Distinct influential mechanisms of the warm pool Madden–Julian Oscillation on persistent extreme cold events in Northeast China

Yitian Qian,Pang-Chi Hsu,Huijun Wang,Mingkeng Duan

Key Laboratory of Meteorological Disaster, Ministry of Education, Nanjing University of Information Science & Technology, Nanjing 210044, China

Keywords:Persistent extreme cold events Northeast China Madden–Julian Oscillation

ABSTRACT This study investigates whether and how the Madden–Julian Oscillation (MJO) influences persistent extreme cold events (PECEs),a major type of natural disaster in boreal winter,over Northeast China.Significantly increased occurrence probabilities of PECEs over Northeast China are observed in phases 3 and 5 of the MJO,when MJOrelated convection is located over the eastern Indian Ocean and the western Pacific,respectively.Using the temperature tendency equation,it is found that the physical processes resulting in the cooling effects required for the occurrence of PECEs are distinct in the two phases of the MJO when MJO-related convection is consistently located over the warm pool area.The PECEs in phase 3 of the MJO mainly occur as a result of adiabatic cooling associated with ascending motion of the low-pressure anomaly over Northeast Asia.The cooling effect associated with phase 5 is stronger and longer than that in phase 3.The PECEs associated with phase 5 of the MJO are linked with the northwesterly cold advection of a cyclonic anomaly,which is part of the subtropical Rossby wave train induced by MJO-related convection in the tropical western Pacific.

1.Introduction

Persistent extreme cold events (PECEs) have occurred frequently in China in recent winters and have had devastating effects on agriculture,transportation,power infrastructure,and human health,attracting much attention from both the public and the scientific community(Peng and Bueh,2011 ;Bueh et al.,2011).For example,in early January 2008,South China experienced PECEs coupled with heavy rain and icy conditions that affected more than 100 million people and led to economic losses of more than 15 billion CNY (Zhao et al.,2008 ;Hong and Li,2009).In January 2018,a widespread PECE brought heavy storms and decreased temperatures in Northeast and North China,especially Inner Mongolia,leading to one of the most severe and persistent weather disasters in China (Liu et al.,2018 ;Qin et al.,2020).

Northeast China is one of the most important regions for food production and is vulnerable to cold damage (Li et al.,2018).Previous studies have mainly focused on the interannual and interdecadal changes in the surface air temperature (SAT) in Northeast China.The interannual variabilities in anomalous SATs in Northeast China are mainly influenced by the East Asian winter monsoon,including the intensification and southeastward extension of the Siberian high (Gong and Ho,2002 ;Liu and Zhu,2020 ;Zhang and Lu,2021 ;Yang and Fan,2022),the East Asian trough and the upper-level westerly jet (Ha et al.,2012 ;Lee et al.,2013).In addition,sea surface temperature anomalies in the tropical Pacific and Atlantic are also related to the interannual changes in the SAT in Northeast China (Jiang et al.,2014).The variabilities in Arctic sea-ice condensation,snow cover over the Mongolian Plateau and the North Pacific Oscillation are related to interdecadal changes in the SAT in Northeast China (Liu and Zhu,2020 ;Yun et al.,2018).Little attention has been paid so far to the variability of PECEs on subseasonal time scales.

The Madden–Julian Oscillation (MJO;Madden and Julian,1971,1972) is one of the most dominant subseasonal variabilities in the tropics and presents as an eastward-propagating circulation anomaly and coherent convective activity along the equatorial region on a time scale of 10–90 days.During the life cycle of the MJO,MJO-related convection in tropical regions induces anomalous circulation and affects the temperature and precipitation in East Asia (Jeong et al.,2005 ;Matsueda and Takaya,2015 ;Liu and Hsu,2019 ;Kim et al.,2020).A symmetrical feature presents as most regions of China experiencing significant cooling in phases 2–4 and warming in phases 6–8 (Jeong et al.,2005).Seo et al.(2016) suggested that adiabatic subsidence caused by a local Hadley circulation following phase 3 of the MJO is a major driver of warming over most regions of East Asia,whereas the cooling process is related to a dipole convection anomaly associated with phase 7 of the MJO (Kim et al.,2020).

