Identification of the atmospheric water sources and pathways responsible for the East Asian summer monsoon rainfall

The East Asian summer monsoon rainfall provides water security and socio‐economic benefit for over 20% of the global population. However, the sources of this rainfall and how it is carried to the East Asian landmass are still uncertain. To address this, atmospheric water sources and pathways associated with the East Asian summer rainfall are identified and quantified in this study using atmospheric water trajectories, calculated with a novel Lagrangian framework. Evaporated water from the East Asian landmass is found to be the major contributor to East Asian rainfall, amounting to local recycling. The results further indicated that the south Indian Ocean is a major non‐local source for rainfall over southern East Asia during June to August. The role of the south Indian Ocean as a source of atmospheric water is one of the major findings of the study and would help in better understanding and predicting the East Asian summer rainfall. Evaporated waters from the Pacific Ocean (particularly the far‐west Pacific Ocean) dominate the non‐local contribution to precipitation over northern East Asia during June to September and over southern East Asian rainfall during September. The spatial structure of the East Asian rainfall is reported to be determined by the atmospheric waters that are evaporated and transported from the non‐local sources. The role of the north Indian Ocean and the South Asian landmass as a source of water for East Asian precipitation is minimal and restricted to southern East Asia. The cross‐equatorial Somali jet and equatorial trade winds associated with the western North Pacific subtropical high are important pathways for East Asian precipitation sourced over the south Indian Ocean and the Pacific Ocean respectively. In contrast, minor roles are attributed to the Bay of Bengal as a source, and midlatitude westerlies as a transport pathway, for East Asian precipitation.


INTRODUCTION
The East Asian summer monsoon (EASM) system during June-September is one of the most influential weather phenomena across the globe as it affects the livelihood of more than 20% of the world's population.It is characterised by the presence of a weather front and an associated rain belt, known as the meiyu or baiu front.The meeting of warm and moist air masses from the south and cold and dry air masses from the north is responsible for the creation of this monsoonal front (Volonté et al., 2022;Yihui & Chan, 2005).During the summer monsoon season, the frontal system and the associated rain belt migrate from the south to the north and give rise to distinct rainy stages (pre-meiyu, meiyu/baiu, and midsummer) in different parts of East Asia (Cheng et al., 2021;Chiang et al., 2017;Yihui et al., 2004).From mid-May, precipitation intensifies over south China and the South China Sea.
During the meiyu/baiu stage (occurs between mid-June and mid-July), the East Asian monsoonal front becomes strongly organised, and stretches from central eastern China to Japan (Kong & Chiang, 2020;Volonté et al., 2022).
The rain band then jumps northwards, spanning northeastern China and Korea, terminating in mid-August.The progression through the EASM stages is known to have links with seasonal northward movement of the westerlies across the Tibetan Plateau (Chiang et al., 2017;Yeh, 1958).
The warm and moist air mass branch associated with the monsoonal front is primarily responsible for the water vapour transport that feeds the East Asian precipitation.It consists of southwesterly monsoonal flow from the Indian Ocean and southeasterly flow from the western edge of the western North Pacific subtropical high (Wang & Chen, 2012).Major changes in atmospheric water content and/or pathways associated with these flows would likely have a significant impact on the East Asian rainfall and consequently affect the livelihood of a large part of East Asian population.For instance, a recent South China drought in July 2022 caused economic damage of CN¥2.73 billion, affecting 5.527 million people and 457,500 ha of land. 1  Identification of the atmospheric water sources and pathways responsible for East Asian summer rainfall has been extensively studied previously from both the Eulerian and Lagrangian viewpoints.Using Eulerian vertically integrated horizontal water vapour fluxes, Simmonds et al. (1999); Zhou and Yu (2005) showed that southeast China receives summer rainfall primarily through the southeastern Asian and Indian monsoon circulations carrying moisture from the South China Sea and the Bay 1 https://wmo.int/media/news/extreme-weather-china-highlights-climate-change-impacts-and-need-early-warnings.
