Abstract:Estimates of climate change remain uncertain—hampering strategic decision making in many sectors. In large part this uncertainty arises from uncertainty in the computational representation of known physical processes. This model component of climate change uncertainty is increasingly being assessed using perturbed model experiments. Some such model perturbations have, for example, led to headline global warming estimates of as much as 12 °C. These experiments consider many differently perturbed versions of a given base model and assess the likelihood of each perturbed model's climate prediction based on how well it simulates present‐day climate. In these experiments, the computational cost of the model assessment is extremely high unless one assumes that the climate anomalies associated with different model perturbations can be combined linearly. Here we demonstrate a different method, which harnesses the power of the data assimilation system to assess directly the perturbed physics of a model. Data assimilation involves the incorporation of daily observations to produce initial conditions (analyses) for numerical weather prediction (NWP). The method used here quantifies systematic initial tendencies in the first few time steps of a model forecast. After suitable temporal averaging, these initial tendencies imply systematic imbalances in the physical processes associated with model error. We show how these tendencies can be used to produce probability weightings for each model that could be used in the construction of probability distribution functions of climate change. The approach typically costs 5% of the cost of a 100‐year coupled model simulation that might otherwise be used to assess the simulation of present‐day climate. Importantly, since the approach is amenable to linear analysis, it could further reduce the cost of model assessment by several orders of magnitude: making the exercise computationally feasible. The initial tendency approach can only assess ‘fast physics’ perturbations, i.e. perturbations that have an impact on weather forecasts as well as climate. However, recent publications suggest that most of the present model parameter uncertainty is associated with fast physics. If such a test were adopted, assessment of the ability to simulate present‐day climate would then only be required for models that ‘pass’ the fast physics test. The study highlights the advantages of a more seamless approach to forecasting that combines NWP, climate forecasting, and all scales in‐between. Copyright © 2007 Royal Meteorological Society

Assessing the predictive impact of factor fixing with an adaptive uncertainty-based approach

A Decision‐Relevant Factor‐Fixing Framework: Application to Uncertainty Analysis of a High‐Dimensional Water Quality Model

A Deconfounding Approach to Climate Model Bias Correction

Quantifying the Impact of Multiple Factors on Air Quality Model Simulation Biases Using Machine Learning

Identifying Observed Factors in Approximate Factor Models: Estimation and Hypothesis Testing

An efficient method for predictive-failure-probability-based global sensitivity analysis

Using numerical weather prediction to assess climate models

Are the Signs of Factor Loadings Arbitrary in Confirmatory Factor Analysis? Problems and Solutions

Factor analysis process monitoring method based on probabilistic information of variables

Statistical Inference for Covariate-Adjusted and Interpretable Generalized Factor Model with Application to Testing Fairness

TESTING FOR STRUCTURAL CHANGES IN FACTOR MODELS VIA A NONPARAMETRIC REGRESSION

Augmented Factor Models with Applications to Validating Market Risk Factors and Forecasting Bond Risk Premia

Parameter Selection for Model Updating with Global Sensitivity Analysis

Global sensitivity and domain‐selective testing for functional‐valued responses: An application to climate economy models

Improving Estimation of Portfolio Risk Using New Statistical Factors

A Factor-Adjusted Multiple Testing Procedure with Application to Mutual Fund Selection

A global sensitivity analysis approach for identifying critical sources of uncertainty in non-identifiable, spatially distributed environmental models: A holistic analysis applied to SWAT for input datasets and model parameters

Effects of error covariance structure on estimation of model averaging weights and predictive performance

Anomaly or Risk Factor? A Stepwise Evaluation

Factor-adjusted Multiple Testing of Correlations

On Time-Varying Factor Models: Estimation and Testing