Accounting & Integrity
Model uncertainty
The error introduced by the structure and assumptions of a model, as distinct from its input parameters.
Two models with identical inputs can produce different results if they represent the system differently. Model uncertainty captures that structural difference, and it is often larger than parameter uncertainty.
It is also harder to quantify, because there is no single correct answer to compare against. Ensemble comparison and validation against independent measurements are the usual approaches.
Frequently asked questions
Can model uncertainty be eliminated?
No. It can be characterised and reduced, but structural choice always remains.
Why does it matter for crediting?
Because it can dominate the total error budget while being invisible in a spreadsheet of parameter ranges.