Teleconnections between the Atlantic Multidecadal Oscillation and eastern China summer precipitation during the Medieval Climate Anomaly and Little Ice Age

Abstract

The teleconnection between the Atlantic Multidecadal Oscillation (AMO) and eastern China summer precipitation (ECSP) for the modern period has been emphasized in the past few decades. We test its stability during the Medieval Climate Anomaly (MCA) and Little Ice Age (LIA) based on the composite results of the median of six models under the Paleoclimate Modeling Intercomparison Project Phase3 (PMIP3) framework. During the MCA, ECSP generally increases in positive AMO phases relative to negative phases, similar to the modern teleconnection, while during the LIA, the precipitation tends to reduce over eastern China, especially in the Yangtze River basin. Decomposing the precipitation change on the basis of a diagnostic moisture budget manifests that the different AMO-related precipitation changes stem from distinct effects of circulation-induced moisture convergence during the two periods. Compared with negative AMO phases, positive AMO phases during the MCA show an anomalous lower-level cyclone and upper-level anticyclone over eastern China that facilitate the upward motion anomaly and precipitation excess. During the LIA, a barotropic anticyclone centers in Northeast China and weakens the high-level westerlies over eastern China; this favors descending and upper-level divergence anomalies and leads to precipitation decreases. The distinct convergence changes are determined by differing propagation paths of the AMO-induced teleconnection wave train during the two periods.

Keywords

Atlantic Multidecadal Oscillation eastern China summer precipitation Little Ice Age Medieval Climate Anomaly simulation

Introduction

Sea surface temperature (SST) variability can modulate the climate system on a range of timescales and lead to profound impacts on the society (e.g. Huang and Sun, 1992; Sutton and Hodson, 2007; Wallace et al., 1998). On the multidecadal timescale, the Atlantic Multidecadal Oscillation (AMO; Kerr, 2000) is one of the major SST variability patterns. Key features of the observed AMO are multidecadal co-variation of SST over the entire North Atlantic with a periodicity on the order of 65 to 80 years (e.g. Deser et al., 2010; Enfield et al., 2001). The AMO is thought to be a natural mode linked to the Atlantic thermohaline circulation (e.g. Delworth and Mann, 2000; Msadek et al., 2011; Zhang and Delworth, 2005), but recent studies highlight external forcings also contribute to its pace and phase changes (e.g. Birkel et al., 2018; Knudsen et al., 2014; Otterå et al., 2010). Although the forcing mechanism still subjects to debate, such a spatially coherent SST variation mode can act as a pacemaker that affects both local and remote climate, such as Atlantic hurricanes (Goldenberg et al., 2001; Trenberth and Shea, 2006), North America and African Sahel rainfall (Hu et al., 2011; Liu et al., 2014; Wang et al., 2013), European temperature (Hong et al., 2017), tropical Pacific state (Dong et al., 2006; Sun et al., 2017a), and India summer climate (Feng and Hu, 2008; Goswami et al., 2006; Joshi and Ha, 2019; Joshi and Pandey, 2011; Joshi and Rai, 2015; Li et al., 2008); it is also reported that the AMO can exert its influence on the East Asian climate, including the eastern China summer precipitation (ECSP; e.g. Fan et al., 2018; Lu et al., 2006; Wang et al., 2009).

ECSP has profound impacts on the national economy and lives, and related drought and flood disasters often cause serious economic losses. Observational studies argue that positive AMO phases generally correspond to excessive ECSP (e.g. Li et al., 2017; Lu et al., 2006; Ning et al., 2017), and this teleconnection comprises two processes. On the one side, AMO-related SST anomalies (SSTAs) can alter the Walker circulation and intertropical convergence zone through the atmospheric bridge; this disturbs the low-latitude western Pacific SST and further the ECSP (Hong et al., 2013; Lu et al., 2006; Sun et al., 2017a). On the other side, the AMO leads to anomalous North Atlantic heating and upper-level divergence, exciting a teleconnection wave train that propagates eastward along the westerly jet to affect the eastern China climate (Gao et al., 2019; Si and Ding, 2016; Sun et al., 2017b).

Note that previous analyses are based on instrumental records that cover only one to two complete AMO cycles. The short time of these records limits a comprehensive understanding of the AMO–ECSP teleconnection, and whether the mechanism behind the two processes holds under different climatic backgrounds is also unclear. To break through the limitation, climate reconstructions have been attempted to extend the AMO index backward for hundreds of years through proxy network, such as tree ring, ice core, coral, sediment, and historical document (Gray et al., 2004; Mann et al., 2009; Wang et al., 2017). Nevertheless, the proxy records suffer from sparsely populating temporally and spatially, varying seasonal sensitivities, and additional nonclimatic information (e.g. Jones et al., 2009; Phipps et al., 2013; Wahl and Smerdon, 2012), therefore affecting the manifestation of the AMO decadal variations. Numerical climate models, as essential tools to study the climatic change and its mechanism, can provide an independent perspective for addressing the issue. In particular, a suite of coupled global climate models have performed the last millennium experiment under the framework of Paleoclimate Modelling Intercomparison Project Phase 3 (PMIP3; Braconnot et al., 2012), and the simulations are proved to be capable of presenting the climate features during the last millennium (PAGES 2K Consortium, 2015; Parsons et al., 2017). Using the simulation data, scientists have discussed some teleconnections during the last millennium, such as El Niño–Southern Oscillation and Australian rainfall (Brown et al., 2016; Lewis and LeGrande, 2015), North Atlantic Oscillation and ECSP (Peng, 2018), and AMO and Iceland–Scotland overflow (Lohmann et al., 2015), but the AMO–ECSP teleconnection remains to be analyzed. Moreover, there are two typical periods during the last millennium, that is, the Medieval Climate Anomaly (MCA; 950–1250) and Little Ice Age (LIA; 1450–1850) (IPCC, 2013). In most proxies and simulations, there are distinct differences between the two periods. The former is featured by a warm and wet climate, while the latter is cold and dry (Ljungqvist et al., 2016; Mann et al., 2009). Their different climate features are largely attributed to external forcings, such as the volcanic eruptions and solar irradiance (e.g. Atwood et al., 2016; Crowley, 2000; Schurer et al., 2013). In this manner, we can use the last millennium simulation outputs during the MCA and LIA to investigate the long-term AMO features and related ECSP changes under different climatic backgrounds.