The cold conditions in East Asia influenced by the MJO and related physical mechanisms have been discussed in many studies.Jeong et al.(2005) indicated that MJO-related convection over the Indian Ocean is correlated with cold anomalies and extreme cold surges in East Asia by inducing strong northerly anomalies.The record-breaking PECEs over Southeast Asia in 2008 were maintained by northerly winds related to stationary MJO convection over the Maritime Continent (Hong and Li,2009).By contrast,with the eastward propagation of the MJO-related convection,the associated upper-level divergence and convergence can act as a Rossby wave source and excite the poleward dispersion of the Rossby wave train (Jin and Hoskins,1995 ;Matthews et al.,2004 ;Tseng et al.,2019),which modulates the atmospheric circulation in different regions of the mid-to high latitudes with a delayed effect (He et al.,2011 ;Seo et al.,2016 ;Abdillah et al.,2018 ;Cui et al.,2021).Previous studies have mainly focused on the modulation of temperature variabilities over East Asia by the MJO,whereas how,and to what extent,the MJO exerts an influence on PECEs in Northeast China needs further investigation.

This paper is organized as follows.Section 2 describes the data and methods.Section 3 discusses the modulation and control of PECEs in Northeast China by the MJO.The final section summarizes our conclusions.

2.Data and methods

2.1. Data

2.1.1.ObservationaldataanddefinitionofPECEs

Gauge-based data from the Asian Precipitation-Highly Resolved Observational Data Integration Towards Evaluation (APHRODITE) gridded SAT over the Asian monsoon region are used to capture the spatial distribution of PECEs in China (Yasutomi et al.,2011).APHRODITE provides high-resolution (0.5° × 0.5°) data and has been widely used for studying extreme events (Hsu et al.,2015,2017).To obtain robust results,the CN05.1 dataset (Wu and Gao,2013),which is derived from more than 2400 daily station records,is used for comparison.The daily SAT anomaly is defined as the SAT relative to the daily climatological mean over the period 1979–2015.

PECEs are commonly regarded as prolonged periods of extremely cold days.Following the definitions of a PECE in previous studies (Peng and Bueh,2011 ;Freychet et al.,2021),we define a PECE as an SAT lower than the fifth percentile for at least three consecutive days.The fifth percentile of the daily SAT is identified for each calendar day in a 15-day window from 1 December to 28 February in the period 1979–2015 for each individual grid point.This gives a total of 540 samples(15 days ×36 winters) for each calendar day.

2.1.2.ReanalysisdataandMJOindices

The daily zonal (u) and meridional (v) winds,verticalp-velocity (ω),temperature (T),and geopotential height (Z) at 21 vertical levels from 1000 to 100 hPa are derived from the ERA5 (fifth major global reanalysis produced by ECMWF) dataset (Hersbach et al.,2020).These data are used to analyze the large-scale fields associated with PECEs in Northeast China and to reveal the physical processes responsible for the variability in temperature in Northeast China.We define winter as 1 December to 28 February and analyze 36 winters from 1979 to 2015.The domain of Northeast China is defined as (40°–54°N,110°–135°E).

The evolution of the MJO cycle is determined using the real-time multivariate MJO (RMM) indices proposed by Wheeler and Hendon(2004).The RMM indices are derived from the empirical orthogonal functions of the 850-and 200-hPa zonal winds and outgoing longwave radiation (OLR) anomalies from the Australian Bureau of Meteorology (http://cawcr.gov.au/staff/mwheeler/maproom/RMM).Days with a large RMM amplitude ((RMM12+RMM22)1/2≥ 1) are defined as active MJO days.To identify the deep convection associated with the different phases of the MJO,we use daily OLR data at a horizontal resolution of 2.5°×2.5° from the polar-orbiting satellites of the National Oceanic and Atmospheric Administration (Liebmann and Smith,1996).The MJO-related signals are extracted by the Lanczos 10–90-day bandpass filtering method (Duchon,1979) with 201 weights.

