The transient climate response (TCR) is useful for quantifying the Earth's short-term (decadal to century timescales) response to an increase in greenhouse gases. The TCR cannot be directly observed, so it is often estimated using the historical temperature record. One important question is the extent to which internal variability confounds these estimates of TCR. To quantify the uncertainty caused by internal variability, we analyze output from a 100-member ensemble of a fully coupled global climate model (Max Planck Institute Earth Science Model, MPI-ESM1.1). Each of the 100 ensemble members has identical historical forcing and model physics but unique initial conditions. The model members consequently evolve differently in time. The TCR estimated from the historical ensemble yields values ranging from 1.34 to 1.9 K, where the spread can be attributed to internal variability alone. We can understand this range using a framework derived from the Earth's energy balance. A majority of the variability observed in the ensemble estimates of TCR is due to heat transport into the deep oceans, ensemble members that transport more heat to the deep ocean have lower TCRs. Another key factor, is the amount of forcing that is radiated back to space. Together, these two factors explain 71% of the variance in the TCR estimates from the large ensemble. Due to the demonstrated effects of internal variability, an estimate of the TCR from the historical record (a single realization out of many possible climate histories) could deviate significantly from the climate system's true value.
