https://orcid.org/0009-0005-6578-9975
ABSTRACT
Extreme precipitation in eastern China (EC) is closely related to the diversity of the decaying phases of El Niño (warm-pool El Niño, i.e., WP El Niño and cold-tongue El Niño, i.e., CT El Niño), but little attention is paid to how the El Niño event variability influences precipitation sources for EC from an isotopic perspective. Stable isotopes are ideal physical tracers that can distinguish different sources of precipitation and quantify their relative contributions to precipitation. Accordingly, this study investigates spatiotemporal variations of water vapor flux and oceanic fraction to precipitation during different ENSO events by an isotopic mixing model. The results show that spatiotemporal patterns of moisture divergence for the decaying phase of WP El Niño are different from that of CT El Niño. The oceanic fraction anomalies present similar spatiotemporal trends with advection fraction anomalies. The spatiotemporal variations of precipitation source anomalies for different El Niño events are closely related to atmospheric circulations, i.e., the intensity and location of the western Pacific subtropical high (WPSH). These findings provide isotopic insights into the precipitation sources by El Niño events in EC. Future studies may further focus on the mechanisms producing extreme precipitation between the two kinds of El Niño.
HIGHLIGHTS
The impacts of different El Niño events on the regional water cycle are estimated from an isotopic perspective.
The anomalies of oceanic moisture in summer precipitation during CT El Niño are more obvious than those in WP El Niño.
These findings point out that CT El Niño should be paid immediate attention to for summer precipitation in eastern China.
INTRODUCTION
El Niño–Southern Oscillation (ENSO) exhibits a considerable degree of diversity in the sea surface temperature anomaly (SSTA) variations and corresponding hydro-meteorological impact. According to the location of SSTA in the Pacific Ocean, ENSO can be classified into eastern Pacific (EP) and central Pacific (CP) events (Li et al. 2014; Veldkamp et al. 2015). For CP events, the SSTA occurs in the EP, which is different from CT El Niño with the largest SSTA in the tropical CP. During most ENSO episodes, the SSTA develops in boreal summer, peaks in winter, and subsequently decays. Thus, summer in the preceding year is regarded as the developing phase, and winter is the mature phase of ENSO events. After the mature winter comes spring and summer, both of which belong to the decaying phase (Kug et al. 2009; Räsänen & Kummu 2013; Wen et al. 2019). A number of studies have revealed the correlations between ENSO events and extreme precipitation. Gershunov & Barnett (1998) estimated the influence of ENSO on intraseasonal extreme precipitation in the Contiguous United States based on observations and model results. The results indicated that El Niño and La Niña play a significant modulating role in extreme precipitation in the United States. Alexander et al. (2009) applied the self-organizing maps (SOMs) to investigate the response of extreme precipitation to ENSO events and showed that very strong precipitation extremes were associated with the first pattern (strong La Niña) and the last pattern (strong El Niño). Costa et al. (2021) used the principal component analysis and cluster analysis techniques to identify the relationships between precipitation extremes and El Niño. They found that extreme events are closely related to the ENSO phases (El Niño, La Niña, and Neutral). Cao et al. (2024) investigated how the diversity of ENSO events impacts the occurrence of extreme precipitation over EP and concluded that during CP El Niño, there is a higher (lower) possibility of extreme precipitation over the Yangtze River (Mei-Yu rainband in China, Baiu in Japan, and Changma in South Korea).
Schema diagram of the three-component mixing model applied for precipitation formation in EC. Precipitation is composed of advection vapor from the upwind sub-region, local evaporation vapor, and transpiration vapor.
Schema diagram of the three-component mixing model applied for precipitation formation in EC. Precipitation is composed of advection vapor from the upwind sub-region, local evaporation vapor, and transpiration vapor.
