Yi Fan , Ke Fan , Zhiqing Xu

a Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters/Key Laboratory of Meteorological Disaster, Ministry of Education/School of Atmospheric Sciences, Nanjing University of Information Science & Technology, Nanjing, China

b School of Atmospheric Sciences, Sun Yat-sen University, Guangzhou, China

c Southern Laboratory of Ocean Science and Engineering (Zhuhai), Zhuhai, China

d Nansen-Zhu International Research Centre, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China

Keywords:Intraseasonal oscillation intensity Representative concentration pathway 1.5 and 2.0°C global warming Western North Pacific

A B S T R A C T The 30—60-day intraseasonal oscillation (ISO) and 10—20-day ISO are two dominant oscillation modes over the western North Pacific during boreal summer. With daily data derived from eight CMIP5 models, changes of the ISO intensities are projected under the 1.5 and 2.0 °C global warming levels under the Representative Concentration Pathway (RCP) 4.5 and RCP8.5 scenarios. Most of the models agree that the ISO intensities increase along a belt region from the south Indochina Peninsula (ICP) to the east to the Philippines. The variation pattern shows little difference between different warming levels or scenarios. Results indicate that the spatial distribution of ISO anomalies is related with the variation of background fields. Enriched lower-level humidity and moist static energy favor the intensity increases of ISOs, which are projected to be larger over the whole western North Pacific,with the most conspicuous changes located over the east to the Philippines for humidity but over the south of the ICP for moist static energy. In contrast, the ISOs over the west to Indonesia and northeast to the Philippines decrease, which is consistent with the local descending motions.

1. Introduction

The sub-seasonal climate variations over the western North Pacific(WNP) in boreal summer are mainly influenced by two types of intraseasonal oscillation (ISO): the 30—60-day ISO and the 10—20-day ISO( Annamalai and Slingo, 2001 ; Li and Wang, 2005 ; Mao and Chan, 2005 ).Generally, the 30—60-day ISO over the WNP represents the activity of the Madden—Julian Oscillation ( Madden and Julian, 1971 , 1972 ), while the 10—20-day ISO is related to the activities of tropical cyclones and some other high-frequency oscillations ( Schrage and Vincent, 1996 ; Teng and Wang, 2003 ). They both exert great influences on monsoonal progress,precipitation, temperature, and extreme weather events over the most densely populated Asian monsoon areas ( Sobel and Maloney, 2000 ;Zhu et al., 2003 ; Li and Wang, 2005 ). Along with global warming,changes in sea surface temperatures (SSTs) will modify the humidity conditions, and vertical motion over the WNP, and therefore modulate the behaviors of ISOs ( Teng and Wang, 2003 ; Deng and Li, 2016 ;Li and Mao, 2016 ; Xu et al., 2017 ). The lower troposphere will be able to contain more moisture under the conditions of SST warming over the WNP. Under this circumstance, the moist static energy (MSE), which favors stronger deep convection ( Li and Mao, 2016 ) over the WNP, will increase under global warming. Therefore, the ISO intensity is sensitive to SST warming and strengthens with the increased moisture and MSE in the planetary boundary layer ( Maloney and Xie, 2013 ; Li and Mao, 2016 ).

Previous studies have mainly focused on how the 30—60-day ISO(i.e., the Madden—Julian Oscillation) evolves under global warming,while limited attention has been paid to the 10—20-day ISO. It has been revealed that the 10—20-day and 30—60-day ISOs show differentresponses to background field changes induced by Pacific SST anomalies ( Wu and Song, 2018 ), and how these two kinds of ISOs respond to the background field changes induced by global warming is hence of great interest. Therefore, the responses of the intensities of both ISOs to global warming is to be emphasized in the current study. Moreover,previous investigations regarding ISO changes under emission scenarios have mainly focused on the whole 21st century, but how the ISOs respond to 1.5 and 2.0 °C global warming has rarely been discussed.According to model projections, more damaging extreme events might occur under 1.5 °C global warming, and would be even worse under 2.0 °C global warming ( Jiang et al., 2016 ; Li et al., 2018 ; Ge et al.,2019 ; Xu and Fan, 2019 ). Therefore, the Paris Agreement reached in 2015 and the IPCC Special Report on 1.5 °C of warming released in 2018 both pointed out the great importance of controlling the global warming level to under 2.0 °C, and even better to under 1.5 °C. Notably, the response of climate systems to global warming is never a linear process.Research on the whole warming period cannot represent the changes at each warming level and may ignore specific important information( Rial et al., 2004 ; Zhou et al., 2018 ). Therefore, it is crucial to investigate how the climate over the WNP, e.g. the 30—60-day and 10—20-day ISOs with essential roles, changes at the two critical warming levels of 1.5 and 2.0 °C.

