5.3 Uncertainty Scoring of the Parameters
93
5.3 Uncertainty Scoring of the Parameters
A feasible way of handling uncertainty in ERA acute is necessary, using the results
from the project development and testing whilst recognizing that many parameter
values may need further refinement in the first years of use. A self-evaluation scoring
system developed by DNV GL (Kruuse-Meyer 2015) was used to score the parameters and provide recommendations on the use of specific parameters. The scoring
was based primarily on the results of sensitivity testing carried out using statistical
and deterministic methods (Sect. 5.2) (Bjørgesæter and Damsgaard-Jensen 2018;
Stephansen and Bjørgesæter 2017).
For each parameter the sensitivity-deciding elements were considered and
assessed within the limits of the knowledge gained in previous project work according
to:
Strength of knowledge (function where it is used): How strong is our confidence
in that the risk function in which the parameter is used is a valid mathematical
representation of the mechanism of impact/restoration?
Belief that the value may deviate from the average assumption: Natural variation
of parameter. Do we believe that the values have a high natural tendency to vary
from the base case (mean)? E.g. if a (standard deviation) (SD) is quantifiable, this
can be used to assess this point.
Sensitivity of function to parameter (sensitivity index): How sensitive is the
model/function to variation in the parameter?
Comments/recommendations on handling to ensure risk is not under-estimated:
Recommended actions for ERA Acute use, data gathering etc.
5.3.1 Surface Compartment
The results of the scoring process based on the results of the deterministic and then
stochastic testing and evaluation of surface compartment parameters are given in
Table 5.1 for the impact parameters and Table 5.2 for the lag- and restoration time
parameters.
93
5.3 Uncertainty Scoring of the Parameters
A feasible way of handling uncertainty in ERA acute is necessary, using the results
from the project development and testing whilst recognizing that many parameter
values may need further refinement in the first years of use. A self-evaluation scoring
system developed by DNV GL (Kruuse-Meyer 2015) was used to score the parameters and provide recommendations on the use of specific parameters. The scoring
was based primarily on the results of sensitivity testing carried out using statistical
and deterministic methods (Sect. 5.2) (Bjørgesæter and Damsgaard-Jensen 2018;
Stephansen and Bjørgesæter 2017).
For each parameter the sensitivity-deciding elements were considered and
assessed within the limits of the knowledge gained in previous project work according
to:
Strength of knowledge (function where it is used): How strong is our confidence
in that the risk function in which the parameter is used is a valid mathematical
representation of the mechanism of impact/restoration?
Belief that the value may deviate from the average assumption: Natural variation
of parameter. Do we believe that the values have a high natural tendency to vary
from the base case (mean)? E.g. if a (standard deviation) (SD) is quantifiable, this
can be used to assess this point.
Sensitivity of function to parameter (sensitivity index): How sensitive is the
model/function to variation in the parameter?
Comments/recommendations on handling to ensure risk is not under-estimated:
Recommended actions for ERA Acute use, data gathering etc.
5.3.1 Surface Compartment
The results of the scoring process based on the results of the deterministic and then
stochastic testing and evaluation of surface compartment parameters are given in
Table 5.1 for the impact parameters and Table 5.2 for the lag- and restoration time
parameters.