Main content area

Estimating chemical soil parameters for sample plots of the Austrian Forest Inventory

Lexer, ManfredJ., Hönninger, Karl, Englisch, Michael
Forstwissenschaftliches Centralblatt 1999 v.118 no.1-6 pp. 212-227
altitude, ecosystems, forest inventory, forest soils, mathematical theory, nitrogen, organic carbon, prediction, regression analysis, simulation models, soil surveys
The spatially explicit application of ecosystem simulation models for large-scale scenario modelling is seriously hampered by a lack of quantitative site data for model initialization. In this contribution an approach based on the Bayesian probability theory is applied to estimate pH-value, organic carbon and nitrogen (total) pools of the upper mineral horizont (0–30 cm) for sample plots of the Austrian Forest Inventory. The predictive models are calibrated from data of 481 sample plots of the Austrian Forest Soil Survey. In a preliminary validation attempt Cohen’s kappa (for ordinal data), linear regression analysis, the modelling efficiency as well as confidence, prediction and tolerance intervals (for metric data) according toREYNOLDS (1984) are applied. It was shown that reliable estimates for chemical soil parameters are possible from quite general site characteristics such as altitude, orientation, slope, topography, soil type and geology.