The study is organized as follows. In section ‘Data and method’, we describe the data and method. In section ‘Model evaluation’, an evaluation about model performances in delineating the observed AMO behavior and AMO–ECSP teleconnection is conducted. In section ‘Results’, we analyze spatial and temporal structures of the AMO, the teleconnection between the AMO and ECSP, and relevant mechanisms during the MCA and LIA based on the last millennium simulation. At last, summary and discussion are presented in section ‘Summary and discussion’.

Data and method

Model and observational data

The study employs coupled global climate models that perform both the historical (1850–2005) and last millennium simulations (850–1850). The historical simulation is used for model evaluation, which is forced by observed time-varying forcings, including natural and anthropogenic aerosol emissions, solar radiation, greenhouse gases, and land use (Taylor et al., 2012). For the last millennium simulation, reconstructed external forcings are imposed, which are composed of volcanic aerosol, solar radiation related to orbital parameters and solar activity, greenhouse gases, and land use (Schmidt et al., 2011). Among 10 models that fulfill the requirement, the MIROC-ESM model is ruled out because of an abnormal warming drift in the last millennium simulation (Sueyoshi et al., 2013). A brief description of the remaining nine models is shown in Supplemental Table S1, available online. Detailed information is presented in Schmidt et al. (2011).

To evaluate the fidelity of the simulated AMO pattern, the global monthly SST from the Hadley Centre Sea Ice and Sea Surface Temperature (HadISST; Rayner et al., 2003) dataset (horizontal resolution 1° × 1°) for 1870–2017 is applied. Furthermore, we use two observation datasets of precipitation for testing modern AMO–ECSP teleconnection. They are the Climatic Research Unit time series version 4.01 (CRU TS v4.01; Harris et al., 2013) dataset of the Hadley Center for 1901–2016 and the Global Precipitation Climatology Center version 7 (GPCC v7; Schneider et al., 2017) dataset of the National Center for Atmospheric Research (NCAR) for 1901–2013. Their horizontal resolutions are both 0.5° × 0.5°. We adopt the full time ranges of the datasets to utilize as much data as possible, so the time ranges of historical simulations (1850–2005) are longer than observations (1870–2017 for SST and 1900–2010s for precipitation), and the results are qualitatively consistent with the patterns during the overlapped time period (not shown). Moreover, the spatial resolution differs among datasets. We interpolate all the datasets to a common latitude-longitude grid of 0.5° × 0.5° using bilinear interpolation to keep uniformity and calculate the multi-model ensemble results.

Method

The AMO index is calculated from the area-weighted North Atlantic basin (0°–60°N and 0°–80°W) SSTA. To reduce the impacts of global signal, such as the anthropogenic global warming, the global mean SSTA over 60°S–60°N is first subtracted from the North Atlantic SSTA before calculating the AMO index (Lyu and Yu, 2017; Trenberth and Shea, 2006). Because the effect of signal over the global ocean is not totally equal to the North Atlantic in some models, an additional de-trending process is imposed to the AMO index to make up the gap. This method to get the simulated AMO index is superior to just calculating the de-trended area-weighted North Atlantic SSTA in PMIP3/CMIP5 models (Lyu and Yu, 2017). Then the index is filtered by an 11-year moving average to highlight the decadal component, as has been done by previous studies (Han et al., 2016; Lu et al., 2006). Periods with the AMO index above 0.5 (less than −0.5) standard deviation of the index series are defined as positive (negative) AMO phases (Luo et al., 2018).

Since positive and negative AMO phases are always relevant to climate variables with opposite polarities, a composite analysis is performed by calculating the mean values in positive AMO phases minus those in the negative ones to show the climate variable anomalies related to the AMO. The composite differences are highly consistent with the results using regression method that calculates the regression coefficient between the AMO index and related climate variables, but the observed changes using composite method are more evident compared with regression method (Supplemental Figures S1–S4, available online). The statistical significance of the composite difference is tested using the standard Student’s t-test.

To scrutinize the propagation of the Rossby wave energy, the horizontal wave activity flux put forward by Plumb (1985) is calculated:

F s = \frac{p \cos ϕ}{2} (\begin{matrix} \frac{1}{a^{2} \cos {}^{2}φ} [{(\frac{\partial ψ^{'}}{\partial λ})}^{2} - ψ^{'} \frac{\partial^{2} ψ^{'}}{\partial {}^{2}λ}] \\ \frac{1}{a^{2} \cos φ} (\frac{\partial ψ^{'}}{\partial λ} \frac{\partial ψ^{'}}{\partial φ} - ψ^{'} \frac{\partial^{2} ψ^{'}}{\partial λ \partial φ}) \end{matrix})

(1)

Here, the vector Fs is the horizontal wave activity flux. In the equation, p = pressure/1000 hPa, and ψ′ is a small perturbation of the streamfunction (the composite positive-minus-negative AMO difference of streamfunction at 200 hPa in this study) relative to its zonal mean. φ, a, and λ represent the latitude, Earth radius, and longitude, respectively.

To unravel the potential mechanism for the AMO-related ECSP change, we carry out a moisture budget analysis (Seager et al., 2010). The equation can be expressed as:

\begin{array}{l} δ \bar{P} \approx δ \bar{E} - \frac{1}{ρ_{w} g} \int_{0}^{p_{s}} \nabla \cdot [(δ \bar{q}) \bar{u}] d p - \frac{1}{ρ_{w} g} \int_{0}^{p_{s}} \nabla \cdot [(δ \bar{u}) \bar{q}] d p - \frac{1}{ρ_{w} g} \int_{0}^{p_{s}} \nabla \cdot δ (\bar{q^{'} u^{'}}) d p \\ δ T H δ D Y δ T E \end{array}

(2)

In equation (2), the notation δ represents the composite positive-minus-negative AMO differences of given variables. Overbars indicate monthly means, and primes indicate departures from the monthly mean. ρ_w is the density of water, g is the gravitational acceleration, p_s is surface pressure, q is specific humidity, and u is the horizontal wind vector. Terms related to changes in q are treated as the thermodynamic contributor (δTH), and terms involving changes in u are viewed as the circulation dynamic contributor (δDY). Thus, the precipitation changes (δP) are decomposed into changes in evaporation (δE), transient eddy (δTE, represented by the residual term of the equation), as well as thermodynamic (δTH) and circulation dynamic (δDY) processes.

