REGIONALIZATION OF CRITICAL LOADS UNDER UNCERTAINTY

The steady-state model PROFILE was used to perform Monte Carlo simulat ions of critical loads of acidity and exceedances of forest soils for 128 sites in the province of Scania, southern Sweden. Statistical test s showed that 100 sites had normal distributed critical loads and exce edances and that the variance of these parameters was statistically eq ual for all sites. Pooled estimates df the standard deviation was 0.19 and 0.31 kmol(c) ha(-1) yr(-1) for the critical loads and exceedances , respectively. Introduction of uncertainties, expressed as confidence intervals, in the cumulative distribution function for critical loads showed that overlaps between percentiles were substantial. The 5%-ile was systematically equal to the 57%-ile using a 67% confidence interv al and equal to the 87%-ile when a 95% confidence level was chosen. Th e overlaps of percentiles cause a reduction of acidic deposition accor ding to the mean value of the 5%-ile to protect only 68% of the ecosys tem area with an 84% probability and not a guaranteed protection of 95 % as if uncertainties did not exist. Thus, uncertainties make it possi ble to advocate reductions to levels of deposition below the 5%-tile o f critical loads.