Abstract

Reinds GJ, de Vries W (2010) Uncertainties in critical loads and target loads of sulphur and nitrogen for European forests: Analysis and quantification. Science of the Total Environment, 408 (8): 1960-1970. [10.1016/j.scitotenv.2009.12.001]

Keywords

Bayesian calibration Forest soils Critical loads Target loads Uncertainties

Abstract

An analysis of the uncertainties in critical loads and target loads of sulphur (S) and nitrogen (N) for 182 European forest soils was carried out using the Very Simple Dynamic (VSD) model. The VSD model was calibrated with a Bayesian approach using prior probability functions for model parameters based on literature data, data from 200 Dutch forest sites and from simulated denitrification rates from a detailed ecosystem model. The calibration strongly improved the fit of the model to observed soil and soil solution concentrations, especially for pH and base saturation. Calibration also narrowed down the ranges in input parameters. The uncertainty analysis showed which parameters contribute most to the uncertainty in the critical loads and target loads. Base cation weathering and deposition and the parameters describing the H-Al equilibrium in the soil solution determine the uncertainty in the maximum critical loads for S, CLmax(S), when based on the aluminium to base cation (Al/Bc) criterion. Uncertainty in CLmax(S) based on an acid neutralizing capacity (ANC) criterion is completely determined by base cation inputs alone. The denitrification fraction is the most important source of uncertainty for the maximum critical loads of N, CLmax(N). N uptake and N immobilisation determine the uncertainties in the critical load for N as a nutrient, CLnut(N). Calibration of VSD reduced the uncertainty: the coefficient of variation (CV) was reduced for all critical loads and criteria. After calibration, the CV for CLmax(S) was below 0.4 for almost all plots; however for CLmax(N) high values occurred for plots with high denitrification rates. Model calibration also improved the robustness of target load estimates: after calibration, no target loads were needed in any of the simulations for 40% of the plots, with the uncalibrated model there was a positive probability for the need of a target load for almost all plots.

LWF Classification

Network: LWF, Category: ISI,