Counterfactual climate data

Counterfactual climate data represents what the climate would have been in the absence of anthropogenic forcing — a hypothetical “world without climate change.” These datasets are essential for climate impact attribution, where the goal is to quantify the fraction of observed impacts (droughts, floods, heat waves) attributable to human-caused climate change versus natural variability.

The ATTRICI algorithm

ATTRICI (ATTRibuting Impacts of Climate change with an Iterative method) is an algorithmic framework for constructing counterfactual climate datasets from historical observational archives. The method was developed at the Potsdam Institute for Climate Impact Research (PIK) as part of the Inter-Sectoral Impact Model Intercomparison Project (ISIMIP).

The algorithm works by:

1. Quantifying the long-term trend in climate variables attributable to anthropogenic forcing (using CMIP model ensembles to separate forced trends from natural variability)
2. Removing this trend from the historical record while preserving the observed variability, spatial patterns, and inter-variable consistency
3. Producing a counterfactual time series that retains the weather sequences of the real world but without the anthropogenic climate trend

Key properties of the counterfactual data:

• Physical consistency: Temperature, Precipitation, humidity, and radiation fields remain internally consistent after trend removal
• Preserved variability: Natural climate variability (ENSO events, volcanic eruptions, internal decadal variability) is retained
• Probabilistic framework: Uncertainty in the forced trend is propagated from the CMIP ensemble

Applications

Counterfactual datasets enable controlled impact attribution experiments:

• Run an impact model (hydrology, agriculture, health) with both the historical and counterfactual climate
• The difference in outcomes quantifies the impact attributable to anthropogenic climate change
• This approach is used across ISIMIP sectors including water resources, agriculture, ecosystems, and health

Research context

The ATTRICI framework was developed during work at PIK (2019), where the focus was on designing robust statistical methods for large-scale climate datasets. The algorithm bridges statistical climate science and impact modeling, providing the counterfactual baseline that impact models need for attribution studies.

See also: ERA5 reanalysis, Dynamical downscaling