2.2. Methods

2.2.1.Temperaturebudgetequation

To elucidate the physical processes controlling the variability of temperature over Northeast China associated with the MJO,we analyze the temperature budget equation on an intraseasonal (10–90-day) time scale.Eq.(1) shows that the intraseasonal temperature anomaly tendency at each pressure level is affected by horizontal temperature advection,the adiabatic process associated with vertical motion and static stability,and diabatic heating,including small-scale physical processes:

In Eq.(1),a curly brace indicates the intraseasonal components associated with the variability of the MJO on a 10–90-day timescale,Tis the temperature in each pressure level,tis the time in seconds,Vis the horizontal velocity vector,? is the horizontal gradient operator,ωis verticalp-velocity,Qis atmospheric apparent heat source,andσis the static stability,calculated as

whereRis the gas constant,Pis the pressure (Pa),andCpis the specific heat at constant pressure.

To determine the temperature variability associated with the MJO activity,the variables can be decomposed into mean and anomaly fields,and the anomaly fields include MJO and non-MJO components,

The overbar denotes the seasonal-mean state,while the prime represents the deviations of the mean state.With our focus on the influence of the MJO on PECEs,we do not consider the interannual changes in the seasonal-mean components and instead use the climatological winter (DJF)-mean during the period 1979–2015.The curly brace and the asterisk symbols represent the MJO and non-MJO components,respectively.Note that {X′} is equal to {X},because {X* } is close to zero and negligible.The temperature budget on an intraseasonal timescale can therefore be further decomposed into nine terms:

Fig.1.(a) Climatological mean,(b) standard deviation,and (c) the fifth percentile of the SAT in DJF over East Asia during the period 1979–2015 (units: °C).The red contours in (a) and (c) represent 0°C.(d) The climatological mean frequency of PECEs in winter (units: days per season).The red box in part (d) indicates Northeast China (40°–54°N,110°–135°E).

uandvin Eq.(4) are the zonal and meridional winds,respectively.The first five terms on the right-hand side of Eq.(4) are the zonal and meridional temperature advection of the mean temperature by the intraseasonal wind anomaly,and of the intraseasonal temperature by the mean wind,and the nonlinear interaction between the two anomalies at the intraseasonal time scale,respectively.The subsequent three terms are the adiabatic heating effects,which correspond to the interactions between the intraseasonal temperature anomaly and the mean vertical velocity,between the mean temperature and the intraseasonal vertical velocity anomaly,and the nonlinear interaction between the two anomalies at the intraseasonal time scale.The last term represents the intraseasonal component of diabatic heating.

2.2.2.Definitionofpercentagechangesintheoccurrenceprobabilityof PECEs

To investigate the modulation of the MJO on the occurrence probability of PECEs in each MJO phase,we define the percentage change in the occurrence probability of a PECE (CPi) in each MJO phase relative to the non-MJO period:

wherePiandPnon-MJOare the occurrence probabilities of PECEs in the each active MJO phase and during the non-MJO period,respectively.The non-MJO phase is defined as an RMM amplitude (RMM12+RMM22)1/2<1.

3.Results

Fig.1 (a,b) shows the climatological mean and standard deviation of the DJF SAT over East Asia during the period 1979 -2015.Overall,the spatial distribution of the winter-mean temperature shows a prominent south–north contrast in East Asia.Daily average temperatures＞0°C are found in South Asia,Southeast Asia,south of 35°N in China,and in southern Japan (Fig.1 (a)).The 0°C contour in East China is around the latitude of 35°N,and the climatological winter-mean temperature over Northeast China is -18.4°C.Fig.1 (b) shows the distribution of the standard deviation of the DJF SAT.A maximum center of temperature anomaly variability occurs over Northeast China,indicating significant variabilities in this region (Fig.1 (b));relatively smaller variabilities are found in Southeast China and Northwest China.

