Dry/wet responses across China to ENSO events and their relationship to the PDO: an event coincidence analysis of historical records

Abstract

This study explores the dry and wet responses across China to ENSO events and the relationship of these responses to the Pacific Decadal Oscillation (PDO), motivated by the need to understand the complex interactions between these climatic phenomena and their impacts on regional precipitation patterns. Utilizing Event Coincidence Analysis (ECA) on historical Dryness/Wetness Index (DWI) records and ENSO chronologies, we identify distinct regional patterns in precipitation responses. Our results show that in North China, positive PDO phases exacerbate dry conditions during ENSO events, while negative PDO phases lead to wetter conditions. Conversely, in the Yangtze River basin, positive PDO phases are associated with increased wet conditions during ENSO events, and negative phases with drier conditions. These findings have significant implications for disaster prevention and mitigation, particularly in managing drought and flood risks. Additionally, our study highlights the importance of considering PDO phases in long-term climate studies and demonstrates the value of ECA in analyzing climate records with missing data. This research provides a new perspective on the dynamic interplay between ENSO, PDO, and regional precipitation patterns, offering a novel metric for studying climate event relationships.

Graphical abstract

Keywords

climate change / event coincidence analysis / teleconnections / flood and drought / ENSO / PDO

1 Introduction

East Asia, home to over one billion people, experiences seasonal reversals in prevailing wind directions. During the rainy season, the East Asian summer monsoon (EASM) delivers substantial moisture from neighboring seas. The EASM is characterized by significant year-to-year fluctuations, often leading to water abundance deficits or surpluses (Ding, 1994). These precipitation extremes, resulting in droughts or floods, severely impact the environment, infrastructure, and society. Rainfall-related hazards in China rank first and second globally in terms of the total people affected and total damage, respectively, among 16 extreme atmospheric hazards (Steptoe et al., 2018). Therefore, identifying the factors influencing the EASM and improving its predictability are crucial for the well-being of this densely populated region.

The El Niño-Southern Oscillation (ENSO) (McPhaden et al., 2006) is the primary interannual variability of the tropical Pacific Ocean. It significantly impacts the climate beyond its region of origin (Ropelewski and Halpert, 1987; Ropelewski and Halpert, 1989). East Asia is particularly affected by ENSO (Huang and Wu, 1989; Chan and Zhou, 2005; Wang et al., 2022), with a strong ENSO-EASM teleconnection. Many drought and flood events in East Asia are closely linked to ENSO events. The region's complex dry and wet responses to ENSO events depend on whether it is an El Niño or La Niña event, as well as the developing or decaying phase of the ENSO cycle (Huang and Wu, 1989; Wu et al., 2003).

The Pacific Decadal Oscillation (PDO; Mantua et al., 1997) is another prominent large-scale climate factor influencing the EASM (Ma, 2007; Qian and Zhou, 2014). The PDO, a low-frequency oscillation mode in North Pacific Sea surface temperatures, differs from ENSO in active timescales, with the PDO operating on decadal scales and ENSO on interannual scales. Consequently, the PDO can be viewed as the background on which fast ENSO fluctuations are superimposed. While the effects of ENSO and PDO on China’s rainfall have been separately examined, the effect of the PDO on the ENSO-EASM teleconnection remains poorly understood. This study focuses on the impact of ENSO on the EASM and how this impact is influenced by the PDO. From a hazard prevention perspective (Li et al., 2020), we are particularly interested in whether the drought/flood hazards associated with each phase of an ENSO event are exacerbated or alleviated in certain phases of the PDO.

Most studies on China’s dry/wet responses to ENSO events are temporally confined to periods starting around the 1950s, as instrumental meteorological observations were scarce in China before then. Consequently, the time span for analysis is limited to a few PDO cycles. In this study, we extend our analysis back to 1871 CE using an atlas of the historical dryness/wetness index (DWI) in China, compiled from historical writings, along with several reconstructed ENSO chronologies. By doing so, we extend the data span to over a century, encompassing several different PDO phases, thus providing a longer record for analyzing the dry/wet response to ENSO events affected by the PDO.

2 Data

This study utilizes six ENSO chronologies, dating back to 1871 CE (Table S1). Three chronologies are based on instrumental observations (Rasmusson and Carpenter, 1983; Kiladis and Diaz, 1989; Allan et al., 2003), while two rely on proxy data (Quinn and Neal, 1992; Whetton and Rutherfurd, 1994; Ortlieb, 2000). The sixth chronology integrates both atmospheric and oceanic signals (Gergis and Fowler, 2005). The PDO index time series, sourced from Kaplan et al. (2000), covers the latter half of the 19th century and the entire 20th century.