In the study, the circulation dynamic term is further divided into two components:

δ D Y = δ D Y_{A} + δ D Y_{D}

(3)

δ D Y_{A} = \frac{1}{ρ_{w} g} \int_{0}^{p_{s}} (δ \bar{u} \cdot \nabla \bar{q}) d p

(4)

δ D Y_{D} = - \frac{1}{ρ_{w} g} \int_{0}^{p_{s}} (\bar{q} \nabla \cdot δ \bar{u}) d p

(5)

The two components describe moisture advection (δDY_A) and convergence (δDY_D) via circulation change.

Model evaluation

First, the skill of the nine PMIP3 models to simulate the observed AMO characteristics is assessed by comparing the historical simulations with observations. Spatially, the composite difference of observed North Atlantic SST exhibits positive anomalies over the entire basin for positive AMO phases relative to negative ones; maximum SSTA centers over the subpolar region, and a secondary warm zone is in the northern subtropical Atlantic, west of Northwest Africa coast (Figure 1a). In PMIP3 models, the positive North Atlantic SSTA pattern is generally reproduced, although there are somewhat discrepancies (Supplemental Figure S1, available online). The simulated high-latitude warm SSTA locates more southward and eastward than the observed one; in several models, the subtropical warmth weakens or disappears; in CCSM4, significant positive SSTAs only emerge in high latitudes, whereas negative ones occupy the middle-latitude western North Atlantic. To quantify the extent to which the individual models are capable of reproducing the spatial structure of AMO, we calculate the spatial correlation coefficients (SCC), normalized standard deviation, and normalized centered root-mean-square error (CRMSE) of simulated SSTA relative to observation over the North Atlantic basin (0°–60°N and 0°–80°W). The Taylor diagram (Supplemental Figure S5, available online; Taylor, 2001) shows that the SCCs range from 0.21 to 0.85, the normalized standard deviations range from 0.44 to 1.30, and the normalized CRMSEs range from 0.62 to 1.10. This indicates that the ability of the models differs from one another. Most models are capable of describing the AMO-related SSTA. However, CCSM4 does not perform well considering the SCC (0.21) and CRMSE (1.10), and IPSL-CM5A-LR shows the smallest normalized standard deviation (0.44).

Figure 1.

Spatial patterns of the AMO (°C) in observation (a, HadISST, 1870–2017) and six-model median (b, 1850–2005) during the instrumental period. The dots in (a) indicate areas with statistically significant anomalies at the 90% confidence level, and gray oblique lines in (b) represent regions where at least four out of the six models share the same sign of change.

In regard to the temporal aspect, the power spectrum features two prominent multidecadal periodicities of 35 and 70 years for the observed AMO index (Supplemental Figure S6j, available online), consistent with earlier studies (Deser et al., 2010; Kerr, 2000). The models also exhibit multidecadal AMO cycles. One spectral peak is at about 20 years, and a longer one peaks at 40 to 70 years. The latter is not significant in some models, such as CCSM4, FGOALS-s2, and IPSL-CM5A-LR (Supplemental Figure S6a–i, available online). Note that we mainly focus on the decadal cycles and hence do not discuss the potential interannual periodicities, which are absent in the power spectrum because of an 11-year moving average to the AMO index. Also notice that the observed AMO phase shifts at about 1899, 1928, 1965, and 1997, consistent with previous studies (Enfield et al., 2001; Feng and Hu, 2008; Trenberth and Shea, 2006), but the simulated AMO shows random time series with inconsistent sequences of AMO regime among models (Supplemental Figure S7, available online). This originates from diverse natural variability phases in individual models (Zhang and Wang, 2013).

Further, the composite positive-minus-negative AMO difference of the ECSP is calculated. In positive AMO phases, more precipitation appears over most of eastern China in observations, while deficit is restricted to Bohai and East China Sea coast (Figure 2a and b), accordant with previous studies (Li et al., 2017; Ning et al., 2017; Qian et al., 2014). Quantitatively, the composite difference of precipitation averaged over eastern China (22°–40°N, 110°–122°E) is 0.15 (0.20) mm day⁻¹ in the CRU (GPCC) dataset. Six out of the nine models reproduce excessive precipitation over eastern China, although the anomalous magnitude (0.03–0.27 mm day⁻¹) and sub-regional pattern vary with models (Supplemental Figure S3a, c–e, h, and i, available online). Three models, CCSM4, HadCM3, and IPSL-CM5A-LR, show broadly negative precipitation anomalies in positive AMO phases (Supplemental Figure S3b, f, and g, available online). It is understandable that CCSM4 and IPSL-CM5A-LR can not capture the AMO–ECSP teleconnection considering they show some inadequacies in reproducing the spatial and temporal AMO features, especially there are significantly colder North Atlantic SSTAs in CCSM4 but much warmer in IPSL-CM5A-LR. As to HadCM3, it fails to simulate the La Niña-like SSTA pattern in Pacific during the positive AMO phases as other models and observation do (not shown), and the Pacific SSTA pattern is greatly relevant to ECSP.

Figure 2.

Composite positive-minus-negative AMO differences of precipitation (mm day⁻¹) over eastern China in observations (a, CRU, 1901–2016; b, GPCC, 1901–2013) and six-model median (c, 1850–2005) during the instrumental period. The dots in (a) and (b) indicate areas with statistically significant anomalies at the 90% confidence level, and gray oblique lines in (c) represent regions where at least four out of the six models share the same sign of change.

