When Does Climate Shift Emerge in the Future beyond the Historical Variability of Precipitation?

'Tipping point' is defined as thresholds for abrupt and irreversible change in the fields of climate change, biodiversity and economics. We have estimated the tipping point of precipitation at each grid point for global domain, using 18 Atmosphere-Ocean General Circulation Models (AOGCMs) participated in the phases 5 of the Coupled Model Intercomparison Project (CMIP5). We used precipitation of pre-industrial control simulation for more than 250 years, historical experiment from 1850 to 2005 for 156 years and future Representative Concentration Pathway (RCP) 8.5 scenario experiment from 2006 to 2100 for 96 years. We defined 'tipping year' after year 2006 as the time of year when the projected 10-year mean climatology at each grid point moves to a state continuously outside the range of historical variability before year 2006. With respect to each numerical experiments by AOGCMs, we augmented the set of 10,000 artificial time series which were distributed around the precipitation simulated by AOGCMs for historical and RCP8.5 experiments. Artificial precipitations were generated by the Monte Carlo method, considering the natural variability of model atmosphere which was estimated from the pre-industrial control simulation of each AOGCMs. Total ensemble size amount to 180,000 = 18 models x 10,000 time series. This large ensemble size enables us to estimate probability distribution function (PDF) of tipping year at each grid point.
Tipping years of 10-year mean annual precipitation are earlier in high latitudes and central tropical Pacific where the increase of precipitation in the end of 21st century is large. In contrast, tipping year does not exist over subtropical high regions where precipitation decreases in the end of 21st century. Tipping year of intense precipitation exist globally including subtropical high regions. The probability of emergence of tipping year before around year 2050 is estimated above 60% over most of high latitudes and about 30% in the tropics. We proposed two kinds of reliability measure for future projection at each grid point; the standard deviation of tipping years and the probability of emergence of tipping year.