The historical precipitation data for China used in this study comes from the “Yearly Charts of Dryness/Wetness in China for the Last 500-Year Period” (CMA, 1981), an atlas of the historical dryness/wetness index (DWI) spanning 1470 to 1979 CE. This atlas was compiled by synthesizing over 2200 local chronicles from 120 sites across China (Table S2). Historical writings are invaluable for supplementing proxy data, especially for inferring past climate states in the absence of instrumental measurements (Brázdil et al., 2018; Nash et al., 2021). Instrumental observations from the second half of the 20th century were incorporated to form a coherent data set (Zhang and Liu, 1993; Zhang et al., 2003). This historical DWI atlas has been widely used in studies of historical floods and droughts in China (e.g., Song, 2000; Qian et al., 2003; Jiang et al., 2006; Qian et al., 2012; Ji et al., 2015; Wang et al., 2022).

In this data set, the dry/wet state of a specific station is denoted by a numerical value between 1 and 5, with lower values indicating wetter conditions and higher values indicating drier conditions (Table S3; CMA, 1981; Zhang and Liu, 1993; Zhang et al., 2003). Out of the 120 sites, only 38 have complete dry/wet index records for the entire 130-year period from 1871 to 2000 (Fig. S1). Sites in densely populated East China have records spanning at least 100 years, while sites with more missing records are generally located in less populated regions, such as remote inland western China and north-eastern China.

3 Method

Historical records are not available for every year within the study period, and many sites have a significant number of missing DWI values. To address this, we selected a novel statistical method: Event Coincidence Analysis (ECA; Donges and Schleussner, 2016; Siegmund et al, 2017; Fig.1). This method is particularly suited for analyzing time series with missing values. It allows us to formally study the inter-relationship between ENSO/PDO and dry/wet responses. ECA is a technique capable of determining the degree of association or coincidence between two or more time series of events. It has been widely applied in various research fields, including social sciences (Galbraith et al., 2020), epidemiology (Latinne and Morand, 2022), and ecology (Siegmund et al., 2016).

3.1 Definition of events

In this study, an ENSO event is defined as either an El Niño or a La Niña event as indicated in the ENSO chronologies. A dry event is defined as a DWI of 4 or 5, while a wet event is defined as a DWI of 1 or 2. The ECA procedure counts the total numbers of ENSO events (N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}) and dry/wet events (N_{\mathrm{D}/\mathrm{W}}) in their respective time series (t^{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}} and t^{\mathrm{D}/\mathrm{W}}). By analyzing the timings of these events in the time series, ECA determines how often ENSO events are followed by dry/wet events within a specified delay window (τ), which can be zero in the case of simultaneous occurrence. The coincidence rate (CR) is calculated as follows:

(1)r\left(\tau \right)=\frac{1}{N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}}{\sum }_{j=1}^{N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}}\Theta \left\{{\sum }_{i=1}^{N_{\mathrm{D}/\mathrm{W}}}\delta \left[\left(t_i^{\mathrm{D}/\mathrm{W}}-\tau \right)-t_j^{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}\right]\right\},

where \Theta \left(x\right) is the Heaviside function, defined to be equal to 1 when x is positive, and 0 when x is nonpositive; \delta \left(x\right) is the Dirac delta function, defined to be equal to 1 when x = 0, and 0 when x\ne 0. We switch the \tau value between 0 and 1, which corresponds to the dry/wet response in the developing and decaying year of an ENSO event, respectively.

3.2 Significance testing

Within the ECA framework, a coincidence is defined, for the purposes of this study, as the simultaneous occurrence of dry/wet and ENSO events in the same year (\tau =0), which indicates an ENSO event is developing, or a dry/wet event is occurring in the year following an ENSO event (\tau =1), indicating an ENSO event is decaying. The probability that a dry/wet event follows an ENSO event within an effective time span (T-\tau) is given by

(2)p=\frac{1}{T-\tau },

where T is the overlapping time span of the two time series. It follows that the probability of an ENSO event is followed by at least one of the N_{\mathrm{D}/\mathrm{W}} dry/wet events is

(3)1-{\left(1-p\right)}^{N_{\mathrm{D}/\mathrm{W}}}.