The diversity of ENSO would trigger different atmospheric circulations over EC in summer. Compared with developing summer, the western Pacific subtropical high (WPSH) in the decaying phases is more significant and extends westward (Xue et al. 2018). During the decaying phase, the WPSH in WP El Niño tends to be weaker than that in CT El Niño (Yuan et al. 2012). Moreover, WPSH could stretch to northern China in the decaying phase of WP El Niño, but only to southern China in the decaying phase of CT El Niño (Feng et al. 2011). Driven by the diverse atmospheric circulations, EC precipitation experiences large spatiotemporal variations during different types and phases of ENSO (Ren & Jin 2011; Feng et al. 2014; Zhang et al. 2016). For example, Li et al. (2014) compared the impacts of different El Niño events on water vapor transport over eastern China and found that the precipitation over the Yangtze River Basin presents a decreased tendency in developing summer but positive anomalies in decaying summer. Cao et al. (2017) investigated the influence of different ENSO types and phases on the rainy season in China and concluded that CT El Niño has a larger impact on China's precipitation than WP El Niño in both developing and decaying phases. However, such studies mainly focused on the impacts of different ENSO events on EC precipitation, rather than on the moisture sources of EC precipitation, which is more important for us to understand ENSO-induced precipitation. Furthermore, Feng et al. (2011) pointed out that the impact of El Niño on EC precipitation is more significant in the decaying phase than in the developing phase. In view of the above considerations, our study focuses on investigating how CT and WP El Niño in the decaying phase individually affect the oceanic sources of EC precipitation.
To further classify the relative contributions of precipitation sources, three kinds of methods, i.e., analytical model, numerical model, and physical tracers method, have been widely used (Galewsky et al. 2016; Wang et al. 2016; Wei & Lee 2019; Zhu et al. 2013). Compared with the analytical model and the numerical model, the physical tracers method has become increasingly popular in practical applications owing to relatively lower uncertainty (Lee et al. 2007; Yoshimura 2015; Hu et al. 2018; Shi et al. 2022). The physical tracers method usually applies stable isotopes in water vapor as tracers to quantify the role of oceanic sources in precipitation. This method assumes that precipitating vapor is composed of transpiration, evaporation, and advection vapor, and the isotopic composition of each water vapor source is unique or statistically different (Clark & Fritz 1997; Genereux 1998; Peng et al. 2020a; Chen et al. 2023). The isotopic composition of precipitation is closely related to the moisture sources and their fractions to precipitation. Based on deuterium excess, Froehlich et al. (2008) identified recycled moisture of precipitation and found that the isotopic results are in good agreement with results from other approaches. With the physical tracers method, Wang et al. (2016) quantified the contribution of recycled moisture to precipitation in oases of arid central Asia and found that the recycled moisture fractions are approximately 16.2% for large oases and less than 5% for small oases. Using oxygen as the tracer, Peng et al. (2020b) apportioned the spatiotemporal contributions of oceanic moisture to summer precipitation and showed that the oceanic moisture fraction of precipitation decreased from June to August in Northern China. Generally, this method can be classified into two categories on the basis of the number of precipitation sources: the three-component mixing model and the two-component mixing model. The two-component mixing model usually neglects the contribution of transpiration in practical case studies, thus a three-component mixing model is more desirable to EC where transpiration is indispensable in summer (Gao et al. 2000).
Accordingly, this study aims to assess the impacts of the decaying phase of ENSO events on regional water cycles from an isotopic perspective and explain the potential mechanisms. It is organized as follows. A description of the study area and data are presented in Section 2. Classification of the ENSO event, the three-component mixing model, evaluation index, and methods are introduced in Section 3. The ENSO effects on water vapor transport and precipitation sources for EC, as well as uncertainty estimation in the results, are investigated in Section 4. Conclusions are summarized in Section 5.
MATERIALS
Study area
Location map and sub-regions of eastern China (EC), as well as the distribution of isotope stations.
Location map and sub-regions of eastern China (EC), as well as the distribution of isotope stations.