of Bengal respectively, whereas the moisture transport over northeast China has been dominated by midlatitude westerlies.A study by Wang and Chen (2012) also confirms outcomes of the aforementioned studies regarding the atmospheric moisture source and transport associated with precipitation over southeast China.Xu et al. (2008) performed correlation analysis between the precipitation and the vertically integrated water vapour flux to reveal that a latitudinal band extending from the Bay of Bengal, across the Indochina Peninsula and South China Sea to the Philippines is the source of moisture for summer precipitation in the Yangtze river basin (YRB) of China.The use of Eulerian integrated water vapour transport enabled Chow et al. (2008) to show that the northern South China Sea is a region of primary focus for the early summer rainfall over China as it provides a major pathway for southwesterly atmospheric vapour transport associated with the Indian summer monsoon and the southeasterly vapour transport associated with the western North Pacific subtropical high (WNPSH).
By analysing backward air-mass trajectories calculated using the three-dimensional Lagrangian particle dispersion algorithm FLEXPART (Stohl & James, 2004) and National Centers for Environmental Prediction (NCEP) final operational global analysis data, Bin et al. (2013) identified four major moisture source regions-namely, the South China Sea, the Indian peninsula and the Bay of Bengal, the East China Sea, and the Arabian Sea-responsible for precipitation over the lower YRB during 2004YRB during -2009. .Using the same Lagrangian model but with data from the European Centre for Medium-Range Weather Forecasts (ECMWF), Drumond et al. (2011) revealed that the moisture contributions from the East and South China Seas are important for precipitation over eastern and southeastern China, whereas the Bay of Bengal and the Arabian Sea are the primary source of moisture for rainfall over southern and central China during 2000-2004. Sun and Wang (2015) employed FLEXPART driven by NCEP data during 2000-2009 over three sub-regions of East China.Their study suggested that the Indian Ocean and the western Pacific Ocean are the major moisture contributors for the summertime rainfall over southern China, whereas evaporation over land is found to be the key moisture contributor for precipitation in the middle and lower YRB and northern China.The dominant role of land evaporation as a moisture source for the YRB precipitation had also been noted by Fremme and Sodemann (2019) and Cheng and Lu (2020).The Fremme and Sodemann (2019) study had used the FLEXPART model driven by the ECMWF Reanalysis (ERA)-Interim dataset during 1980-2016, and Cheng and Lu (2020) implemented the Dynamic Recycling Model (DRM; Dominguez et al., 2006) driven by ERA v5 (ERA5) data during 1979-2018.However, a study by Wang et al. (2018) using a modified DRM (Martinez & Dominguez, 2014) forced with 32 years of NCEP data  noted the dominance of oceanic evaporation compared with terrestrial evaporation over the mid-lower reaches of the YRB.Guo et al. (2019) used the WAM2layers ( Van der Ent et al., 2013) and ERA-Interim from 1979 to 2015 to identify the moisture sources for EASM.Their results highlighted the importance of tropical oceans as the primary moisture source region for precipitation over southeastern East Asia, whereas extratropical land sources are key for the other East Asian subregions.The use of the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT; Draxler & Hess, 1998) platform with NCEP data by Shi et al. (2020) indicated that the moisture channel migrates from the west Pacific Ocean to the Indian Ocean and back to the Pacific Ocean as the weather system moves from the pre-monsoon stage in South China to monsoon in South China and YRB and finally in North China for the terminal stage.A recent study by Cheng et al. (2021) highlighted the importance of terrestrial moisture sources and Somali and South Asian moisture channels for the East Asian rain belt.This study was conducted with the ERA5 data during 1981-2018 and the semi-Lagrangian DRM model.