Briefly, models can generally capture the spatial structure and multidecadal cycles of the observed AMO, although there exist certain inadequacies. Six models portray the pluvial condition over eastern China during positive AMO phases as the observation shows. For this reason, only those six models are applied to the further analyses. As shown in Figures 1 and 2, the median of the six models well reproduce the observed positive North Atlantic SSTA and ECSP excess during positive AMO phases relative to negative ones, and thus the six-model median results are emphatically analyzed in following sections. We use the multi-model median instead of multi-model mean to measure the center tendency of multiple models. This is because we need to reduce the influence of obvious outliers when only six models results are synthesized, and the median is resistant to outliers (Li et al., 2012).

Results

AMO features and AMO–ECSP teleconnection during the MCA and LIA

Figure 3 displays the six-model median of the AMO spatial pattern during the MCA and LIA. The composite difference of North Atlantic SST exhibits basin-scale positive values during the MCA and LIA. Largest SSTA occurs over 40°N–60°N, similar to that during the historical period. There are also discrepancies in the SST changes between the two periods; the MCA exhibits a comma-like structure with least warming in middle latitudes, while the other period shows more uniform and stronger changes. The warmer SSTAs during the two periods are well portrayed by individual models (Supplemental Figures S8 and S9, available online); only BCC-CSM1-1 shows significant colder SSTAs in the middle latitudes during the MCA. As for the temporal dimension, most models show a significant peak at around 40 years and a minor peak at 20 years during the MCA (Supplemental Figure S10a–f, available online); whereas during the LIA, the main periodicities are less uniform, with a shorter cycle at 20 to 30 years and a longer one at 40 to 80 years (Supplemental Figure S10g–l, available online). Overall, the AMO features during the MCA and LIA are little different, with similar coherent North Atlantic SSTAs and multidecadal cycles.

Figure 3.

Spatial patterns of the AMO (°C) in six-model median during the (a) MCA and (b) LIA. Gray oblique lines represent regions where at least four out of the six models share the same sign of change.

The AMO–ECSP teleconnections during the MCA and LIA are then examined based on the composite difference of ECSP (Figure 4). During the MCA, positive AMO phases are accompanied by generally positive ECSP anomalies in the six-model median, which is also the characteristic in the historical period; regionally, the Yangtze River basin and South China show largest precipitation excess, and the deficit only exists in part of North China (Figure 4a). By contrast, the precipitation in positive AMO phases during the LIA tends to reduce over eastern China, especially in the Yangtze River basin (Figure 4b). Individual models also hold the pattern of generally excessive precipitation over eastern China and notable excess in the Yangtze River basin during the MCA (Supplemental Figure S11, available online). During the LIA, the precipitation deficit in the Yangtze River basin is more or less reproduced in five models, except for MRI-CGCM3 (Supplemental Figure S12, available online).

Figure 4.

Composite positive-minus-negative AMO differences of precipitation (mm day⁻¹) over eastern China during the (a) MCA and (b) LIA. Gray oblique lines represent regions where at least four out of the six models share the same sign of change.

Moisture budget analysis

The moisture budget components of AMO-related ECSP anomalies during the MCA and LIA are depicted in Figures 5 and 6. During the MCA, the largest part of precipitation changes are linked to circulation dynamic and transient eddy items. The circulation dynamic change is similar in pattern to the precipitation one, with negative anomalies over part of North China but positive in other regions (Figure 5a), while the transient eddy term shows overall positive values over eastern China (Figure 5d). During the LIA, circulation dynamic and transient eddy items also dominate the precipitation anomalies. The contribution of the transient eddy term is consistent with that during the MCA, promoting the precipitation excess (Figure 6a), but the circulation dynamic term mainly facilitates a precipitation deficit (Figure 6d). In addition, the thermodynamic term shows notable positive values over the coast of Bohai during the LIA. Because different changes between the two periods mainly appear in circulation dynamic term, that term is further broken down into two subterms, the circulation-induced moisture advection and convergence terms (Figure 7). Changes in the moisture advection term are similar between the two periods, with positive anomalies existing in North China and negative from the Yangtze–Huai River Valley to South China (Figure 7a and c). For the convergence term, the anomalous magnitude is larger than the advection term, and the spatial patterns are distinct between the MCA and LIA, that is, positive anomalies in eastern China except for North China during the MCA and negative excepting part of South China during the LIA (Figure 7b and d). Therefore, the circulation-induced moisture advection, convergence, and transient eddy terms all play important roles in the AMO-related ECSP changes during the MCA and LIA, but only the moisture convergence term shows quite dissimilar anomalies during the two periods and thus is the primary reason for different ECSP changes.

Figure 5.

Items of the moisture budget equation for the AMO-related ECSP changes (mm day⁻¹) during the MCA: (a) circulation dynamic item, (b) thermodynamic item, (c) evaporation item, and (d) residual item (representing the transient eddy item). Gray oblique lines represent regions where at least four out of the six models share the same sign of change.

Figure 6.

Same as Figure 5, but during the LIA.

Figure 7.

Subitems of the circulation dynamic item in the moisture budget equation (mm day⁻¹): moisture advection item and moisture convergence during the (a and b) MCA and (c and d) LIA. Gray oblique lines represent regions where at least four out of the six models share the same sign of change.

Two paths linking the AMO and ECSP

How the AMO leads to different composite circulation convergence and in turn ECSP changes during the MCA and LIA remains to be resolved. Thus, the effects of the low-latitude air–sea interactions and teleconnection wave train that may relate the AMO to ECSP are further examined. On the one side, the path of the AMO influencing the ECSP through disturbing the Walker circulation and further air–sea interactions is shown in Figure 8. The composite difference of upper-level velocity potential shows a zonal dipole pattern, with anomalous convergence over the eastern to central Pacific and divergence from western Pacific to Atlantic. This pattern is similar between the MCA and LIA (Figure 8a and b). In the lower troposphere, there is noticeable divergence and high pressure anomaly in the eastern to central Pacific, while the western Pacific to Atlantic is marked by convergence and low-pressure anomaly during the two periods, opposite to the upper troposphere (Figure 8c and d). As such, there is an anomalous low-pressure (cyclonic) center near Taiwan in the lower level (Figure 9a and b). In the northwest side of the anomalous cyclone, eastern China is affected by the associated northeasterly anomalies, which are unfavorable to the moisture inflow to eastern China. Similarly, the anticyclonic anomaly centering in Bohai Bay (40°–45°N and 125°E) brings more water vapor to North China through anomalous southeasterly winds. This explains the similar effects of the moisture advection term during the MCA and LIA, which increase the precipitation over North China and reduce it for the remaining parts of eastern China (Figure 7a and c).

