Reply to Fleming: Not a gossamer web

We recently performed a thought experiment with model simulations of historical climate change (1). We posed this question: If the capability to accurately monitor changes in atmospheric temperature had existed in 1860, when could scientists have first identified a human fingerprint on climate? Our focus was on temperature changes from the surface to the top of the stratosphere. We found that in the mid- to upper stratosphere, from roughly 25 to 50 km in altitude, temperature changes caused by greenhouse gas increases from both fossil fuel burning (2) and land surface modification (3) could have been identified as early as 1885, in accord with results from related work (4).

Fleming argues that our study is “merely entertaining” and of limited scientific value (5). We respectfully disagree. He points out (as we did in our paper) that the ability to monitor stratospheric temperature did not exist in 1860 and that our study is therefore just “counterfactual spinning of gossamer webs.”

Counterfactuals are valuable and routinely used components in the web of knowledge. A robust way to identify the effects of X is to consider a control without X. For example, X could be human-caused changes in atmospheric CO₂. It could be the adoption of the Montreal protocol to limit emissions of ozone-depleting substances (6). It could even be the importance of COVID vaccines on a global scale. In our case, X is the inclusion of something we did not have in the real world—the ability to reliably monitor changes in atmospheric temperature from 1860 onward.

Our thought experiment helped us to understand the size and structure of human-caused atmospheric temperature “signals” relative to the “noise” of natural climate variability. We did not know a priori that because the signal-to-noise properties of the mid- to upper stratosphere are very favorable, even a small CO₂-driven stratospheric cooling signal could have been identifiable by 1885 (with global measurements) and by 1894 (with more limited regional measurements). Nor did we know a priori how changes in solar and volcanic activity from 1860 to 2024 would affect signal detection.

In summary, we do not share Fleming’s view that our thought experiment is insubstantial “gossamer.” Rather, it has key implications for the real world. The results reveal the critical importance of the stratosphere for detecting the signal of human-caused climate change. The stratosphere is a “canary in the coalmine” for identifying this signal. We show that anthropogenic climate change is not just a 20th-century phenomena. It was detectable in the late 19th century. And in an era when the funding and support for climate monitoring are being dismantled, our findings highlight the critical importance of continued measurements of stratospheric temperatures for understanding current and future climate change.