The difference in results from these various studies may arise for a variety of reasons, encompassing the model used, the data sources, and the horizontal, temporal, and vertical resolutions of the data, to the definition of the moisture source and sink regions.It is important to note that the Eulerian integrated vapour transport method can only indicate moisture transport pathways and not the sources of the moisture.The WAM2layers model represents the atmosphere as two layers, whereas the semi-Lagrangian model DRM eliminates the vertical dependency altogether.These simplifications increase the computing efficiency but significantly underrepresent the moisture transport at different vertical levels.The three-dimensional Lagrangian model FLEXPART and the HYSPLIT model can only trace the humidity along the air-mass trajectories and not the atmospheric water itself.
In this study, to precisely trace the atmospheric water, the horizontal and vertical water-mass transports are calculated using the atmospheric water-mass conservation equation (Dey & Döös, 2019) and incorporated in a three-dimensional Lagrangian particle tracing model, TRACMASS (Dey et al., 2023;Dey & Döös, 2020;Dey & Döös, 2021).The unique characteristics of the TRACMASS simulated atmospheric water trajectories are that each trajectory represents atmospheric mass transport of water (in kilograms per second) for trajectories originating and terminating at the surface, due to the evaporation and precipitation respectively.This is not possible to achieve in other Lagrangian models, which trace humidity along air trajectory pathways.The atmospheric water tracing version of TRACMASS is thus unique and has started to provide new insights into classical climate riddles; for example, the dominance of midlatitude westerlies in the Atlantic-to-Pacific atmospheric water transport (Dey & Döös, 2020).The objectives of the present study are to use TRACMASS for identification of the atmospheric water sources, transport pathways, and the associated variability of East Asian rainfall during June to September of 1999-2018.Section 2 outlines the working principles of the atmospheric water tracing version of TRACMASS, the equations used to calculate the diagnostic quantities, the Lagrangian model configuration implemented in this study, and the details of the ERA-Interim product.In Section 3 we explain our results.Section 4 highlights the conclusions obtained from our study and the associated discussions.

Lagrangian particle tracking model and computed diagnostics
The three-dimensional mass-conserving Lagrangian particle tracing algorithm TRACMASS v7.0 (Aldama-Campino et al., 2020;Döös, 1995) is used in the present study to identify the water sources and pathways responsible for East Asian rainfall.TRACMASS has been used to trace water in the ocean (Berglund et al., 2017;Berglund et al., 2021;Berglund et al., 2022;Döös et al., 2008) and air in the atmosphere (Kjellsson & Döös, 2012;Schumacher et al., 2022).Recently, TRACMASS was modified to trace atmospheric water (Dey & Döös, 2020).Full details of the working principles of the atmospheric water-tracing version of TRACMASS and its advantages and limitations can be found in Dey and co-workers (Dey & Döös, 2020;Dey & Döös, 2021;Dey et al., 2023).The atmospheric water tracking capability of TRACMASS was implemented in a recent series of studies to understand Atlantic-to-Pacific water transport through the atmosphere (Dey & Döös, 2020), South Asian summer monsoon precipitation (Dey & Döös, 2021), and the global water cycle (Dey et al., 2023).The results from these studies provide new insights into our understanding and highlight the complexity of the water transport in the atmosphere.A detailed assessment of the numerical schemes that are used to calculate trajectory routes in TRACMASS can be found in Döös et al. (2017).
TRACMASS trajectories are based on the Eulerian mass transport of water.The Eulerian zonal (U) and meridional (V) atmospheric water transports at time step n are calculated following Dey and Döös (2019, 2021) where i, , and k are the zonal, meridional, and vertical model indices, q (kg −1 ⋅kg −1 ) is the specific humidity, and u and v (m⋅s −1 ) are the zonal and meridional wind velocities respectively.The zonal and meridional grid distances are labelled as Δx and Δy respectively, g (= 9.81 m⋅s 2 ) is the gravitational acceleration, and Δp (Pa) is the thickness between pressure levels.Thus, the horizontal transports have units of kilograms per second of atmospheric water.The vertical mass transport of water W is then obtained from the atmospheric mass conservation of water: where Δt is the time interval between model/reanalysis stored fields.The vertical water transport at the highest level of the atmosphere is assumed to be zero and equal to evaporation E minus precipitation P at each vertical model level.