Figure 8.

Same as Figure 4, but for velocity potential (shading, 10⁵ m² s⁻¹) and divergent wind (vectors, m s⁻¹) at 200 hPa and sea level pressure (shading, Pa) and horizontal wind (vectors, m s⁻¹) at 850 hPa during the (a and c) MCA and (b and d) LIA.

Figure 9.

Same as Figure 4, but for horizontal wind (m s⁻¹) at 850 hPa and 200 hPa and zonal wind at 200 hPa during the (a, c, and e) MCA and (b, d, and f) LIA. In (e) and (f), gray oblique lines represent regions where at least four out of the six models share the same sign of change, and the structure of upper-tropospheric westerlies at 200 hPa are indicated by the contours of zonal wind equal to 10, 18, and 26 m s⁻¹.

On the other side, the AMO-related eastward propagating wave train is expressed by composite differences of geopotential heights at 200, 500, and 700 hPa (Figures 10 and 11). During the MCA, the upper level is mainly characterized by positive height anomalies in the Northern Hemisphere for positive AMO phases compared with negative phases (Figure 10c). A belt of anomalous positive height nodes surround the mid-latitude regions, forming a circumglobal teleconnection (CGT) wave pattern (Ding and Wang, 2005). The anomalous height nodes exist from the lower to upper troposphere, demonstrating equivalent barotropic structures (Figure 10). This teleconnection wave train could be stimulated by the AMO-related upper-level divergence over North Atlantic (Figure 8a), as discussed by previous literatures (Sun et al., 2015, 2017b). During the LIA, the height anomaly nodes also form a barotropic CGT wave pattern (Figure 11). However, its positive anomalies are stronger than that during the MCA, and the height anomaly nodes in Eurasia lie more southward than during the MCA. To exhibit a clearer CGT pattern, we calculate the geopotential height eddy (deviation from the zonal mean) anomaly and the wave active flux at 200 hPa (Figure 12). Over East Asia, the positive height anomaly node centers in the east of Lake Baikal (around 50°N and 125°E) during the MCA (Figure 12a), but its LIA counterpart centers in around 40°N and 115°E (Figure 12b). Together with the height anomaly, the related anticyclonic anomaly during the LIA also extends more southward than that during the MCA (Figure 9c and d). During the LIA, the anticyclone leads to anomalous easterlies over 20°–40°N. Thus, climatological westerlies over eastern China are weakened, and the maximum weakening lies along the Meiyu region (Figure 9f). This favors anomalous upper-level zonal convergence, consequently descending, and ultimately deficient precipitation over eastern China. During the MCA, the easterly anomalies related to the anticyclone are along 35°–45°N and the east of 120°E (Figure 9e). Therefore, their impacts on eastern China are only limited to North China where precipitation decreases. In the south of North China, there is an upper-level anomalous anticyclone as the counterpart of the anomalous cyclone centering near Taiwan in the lower troposphere, forming a baroclinic pattern. This anomalous baroclinic system impels lower-level convergence, upper-level divergence, and upward motion anomalies, which are beneficial to precipitation excess. Altogether, the AMO-related CGT wave train affects the convergence field over eastern China, and dissimilar spatial locations of the geopotential height nodes during the MCA and LIA lead to different convergence anomalies over East Asia, explaining different contributions of the moisture convergence term during the two periods.

Figure 10.

Composite positive-minus-negative AMO differences of geopotential height (m) at (a) 700 hPa, (b) 500 hPa, and (c) 200 hPa during the MCA. Gray oblique lines represent regions where at least four out of the six models share the same sign of change.

Figure 11.

Same as Figure 10, but during the LIA.

Figure 12.

Composite positive-minus-negative AMO differences of geopotential height eddy (m) and wave activity fluxes (vector, m² s⁻²) at 200 hPa during the (a) MCA and (b) LIA. Only the wave activity fluxes greater than 1 m² s⁻² are drawn.

Summary and discussion

This study investigates the teleconnection between the AMO and ECSP during the MCA and LIA using the last millennium simulation outputs from PMIP3 models. Most models are capable of reproducing the observed AMO features and the positive ECSP anomalies during modern positive AMO phases. However, there are also models that can not well portray the observed AMO-related SST pattern and multidecadal cycles, which generally makes the models hard to capture the AMO–ECSP teleconnection. Therefore, models that perform well are selected out for the analysis. According to the median of the six selected models, the AMO–ECSP teleconnections are different between the MCA and LIA, although the states of the AMO are similar during the two periods. During the warm MCA, eastern China is dominated by positive precipitation anomalies in positive AMO phases, similar to the observational period; exceptions occur in North China where precipitation reduces during the MCA. During the cold LIA, precipitation tends to decrease over eastern China, and an increase only occurs in South China and Henan Province. A diagnostic moisture budget analysis reveals that the precipitation changes are jointly induced by transient eddy, circulation-induced moisture convergence and moisture advection terms, and the last term is the chief source for distinct precipitation changes during the two periods.

Two paths are involved in the ECSP changes during the two periods. Firstly, the AMO-related North Atlantic SSTA disturbs the basic state of low-latitude Pacific through atmospheric bridge and leads to a lower-level anomalous cyclone centering near Taiwan in positive AMO phases, obstructing the moisture inflowing to eastern China. This process is accordant between the MCA and LIA. Secondly, the warm North Atlantic SSTA in positive AMO phases induces a CGT wave train that spreads along the middle latitudes. During the LIA, one of the anomalous positive height nodes of the wave train centers at 40°N and 115°E. Related to the height anomaly, there are easterly anomalies weakening the climatological westerlies over eastern China. This brings downward motion, upper-level convergence, and ultimately negative precipitation anomalies. During the MCA, the wave path over Eurasia locates more northward. Therefore, the related anomalous easterlies have little influence on eastern China. An anomalous high pressure in the upper level and its paired low pressure in the lower level control eastern China, and this structure promotes anomalous lower-level convergence and upper-level divergence, leading to excessive precipitation. This means that the different locations of the CGT wave train lead to different circulation convergence anomalies over East Asia, thus contributing to the distinct effects of the circulation-induced moisture convergence term during the two periods.