The key questions that the present study seeks to answer are: (1) How will the intensity of the 30—60-day and 10—20-day ISOs change under 1.5 and 2.0 °C global warming? (2) What are the possible mechanisms responsible for the ISO intensity changes? Following this introduction, the data and methods used in the study are described in Section 2 . The projected changes in the intensity of the 30—60-day and 10—20-day ISOs under 1.5 and 2.0 °C global warming are presented in Section 3.1 . The possible contributions of changing backgrounds to the anomalous 30—60-day ISO and the 10—20-day ISO intensities are analyzed in Section 3.2 .Finally, a summary and discussion are provided in Section 4.

2. Data and methods

The reanalysis daily datasets are from the NCEP—DOE Reanalysis 2 dataset ( https://www.esrl.noaa.gov/psd/data/gridded/data.ncep.reanalysis2.pressure.html ). Considering that the models with relatively poor skill in reproducing the historical result may also fail to capture the projected changes, 8 out of the 20 CMIP5 models ( https://esgfnode.llnl.gov/search/cmip5/ ) with reasonable ability, whose pattern correlations between simulations and reanalysis is statistically significant and larger than 0.68 in simulating the historical summer ISO intensities over the area (10°S—20°N, 80°—160°E), are chosen for analysis (the definition of ISO intensity is introduced below). The selected models are:BNU-ESM, CNRM-CM5, GFDL-CM3, GFDL-ESM2G, GFDL-ESM2M, IPSLCM5A-LR, IPSL-CM5A-MR, and MPI-ESM-MR ( Table 1 ). By comparing the model simulations with the reanalysis (figure not shown), it is found that the selected models can capture the patterns of the climatological 30—60-day and 10—20-day ISO intensities, SST, and lower-tropospheric moisture during 1979—2005 well.

Table 1 Details of the CMIP5 models used in this study.

According to Vautard et al. (2014) , historical warming of the global mean surface temperature is calculated by the base period (1971—2000)minus the pre-industrial period (1881—1910), which equals 0.46 °C, and the 1.5 °C (2.0 °C) warming world is identified by the projected 30-year running mean minus the base period (1971—2000) plus the historical warming equal to 1.5 °C (2.0 °C). All the projected changes are calculated compared to the base period of 1971—2000.

Generally, outgoing longwave radiation (OLR) data are needed to calculate the ISO intensity. However, OLR is not provided in the CMIP5 model projections ( https://esgf-node.llnl.gov/search/cmip5/ )and therefore the 30—60-day ISO intensity (ISO(30—60))is calculated as

wher e UA1 (VA) stands for a 29-day running mean minus a 61-dayrunningmeanofzonal(meridional)windanomalies,andUA(VA) represents a 9-day running mean minus a 21-day running mean of zonal (meridional) wind anomalies. This method has been proven to be reliable in reflecting the interannual variations and climatological distributions of the ISO intensities over the WNP ( Wu and Song, 2018 ).The average kinetic energy during June—July—August represents the intensity of the ISO in summer.

The MSE is calculated as where

T

is temperature,

C

is the specific heat of air,

L

is the latent heat of evaporation,

q

is the mixing ratio of water vapor,

g

is gravitational acceleration, and

z

is height.