Data
The isotopic dataset consists of oxygen-18 and deuterium composition of precipitation. The oxygen-18 compositions of precipitation were collected from Peng et al. (2020b), which generate reliable and spatiotemporally continuous precipitation oxygen isoscape for EC. The deuterium compositions of precipitation were obtained by bias-correcting simulations from the isotope-enabled global climate model Laboratoire de Météorologie Dynamique GCM (LMDZiso). The resolution of LMDZiso (50–60 km) is relatively higher than that of other iGCMs and the timespan of LMDZiso (1979–2016) is relatively longer. Linear scaling (LS) and distribution translation (DT) are applied in this study because they exhibit similar performances in correcting isotopic simulations (Peng et al. 2020a). The isotopic compositions of precipitation are at a spatial resolution of 50–60 km.
Meteorological data for this study contain temperature, precipitation, wind speed, wind direction, air pressure, precipitable water content, and relative humidity. All these data are collected from the monthly 0.5° × 0.5° gridded dataset of the China Meteorological Data Service Center (CMDC) (http://data.cma.cn/en) and bilinearly interpolated into a resolution of 50–60 km to be in line with the LMDZiso grid.
METHODS
To investigate the impact of ENSO types and phases, we identify two different types of ENSO from 1979 to 2016 by the variation of SSTA. A three-component mixing model is applied to apportion the spatiotemporal contributions of oceanic moisture. Results during the same type of ENSO years are averaged to obtain the mean values. The fraction anomaly index is adopted to evaluate the spatiotemporal variations of the contribution of advection or oceanic moisture to precipitation under different ENSO events. Gaussian first-order approximation is applied to assess uncertainty in the results.
Classification of ENSO regimes
We calculate the December–February averaged NCT and NWP from 1979 to 2016 and then select the years when NCT and NWP are larger than one standard deviation to represent the developing phases of CT and WP El Niño, respectively. The year after the mature phases is regarded as the decaying year. In total, there are five WP events in the decaying phase and four CT events in the decaying phase from 1979 to 2016. The decaying years of WP and CT El Niño are dominated by the two types of ENSO, as listed in Table 1.
The decaying years of warm-pool (WP) El Niño and cold-tongue (CT) El Niño during 1979–2016
| Types | Decaying years | ||||
|---|---|---|---|---|---|
| WP | 1988 | 1991 | 1995 | 2005 | 2010 |
| CT | 1983 | 1992 | 1998 | 2016 | |
| Types | Decaying years | ||||
|---|---|---|---|---|---|
| WP | 1988 | 1991 | 1995 | 2005 | 2010 |
| CT | 1983 | 1992 | 1998 | 2016 | |
Three-component mixing model
According to Equations (1) and (2), isotopic compositions of precipitating vapor, evaporation vapor, transpiration vapor, and advection vapor are necessary for calculating the contribution of oceanic moisture to precipitation.
Relationships between simulated hydrogen isotope in precipitation and in soil water for SC (a), YZ (b), NC (c), and NEC (d) in summer.
Relationships between simulated hydrogen isotope in precipitation and in soil water for SC (a), YZ (b), NC (c), and NEC (d) in summer.
and 
represent the amount of precipitable water in the sub-region denoting i (product sub-region) and i − 1 (upwind sub-region), respectively.Fraction anomaly index
and 
represent the average fraction of advection or oceanic moisture to precipitation in ENSO and non-ENSO years, respectively.Gaussian first-order approximation
, 
, and FTr (Phillips & Gregg 2001; Zhang & Liu 2016). This method uses standard errors of isotopic composition for advection vapor, precipitating vapor, evaporation vapor, and transpiration vapor to estimate the 95% confidence intervals (CIs) for the mean proportion estimates as
represents the variance of each source, and the subscripts Adv, Ev, Tr, and PV represent the advection vapor, evaporation vapor, transpiration vapor and precipitating vapor, respectively, which are assumed to be independent.
Besides, approximate 95% CIs and standard errors for the mean proportion (F) are calculated as 
for uncertainty estimates.
RESULTS AND DISCUSSION
Moisture transport and its divergence
Spatiotemporal variations of regional mean water vapor flux (kg m−1 s−1; vector) and its divergence anomalies (10−6 kg m−2 s−1; shading) from surface to 300 hPa for eastern China during WP (a) and CT El Niño (b) decaying summer.
