Fig 1.
Part A shows the Cumulative Impact assessment methodology; Part B refers to the general analysis of uncertainty based on the Walker et al. [13] uncertainty matrix and respective uncertainty quantification methods applied: UD = uncertainty description (level 1), SQ—Semi-Quantitative analysis (level 2); numerical uncertainty analysis (UA) and sensitivity analysis (SA) (level 3). In Part C, the contributors in the different modelling phases are reported.
Fig 2.
The Adriatic and Ionian region.
Fig 3.
Scheme of the three-level analysis of uncertainty.
Modelers initially describe uncertainty in level 1. Semi-quantitative analysis is performed in level 2, while statistical quantitative analysis in level 3.
Table 1.
Locations of uncertainty.
Table 2.
Descriptors of uncertainty.
Fig 4.
Combinations between level and nature of uncertainty give place to 6 different types of uncertainty magnitudes; elaborated from [10,13].
Table 3.
Factor groups, factors and factor ranges applied in the Monte Carlo (MC) simulations in this study.
Fig 5.
Cumulative impacts scores derived from the baseline run for the AIR.
Cumulative impact scores varies from 0.0 (no impact) to 8.5; most impacted areas are indicated in blue frames.
Fig 6.
Contribution of human uses (U) to the CI scores for the AIR.
“Use presence” represents the percentage of the AIR where the use is located; “impacted cells” represents the percentage (%) of cells that are impacted by the use, considering the distance model at which the pressure takes place; “scores” represents the contribution to the total CI score of the use in percentage.
Fig 7.
Ranking of the environmental components (E) that are majorly affected by the CI scores in the AIR.
“Presence of environmental components” represent the percentage of the total cells where E is located, “impacted cells” represents the percentage of cells where E is located that are impacted by U, “score” represents the contribution to the total CI score deriving from E.
Fig 8.
Data availability index (DAI) for the AIR.
Dark blue indicates where all data sets are available.
Fig 9.
Local sensitivity confidence index (LSCI) for the AIR.
LSCI = 1.00 (in dark blue) indicates where the LSCI is higher, meaning that the confidence in sensitivities judgement from experts is high; LSCI = 0.00 (in dark red) indicates where the LSCI is lower, meaning that the confidence in sensitivities judgement from experts is low.
Fig 10.
Distribution of the rate of uncertainty in percentage per location (a) and per level and nature (b).
Fig 11.
Relative weight of uncertainty per sub-location according to the 5 uncertainty descriptors.
Ranking of sub-locations according to the relative weight of uncertainty.
Fig 12.
Uncertainty analysis of four input factors groups.
The spatial distribution of the coefficient of variation expressed (CV)–resulting from the Monte Carlo simulation of the four input factors groups of i) sensitivity score errors, ii) pressure distance errors, iii) stressor combination factor and iv) response factor—is reported, from lower (dark green) to higher values (orange).
Fig 13.
High- and low-impact areas according to the results of 50% to 100% of the Monte Carlo simulations in the Italian Adriatic area.
The maps show the percentage of how often each grid cell was in the most and least impacted 25% (a) and 10% (b) of the Italian Adriatic region. The red gradient refers to most impacted, the green one to least impacted for the percentage (between 50% to 100%) of simulation runs.
Table 4.
Mean values of the first order index (S1) and total order index (ST).
Fig 14.
Distribution of first-order index (a) and total-effect index (b) for each grid cell.
Table 5.
Mean value of second-order interaction between pair of input factors.
Fig 15.
Distribution of sums of first order indexes.