The probability that K coincidences occur in a binomial process is then given by

(4)\begin{array}{rl}& P\left(K;N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}},1-{\left(1-p\right)}^{N_{\mathrm{D}/\mathrm{W}}}\right)\\ & =\left(\underset{\mathrm{K}}{N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}}\right){\left(1-{\left(1-p\right)}^{N_{\mathrm{D}/\mathrm{W}}}\right)}^K{\left({\left(1-p\right)}^{N_{\mathrm{D}/\mathrm{W}}}\right)}^{N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}-K}.\end{array}

The p-value of observing K_{\mathrm{o}\mathrm{b}\mathrm{s}}coincidences can be derived as

(5)P\left(K\ge K_{\mathrm{o}\mathrm{b}\mathrm{s}}\right)={\sum }_{K=K_{\mathrm{o}\mathrm{b}\mathrm{s}}}^{N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}}}P\left(K;N_{\mathrm{E}\mathrm{N}\mathrm{S}\mathrm{O}},1-{\left(1-p\right)}^{N_{\mathrm{D}/\mathrm{W}}}\right).

While Poisson processes reasonably describe the dry/wet events in the DWI time series, they cannot describe the ENSO time series as they contain some level of autocorrelation. Thus, we adopt the Iterative Amplitude Adjusted Fourier Transform (IAAFT) algorithm (Schreiber and Schmitz, 2000) when generating surrogates of the ENSO time series. The surrogate time series generated by the IAAFT algorithm share the amplitude and spectrum characteristics with the original ones. This is achieved with an iterative scheme that alternates between two steps until convergence is achieved. Starting with a random shuffle of the original time series, the first step involves a Fourier transform where the amplitudes are adjusted to match those of the original series while preserving the phases of the shuffled data. In the second step, the resulting series undergoes a rank-ordering procedure to match the amplitude distribution of the original data. The series is then rank-ordered to match the original amplitudes. This process continues until either a fixed point is reached or the discrepancy between consecutive iterations falls below an empirically specified tolerance threshold. The IAAFT method overcomes the limitation of the simple Amplitude Adjusted Fourier Transform (AAFT) algorithm, which can introduce a bias toward a flatter spectrum in strongly correlated sequences. This makes IAAFT particularly suitable for the analysis in the present study, as it ensures the surrogate time series maintain the essential statistical properties of the original data while randomizing the temporal structure, providing a robust null model for testing the significance of the observed patterns. Each site undergoes a significance test between its dry/wet event time series and each of the six ENSO chronologies. The significance level (p-value) of the ECA results for each pair of time series is computed accordingly.

For each site, a wet or dry CR time series to each ENSO phase is calculated within a sliding 31-yr window. To guard against possible adverse effect of small sample size on the statistical results, the data obtained with fewer than 10 ENSO events are excluded from analysis. The correlation coefficient of the CR time series with the PDO index time series is calculated for each site and each ENSO phase. Sites with similar responses to ENSO events are grouped and their CRs are averaged to obtain regionally mean CR time series, which are then compared with the PDO time series to explore possible relations between them. The three regions of Eastern China under study include North China (NC; 32°–40° N, 108°–121° E), Yangtze River (YR; 28°–32° N, 108°–121° E), and South China (SC; 21°–28° N, 108°–121° E).

4 Dry/wet responses to ENSO events

4.1 Responses to El Niño Events

Fig.2 shows the spatial distribution of dry and wet CRs across China during the developing phase (year 0) of El Niño events. The left column represents wet responses, while the right column shows dry responses. Each row corresponds to one of the six ENSO chronologies (GF05, AL03, KD89, QNO, RC83, and WR94). Statistically significant CRs (p < 0.05) are denoted by triangular markers, while square markers indicate sites without significant relationships. The filled colors at each site indicate the corresponding CR values. This multi-chronology visualization reveals a few consistent spatial patterns in the ENSO-precipitation relationship, particularly highlighting regions where El Niño events demonstrate robust statistical associations with either dry or wet conditions across different ENSO reconstructions. Notably, wet CRs are unanimously in all ENSO reconstructions much greater than dry CRs in northeast China. Furthermore, the association between wet conditions and a developing El Niño is significant on the p < 0.05 level at all stations in northeast China in 4 of the 6 ENSO reconstructions (namely, AL03, KD89, RC83, and WR94). In contrast to northeast China, there appears a tendency to dry conditions in north China along with a developing El Niño. Dry CRs are in general greater than wet CRs, and the association between dry conditions and developing El Niños at many stations is significant on the p < 0.05 level, in all ENSO reconstructions except QNO; the relative inability of QNO to discern the tendency to dry conditions might stem from the fact that the ENSO chronology by QNO is compiled only based on records from South America (Peruvian records in majority). Fig.3(a) summarizes the number of ENSO chronologies showing highly significant (p < 0.01) associations with wet or dry conditions for each site, providing a measure of consistency across different ENSO reconstructions. The consistently wet northeast China and dry north China patterns suggest a strong regional dependence in the precipitation response to El Niño forcing during its developing phase, with the consistency across multiple ENSO chronologies reinforcing the reliability of these spatial patterns.