To provide an objective analysis, variation percentage is employed to investigate how much the ISO intensity and background fields change compared to the historical value. The variation percentage is calculated as

where

V

is the variation percentage,

V

is the value of under global warming and

V

h is the value in the historical period.

To examine the credibility of projection results from the ensemble, the signal-to-noise ratio (SNR) is employed according to Zhu et al. (2020) :

where

x

indicates the result derived from each single model,

x

represents the multi-model ensemble median, and

n

denotes the ensemble size. An SNR larger than one implies that the signal is greater than the noise, representing the considerable consistency and reliability of the projection.

3. Projected results

The variation percentages in the ISO intensities under four conditions —(a) 1.5 °C global warming under the Representative Concentration Pathway (RCP) 4.5 scenario; (b) 1.5 °C global warming under the RCP8.5 scenario; (c) 2.0 °C global warming under the RCP4.5 scenario;and (d) 2.0 °C global warming under the RCP8.5 scenario —are illustrated in Fig. 1 .

Fig. 1. Percentage changes in the multi-model ensemble median of summer (a—d) 30—60-day ISO and (e—h) 10—20-day ISO intensity (units:%) under (a, e) 1.5 °C global warming and RCP4.5, (b, f) 1.5 °C global warming and RCP8.5, (c, g) 2.0 °C global warming and RCP4.5, and (d, h) 2.0 °C global warming and RCP8.5. Dotted areas indicate more than 75% of the models agree on the sign and crossed areas indicate the SNR is greater than 1.0.

3.1. Projected changes in the 30-60-day ISO intensity

Under 1.5 °C global warming with the RCP4.5 scenario ( Fig. 1 (a)),the changes in the 30—60-day ISO intensity display an increase along a belt covering the south of the Indochina Peninsula (ICP), the South China Sea (SCS), and the south of the Philippine Sea with percentage changes of around 4%—14%, but a decrease located over the tropical regions west and east to Indonesia with percentage changes of around 6%—16%. The corresponding changes in percentage for the RCP8.5 scenario under 1.5 °C global warming, exhibiting two increasing centers over the south of the ICP and east to the Philippines, range from 4% to 20% and a decreasing center over the west of Indonesia ranging from 5% to 16% ( Fig. 1 (b)).

As for 2.0 °C global warming, the 30—60-day ISO intensity variation percentages ( Fig. 1 (c, d)) have similar patterns to those under 1.5 °C global warming ( Fig. 1 (a, b)). The considerable changes under 2.0 °C global warming are located over the southern part of the ICP and east to the Philippines with amplitude around 4%—18% for RCP4.5 ( Fig. 1 (c))and 4%—20% for RCP8.5 ( Fig. 1 (d)). In contrast, the most conspicuous decreasing center is mainly located over the west to Indonesia. However, the SNRs are projected to be less than 1 in Fig. 1 (a, c), indicating large uncertainties among the model simulations regarding the increase of 30—60-day ISO intensity under 1.5 °C global warming with RCP4.5 and RCP8.5, and 2.0 °C global warming with RCP4.5. In contrast, under 2.0 °C global warming with the RCP8.5 scenario, credible increases in the 30—60-day ISO intensity are detected over the southern part of the ICP and the SCS.

Fig. 2. Percentage changes in the multi-model ensemble median of (a)—(d) the specific humidity at the 1000-hPa level, (e)—(h) the MSE between 1000 and 850 hPa,and (i)—(l) the vertical speed at 500 hPa (positive indicates descending motion and negative indicates ascending motion) in summer under (a, e, i) 1.5 °C global warming and RCP4.5, (b, f, j) 1.5 °C global warming and RCP8.5, (c, g, k) 2.0 °C global warming and RCP4.5, and (d, h, l) 2.0 °C global warming and RCP8.5. Dotted areas indicate more than 75% of the models agree on the sign and crossed areas indicate the SNR is greater than 1.0. For specific humidity, all the plotted areas,more than 75% of the models agree on the sign.

