Quantifying stochastic uncertainty in detection time of human-caused climate signals

Large initial condition ensembles of a climate model simulation provide many different realizations of internal variability noise superimposed on an externally forced signal. They have been used to estimate signal emergence time at individual grid points, but are rarely employed to identify global fingerprints of human influence. Here we analyze 50- and 40-member ensembles performed with 2 climate models; each was run with combined human and natural forcings. We apply a pattern-based method to determine signal detection time t(d) in individual ensemble members. Distributions of t(d) are characterized by the median t(d){m} and range t(d){r}, computed for tropospheric and stratospheric temperatures over 1979 to 2018. Lower stratospheric cooling-primarily caused by ozone depletion-yields t(d){m} values between 1994 and 1996, depending on model ensemble, domain (global or hemispheric), and type of noise data. For greenhouse-gas-driven tropospheric warming, larger noise and slower recovery from the 1991 Pinatubo eruption lead to later signal detection (between 1997 and 2003). The stochastic uncertainty t(d){r} is greater for tropospheric warming (8 to 15 y) than for stratospheric cooling (1 to 3 y). In the ensemble generated by a high climate sensitivity model with low anthropogenic aerosol forcing, simulated tropospheric warming is larger than observed; detection times for tropospheric warming signals in satellite data are within t(d){r} ranges in 60% of all cases. The corresponding number is 88% for the second ensemble, which was produced by a model with even higher climate sensitivity but with large aerosolinduced cooling. Whether the latter result is physically plausible will require concerted efforts to reduce significant uncertainties in aerosol forcing.