Spatiotemporal variations of regional mean water vapor flux (kg m−1 s−1; vector) and its divergence anomalies (10−6 kg m−2 s−1; shading) from surface to 300 hPa for eastern China during WP (a) and CT El Niño (b) decaying summer.
Advection vapor contribution anomalies to precipitation
Spatiotemporal variations of contribution anomalies of advection vapor to the precipitation over eastern China during WP (a) and CT El Niño (b) in decaying summer. Red points represent negative anomalies. Blue points represent positive anomalies. The size of the point represents the intensity of the anomaly. The bigger size means a larger anomaly in the specific sub-region.
Spatiotemporal variations of contribution anomalies of advection vapor to the precipitation over eastern China during WP (a) and CT El Niño (b) in decaying summer. Red points represent negative anomalies. Blue points represent positive anomalies. The size of the point represents the intensity of the anomaly. The bigger size means a larger anomaly in the specific sub-region.
The spatiotemporal patterns of advection moisture fraction under different ENSO events are similar to the moisture divergence anomalies in Section 4.1 with opposite signs and the variations of precipitation amount anomalies by previous research. As concluded in Feng et al. (2011), the signals of advection fraction during the decaying phase of WP El Niño are weaker than CT El Niño. This means that decaying CT El Niño in the decaying phase has a larger effect on precipitation and its advection source, compared with WP El Niño. Besides, WP and CT El Niño in the decaying phase show different spatial patterns of precipitation anomalies. SC shows dry signals during CT El Niño and wet signals during WP El Niño in the decaying phase (Li et al. 2014). The consistency between advection fraction anomalies and precipitation anomalies is because advection vapor dominates EC summer precipitation during ENSO events, thus its contribution anomalies well capture precipitation anomalies. Furthermore, the results based on the three-component mixing model are highly consistent with those of moisture divergence anomalies, which also prove the reliability of isotopic analysis. However, precipitation anomalies (ranging from −30 to 20%) during different ENSO events are more obvious than advection fraction anomalies (ranging from −7 to 12%). This might be because the signals of precipitation anomalies affected by ENSO events contain advection, evaporation, and transpiration anomalies, thus the precipitation anomalies should be bigger than either precipitation source anomalies.
Oceanic moisture contribution anomalies to precipitation
Spatiotemporal variations of contribution anomalies of oceanic moisture to precipitation over eastern China during WP and CT El Niño decaying summer. Red points represent negative anomalies. Blue points represent positive anomalies. The size of the point represents the intensity of the anomaly. The bigger size means a larger anomaly in the specific sub-region.
Spatiotemporal variations of contribution anomalies of oceanic moisture to precipitation over eastern China during WP and CT El Niño decaying summer. Red points represent negative anomalies. Blue points represent positive anomalies. The size of the point represents the intensity of the anomaly. The bigger size means a larger anomaly in the specific sub-region.
The spatiotemporal variations of contribution anomalies of oceanic moisture to precipitation might be explained by the intensity and location of WPSH. The variations of WPSH could be illustrated by the western North Pacific (WNP) anti-cyclone anomaly. For the decaying phase of WP El Niño, the WPSH extends northward with WNP anti-cyclone moving around to 40°N (Sun & Ying 1999; Zhou et al. 2005). This induces more oceanic moisture transport to NC which is located northwest of WPSH and southwest winds carry more oceanic moisture driven by WNP anti-cyclone. For the decaying phase of CT El Niño, the WPSH retreats at a low level with WNP anti-cyclone only stretching to 25°N (Feng et al. 2011). The WNP anti-cyclone confines oceanic moisture transport, which causes negative oceanic fraction anomaly to appear in SC. Furthermore, the prominent oceanic anomalies during CT El Niño can be explained by the relatively stronger anti-cyclone during CT El Niño, which enhances oceanic moisture transport (Wang et al. 2000; Yuan et al. 2012; Cao et al. 2017).