The spatial distribution of the extreme cold threshold of the fifth percentile at individual grid points in winter is shown in Fig.1 (c).The spatial pattern of the SAT at the fifth percentile is similar to the distribution of the winter-mean SAT (Fig.1 (a)).The area-averaged mean of the fifth percentile over Northeast China is -26.1°C,7.7°C lower than the climatological winter-mean temperature.The 0°C contour for the fifth percentile of temperature shifts south around the latitude of 30°N.Fig.1 (d) shows the geographical distribution of the climatological frequencies of PECE days in winter.Two centers of the maximum number of PECE days occur in East China: one in Southwest China and the other in Northeast China (Fig.1 (d)).Climatologically,the frequency of PECEs over Northeast China in winter is about 2.5–3.2 days per season.

Fig.2.(a–h) Percentage changes in the probability of PECEs in eight MJO phases (CP i ;units: %).Green hatching denotes regions with 99% confidence level based on the Monte Carlo test.The red boxes indicate Northeast China (40°–54°N,110°–135°E).(i) Area-averaged CP in eight MJO phases over Northeast China (units: %).

To elucidate the influence of the MJO on PECEs in Northeast China,Fig.2 shows the percentage change in the occurrence probability of PECEs at each grid point during different MJO phases with respect to the non-MJO period.The spatial distribution of the occurrence probabilities of PECEs are considerably modulated by the life cycle of the MJO (Fig.2).When MJO convection is initiated over the western Indian Ocean,the occurrence probability of PECEs is decreased in Central and Northeast China compared with the non-MJO period (Fig.2 (a,i)).When the MJO strengthens and propagates eastward to the central and eastern Indian Ocean in phases 2 and 3,the probability of the occurrence of PECEs in Northeast China increases by 16% and 128%,respectively(Fig.2 (b,c,i)).When the MJO convection moves to the Maritime Continent,the occurrence probability of PECEs in Northeast China decreases(7%) and is comparable with that in the non-MJO period (Fig.2 (d,i)).The enhanced convective anomaly then passes through the Maritime Continent and is located over the western Pacific in phases 5 and 6.The occurrence probabilities of PECEs over Northeast China are enhanced by 116% and 68% in phases 5 (Fig.2 (e,i)) and 6 (Fig.2 (f,i)),respectively.The MJO-related convective activity weakens as the MJO convection moves further east to the central Pacific in phases 7 and 8.The occurrence probabilities of PECEs become negative over almost all of East China relative to the non-MJO period (Fig.2 (g–h,i)).

To verify the robustness of the enhanced occurrence probabilities of PECEs in phases 3 and 5 of the MJO,we use a different definition of PECEs by relaxing the threshold of extreme temperature to the 15th percentile but focusing on a longer persistence time of seven days.The composites of these two definitions of the occurrence frequency of PECEs(SAT lower than the fifth percentile for at least three consecutive days and SAT lower than the 15th percentile for seven consecutive days) in the MJO phases are almost identical,regardless of the choice of temperature threshold and persistent days (not shown).Similar results are found in the composite results of the PECE occurrence probabilities in each MJO phase when we use the SAT derived from the CN05.1 dataset(not shown).The composites of SAT anomalies averaged over Northeast China also show large negative values in MJO phases 3 and 5 (not shown),indicating that these two MJO phases favor the occurrence of cold conditions regardless of whether the statistics are for individual cases (PECEs).Overall,the occurrence probability of PECEs in Northeast China is significantly increased in both phases 3 and 5 of the MJO,during which the MJO convection is vigorous over the warm pool.Whether the cooling effects over Northeast China in these two phases are dominated by similar (or different) processes linked with warm pool convective anomalies will be addressed in future work.

Fig.3.Area-averaged 10 -90-day temperature anomaly (bars;units: K) and temperature tendency anomaly (lines;units: K d-1) at 925 hPa from day -10 to day 10 relative to the occurrence of RMM in (a) phase 3 and (b) phase 5 based on the composites of days with an RMM amplitude ＞ 1.Green dashed lines in (a) and(b) indicate the cooling processes of (a) phase 3 (day -7 to day -1) and (b) phase 5 (day -5 to day 2).The (c,d) 10 -90-day-filtered and (e,f) scale-decomposed temperature budget (units: K d-1) at 925 hPa over Northeast China during (c,e) day -7 to day -1 of RMM phase 3 and (d,f) day -5 to day 2 of RMM phase 5 based on the composites of days with an RMM amplitude ＞ 1.