It is still unclear why the AMO-induced CGT wave train trajectory locates more southward during the LIA than during the MCA. The different climatological basic flows between the two periods may act an important role, because basic flow can restrain the Rossby wave path. Specifically, the climatological westerly jet is accompanied by large vorticity gradients, which can form a waveguide to limit the meridional dimension of the wave train and trap it in the jet stream waveguide (Hoskins and Ambrizzi, 1993). This is particularly true for the CGT wave train (Ding and Wang, 2005). Therefore, we compare the climatological states of the summer westerlies between the two periods. It is shown that the zonal wind displays a dipole anomaly pattern; strengthened westerlies are in the westerly zone and its south and weakened ones in its north during the LIA compared with the MCA, which is particularly remarkable over Eurasia (Supplemental Figure S13a, available online). This indicates an enhancement and equatorial shift of the summer westerlies over Eurasia during the LIA, concurring with previous reconstructions and simulations (e.g. Chen et al., 2010; Jiang et al., 2020; Shi et al., 2016). As the westerlies show an equatorial movement, the AMO-related wave train could also shift southward. The movement of westerlies can be explained by the thermal wind relationship. Compared with the MCA, the temperature in the troposphere is lower during the LIA, and the cooling over Eurasian continent is strongest (Supplemental Figure S13b, available online). Accordingly, the meridional temperature gradient increases in the south of the cooling center at about 45°N and decreases in its north (Supplemental Figure S13c, available online), consistent with the changes of westerlies.

This study indicates the teleconnection between the AMO and ECSP is unstable under different climatic backgrounds. The average temperature is similar between the MCA and observational period, and the AMO–ECSP teleconnections during the two periods are broadly consistent. However, as the climatological temperature decreases during the LIA, the upper-level westerlies in the Northern Hemisphere move southward, which benefits a southward motion of the AMO-related wave train, and affects the AMO–ECSP teleconnection. This has implication for predicting the AMO-related climatic changes. With the global warming persisting, it is suggested that the westerlies tend to shift poleward in the future (e.g. Lorenz and DeWeaver, 2007; Rivière, 2011; Yin, 2005). We may expect an accompanying poleward shifting of the AMO-related wave train and associated ECSP changes. Moreover, in our study, some models have difficulties in reproducing the AMO spatial pattern and its multidecadal periodicities. This implies some processes relevant to the formation of the AMO may be not properly simulated by models, which needs to be further studied for a better simulation of the AMO-related climatic changes.

Supplemental Material

20200608-supplementary_material – Supplemental material for Teleconnections between the Atlantic Multidecadal Oscillation and eastern China summer precipitation during the Medieval Climate Anomaly and Little Ice Age

Supplemental material, 20200608-supplementary_material for Teleconnections between the Atlantic Multidecadal Oscillation and eastern China summer precipitation during the Medieval Climate Anomaly and Little Ice Age by Xuecheng Zhou, Xianmei Lang and Dabang Jiang in The Holocene

Footnotes

Acknowledgements

We sincerely thank the two anonymous reviewers for their insightful comments and suggestions to improve this manuscript. We also acknowledge the climate modeling groups participating in the CMIP5/PMIP3 for producing and sharing their model outputs.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the National Key R&D Program of China (2017YFA0603302 and 2017YFA0603404) and the National Natural Science Foundation of China (41625018).

ORCID iDs

Xianmei Lang

Dabang Jiang

Supplemental material

Supplemental material for this article is available online.

References

Atwood

Frierson

DMW

, et al. (2016) Quantifying climate forcings and feedbacks over the last millennium in the CMIP5–PMIP3 models. Journal of Climate 29(3): 1161–1178.

Birkel

Mayewski

Maasch

, et al. (2018) Evidence for a volcanic underpinning of the Atlantic Multidecadal Oscillation. npj Climate and Atmospheric Science 1(1): 24.

Braconnot

Harrison

Kageyama

, et al. (2012) Evaluation of climate models using palaeoclimatic data. Nature Climate Change 2(6): 417–424.

Brown

Hope

Gergis

, et al. (2016) ENSO teleconnections with Australian rainfall in coupled model simulations of the last millennium. Climate Dynamics 47(1–2): 79–93.

Chen

Holmes

(2010) Moisture changes over the last millennium in arid central Asia: A review, synthesis and comparison with monsoon region. Quaternary Science Reviews 29(7–8): 1055–1068.

Crowley

(2000) Causes of climate change over the past 1000 years. Science 289(5477): 270–277.

Delworth

Mann

(2000) Observed and simulated multidecadal variability in the Northern Hemisphere. Climate Dynamics 16(9): 661–676.

Deser

Alexander

Xie

, et al. (2010) Sea surface temperature variability: Patterns and mechanisms. Annual Review of Marine Science 2: 115–143.

Ding

Wang

(2005) Circumglobal teleconnection in the Northern Hemisphere summer. Journal of Climate 18(17): 3483–3505.

10.

Dong

Sutton

Scaife

(2006) Multidecadal modulation of El Nino–Southern Oscillation (ENSO) variance by Atlantic Ocean sea surface temperatures. Geophysical Research Letters 33(8): L08705.

11.

Enfield

Mestas-Nunez

Trimble

(2001) The Atlantic Multidecadal Oscillation and its relation to rainfall and river flows in the continental U.S. Geophysical Research Letters 28(10): 2077–2080.

12.

Fan

, et al. (2018) ENSO–South China Sea summer monsoon interaction modulated by the Atlantic Multidecadal Oscillation. Journal of Climate 31(8): 3061–3076.

13.

Feng

(2008) How the North Atlantic Multidecadal Oscillation may have influenced the Indian summer monsoon during the past two millennia. Geophysical Research Letters 35(1): L01707.

