Effects of data uncertainty in Swedish critical load assessment for forest soils.

Abstract

The effects of input data uncertainty on the critical loads andexceedance estimates for Swedish forest soils was assessed usingMonte Carlo simulations of the PROFILE model. The study focuseson the effects of data uncertainty on the 5%-ile critical loadat 150 × 150 km resolution and the 95%-ile exceedance at150 × 150 km and 50 × 50 km resolution.The results indicate that datauncertainty limits the possibility to differentiate grid cellson 150 × 150 km resolution. The confidence interval for agiven percentile can generally be reduced if the uncertaintiesin calculated critical loads are addressed simultaneously forall sites in a grid cell. The resulting best estimates of the5%-ile critical load were found to be lowered, therebyadvocating larger deposition reductions to comply with a givengap closure of exceedance. The results further indicate that thenumber of sites within the grid cells is important for the rangeof the confidence interval for a given percentile.Re-aggregation of exceedance estimate in 50 × 50 km gridcells showed that differentiation may be improved as compared to150 × 150 km resolution. For 70% of the grid cells on 50× 50 km resolution, the confidence interval forcalculated exceedance covers both negative and positive values.