We determine the temporal evolution of the 10–90-day temperature anomaly and the temperature anomaly tendency near the surface(925 hPa) over Northeast China from day -10 to day 10 for phases 3(Fig.3 (a)) and 5 (Fig.3 (b)).Day 0 for each phase is defined based on the composite of the days with an RMM amplitude ≥ 1.The negative temperature anomaly for phase 3 occurs from day -4 to day 3 and the cooling effect of the negative temperature tendency occurs from day -7 to day -1 of phase 3–that is,prior to day 0 of phase 3.The SAT anomaly therefore reaches the minimum value (-0.32°C) on day 0 of phase 3.By contrast,the negative SAT anomaly in phase 5 is from day -2 to day 6 and is longer than that in phase 3.The cooling effect of the negative temperature tendency is from day -5 to day 2 in phase 5.The SAT anomaly in Northeast China therefore decreases after day 0 (-0.38°C)and reaches a minimum of -0.56°C on day 2.A stronger cooling effect occurs in phase 5 (-0.56°C) than in phase 3 (-0.32°C),although the enhanced occurrence frequencies of PECEs at day 0 for phases 3 (86%)and 5 (74%) are comparable (Fig.2).

To reveal the key physical processes associated with these phases in increasing the occurrence probability of PECEs in Northeast China,we compare the results of the temperature budget equation in Eq.(1) during the cooling periods of phases 3 (Fig.3 (c)) and 5 (Fig.3 (d)).Fig.3 (c)shows that the key processes contributing to the occurrence of PECEs can be attributed to the adiabatic cooling associated with the ascending motion of the MJO and diabatic cooling.By contrast,the most important factor in enhancing the occurrence probability of PECEs in phase 5 is cold advection (light bar).The adiabatic cooling effect (blue bar) is canceled out by diabatic heating (green bar) (Fig.3 (d)).

To further elucidate the distinct mechanisms of the cooling effects associated with RMM phases 3 and 5,we analyze the scale-decomposed temperature budgets during the cooling periods of phases 3 and 5(Fig.3 (e,f)).The opposite effects of the contributions to the occurrence of PECEs between phases 3 and 5 are mainly from the meridional temperature advection of the mean meridional temperature by the intraseasonal meridional wind anomalyfollowed by the adiabatic process associated with the intraseasonal vertical motion and mean temperature ({ω}ˉσ) and the meridional temperature advection of the intraseasonal meridional temperature by the mean meridional wind

During the cooling period associated with phase 3 (day -7 to day-1),the MJO-related convection located in the eastern Indian Ocean induces a southwest–northeast-oriented Rossby wave train at the upper level of 200 hPa.This wave train is characterized as a low-pressure anomaly over East Asia and two high-pressure anomalies are located offthe equator to the northwest of the convection and in the central North Pacific (Fig.4 (a),black contours).The quasi-barotropic low-pressure anomaly system at 850 and 200 hPa over East Asia has been identified in previous studies (Jeong et al.,2005 ;Song and Wu,2019 ;Kim et al.,2020).Jeong et al.(2005) suggested that dry and cold air from southern Siberia is advected to northern and central China when there is a lowpressure anomaly over East Asia.Northeast China is to the east of this cyclonic anomaly (Fig.4 (a),shading).The intraseasonal southerly and southeasterly warm wind anomalies decrease the occurrence probability of PECEs in Northeast China (Fig.4 (b),vectors),whereas the ascending motion and divergent wind east of the cyclonic anomaly contribute to adiabatic cooling in Northeast China (Fig.4 (c)).The advection of the intraseasonal meridional temperature by the mean northwesterly winds contributes to a small positive temperature anomaly in Northeast China(Fig.4 (d)).