These ECA findings align with previous observational studies that have noted an increase in drought occurrences in north China during the developing phase of El Niño events (Huang and Wu, 1989; Jiang et al., 2017; Chen et al., 2019), which reported significant precipitation decreases in this region during El Niño years compared to ENSO-neutral years. Moreover, historical data from tree-ring cellulose δ¹⁸O chronologies suggest that extremely dry years over the past ~200 years have often coincided with El Niño events (Li et al., 2011).

Fig.4 presents the spatial distribution of dry and wet CRs across China during the decaying phase (year 1) of El Niño events in a way similar to Fig.2. The spatial patterns of precipitation anomalies exhibit considerable regional variation across China. North-east China has no dominant tendency to wet or dry conditions. In north China a strong tendency toward dry conditions appears in the GF05 and AL03 chronologies, but is obscured in other chronologies. The Yangtze-Huaihe valley exhibits mixed wet and dry signals. In the middle and lower reaches of the Yangtze River and the southern basins, wet CRs become increasingly dominant across all ENSO chronologies. This suggests a heightened risk of flood in these agriculturally critical areas during the decaying phase of El Niño. South-east China exhibits a mixed response. The number of ENSO chronologies displaying highly significant (p < 0.01) associations with wet or dry conditions are summarized in Fig.3(b) for each site. The responses to decaying El Niños are in general mixed throughout China. However, a consistent drying tendency is found in the Yangtze-Huaihe valley where it tends to be dry, and in contrast, a wetting tendency in the middle and lower reaches of the Yangtze River and their southern basins. These ECA results qualitatively agree with analyses of precipitation anomalies observed during El Niño decaying summers in the instrumental period (Huang and Wu, 1989; Gao et al., 2006; Huang et al., 2012; Zhang et al., 2017a; Sun et al., 2021).

4.2 Responses to La Niña Events

The developing phase of La Niña events generally results in widespread higher wet CRs than dry CRs across China in all the three La Niña chronologies (Fig.5). In contrast to developing El Niños, north China’s response to developing La Niñas is characterized by predominantly wet conditions, with dry responses occurring at only a few isolated sites (Fig.3(c)). The majority of sites within the entire Yangtze River basin exhibit wet responses (Fig.3(c)), and this pattern extends to south China, with a few exceptions along the southern coast. These findings are consistent with tree-ring cellulose δ¹⁸O chronologies from the North China Plain, which date back to 1784 CE (Li et al., 2011), and a 657-year tree ring width chronology from the southern Taihang Mountains (Zhang et al., 2017a), both of which align wet spells with La Niña occurrences.

La Niña events in their decaying phase show the least organized impact among the four ENSO phases (Fig.3(d) and Fig.6). The dry/wet responses are mixed in most regions of China. Nevertheless, a slight tendency to dry conditions is still discernible in the Yangtze-Huaihe valley and the middle and lower reaches of the Yangtze River.

Overall, these findings highlight the complex and regionally varied nature of dry/wet responses to ENSO events across China, emphasizing the significant influence of both El Niño and La Niña phases on precipitation patterns.

5 Relationship of dry/wet responses to ENSO events and the PDO

The correlation coefficients between the 31-yr moving averaged CR time series and the PDO index (Kaplan et al., 2000; Fig.7(a)) are illustrated for each site in eastern China in Fig.8. To study geographical variations of PDO modulation on dry/wet CRs, the CR time series for sites in eastern China are averaged across three regions (NC, YR, and SC) as described in the method section. The regionally averaged CR time series are plotted and compared with the PDO index in Fig.7. Tab.1 lists the correlation coefficients between the time series of regionally averaged CR and the PDO index along with their respective levels of significance. The CR time series of dry/wet response to ENSO events display significant temporal variability and prominent non-stationary behavior (Fig.7(b)–Fig.7(e)). Notably, these temporal variations in dry/wet CRs are in some cases closely associated with those of the PDO index (Fig.7(a)), as further described below.