To map the atmospheric water transport pathways from the simulated three-dimensional water trajectories, a vertically integrated zonal water flux F x i, and meridional water flux F y i, water flux are calculated: Δx i, . (5) ) is the atmospheric water transport by the trajectory indexed m through the meridional-vertical grid face and T y i,,k ′ ,m is through the zonal-vertical grid-box face.The highest vertical level of the atmosphere is denoted as k ′ = kz (for the ERA-Interim data, this is at 0.1 hPa).The vertically integrated horizontal water flux is thus which has units kg⋅m −1 ⋅s −1 .The residence times  i, of the atmospheric waters at their net evaporation points were computed as which defines the time spent by the atmospheric waters between the net precipitation and net evaporation.Here, T z i,,m is the Lagrangian water transport through the surface, M is the total number of water trajectories, and t P and t E denote times when the trajectories precipitate and evaporate respectively.An age of atmospheric water  i, along trajectories from precipitation locations is additionally calculated as where t m is time when the mth trajectory crosses any grid-box face, with a mass transport of T i,,m .

Lagrangian model set-up and data used
To identify the water sources and pathways responsible for the East Asian rainfall, the Lagrangian trajectories were initialised at each grid point over the entire surface of East Asia (Figure 1) every 6 hr where P > E-calculated from Equation (3)-and then advected backward in time by the three-dimensional horizontal mass transport of water-computed using Equations ( 1) and (2)-and followed until they return to the surface where E > P. The East Asian domain is defined as the land areas between 21 • N-45 • N and 102 • E-144 • E (Figure 1).These trajectories were started at the surface every 6 hr during June to September of 1999-2018.The total number of atmospheric water trajectories started was close to 5 million and were simulated for a maximum of 1 year.Only 2.5% of the water trajectories remained in the atmosphere after 1 ear, and were not considered.Note, since backward trajectories are simulated, the terminology used throughout the article is such that starting points of the trajectories are net precipitation locations and ending points are net evaporation regions.
The three-dimensional atmospheric water transports were computed using the ERA-Interim reanalysis data (Dee et al., 2011).The horizontal resolution of the ERA-Interim data is approximately 80 km, the temporal resolution is 6 hr, and it has 60 hybrid sigma-pressure levels in the vertical direction.The vertical level ranges from the surface to 0.1 hPa (a height of about 65 km).The atmospheric reanalysis product ERA5 (Hersbach et al., 2020) could not be used in the present study, since TRACMASS uses data on model levels to conserve mass.The ERA5 data on the native model grid is vast in volume due to higher F I G U R E 1 Randomly chosen atmospheric water trajectories obtained from the backward tracing of the East Asia summer rainfall.
These trajectories are a small subset of a total of 5 million trajectories.The green dots are the starting (net precipitation) positions and blue dots indicate the ending locations (net evaporation) of the trajectories.Basins were defined to select specific groups of trajectories (based on the ending positions) and termed as the south Indian Ocean (SIO), north Indian Ocean (NIO), South Asia (SA), East Asia (EA), Pacific Ocean (PAC), Atlantic Ocean (ATL), and global landmass except SA and EA.spatial (≈31 km), vertical (137 hybrid model levels), and temporal resolution (1 hr) than ERA-Interim, and it would have exceeded capacity for the present study, which uses data spanning a 20-year (1999-2018) time period.Note that the next-generation atmospheric reanalysis product ERA5 requires ≈80 times higher data storage than ERA-Interim does and confronts the community with a big data and resource problem (Hoffmann et al., 2019).
The atmospheric water transport connection between the East Asian landmass and the other oceanic and land basins was derived by sorting different classes of trajectories based on their starting (from net precipitation) and ending (at net evaporation) locations.Since TRACMASS uses a mass-conserving algorithm, a trajectory that starts with a given mass transport of water will keep it unchanged throughout the journey.Thus, the evaporative transport from a particular basin to the East Asian landmass can be viewed as its contribution to the rainfall (since E = P).In the present study, the starting locations of the trajectories were over the East Asian landmass.The ending points of the water trajectories were classified into the four ocean basins and global landmass except East Asia and South Asia, as defined in Figure 1.The ocean basins are termed the south Indian Ocean (SIO), north Indian Ocean (NIO), Pacific Ocean, and Atlantic Ocean.

