Time of Emergence of Annual Mean Precipitation is Long Due to Natural and Observational Limitations

The annual mean precipitation is a key metric of the regional and global hydrologic cycle. Comparing CMIP6 models, we find that future projections of mean precipitation are robust over much of the globe. Despite this robust signal, we show that we cannot expect sustained, statistically-significant changes in mean precipitation until far into the future. Under SSP5-8.5 forcing, the change in annual mean precipitation is only statistically-significant for less than one-third of the planet by 2080. Forced changes are expected to emerge the earliest over the high latitudes and parts of the tropics, though even these regions do not emerge until mid-century. Model uncertainty in this time of emergence arises primarily from intermodel spread in precipitation sensitivity, though intermodel spread in the global mean surface temperature response and precipitation variability are also important contributing factors. Despite this spread between models, the projected time of emergence of mean precipitation change is long in all models because of precipitations high internal variability. Moreover, inherent observational limitations further contribute to the long time of emergence. Specifically, we quantify how small observational periods increase statistical uncertainty; relatively recent observational datasets decrease the anthropogenic signal; and our ability to only look forward in time creates an uncertainty of whether apparent emergence is just natural variability. This work quantifies the factors that limit our ability to ascertain changes in mean precipitation observations within this century, and underscores the need to develop better metrics for measuring mean state hydrologic cycle changes.