Critical loads - assessment of uncertainty

Sammanfattning: The effects of data uncertainty in applications of the critical loads concept were investigated on different spatial resolutions in Sweden and northern Czech Republic. Critical loads of acidity (CL) were calculated for Sweden using the biogeochemical model PROFILE. Three methods with different structural complexity were used to estimate the adverse effects of SO2 concentrations in northern Czech Republic. The methods were employed to derive critical levels of SO2. Data uncertainties in the calculated critical loads/levels and exceedances (EX) were assessed using Monte Carlo simulations. Uncertainties within cumulative distribution functions (CDF) were aggregated by accounting for the overlap between site specific confidence intervals. Aggregation of data uncertainties within CDFs resulted in lower CL and higher EX best estimates in comparison with percentiles represented by individual sites. Data uncertainties were consequently found to advocate larger deposition reductions to achieve non-exceedance based on low critical loads estimates on 150x150 km resolution. Input data was found to impair the level of differentiation between geographical units at all investigated resolutions. Aggregation of data uncertainty within CDFs involved more constrained confidence intervals for a given percentile. Differentiation as well as identification of grid cells on 150x150 km resolution subjected to EX was generally improved. Calculation of the probability of EX was shown to preserve the possibility to differentiate between geographical units. Re-aggregation of the 95%-ile EX on 50x50 km resolution generally increased the confidence interval for each percentile. The majority of the grid cells on 50x50 km resolution could not be unambiguously classified as exceeded or non exceeded. Significant relationships were found between forest decline and the three methods addressing risks induced by SO2 concentrations. Modifying SO2 concentrations by accounting for the length of the vegetation period was found to constitute the most useful trade-off between structural complexity, data availability and effects of data uncertainty. Data uncertainties in CL and EX estimates were found to be efficiently mitigated by reducing data uncertainty in the critical limit of the chemical criteria - the BC/Al ratio. The distributed CL and EX assessment on local level in Sweden was found to be efficiently improved by enhancing the resolution of the underlying vegetation map.