RESULTS
A few simulated atmospheric water trajectories are randomly chosen and plotted in Figure 1.Since backward trajectory calculation has been launched from the net precipitation regions of East Asia, the green dots indicate the starting positions of the trajectories where P > E and blue dots are the ending locations where E > P.
This small subset of trajectories hints that the atmospheric waters from the Indian Ocean come under the influence of the cross-equatorial Somali jet, and the evaporated waters from the Pacific Ocean flow with the equatorial trade winds, both before deposition over East Asia as precipitation.To obtain a full knowledge of the evaporative sources, we account for all the trajectories that reached the surface, presented in Figure 2. Additionally, Figure 3 indicates the spatial distribution of the rainfall over East Asia due to different atmospheric water source regions.The net precipitation over East Asia due to different sources and their contribution has been computed by summing the selected trajectories that were grouped as per their starting (net precipitation over the East Asia) and ending positions (evaporated from a basin as defined in Figure 1) and presented in Figure 4.
The results show that the East Asia landmass itself provides the majority of the essential water for the East Asia precipitation across June to September through local recycling (Figure 2) with the same spatial spread and little intra-seasonal variability (Figure 3, row 2).The absolute values of the precipitation due to the evaporated waters from the East Asia landmass (in the range 0.13-0.14Sv) also confirm this outcome (Figure 4).However, the percentage contribution increases from 37% in June to 48% in September (Figure 4) due to the reduction in contribution from other sources.A split between the southern East Asia The NIO and the South Asian landmass contributes only to the southern East Asia precipitation (Figure 3, rows 3 and 6).The NIO waters (primarily the Arabian Sea) provide 6% of the southern East Asia rainfall in June, reduced to 3% in July and vanishing during August and September (Figure 4).The South Asia contribution ranges from 8% in June to 4% in September.As previous studies have used different regions, datasets, Lagrangian tracing models, and time periods, the local recycling estimates are not readily comparable.The local recycling time-scale is calculated, using Equation (7), to be less than 4 days over the East Asia (Figure 5).
The SIO evaporative sources are concentrated between 10 • S and 30 • S and increase from June to July but start to decay from August onward (Figure 2).This characteristic matches well with the strength of the cross-equatorial water transport as presented in Figure 6.The SIO is the second major atmospheric water source region during June and July, with a contribution of 0.07 Sv (20%) in June and 0.08 Sv (22%) in July (Figure 4).It holds the third place in August and September, when the cross-equatorial transport weakens (Figure 6, top).Since East Asia has distinct rainy stages at different locations, it is important to understand the locations of precipitation due to the various source regions.The evaporated waters from the SIO provide rainfall primarily over southern East Asia in June (20%) (Figure 3, row 4).In July it extends over northern East Asia with a contribution of 6%, but its share over southern East Asia reduced to 16%.The SIO further reduces its share in August, with 14% over southern East Asia and 3% over northern East Asia.In September, the evaporated waters from the SIO provide rainfall only over southern East Asia (11%).Note that the spatial structure of the total net precipitation over southern East Asia during June to August (Figure 3, row 1) is majorly determined by the evaporative waters that are evaporated and travelled from the SIO (Figure 3, row 4).The residence time of the SIO waters mapped at their net evaporative points, calculated using Equation ( 7), indicates that the atmospheric waters take different times to reach East Asia depending on the location of net evaporation (Figure 5).The closer the net evaporation location is to East Asia, the less time needed to reach East Asia.However, the drawback of residence time mapped at evaporative points is that no subsequent time information is known, as the atmospheric waters leave the source location and start to move with the flow.For