Seasonal Prediction Of Spatial Distribution Of Extreme Precipitation Days In Summer In Eastern Chin

Extreme precipitation over Eastern China is expected to become more frequent under the context of global warming.However,the prediction skills and predictability of the spatial distribution of extreme precipitation days(EPDs)over eastern China remain unclear.To address this issue,we analyze the physical mechanisms of the leading modes of summer EPDs in southern China(SC,south of 31°N,east of 110°E)and northern China(NC,north of 31°N,east of 110°E),and establish a set of physical–based empirical model(P–E model)to predict the distribution of EPDs in southern China and northern China,respectively.Finally,we estimate the predictability of the distribution of EPDs by employing predictable mode analysis method.The results are revealed the following:1)Significant seasonal differences are found in extreme precipitation days over eastern China.In May and June,the center of the area with the maximum number of EPDs confined to southern China(south of 31°N),while the area with the maximum number of EPDs is shifted to northern China(north of 31°N)during July and August.2)The first EOF mode displays a uniform pattern,while the second EOF mode presents a meridional dipole pattern of summer(May–June)EPDs in southern China.These two modes are significantly independent from other modes and explained 49% of the total variance.The uniform pattern of the first mode is associated with the tropical western North Pacific anomalous anticyclone(TWPAC),whereas the meridional dipole pattern of the second mode is related to the subtropical western North Pacific anomalous anticyclone(SWPAC).The spring North Atlantic triple sea surface temperature(SST)pattern,the northern North America surface air temperature,and the snow depth over central Siberia contribute to the variation of the first mode and TWPAC,whereas the spring extratropical North Atlantic dipole SST pattern and the spring change of sea ice concentration over the Barents and Kara seas are physically linked with the variation of the second mode and SWPAC.Based on these predictors,the established P–E model demonstrated significant skill(TCC=0.61/0.86)with regard to the principal components of the first two modes,and the spatial distribution of EPDs in southern China is reconstructed.The areal mean temporal correlation coefficient skill for the independent prediction period(2011–2021)is 0.34,while for the spatial correlation coefficient skill averaged over the entire period(1979–2021)is 0.28.3)The first EOF mode of summer EPDs displays a uniform pattern,the second EOF mode presents a meridional dipole pattern,while the third mode displays a zonal dipole pattern.These three modes are significantly independent from other modes and explained 45.9% of the total variance.The uniformly positive EPDs anomalies in northern China of the first EOF mode is controlled by the anomalous cyclone over northeast China and the anomalous anticyclone over tropical western North Pacific.The second mode with negative(positive)EPDs anomalies over north China(northeastern China)is mainly related to the strong anomalous anticyclone over eastern Siberia and the local anomalous cyclone over the Yellow Sea and Bohai Sea.The third mode shows negative(positive)EPDs anomalies over the western part of northern China(the eastern part of northern China),which is linked to the anomalous anticyclone over Mongolia and southern Sea of Japan.Based on the understanding of the physical mechanism of the simultaneous correlation,we select two predictors for PC1: the tropical north Atlantic dipole SST tendency and the tendency of change in Sea Ice Concentration in Beaufort Sea.For PC2,two predictors are found.The first is PMM-like SST pattern tendency and the second is SST warming tendency over equatorial central Pacific.Two predictors for PC3 are selected: the SST warming tendency over north Atlantic and the decreased tendency of snow depth over Balkan Peninsula.Based on these predictors,a set of P–E models are established.The results show that P–E model can predict these three modes with promising skill(TCC=0.63/0.61/0.60).Thus,they can be regarded as the predictable modes.The areal mean temporal correlation coefficient skill for the independent prediction period was 0.32,while for the spatial correlation coefficient skill averaged over the entire period was 0.24.