Fig.4.(a) The 10–90-day band-passed geopotential height anomalies at 850 hPa (shading;units: gpm) and 200 hPa (contours;interval of 10 gpm;dashed (solid)contours for negative (positive) anomalies;units: gpm) and spatial distribution of the negative OLR anomaly (green contours;units: W m-2 ;starting from -4 with intervals of -4 W m-2).(b) The 10–90-day band-passed horizontal wind anomalies (vectors;units: m s-1) and climatological temperature (shading;units: °C) at 925 hPa.(c) The 10–90-day meridional averaged (40°–54°N) vertical motion (shading;units: 10-2 Pa s-1) and vertical and zonal winds (vectors;units: m s-1);the vertical green lines indicate the zonal range of Northeast China (110°–135° E).(d) The climatological horizontal winds (vectors;units: m s-1) and 10–90-day temperature anomaly (shading;units: K) at 925 hPa composited for day -7 to day -1 of phase 3.(e -h) As in (a -d) but for the large-scale field composites for day -5 to day 2 of phase 5;the green boxes indicate Northeast China (40°–54°N,110°–135°E).

By contrast,during the cooling period of phase 5,the MJO-related convection propagating to the western Pacific also induces a Rossby wave train in East Asia,but with a larger magnitude than that in phase 3.The Rossby wave train is characterized by a meridional dipole wave pattern with a high-pressure anomaly in South China and a low-pressure anomaly in North Asia at 200 hPa (Fig.4 (e),black contours).Anomalous high–low–high geopotential systems are found in Asia at low levels of the troposphere.The low-pressure anomaly centered in southeast Japan is between the intensified Siberian high to the east of Lake Baikal and a high-pressure anomaly in Alaska (Fig.4 (e),shading).The cyclonic anomaly occupies Northeast Asia and deepens the East Asian trough,bringing northwesterly cold advection to Northeast China and enhancing the occurrence probability of PECEs (Fig.4 (f)).The northwest–southeast temperature gradient advected by the mean northwesterly winds contributes to the negative temperature anomaly in Northeast China (Fig.4 (h)),whereas subsiding motion associated with the MJO in Northeast China results in a positive temperature anomaly (Fig.4 (g)).

4.Conclusions

This study examines the influence of the MJO on the occurrence probability of PECEs in Northeast China.PECEs can result in severe damage to agriculture,transportation,power infrastructure and human health in boreal winter.By compositing the occurrence probabilities of PECEs in different phases of the MJO,we find that the occurrence probabilities of PECEs increase significantly to 128% and 116% in phases 3 and 5,respectively,of the MJO.Our analyses show that the physical mechanisms underlying the influence of the MJO on the enhanced occurrence probabilities of PECEs in these two MJO phases are distinct.

The cooling effects associated with phases 3 and 5 of the MJO are closely related to the two different physical processes associated with the cyclonic anomaly induced by the MJO convection propagating along the equator.In phase 3,the MJO convection is in the eastern Indian Ocean and induces a cyclonic circulation in East Asia.The ascending motion east of this low-pressure center results in adiabatic cooling and enhances the occurrence probability of PECEs in Northeast China.When the MJO convection propagates into the western Pacific,the induced cyclonic anomaly also moves to Northeast Asia and deepens the East Asian trough.The strong northwesterly winds of the intensified East Asian trough contribute to the enhanced occurrence probability of PECEs in Northeast China in phase 5 of the MJO.

Previous studies have suggested that the MJO provides suitable conditions for the enhanced occurrence of cold events (Jeong et al.,2005 ;Song and Wu,2019).In addition to the impact of the MJO,the occurrence probability of PECEs in East Asia is also affected by other variabilities in extratropical regions,such as the Siberian high (Liu and Zhu,2020 ;Zhang and Lu,2021 ;Yang and Fan,2022) and the Arctic Oscillation (Song and Wu,2018,2019).Whether and how to utilize the MJO and the other intraseasonal variabilities of extratropical systems as potential sources of subseasonal predictability to predict the occurrence frequency of PECEs in Northeast China will be addressed in future work.(Eqs.(2),(3),and (5)).

Declaration of Competing Interest

No potential conflict of interest was reported by the authors.

Acknowledgments

The authors thank the two anonymous reviewers for their help in improving this paper.

Funding

This work was supported by the National Natural Science Foundation of China [grant number 42088101 ],the National Postdoctoral Program for Innovative Talent of China [grant number BX2021133] and the China Postdoctoral Science Foundation of No.70 General Fund [grant number 2021M701753].