by L. Lueken, July 31, 2025 in ClimateChangeDispatc

A recent post at Phys.org claims that a recent attribution study shows that climate change made April 2022’s flooding in South Africa “significantly” worse. [emphasis, links added]

This is an unfalsifiable (not able to be proven or disproven experimentally or observationally) claim that ignores the complexities of weather and relies on distinctly unreliable computer modeling.

The article, titled “Climate change significantly worsened deadly 2022 Durban floods, study shows,” goes over an attribution study that focused on flooding in Durban, South Africa, three years ago.

Phys.org claims the study “shows that rainfall during the storm of 11–12 April 2022 was between 40 percent and 107 percent heavier than it would have been in a cooler, preindustrial climate.”

How do they know this? They don’t; rather, they claim it based on computer model outputs.

Unlike most coverage of attribution science, Phys.org vaguely hints at the fact that the modeling is less than bulletproof, explaining that the models “simulated the storm in both today’s warmed climate and a counterfactual world without human-induced global warming.”

Climate Realism has explained at length why attribution modeling is not evidence, but it may be helpful to point out that Phys.org is only half right here.

They indeed used a counterfactual world with no warming, but the warmed model is also counterfactual. A number of assumptions, some more robustly backed by available data and evidence than others, go into modeling the “current world.”

Statistician Dr. William Briggs has what I consider the best simple summary of how attribution modeling works:

…