Simulations And Projections Of Interannual Variability Of Global Monsoon By CMIP5 Models

Simulations and future projections of year-to-year variation in global monsoon precipitation(GMP),defined as the summer precipitation amount per unit area within the global monsoon domain,are investigated using the 33 models that participated in Phase 5 of the Coupled Model Intercomparison Project(CMIP5).The interannual standard deviation of GMP,which represents the amplitude of year-to-year variability in monsoon rainfall,is simulated well by the multi-model ensemble(MME)mean of the 33 models and of the 'best five(B5)models.The B5 models show superior skills in reproducing the climatological monsoon precipitation and its interannual variability.The interannual standard deviations of 25-year GMP in the present day(1979-2005)derived from the MME average of the 33 models and of the B5 models are 0.17 and 0.15 mm day-1 per unit area,respectively,both being close to the observation(0.16 mm day?1 per unit area).Consistent with the observed ENSO-GMP relationship,the simulated variability of GMP is negatively correlated with the sea surface temperature in the Nino areas at the interannual timescale.Projections by the 33 CMIP5 models under the RCP4.5 scenario indicate that wet and dry monsoon years would occur more intensively in a warmer climate.Around 67%(20 out of the 33)of the models show enhanced interannual variability of GMP by the end of the 21st century(2072-2098)compared to the period 1979-2005.The MME of the B5 models suggests that the amplitude of projected GMP variability would increase by?20%from present day to the late 21st century.However,the increasing rate of GMP variability detected by the 33 models' MME is relatively small(?6%),which is even smaller than the inter-model standard deviation,suggesting the amplitude changes in GMP variability feature non-negligible uncertainty among the 33 models.Under global warming,the interannual variation of GMP would still be tightly connected with ENSO,although the projected amplitude changes in ENSO variability are noticeably diverse across the 33 CMIP5 models.The enhanced amplitude of interannual GMP variability may be attributable to the increase in mean-state moisture associated with global warming.The dynamic effect related to changes in the variability of monsoon circulation is small and less robust among the 33 CMIP5 models.Among the global monsoon systems,the western North Pacific monsoon(WNPM),the Australian monsoon(AM)and Indian monsoon(IM)show clear phase relationships at the interannual timescale.A strong WNPM during boreal summer often follows a strong AM in the previous austral summer but leads a weak AM in the following austral summer.Besides,a significant simultaneous negative correlation appears between WNPM and IM.We evaluated the ability of the CMIP5 models to simulate the phase relationship between the three regional monsoons.The simultaneous negative correlation between WNPM and IM can be simulated by the CMIP5 models.However,the CMIP5 models showed limited ability in simulating the phase relationship between WNPM and AM.This might be attributed to the poor skill in reproducing the ENSO and the western northern Pacific anticyclone,which are the keys for connecting the Asian and Australian monsoon.