Uncertainty analysis
Mean uncertainty contributions from the precipitating vapor, advection vapor, evaporation vapor, and transpiration vapor to the variances of advection proportion estimates in summer over SC (a), YZ (b), and NC (c).
Mean uncertainty contributions from the precipitating vapor, advection vapor, evaporation vapor, and transpiration vapor to the variances of advection proportion estimates in summer over SC (a), YZ (b), and NC (c).
Sample numbers and standard errors of advection fraction, evaporation fraction, and transpiration fraction to summer precipitation over SC, YZ, and NC
| Sub-region | NPV; NEv; NAdv; NTr | SEfEv | SEfAdv | SEfTr |
|---|---|---|---|---|
| SC | 140 | 0.02 | 0.04 | 0.03 |
| YZ | 233 | 0.01 | 0.02 | 0.01 |
| NC | 156 | 0.02 | 0.04 | 0.01 |
| Sub-region | NPV; NEv; NAdv; NTr | SEfEv | SEfAdv | SEfTr |
|---|---|---|---|---|
| SC | 140 | 0.02 | 0.04 | 0.03 |
| YZ | 233 | 0.01 | 0.02 | 0.01 |
| NC | 156 | 0.02 | 0.04 | 0.01 |
Approximate 95% CIs for advection fraction, evaporation fraction, and transpiration fraction to precipitation in summer over the sub-regions of EC.
Approximate 95% CIs for advection fraction, evaporation fraction, and transpiration fraction to precipitation in summer over the sub-regions of EC.
CONCLUSIONS
In this study, we investigate how the decaying phase of WP and CT El Niño affect summer precipitation sources over EC from an isotopic perspective. Frequently used CT and WP indexes are employed to classify the two kinds of ENSO events. Water vapor flux, during the decaying phase of WP and CT El Niño, is compared. The three-component mixing model is applied to quantify the spatiotemporal variations of advection and oceanic fractions of regional precipitation during the two kinds of ENSO events. Gaussian first-order approximation, standard error, and 95% CIs are used to verify the accuracy of estimated contributions.
It is found that compared with CT El Niño, the moisture divergence in the decaying phase of WP El Niño shows almost opposite signals over EC with a smaller magnitude for WP El Niño. Spatiotemporal patterns of advection fraction to precipitation during WP El Niño are different from those during CT El Niño. The spatiotemporal pattern of advection fraction anomaly is similar to that of moisture divergence anomaly with opposite signs and to that of oceanic fraction anomaly. The ENSO-induced anomalies can be explained by the strength and location of the WPSH. The largest uncertainty of the estimated anomalies comes from the precipitating vapor, owing to higher variances of the isotopic composition of precipitating vapor. 95% CIs for the estimates are narrow and the standard errors are low, which indicates the robustness of the estimated proportions. These findings provide new insight into how the variations of ENSO in the decaying phase affect the regional water cycle in EC.
ACKNOWLEDGEMENTS
This work was partially supported by the National Natural Science Foundation of China (Grant No. 52109007), the Natural Science Foundation of Hubei Province (Grant No. 2020CFA100), the Natural Science Foundation of Chongqing, China (Grant No. cstc2021jcyj-msxm2426), and the Scientific and Technological Research Program of Chongqing Municipal Education Commission (Grant No. KJQN2021007). The authors would like to thank Camille Risi for providing the gridded δOP and δHP for eastern China at a spatial resolution of 50–60 km. The authors also acknowledge Prof. Xuefa Wen from the Chinese Academy of Sciences (CAS) for providing the station observed δOP and δHP for the study area. The numerical calculations in this paper have been done on the supercomputing system in the Supercomputing Center of Wuhan University.
AUTHOR CONTRIBUTIONS
P.P.: conceptualization, methodology, software, and writing. Y.Z.: computation, visualization, and editing of the manuscript. J.C.: supervision, writing – review. X.J.Z.: conceptualization and editing of the manuscript. X.L.: conceptualization and writing – review. D.X.: formal analysis and validation.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.
%20cropped.png?versionId=7143)