5.1 Developing phase of El Niño

During the developing phase of an El Niño event, the influence of the PDO is evident across all three regions of Eastern China (Fig.7(b), Fig.8(a), and Fig.8(e); Tab.1). In the YR and SC regions, significantly positive (negative) correlations are observed between wet (dry) CRs and the PDO index (Tab.1). This indicates that during developing El Niño events, these regions are wetter with a positive PDO index than with a negative one. In NC, the PDO index shows a significant, negative correlation with wet CRs, while its correlation with dry CRs is positive but statistically insignificant. This suggests that NC is less wet during a positive PDO phase. Overall, the effect of PDO phase in NC contrasts with that in YR and SC.

5.2 Decaying phase of El Niño

In the decaying phase of El Niño events, significant influences of the PDO on dry/wet CRs are primarily confined to the NC and YR regions (Fig.8(b) and Fig.8(f)), as also suggested by the high correlation coefficients in Tab.1. In NC, the time series of dry (wet) CRs associated with a decaying El Niño display a positive (negative) correlation with the PDO index (Fig.7(c); Tab.1), with correlation coefficients of 0.45 (−0.49) and significant at the 0.001 level. This indicates that during a positive PDO phase, the likelihood of drought (flood) in NC in response to a decaying El Niño is significantly increased (reduced). In contrast, the pattern of PDO influence is reversed in YR. There, the time series of wet (dry) CRs associated with a decaying El Niño exhibits a positive (negative) correlation with the PDO index, with correlation coefficients of 0.65 (−0.27) at the 99.9% confidence level. This suggests that during a positive PDO phase, the chance of floods (droughts) in YR in response to a decaying El Niño is significantly increased (decreased).

Overall, a positive (negative) PDO phase tends to exacerbate (alleviate) drought hazards in NC and flood hazards in YR associated with an El Niño event in both its developing and decaying phases (Fig.2 and Fig.3). The latitudinal boundary demarcating the opposing signs of correlation between the NC and YR regions (Fig.8(b) and Fig.8(f)) is situated somewhat farther north (34° N) during the El Niño decaying phase compared to the other three ENSO phases.

5.3 Developing phase of La Niña

During the developing phase of La Niña events, NC tends to be drier with a positive PDO index than with a negative one. The time series of the wet (dry) CRs correlates negatively (positively) with the PDO index (Fig.7(d), Fig.8(c), and Fig.8(g); Tab.1). In YR, the impact of the PDO is less pronounced, with a weakly positive yet significant correlation observed between the dry CRs and the PDO index. This suggests that YR may experience slightly drier conditions when the PDO index is positive. No significant correlation is found between the PDO index and dry/wet CR time series in SC.

5.4 Decaying phase of La Niña

During the decaying phase of La Niñas, the PDO’s impact is significant only on wet CRs in NC and YR (Fig.7(e), Fig.8(d), and Fig.8(h); Tab.1), bearing a negative (positive) correlation in the former (latter) region. The overall impact is drier (wetter) conditions in NC and wetter (drier) conditions in YR when the PDO index is positive (negative).

These findings underscore the complex interplay between ENSO phases and PDO in influencing dry and wet conditions across different regions of China. The PDO significantly influences the regional response to ENSO events, highlighting the importance of considering both phenomena in understanding climate variability and its impacts.

6 Discussion and conclusions

The application of Event Coincidence Analysis (ECA) on historical DWI records and ENSO chronologies has provided valuable insights into the dry/wet responses across various sites in China during different ENSO phases. This study also offers novel interpretations of historical records, particularly regarding the previously unresolved issue of PDO influence on the ENSO-EASM relationship. Our findings indicate that a positive (negative) PDO index leads to more frequent dry (wet) conditions in North China, regardless of the ENSO phase. In contrast, the Yangtze River basin experiences more frequent wet (dry) conditions during both phases of El Niño events when the PDO index is positive (negative). More insights are provided thanks to the inclusion of historical records prior to the instrumental era, which significantly lengthens the period for analysis by more than one PDO cycle. It is worth noting that the low correlations observed in certain regions, such as the South China, suggest that other factors might moderate the PDO’s influence on regional precipitation responses to ENSO.

