An overview of methods to evaluate uncertainty of deterministic models in decision support

By Laura Uusitalo, Annukka Lehikoinen, Inari Helle, Kai Myrberg. Environmental Modelling & Software, 2015

531 citations recorded by the source corpus.

There is an increasing need for environmental management advice that is wide-scoped, covering various interlinked policies, and realistic about the uncertainties related to the possible management actions. To achieve this, efficient decision support integrates the results of pre-existing models. Many environmental models are deterministic, but the uncertainty of their outcomes needs to be estimated when they are utilized for decision support. We review various methods that have been or could be applied to evaluate the uncertainty related to deterministic models' outputs. We cover expert judgement, model emulation, sensitivity analysis, temporal and spatial variability in the model outputs, the use of multiple models, and statistical approaches, and evaluate when these methods are appropriate and what must be taken into account when utilizing them. The best way to evaluate the uncertainty depends on the definitions of the source models and the amount and quality of information available to the modeller. We review different types of uncertainty present in environmental modelling.We review methods to evaluate uncertainty related to model results.Best way to evaluate uncertainty depends on the models and available information.

Subfields: Environmental Engineering, Ocean Engineering, Water Science and Technology

View Paper: https://www.semanticscholar.org/paper/6878bf9972ab1427789400afcae9d0a3683b6cb4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Environmental Engineering

1. Annukka Lehikoinen
2. Laura Uusitalo

Citations in the corpus

• A Catalogue of Marine Biodiversity Indicators (2016, background): “Indicator name Developmental status DPSIR stage Main attribute Monitoring time series Abundance of bioengineering species Operational State Biotope features Accumulated cover of submerged vascular plants Operational State Target groups Areal extent of eelgrass Operational State Biotope features 1990”
• A survey on missing data in machine learning (2021, methodology): “Multiple imputation is where the distribution of the observed data is utilized to approximate numerous values that reflect the uncertainty around the true value, and this method was mostly implemented to solve the limitations of single imputation [57].”
• How to value biodiversity in environmental management (2015, background): “Environmental management problems are typically complex nd multidisciplinary, involving various unavoidable trade-offs and ncertainties (Uusitalo et al., 2015) in informed decision-making.”