14.

Gao

Yang

Gong

, et al. (2019) Footprints of Atlantic Multidecadal Oscillation in the low-frequency variation of extreme high temperature in the Northern Hemisphere. Journal of Climate 32(3): 791–802.

15.

Goldenberg

Landsea

Nunez

AMM

, et al. (2001) The recent increase in Atlantic hurricane activity: Causes and implications. Science 293(5529): 474–479.

16.

Goswami

Madhusoodanan

Neema

, et al. (2006) A physical mechanism for North Atlantic SST influence on the Indian summer monsoon. Geophysical Research Letters 33(2): L02706.

17.

Gray

Graumlich

Betancourt

, et al. (2004) A tree-ring based reconstruction of the Atlantic Multidecadal Oscillation since 1567 AD. Geophysical Research Letters 31(12): L12205.

18.

Han

Luo

, et al. (2016) Simulation by CMIP5 models of the Atlantic Multidecadal Oscillation and its climate impacts. Advances in Atmospheric Sciences 33(12): 1329–1342.

19.

Harris

Jones

Osborn

, et al. (2013) Updated high-resolution grids of monthly climatic observations – the CRU TS3.10 Dataset. International Journal of Climatology 34(3): 623–642.

20.

Hong

Kang

Choi

, et al. (2013) Climate responses in the tropical Pacific associated with Atlantic warming in recent decades. Asia-Pacific Journal of Atmospheric Sciences 49(2): 209–217.

21.

Hong

(2017) Amplified summer warming in Europe–West Asia and Northeast Asia after the mid-1990s. Environmental Research Letters 12(9): 094007.

22.

Hoskins

Ambrizzi

(1993) Rossby wave propagation on a realistic longitudinally varying flow. Atmospheric Sciences 50(12): 1661–1671.

23.

Feng

Oglesby

(2011) Variations in North American summer precipitation driven by the Atlantic Multidecadal Oscillation. Journal of Climate 24(21): 5555–5570.

24.

Huang

Sun

(1992) Impacts of the tropical western Pacific on the East Asian summer monsoon. Journal of the Meteorological Society of Japan 70(1B): 243–256.

25.

IPCC (2013) Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge: Cambridge University Press, p. 1535.

26.

Jiang

Yan

Wang

(2020) Variation of the summer Asian westerly jet over the last millennium based on the PMIP3 simulations. The Holocene 30(2): 332–343.

27.

Jones

Briffa

Osborn

, et al. (2009) High-resolution palaeoclimatology of the last millennium: A review of current status and future prospects. The Holocene 19(1): 3–49.

28.

Joshi

(2019) Fidelity of CMIP5-simulated teleconnection between Atlantic Multidecadal Oscillation and Indian summer monsoon rainfall. Climate Dynamics 52(7–8): 4157–4176.

29.

Joshi

Pandey

(2011) Trend and spectral analysis of rainfall over India during 1901–2000. Journal of Geophysical Research: Atmospheres 116: D06104.

30.

Joshi

Rai

(2015) Combined interplay of the Atlantic Multidecadal Oscillation and the Interdecadal Pacific Oscillation on rainfall and its extremes over Indian subcontinent. Climate Dynamics 44(11–12): 3339–3359.

31.

Kerr

(2000) A North Atlantic climate pacemaker for the centuries. Science 288(5473): 1984–1985.

32.

Knudsen

Jacobsen

Seidenkrantz

, et al. (2014) Evidence for external forcing of the Atlantic Multidecadal Oscillation since termination of the Little Ice Age. Nature Communications 5: 3323.

33.

Lewis

LeGrande

(2015) Stability of ENSO and its tropical Pacific teleconnections over the last millennium. Climate of the Past 11: 1347–1360.

34.

Jin

(2012) Adaptive trimmed mean as a location estimate. Journal of Systems Science & Complexity 25(5): 973–986.

35.

Perlwitz

Quan

, et al. (2008) Modelling the influence of North Atlantic multidecadal warmth on the Indian summer rainfall. Geophysical Research Letters 35(5): L05804.

36.

Ding

(2017) Interdecadal variability of the Afro-Asian summer monsoon system. Advances in Atmospheric Sciences 34(7): 833–846.

37.

Liu

Chiang

Chou

, et al. (2014) Atmospheric teleconnection mechanisms of extratropical North Atlantic SST influence on Sahel rainfall. Climate Dynamics 43(9–10): 2797–2811.

38.

Ljungqvist

Krusic

Sundqvist

, et al. (2016) Northern Hemisphere hydroclimate variability over the past twelve centuries. Nature 532(7597): 94–98.

39.

Lohmann

Mignot

Langehaug

, et al. (2015) Using simulations of the last millennium to understand climate variability seen in palaeo-observations: Similar variation of Iceland–Scotland overflow strength and Atlantic Multidecadal Oscillation. Climate of the Past 11(2): 203–216.

40.

Lorenz

DeWeaver

(2007) Tropopause height and zonal wind response to global warming in the IPCC scenario integrations. Journal of Geophysical Research: Atmospheres 112: D10119.

41.

Dong

Ding

(2006) Impact of the Atlantic Multidecadal Oscillation on the Asian summer monsoon. Geophysical Research Letters 33(24): L24701.

42.

Luo

Gao

, et al. (2018) The connection between the Atlantic Multidecadal Oscillation and the Indian summer monsoon in CMIP5 models. Climate Dynamics 51(7–8): 3023–3039.

43.

Lyu

(2017) Climate impacts of the Atlantic Multidecadal Oscillation simulated in the CMIP5 models: A re-evaluation based on a revised index. Geophysical Research Letters 44(8): 3867–3876.

44.

Mann

Zhang

Rutherford

, et al. (2009) Global signatures and dynamical origins of the Little Ice Age and Medieval Climate Anomaly. Science 326(5957): 1256–1260.

45.

Msadek

Frankignoul

LZX

(2011) Mechanisms of the atmospheric response to North Atlantic multidecadal variability: A model study. Climate Dynamics 36(7–8): 1255–1276.