instance, the residence time for an evaporative location over the SIO during June is 20 days (i.e., time to reach East Asia), but we have no knowledge about the time at each downstream location as it moves away from the source region with the prevailing flow and closer to the target area.The along-trajectory age of the water can provide this information, and is thus computed using Equation ( 8) and presented in Figure 7.The age of the water in Figure 7 precisely demonstrates the time left to reach East Asia from each location through which a trajectory passes.The residence time map and age of the water will be similar over the net evaporation locations.The evaporated waters from the SIO take less time to reach East Asia in July compared with in June (Figure 5).The age of the water also confirms this (Figure 7, top row).The age also indicates that the atmospheric waters from the SIO takes around 7-13 days to reach East Asia once they are in the NIO.During August and September the residence time and age of the SIO evaporated waters increases.The age of the waters also increases over the Arabian Sea during this time and indicates the sluggishness of cross-equatorial flow.The Pacific Ocean evaporative sources are located primarily along the pathways of the equatorial trade winds (associated with the WNPSH), with maximum magnitude areas located in the west Pacific Ocean including the northern part of South China Sea, the East China Sea, the Yellow Sea, and the Sea of Japan (Figure 2).The contribution of the Pacific Ocean as a source to the total East Asia precipitation strengthens from June (0.05Sv, 14%) to July (0.07 Sv, 19%) (Figure 4).In June, the Pacific Ocean provides 10% rainfall over southern East Asia and 4% in northern East Asia, whereas during July the contribution to southern East Asia decreases to 6% and increases to 13% over northern East Asia (Figure 3, row 5).The atmospheric water transport pathways suggest that, during July, the water transport associated with the trade winds enhances over the Pacific Ocean compared with the month of June, which increases its contribution to northern East Asia rainfall (Figure 6,bottom).It is interesting to note that the evaporated waters from the Pacific Ocean not only come under the influence of the equatorial trade winds, associated with the WNPSH, but some also cross over to the Indian Ocean and merge with the cross-equatorial flow.In August, the Pacific increases (left panel) and their percentage contribution (right panel).This has been computed from the trajectories that were sorted as per their starting (net precipitation) and ending (net evaporation) positions.Since TRACMASS uses a mass-conserving algorithm, a trajectory that starts with a given mass transport of water will keep it unchanged throughout the journey.Thus, the evaporative transport from a particular basin to the EA landmass can be viewed as its contribution to the rainfall (since E = P).NIO: north Indian Ocean; SIO: south Indian Ocean; PAC: Pacific Ocean; SA: South Asia.

F I G U R E 5
Residence time (days) of the atmospheric waters mapped at their net evaporative locations during June to September.To focus on the primary evaporative regions, only those points were plotted where the monthly mean value of net evaporation in Figure 2 exceeds 0.03 mm⋅day −1 .its share over southern East Asia to 12% and reduces its contribution to 8% in northern East Asia.This characteristic continues until September, with a contribution of 15% in southern East Asia and 7% over northern East Asia.The Pacific is the second major rainfall provider to East Asia during August and September (Figure 4).As the cross-equatorial flow weakens during August-September, the tendency of the Pacific waters to merge with the cross-equatorial flow also reduces (Figure 6, bottom).The results further outline the importance of the Pacific Ocean sources (Figure 3, row 5) in setting the spatial structure of the total net precipitation over northern East Asia during F I G U R E 6 Vertically integrated atmospheric water flux (shaded; kg⋅m −1 ⋅s −1 ) during June to September.This has been calculated from Lagrangian trajectories using Equation ( 6).The top row is for the waters evaporated from the south Indian Ocean (SIO) and precipitated over East Asia, and the bottom row is correspondingly for Pacific Ocean (PAC) waters.The flux directions are given by green streamlines.