Examining whether the effects of ENSO and PDO phases on dry/wet conditions align can offer valuable insights for disaster prevention and mitigation. For instance, in North China, where a decaying El Niño increases drought probability, and a positive PDO exacerbates this condition, drought preparedness should be prioritized during a positive PDO phase. Conversely, with a negative PDO, the effects of ENSO and PDO tend to cancel each other out, reducing the urgency of drought preparedness. In the Yangtze River basin, similar considerations apply for flood hazards, with a positive PDO increasing the risk during a decaying El Niño phase.

The PDO influence of ENSO impacts on dry/wet conditions observed in this ECA study may help explain decadal changes in precipitation patterns in East Asia. Our results suggest that the PDO has opposite impacts on dry/wet responses to ENSO depending on its phase. Therefore, when studying long periods that include alternating PDO phases, it is crucial to consider these opposing effects, as they might blur or cancel each other out (Chan and Zhou, 2005).

The PDO modulation on the effects of dry/wet responses in China to ENSO events is speculated to be associated with a combination of teleconnection patterns and atmospheric circulation dynamics. The Pacific–Japan/East Asian–Pacific (PJ/EAP) pattern plays a key role, wherein positive PDO phases induce an anomalous meridional wave train along the East Asian coast, leading to dry conditions in NC via an anticyclonic circulation anomaly (Qian and Zhou, 2014). On the other hand, analyses of the observations in the instrumental era (after 1950 CE) have identified the role of the western North Pacific subtropical high (WNPSH) and the associated western North Pacific anomalous anticyclone (WNPAC) in conveying ENSO impacts to East Asia (Yang et al., 2017; Zhang et al., 2017b). During positive PDO, the WNPSH is strengthened, intensifying anticyclonic circulation (WNPAC) over the western North Pacific. This enhances moisture transport and precipitation in YR while suppressing rainfall in regions to the north (Lee et al., 2019). This coordinated interaction between the PJ/EAP pattern and the WNPSH might further amplify the north–south contrast of rainfall responses. The present paper presents analysis results that align with these candidate mechanisms, and points to an avenue of future research on the relative roles of them in ENSO-EASM teleconnection in the historical period before 1950; such research can be carried out using state-of-the-art reconstructions of the past state of the atmosphere and ocean.

Documentary records offer valuable climate information, especially when instrumental observations or proxy records are unavailable (Brázdil et al., 2018). However, historical records often contain gaps. Without making assumptions with certain level of subjectiveness, it is impossible to form continuous time series from gappy data. Interpreting conventional correlation analysis results of gappy data is somewhat difficult. In contrast, ECA focuses on extreme events meeting specific criteria, allowing analysis of selected data subsets and thus circumventing the missing data problem. This study demonstrates how ECA can uncover the hidden value of climate records with missing data. However, caveats must be taken as the bias brought in by the missing data cannot be excluded. In the present study, we deem such bias as minimal since the authors of the chronicles from which the DWI atlas was compiled practiced the principle of recording the abnormal instead of the normal in order to maintain brevity of historical records (Fang et al., 2024), and the missing records are mostly years with normal level of rainfall. Further research is also needed to assess the impact of data set variability on the study’s conclusions, as inconsistencies among ENSO data sets may introduce some degree of uncertainty.

Lastly, the successful application of ECA on historical DWI and ENSO records suggests that such methodology can be used in more areas of paleoclimate research. Information extracted from tree rings, ice cores, corals, stalagmites, and ocean and lake sediments, as well as historical documents, are often in separate segments instead of long, continuous records. ECA can be used to distill insights from what is available in the segments. The present study also suggests that coincidence rate, the core concept of ECA, can serve as a new metric for quantitatively assessing inter-relationships between climate and weather events. By varying the time lag coefficient in ECA’s mathematical formulation, both instantaneous and lagged coincidences can be studied, opening new avenues for investigating lead-lag relationships between climate and weather events.