Fig. 3. Percentage changes in the multi-model ensemble median of the pressure—longitude airflow in summer (average between 0°—5°N, stream lines, zonal wind;contour plot, vertical speed) under (a) 1.5 °C global warming and RCP4.5, (b) 1.5 °C global warming and RCP8.5, (c) 2.0 °C global warming and RCP4.5, and (d)2.0 °C global warming and RCP8.5. Values inside the black curve indicates more than 75% of the models agree on the sign, and the colored area indicates the SNR is greater than 1.0.

3.2. Projected changes in the 10-20-day ISO intensity

The changes in the 10—20-day ISO intensity for all the global warming and RCP scenarios are displayed in Fig. 1 (e—h). Under 1.5 °C global warming with the RCP4.5 scenario ( Fig. 1 (e)), most of the models agree that the 10—20-day ISO intensity strengthens over the area (5°—20°N,90°—150°E), with the largest center of increase (18%) located around(12°N, 105°E) and the largest center of decrease (18%) over the west to Indonesia. With regard to 1.5 °C global warming under the RCP8.5 scenario, increasing intensities of the 10—20-day ISO are projected to be more conspicuous over the WNP, with a peak magnitude of increase of 20% over the east to the Philippines and a decrease over the west to Indonesia with a magnitude of around 18% ( Fig. 1 (f)). The distribution of changes in the 10—20-day ISO intensity under 2.0 °C global warming shows little difference to that under 1.5 °C global warming ( Fig. 1 (g)and (h)). According to the SNRs, credible changes under 2.0 °C global warming are located around (15°N, 105°E) with both scenarios.

Compared to the changing magnitudes of the 30—60-day ISO, the region with a reliable increase (SNR

＞

1.0) of 10—20-day ISO is generally larger. Credible increases are mainly detected over the southwest part of the ICP for 2.0 °C global warming under RCP4.5 and RCP8.5, and 1.5 °C global warming under RCP4.5, but over the west to the Philippines for 1.5 °C global warming under RCP8.5. This might be attributable to the more sensitive responses of the 10—20-day ISO to variations of certain background conditions ( Wu and Song, 2018 ).

4. Mechanistic analysis

Similar distributions are observed in the variation percentages for the 10—20-day and 30—60-day ISOs. ISO intensities increase over the south of the ICP and the region east to the Philippines, but decrease over the west to Indonesia over (5°S—5°N, 90°—110°E). In this section,possible impacts of the changing background fields on the ISO intensity variations are analyzed.

According to the Clausius—Clapeyron relation, the lower troposphere will be able to contain more moisture along with the warming SSTs over the WNP. Held and Soden (2006) pointed out that, with each 1 °C of SST warming, the atmospheric saturation specific humidity would increase by around 7%. It has been reported that rising moisture in the troposphere might contribute to the larger intensity of ISOs( Subramanian et al., 2014 ). Here, in our study, for both 1.5 and 2.0 °C global warming, the lower tropospheric specific humidity shows an increase over the whole WNP ( Fig. 2 (a—d)), which is reflected by most of the models. The variation percentages of specific humidity are characterized by magnitudes of around 3%—7% under 1.5 °C global warming,and 5%—10% under 2.0 °C global warming, which do not show obvious differences between scenarios. The largest increases are located over the south of the ICP, the SCS, and the east of the Philippines, being in close association with the spatial distribution of the most evident ISO enhancement. Moreover, the lowest increases in amplitude of specific humidity are detected over the areas around Indonesia, which could be related to the decreasing ISO intensities over there. Under this circumstance, the MSE over the WNP will also increase under global warming,and larger MSE tends to provide a favorable condition for stronger convections and ISO intensities ( Maloney and Xie, 2013 ; Li and Mao, 2016 ).As shown in Fig. 1 , there is a common enhancement center of the 10—20-day and 30—60-day ISOs located over the south of the ICP, which could be linked to the increase in lower-level MSE over surrounding areas ( Fig. 2 (e—h)).