F I G U R E 7
The age (time left to reach East Asia in days) of the atmospheric waters.To focus on the age of the atmospheric waters associated with the major transport pathways, this has been calculated at points where the monthly mean value of atmospheric water transport in Figure 6 exceeds 0.6 kg⋅m −1 ⋅s −1 .The top row is for the waters evaporated from the south Indian Ocean (SIO) and precipitated over East Asia, and the bottom row is same but for the Pacific Ocean (PAC) waters.
the June to September months (Figure 3, row 1).It would be misleading to conclude (from Figure 2) that only evaporation from the "far-west" Pacific Ocean is important for East Asia precipitation, rather than that across the wider "rest-of-Pacific" basin.Despite high evaporation rates in the far-west Pacific, these occupy a relatively small horizontal area.Evaporation rates across the wider Pacific responsible for East Asia rainfall are small in magnitude but extend over a much larger area and are cumulatively important.To quantify the relative contributions of far-west Pacific and rest-of-Pacific sources, we area-integrate evaporation west and east of the 150 • E meridian (blue dotted line in Figure 2).The far-west Pacific (west of 150 • E) contributions to the East Asia rainfall are 63% (496 mm⋅day −1 ), 52% (506 mm⋅day −1 ), 56% (560 mm⋅day −1 ), and 63% (598 mm/day) of the total Pacific contribution during June, July, August, and September respectively.In comparison, the rest-of-Pacific contributions are respectively 37% (293 mm⋅day −1 ), 48% (460 mm⋅day −1 ), 44% (434 mm⋅day −1 ), and 37% (350 mm⋅day −1 ).The residence-time map at evaporative regions over the Pacific in Figure 5 and age of the water along the transport pathways (Figure 7, bottom) reveal that, as the water moves closer to the East Asia, the time reduces from more than 28 days to less than 7 days.Note that the 7-13 days time-scale to reach East Asia once the atmospheric waters are in the NIO during June and July is also true for the Pacific evaporated waters.

CONCLUSIONS AND DISCUSSION
The East Asian monsoon region is one of the hotspots of climate change around the globe.Extreme rainfall events and related inundation occur regularly in East Asia (Mori et al., 2021) and are projected to intensify and become more frequent in a warmer future climate (Oppenheimer et al., 2019).Additionally, rising air temperature increases the threat of more frequent and longer heatwaves and droughts across East Asia (Pörtner et al., 2022).Thus, it is important to trace the origin of atmospheric water transported to East Asia and to understand the reasons for its variability.This knowledge of atmospheric water sources will be useful to formulate strategies for water availability and early-warning disaster management across East Asia.
In this study, East Asian precipitation is traced backward in time with Lagrangian atmospheric water trajectories.These water trajectories are calculated using a novel Lagrangian framework and ERA-Interim data spanning 1999-2018.The Lagrangian framework used in this study is able to trace the atmospheric mass transport of water in kilograms per second from the surface evaporative regions to surface precipitating locations (forward tracking) or vice versa (backward tracking).This is not possible to achieve in other three-dimensional Lagrangian moisture-tracing models, which trace specific humidity along air trajectories and not explicitly the atmospheric water from surface to surface.
The results show that local recycling is the major contributor to East Asian rainfall, with a contribution of 37% in June, increasing to 48% in September.The dominance of local cycling has also been noted by Sun and Wang (2015), Cheng and Lu (2020), Cheng et al. (2021).Additionally, local recycling is found to be more dominant over southern East Asia in comparison with northern East Asia.During June and July, the SIO and Pacific Ocean (specifically the far-west Pacific Ocean) become the second major contributor of rainfall over southern and northern East Asia respectively.From June to July, these two sources increase their contribution to rainfall over East Asia, associated with enhancement of cross-equatorial flow over the Indian Ocean and equatorial trade winds (associated with the WNPSH) over the Pacific Ocean.During this time, waters evaporated from the SIO are carried with a cross-equatorial flow, taking between 16 and 28 days (depending on the net evaporative location) to reach East Asia.It has been previously reported that the atmospheric water residence time associated with the inter-basin and inter-hemispheric flow can be higher than the global average residence time of 9 days (Dey et al., 2023;Dey & Döös, 2020;Dey & Döös, 2021).We further note that water evaporated from the SIO traverses the Bay of Bengal, whereas water evaporated from the Bay of Bengal itself contributes little to East Asian rainfall.This is contrary to the previous findings (e.g., Bin et al., 2013;Drumond et al., 2011;Simmonds et al., 1999;Wang & Chen, 2012;Xu et al., 2008;Zhou & Yu, 2005).The results also show the dominant role of non-local sources in setting the spatial structure of the East Asian rainfall during the June to September months.