References

[1] Allan R J, Reason C J C, Lindesay J A, Ansell T J (2003). Protracted’ ENSO episodes and their impacts in the Indian Ocean region.Deep Sea Res Part II Top Stud Oceanogr, 50(12−13): 2331–2347

[2] Brázdil R, Kiss A, Luterbacher J, Nash D J, Řezníčková L (2018). Documentary data and the study of past droughts: a global state of the art.Clim Past, 14(12): 1915–1960

[3] Chan J C L, Zhou W (2005). PDO, ENSO and the early summer monsoon rainfall over south China.Geophys Res Lett, 32(8): L08810

[4] Chen Y, Guo M, Qian H (2019). Effect of ENSO on the annual precipitation anomaly in Guanzhong Basin and southern Shaanxi China. IOP Conf Ser Earth Environ Sci 344: 012090

[5] CMA (China Meteorological Administration) (1981). Yearly Charts of Dryness/Wetness in China for the Last 500-year Period. Beijing: Sino Maps Press

[6] Ding Y (1994) Monsoons over China. Dordrecht: Springer

[7] Donges J F, Schleussner C F, Siegmund J F, Donner R V (2016). Event coincidence analysis for quantifying statistical interrelation-ships between event time series.Eur Phys J Spec Top, 225(3): 471–487

[8] Fang X, Su Y, Zheng J, Xiao L, Wei Z, Yin J (2024). The social impacts of climate change in China over the past 2000 years. Singapore: Springer

[9] Galbraith C, Smyth P, Stern H S (2020). Quantifying the association between discrete event time series with applications to digital forensics.J R Stat Soc Ser A Stat Soc, 183(3): 1005–1027

[10] Gao H, Wang Y, He J (2006). Weakening significance of ENSO as a predictor of summer precipitation in China.Geophys Res Lett, 33(9): L09807

[11] Gergis J L, Fowler A M (2005). Classification of synchronous oceanic and atmospheric El Niño-Southern Oscillation (ENSO) events for palaeoclimate reconstruction.Int J Climatol, 25(12): 1541–1565

[12] Huang R, Chen J, Wang L, Lin Z (2012). Characteristics, processes, and causes of the spatio-temporal variabilities of the East Asian monsoon system.Adv Atmos Sci, 29(5): 910–942

[13] Huang R, Wu Y (1989). The influence of ENSO on the summer climate change in China and its mechanism.Adv Atmos Sci, 6(1): 21–32

[14] Ji Y, Zhou G, Wang S, Wang L (2015). Increase in flood and drought disasters during 1500–2000 in southwest China.Nat Hazards, 77(3): 1853–1861

[15] Jiang R, Xie J, Zhao Y, He H, He G (2017). Spatiotemporal variability of extreme precipitation in Shaanxi province under climate change.Theor Appl Climatol, 130(3−4): 831–845

[16] Jiang T, Zhang Q, Zhu D, Wu Y (2006). Yangtze floods and droughts (China) and teleconnections with ENSO activities (1470–2003).Quat Int, 144(1): 29–37

[17] Kaplan A, Kushnir Y, Cane M A (2000). Reduced space optimal interpolation of historical marine sea level pressure: 1854–1992.J Clim, 13(16): 2987–3002

[18] Kiladis G N, Diaz H F (1989). Global climatic anomalies associated with extremes in the Southern Oscillation.J Clim, 2(9): 1069–1090

[19] Latinne A, Morand S (2022). Climate anomalies and spillover of bat-borne viral diseases in the Asia–Pacific region and the Arabian Peninsula.Viruses, 14(5): 1100

[20] Lee S H, Seo K H, Kwon M (2019). Combined effects of El Niño and the Pacific Decadal Oscillation on summertime circulation over East Asia.Asia-Pac J Atmospheric Sci, 55(1): 91–99

[21] Li Q, Nakatsuka T, Kawamura K, Liu Y, Song H (2011) Hydroclimate variability in the North China Plain and its link with El Niño-Southern Oscillation since 1784 A.D.: insights from tree-ring cellulose δ18O. J Geophys Res Atmos 116: D22106

[22] Li Y, Strapasson A, Rojas O (2020). Assessment of El Niño and La Niña impacts on China: enhancing the early warning system on food and agriculture.Weather Clim Extrem, 27: 100208

[23] Ma Z (2007). The interdecadal trend and shift of dry/wet over the central part of north China and their relationship to the Pacific Decadal Oscillation (PDO). Chinese Sci Bull. 52: 2130-2139

[24] Mantua N J, Hare S R, Zhang Y, Wallace J M, Francis R C (1997). A Pacific interdecadal climate oscillation with impacts on salmon production.Bull Am Meteorol Soc, 78(6): 1069–1079

[25] McPhaden MJ, Zebiak SE, Glantz MH (2006). ENSO as an integrating concept in earth science.Science, 314: 1740–1745

[26] Nash D J, Adamson G C D, Ashcroft L, Bauch M, Camenisch C, Degroot D, Gergis J, Jusopović A, Labbé T, Lin K H E, Nicholson S D, Pei Q, del Rosario Prieto M, Rack U, Rojas F, White S (2021). Climate indices in historical climate reconstructions: a global state of the art.Clim Past, 17(3): 1273–1314