As shown in Fig. 2 (i—l), strong descending anomalies at 500 hPa are detected over the equatorial areas to the west of 135°E and the area to the northeast of the Philippines. The anomalous descent ( Fig. 2 ) shows consistency for all the vertical levels from 1000 to 300 hPa ( Fig. 3 ).Similarly, for most of the warming scenarios, the areas characterized by decreasing 30—60-day and 10—20-day ISO intensities are mainly located in the equatorial areas, especially to the west of 110°E, indicating modulation of the ISO intensity by background vertical motion changes.

5. Conclusions and discussion

The projected changes in the 30—60-day and 10—20-day ISO intensities over the WNP under 1.5 and 2.0 °C global warming have been investigated, as well as the possible responsibility of the background fields for the ISO variations.

· According to the multi-model ensemble median of variation percentages, under both RCP4.5 and RCP8.5, the 30—60-day ISO intensity shows an increase over a belt area from the ICP to the east of the Philippines with global warming of 1.5 and 2.0 °C, which is reflected by most of the models. Distributions of the SNRs indicate that changes under 2.0 °C global warming under the RCP8.5 scenario are more credible over the WNP than the other three scenarios.

· The 10—20-day ISO intensity displays an increase over the area(5°—20°N, 90°E—150°E) under global warming of 1.5 and 2.0 °C.For the RCP4.5 scenario, the largest changes are projected over the west to the ICP. For the RCP8.5 scenario, credible changes are located to the east of the Philippines under 1.5 °C global warming, but over the central western ICP under 2.0 °C global warming.

· The enhanced ISO intensities could be attributable to the combined effects of background fields such as anomalous moisture,MSE and vertical motion. Large centers of increase in specific humidity are located over (10—20°N, 90—150°E), which favors stronger convection. Conspicuous changes in MSE over the ICP could be responsible for consistent enhancement of ISO intensities among the models. Anomalous descending motion over the area (5°S—10°N, 90—130°E) contributes to the decrease in ISO intensities.

· Considering the important roles ISOs play in modulating monsoonal activity and the formation of tropical cyclones, this study inspires how the WNP monsoon and tropical cyclone activities change under global warming. Previous studies have indicated that there are more tropical cyclones formed over the WNP during the active phases of the 10—20-day ISO ( Nakazawa,1986 ), and therefore it is a reasonable hypothesis that the changes in ISO intensity could lead to variation in tropical cyclones over the WNP, which could be an interesting topic for future exploration. Since the subseasonal monsoon activities over the WNP bear a close relationship with the local ISOs,the WNP monsoonal processes might show more frequent abrupt changes with the enhanced intensity of the 30—60-day ISO under both 1.5 and 2.0 °C global warming ( Li and Wang, 2005 ). Therefore, investigations are to be carried out to find out whether the changes in the intensity of ISO further feed back to the local summer climate.

· This study mainly concentrates on the percentage changes of ISO intensities over the WNP, and the actual magnitudes of change show similar variations as the percentages but different distributions (figure not shown). Moreover, the variation pattern in ISO intensity is related to the combined influences of background fields such as anomalous moisture content, MSE, and vertical motion, but the relationships are not simply linear. Complex interactions exist between climate systems of different scales. The changes in ENSO and vertical wind shear distributions also exert great influence on the ISO intensity and propagation ( Cai et al., 2014 ; Wu and Song, 2018 ), which is to be investigated in the future. Besides, even the current stateof-the-art models have insufficiencies in the projection of ISO intensities. As indicated by the SNR, variability among models is larger than the multi-model median of changes over most areas, which reveals the large influence from model uncertainty on the reliability of results. Therefore, further analysis is still needed towards the uncertainty in projection patched by model developments.

Funding

This research was jointly supported by the National Key R&D Program of China [grant number 2017YFA0603802 ], the National Natural Science Foundation of China [grant numbers 41730964 and 41991283 ],and the Innovation Group Project of Southern Marine Science and Engineering Guangdong Laboratory (Zhuhai).

Acknowledgments

We thank the support provided by the Startup Foundation for Introducing Talent of NUIST.