After that, this estimate is smoothed using a Hamming window. The lag L value with the highest correlation coefficient

R was found at approximately T/4 (see Table 2). Thus, it is assumed that both eigenvectors as the T-PCs that define the oscillatory pairs are in quadrature. Fig. 5a presents the partial reconstruction of the nonlinear trend TEN18 (t), based on T-PC1 and T-EOF1, and filtered series REC1[tot] (t), corresponding to the partial reconstruction with the trend and the two oscillatory modes that were detected with SSA. The TEN18 (t) series features a clear ATM Kinase Inhibitor concentration long-term change to positive values from the 1960s, presenting a humid period until 2000 where a negative change is observed, increasing from the year 2008. The PC118 (t) series (Fig. 5a) indicates that the largest and most frequent hydrological droughts in the NEA occurred between 1901 and 1960, with a clearly differentiated period of long duration and high intensity wet EPE between years 1970 and 2005. We note an extended period of negative SPI18 (t) values between 1924 and 1939 interrupted only by positive values in the normal range between the months of December 1930 to June 1932. The most intense drought events were

recorded in the early twentieth century: between January 1906 and June 1912 (mean intensity of SPI18 (t) = −1.65) with 78 consecutive months of duration, and between March 1916 and April 1919 (mean signaling pathway intensity of SPI18 (t) = −2.17) with 38 months of duration. Both events are congruent with La Niña periods according to a historical analysis of Southern Oscillation Index Bacterial neuraminidase (SOI, Trenberth and Hoar, 1996) time series. On the other hand, the most intense wet events were recorded in the last 30 years of the 20th century, with

extraordinary peaks in April 1973 and November 2002. The former is consistent with strong El Niño event while the latter coincides with a moderate El Niño event, defined according to the Oceanic Niño Index (ONI, 3-month mean SST anomaly for the Niño 3.4 region, NOAA/NWS/CPC). It should be emphasized that we used SOI time series for determining the intensity of El Niño/La Niña events before 1950 and, from then on, we use ONI. These two indices are common methods used for determining the intensity of ENSO and have similar shapes but of opposite sign. The SOI is based on the difference between sea level pressures at Tahiti and Darwin, Australia and the ONI is based on sea surface temperatures in the eastern equatorial Pacific Ocean, whose data record starts at 1950. Fig. 4b presents the pattern of correlation of the SPI18 (t) series in each grid point with PC218 (t). A maximum value of positive correlation with relative importance (a18i2 max = 0.56) is situated in the Northwestern corner of the region. The PC218 (t) series accounts for the 9.5% of the total variance and Fig. 5b shows its evolution with time.