[27] Ortlieb L (2000). The documented historical record of El Niño events in Peru: an update of the Quinn record (sixteenth through nineteenth centuries). In: Diaz H F, ed. El Niño and the South Oscillation: Multiscale Variability and Global and Regional Impacts. Cambridge University Press, 207–295

[28] Qian C, Zhou T (2014). Multidecadal variability of North China aridity and its relationship to PDO during 1900–2010.J Clim, 27(3): 1210–1222

[29] Qian W H, Chen D, Zhu Y, Shen H Y (2003). Temporal and spatial variability of dryness/wetness in China during the last 530 years.Theor Appl Climatol, 76(1−2): 13–29

[30] Qian W, Shan X, Chen D, Zhu C, Zhu Y (2012). Droughts near the northern fringe of the East Asian summer monsoon in China during 1470–2003.Clim Change, 110(1−2): 373–383

[31] Quinn WH, Neal VT (1992) The historical record of El Niño events. In: Bradley R S, Jones P D, eds. Climate since AD 1500. London: Routledge, 623–648

[32] Rasmusson E M, Carpenter T H (1983). The relationship between eastern equatorial Pacific sea surface temperatures and rainfall over India and Sri Lanka.Mon Weather Rev, 111(3): 517–528

[33] Ropelewski C F, Halpert M S (1987). Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation.Mon Weather Rev, 115(8): 1606–1626

[34] Ropelewski C F, Halpert M S (1989). Precipitation patterns associated with the high index phase of the southern oscillation.J Clim, 2(3): 268–284

[35] Schreiber T, Schmitz A (2000). Surrogate time series.Physica D, 142(3−4): 346–382

[36] Siegmund J F, Siegmund N, Donner R V (2017). CoinCalc —A new R package for quantifying simultaneities of event series.Comput Geosci, 98: 64–72

[37] Siegmund J F, Wiedermann M, Donges J F, Donner R V (2016). Impact of temperature and precipitation extremes on the flowering dates of four German wildlife shrub species.Biogeosciences, 13: 5541–5555

[38] Song J (2000). Changes in dryness/wetness in China during the last 529 years.Int J Climatol, 20: 1003–1016

[39] Steptoe H, Jones S E O, Fox H (2018). Correlations between extreme atmospheric hazards and global teleconnections: implications for multihazard resilience.Rev Geophys, 56(1): 50–78

[40] Sun L, Yang X Q, Tao L, Fang J, Sun X (2021). Changing impact of ENSO events on the following summer rainfall in eastern China since the 1950s.J Clim, 34(20): 8105–8123

[41] Wang D, Du S C, Jia W (2022). Multiscale variability of China’s historical flood/drought index and precipitation teleconnections with ENSO using wavelet analyses.Theor Appl Climatol, 149(3−4): 1583–1597

[42] Whetton P, Rutherfurd I (1994). Historical ENSO teleconnections in the eastern hemisphere.Clim Change, 28(3): 221–253

[43] Wu R, Hu Z Z, Kirtman B P (2003). Evolution of ENSO-related rainfall anomalies in East Asia.J Clim, 16(22): 3742–3758

[44] Yang Q, Ma Z, Fan X, Yang Z L, Xu Z, Wu P (2017). Decadal modulation of precipitation patterns over eastern China by sea surface temperature anomalies.J Clim, 30(17): 7017–7033

[45] Zhang D, Li X, Liang Y (2003). Continuation (1993–2000) to “Yearly Charts of Dryness/Wetness in China for the Last 500-year Period”.J Applied Meteor Sci, 14(03): 124–133

[46] Zhang D, Liu C (1993). Continuation (1980–1992) to “Yearly charts of Dryness/Wetness in China for the Last 500-year Period”. Meteorology 019(011): 41–4545

[47] Zhang R, Min Q, Su J (2017b). Impact of El Niño on atmospheric circulations over East Asia and rainfall in China: role of the anomalous western North Pacific anticyclone.Sci China Earth Sci, 60(6): 1124–1132

[48] Zhang Y, Tian Q, Guillet S, Stoffel M (2017a). 500-yr. precipitation variability in Southern Taihang Mountains, China, and its linkages to ENSO and PDO.Clim Change, 144(3): 419–432

RIGHTS & PERMISSIONS

Higher Education Press