46.

Ning

Liu

Sun

(2017) Influences of volcano eruptions on Asian summer monsoon over the last 110 years. Scientific Reports 7: 42626.

47.

Otterå

Bentsen

Drange

, et al. (2010) External forcing as a metronome for Atlantic multidecadal variability. Nature Geoscience 3(10): 688–694.

48.

PAGES 2K Consortium (2015) Continentalscale temperature variability in PMIP3 simulations and PAGES 2k regional temperature reconstructions over the past millennium. Climate of the Past 11: 1673–1699.

49.

Parsons

Loope

Overpeck

, et al. (2017) Temperature and precipitation variance in CMIP5 simulations and paleoclimate records of the last millennium. Journal of Climate 30(22): 8885–8912.

50.

Peng

(2018) Simulated interannual teleconnection between the summer North Atlantic Oscillation and summer precipitation in eastern China during the last millennium. Geophysical Research Letters 45(15): 7741–7747.

51.

Phipps

McGregor

Gergis

, et al. (2013) Paleoclimate data–model comparison and the role of climate forcings over the past 1500 years. Journal of Climate 26(18): 6915–6936.

52.

Plumb

(1985) On the three-dimensional propagation of stationary waves. Journal of the Atmospheric Sciences 42(3): 217–229.

53.

Qian

Chen

(2014) Decadal summer drought frequency in China: The increasing influence of the Atlantic Multidecadal Oscillation. Environmental Research Letters 9(12): 124004.

54.

Rayner

NAA

Parker

Horton

, et al. (2003) Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. Journal of Geophysical Research: Atmospheres 108(D14): 4407.

55.

Rivière

(2011) A dynamical interpretation of the poleward shift of the jet streams in global warming scenarios. Journal of the Atmospheric Sciences 68(6): 1253–1272.

56.

Schmidt

Jungclaus

Ammann

, et al. (2011) Climate forcing reconstructions for use in PMIP simulations of the last millennium (v1.0). Geoscientific Model Development 4(1): 33–45.

57.

Schneider

Finger

Meyer-Christoffer

, et al. (2017) Evaluating the hydrological cycle over land using the newly-corrected precipitation climatology from the Global Precipitation Climatology Centre (GPCC). Atmosphere 8(3): 52.

58.

Schurer

Hegerl

Mann

, et al. (2013) Separating forced from chaotic climate variability over the past millennium. Journal of Climate 26(18): 6954–6973.

59.

Seager

Naik

Vecchi

(2010) Thermodynamic and dynamic mechanisms for large-scale changes in the hydrological cycle in response to global warming. Journal of Climate 23(17): 4651–4668.

60.

Shi

Yan

Jiang

, et al. (2016) Precipitation variation over eastern China and arid central Asia during the past millennium and its possible mechanism: Perspectives from PMIP3 experiments. Journal of Geophysical Research: Atmospheres 121(20): 11989–12004.

61.

Ding

(2016) Oceanic forcings of the interdecadal variability in East Asian summer rainfall. Journal of Climate 29(21): 7633–7649.

62.

Sueyoshi

Ohgaito

Yamamoto

, et al. (2013) Set-up of the PMIP3 paleoclimate experiments conducted using an Earth system model, MIROC-ESM. Geoscientific Model Development 6(3): 819–836.

63.

Sun

Kucharski

, et al. (2017a) Western tropical Pacific multidecadal variability forced by the Atlantic Multidecadal Oscillation. Nature Communications 8: 15998.

64.

Sun

Ding

, et al. (2017b) Cold season Africa–Asia multidecadal teleconnection pattern and its relation to the Atlantic multidecadal variability. Climate Dynamics 48(11–12): 3903–3918.

65.

Sun

Zhao

(2015) Remote influence of Atlantic multidecadal variability on Siberian warm season precipitation. Scientific Reports 5: 16853.

66.

Sutton

Hodson

(2007) Climate response to basin-scale warming and cooling of the North Atlantic Ocean. Journal of Climate 20(5): 891–907.

67.

Taylor

(2001) Summarizing multiple aspects of model performance in a single diagram. Journal of Geophysical Research: Atmospheres 106: 7183–7192.

68.

Taylor

Stouffer

Meehl

(2012) An overview of CMIP5 and the experiment design. Bulletin of the American Meteorological Society 93(4): 485–498.

69.

Trenberth

Shea

(2006) Atlantic hurricanes and natural variability in 2005. Geophysical Research Letters 33(12): L12704.

70.

Wahl

Smerdon

(2012) Comparative performance of paleoclimate field and index reconstructions derived from climate proxies and noise-only predictors. Geophysical Research Letters 39(6): L06703.

71.

Wallace

Rasmusson

Mitchell

, et al. (1998) On the structure and evolution of ENSO-related climate variability in the tropical Pacific: Lessons from TOGA. Journal of Geophysical Research: Oceans 103(C7): 14241–14259.

72.

Wang

Kim

, et al. (2013) Northern Hemisphere summer monsoon intensified by mega-El Niño/Southern Oscillation and Atlantic Multidecadal Oscillation. Proceedings of the National Academy of Sciences of the United States of America 110(14): 5347–5352.

73.

Wang

Yang

Ljungqvist

, et al. (2017) Internal and external forcing of multidecadal Atlantic climate variability over the past 1,200 years. Nature Geoscience 10(7): 512–518.

74.

Wang

Luo

(2009) Seasonal response of Asian monsoonal climate to the Atlantic Multidecadal Oscillation. Journal of Geophysical Research: Atmospheres 114(D2): D02112.

75.

Yin

(2005) A consistent poleward shift of the storm tracks in simulations of 21st-century climate. Geophysical Research Letters 32(18): L18701.

76.

Zhang

Wang

(2013) Multidecadal North Atlantic sea surface temperature and Atlantic meridional overturning circulation variability in CMIP5 historical simulations. Journal of Geophysical Research: Oceans 118(10): 5772–5791.

77.

Zhang

Delworth

(2005) Simulated tropical response to a substantial weakening of the Atlantic thermohaline circulation. Journal of Climate 18(12): 1853–1860.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

4.05 MB