transfers. First, the forest soil monitoring data of Saxony were evaluated using variables derived from a digital terrain model, geological information and soil and stand related parameters available in high spatial resolution as independent variables. In a second step the influence of environmental factors on the medium-term crown defoliation was analysed using the modelling results from step 1 (regionalized soil chemical attributes) as additional auxiliary variables. Spatio-temporally limited damaging events were diminished by the plot-wise and temporal aggregation of defoliation values. Unlike soil chemical data, it was not possible to detect clear auto-correlative spatial structures for defoliation measurements due to the influence of stand age. About 70–80% of the total variance of defoliation could be explained by the multiple linear regression models. Methodological limitations and interpretations are discussed. Stand age, with 35–64% of the explained variance, showed by far the highest influence. This confirms the necessity to quantify the influence of the stand age as a systematic and intrinsic natural factor if the relationship between defoliation and stress factors is intended to be examined. The model performance of the regression algorithms was examined using both an independent validation procedure and cross-validation terms. Finally, the modelled “mean level of defoliation” was mapped using the mean stand age of forests in Saxony (60 years) by means of regression equations as scenario models.