A1. t-test
t-test is a common statistical test for the difference between two independent sample means (Wilks, 2005):
where and , and are the sample mean value and variance of two data sets X1(t) and X2(t), respectively, n1 and n2 are sample sizes ofX1 and X2.The sampling distribution for is the t distribution, with, whereis the degrees of freedom. The null hypothesis, calledH0, is that mean values of two data sets are equal. Given the significance level , if H0 is rejected, and is not rejected otherwise.
A2. The Ensemble Empirical Mode Decompositionmethod (EEMD)
The EEMDmethod (Huang and Wu, 2008; Wu and Huang, 2009) is an adaptive and temporally local filter, which is developed based on Empirical Mode Decomposition (EMD). This method is introduced as follows.
(1) add a white noise series to the targeted data;
where is the constructed data, with the targeted data,, plusthe realization of white noise, .
(2) decompose the data with added white noise into intrinsic mode functions (IMFs);
whereis the residue of data , after number of IMFs, , are extracted.are simple oscillatory functions with varying amplitude and frequency.
(3) repeat step 1 and step 2, but with different white noise series ()each time;
(4) obtain the ensemble means of corresponding IMFs of the decompositions as the final result, .
Detailed procedures to obtain the IMFs in step 2 can befound in Wu and Huang (2009).Here an example is provided in Fig. A1, and the interdecadal variabilities of the PDO index is presented by IMF5.
Fig.A1 Components for annual PDO index during 1900-2013 obtained by the EEMD method.
A3.ThePDO
Using NOAA ERSST monthly SST (1900-2013), the leading EOF of SST (removing the long-term trend) in the North Pacific Ocean (20°N -70°N) is given in Fig.A2. The pattern of SST anomalies (Fig.A2a) is the PDO, whilePrincipal Component 1 (PC1, Fig. A2b) is the PDO index.When PC1 is positive, the PDO is in warm phase, and there areanomalously cold SSTs in the central North Pacific, as well as anomalously warm ones along the western coast of the America.The pattern of SST anomalies reversesduring the cold phase of PDO (negative PC1).
Fig.A2(a) The leading EOF and (b) associated time series PC of monthly mean SST during 1900-2013 from NOAA ERSST. The red line in (b) is 108-month running mean of PC1.
A4. Climatological circulation and moisture in April, June and August over East China
As shown in Fig. A3, there are climatological prevailingnorthwesterly winds over North China at 500 hPain April and June, which change towesterly ones in August. In South China, there are westerly winds in April, which change to southwesterly ones in June and southeasterly ones in August along the northwestern and western flanks of the West Pacific Subtropical High, respectively.
Correspondingly, the climatological moisture flux and divergence show that the moisturefrom the Bay of Bengal and South China Sea converges over South China in April, which suggests the location of climatological precipitation; while the convergence is strengthened and move northward in June with the moisture from the Indian Ocean and South China Sea, and the moisture from East China Sea converges weakly over North China; In August, the strong convergence zone moves further northward along the northwestern flank of the West Pacific Subtropical High, and there isstill a weak convergence zone over North China with the moisture from the East China Sea.
Fig.A31958-2001 mean of(left) 500 hPa geopotential height (0.1 m contour), winds (m s−1, vector), (right) moisture flux (g cm−1 s−1, vectors) and moisture divergence (10−4 mm d−1, shaded) integrated from the surface to 700 hPa in (a, b) April, (c, d) June and (e, f) August from the ERA40 Reanalysis. Gray areas are theTibetan Plateau.
1