The age of the atmospheric water evaporated from the SIO shows that it takes 7-13 days to reach East Asia once they are in the core of the cross-equatorial flow over the NIO.Waters evaporated from the Pacific Ocean take anything from less than 7 days to more than 30 days (depending on the net evaporative location) to reach East Asia when carried by the trade winds.Some of these waters are carried by the cross-equatorial flow via the Indian Ocean pathway.During August, the SIO is still the second major source of water for southern East Asian rainfall, but over wider East Asia the Pacific Ocean dominates as a non-local source due to its higher contribution to northern East Asian precipitation.The actual amount of rainfall attributed to SIO evaporation is reduced in August due to the weakening of the cross-equatorial flow, which can be seen in the residence time and age maps (higher values).In September, the Pacific Ocean becomes the dominant non-local source for both the southern and northern East Asian precipitation.These characteristics of evaporated waters from the SIO and Pacific Ocean are clearly illustrated in Figure 8, conforming with northward and southward migration of the monsoonal front between southern and northern East Asia.Waters evaporated from the NIO and the South Asian landmass contribute very little to the East Asia monsoon precipitation.Reanalysis data show that increased moisture over the NIO, related to excessive warming-driven evaporation, over the last few decades mainly falls as precipitation over the ocean (especially over the Bay of Bengal), contributing little to moisture transport across southeast Asia (Skliris et al., 2022).
A key finding of this study is the role of the SIO as a source of water for East Asian rainfall, which has only been noted in a limited number of previous studies (e.g., He et al., 2007;Tao, 1987;Wang et al., 2018).This outcome will have consequences in terms of understanding and predicting East Asian summer rainfall.For instance, as the SIO is also a major contributor to South Asian monsoon rainfall (Dey & Döös, 2021), it may happen in some years that the South Asian landmass attracts more water from the SIO, which will then create a drought-type situation over East Asia (particularly over southern East Asia).In this regard, it is important to note that during the July 2022 South China drought the rainfall over the Indian subcontinent was above average (https://internal.imd.gov.in/press_release/20221001_pr_1849.pdf)and lower than average over southern East Asia (https://public.wmo.int/en/media/news/extreme-weather-china-highlights-climate-change-impacts-and -need-early-warnings).However, we acknowledge that a dedicated study is required to fully understand the contrasting behaviour of the July 2022 rainfall over the Indian subcontinent and East Asia, which lies outside the scope of the present study.
The results show that evaporated waters from the SIO and Pacific Ocean travel with cross-equatorial flow and easterly trade winds respectively before precipitating over the East Asian landmass.The Pacific Ocean (particularly the northern part of the South China Sea, the East China Sea, the Yellow Sea, and the Sea of Japan) as a moisture source, and trade winds as the transport mechanism for East Asia summer rainfall, are well documented in previous literature (e.g., Bin et al., 2013;Simmonds et al., 1999;Xu et al., 2008;Zhou & Yu, 2005) (Chen et al., 2020;Geen et al., 2023;Jiang et al., 2020;Xin et al., 2020).The results obtained from the present study, using observationally constrained reanalysis data, could therefore be used to provide insight for assessing behaviour in climate models.For example, if any climate models are simulating a dry/wet bias structure over southern East Asia during July similar to the precipitation structure observed due to the SIO sources (Figure 3, row 4) then it would hint towards the model deficiency in simulating the cross-equatorial flow and/or evaporation over the SIO.
We further establish that the midlatitude westerlies are not responsible for atmospheric water transport to East Asia during summer monsoon months, contrary to some previous results (e.g., Simmonds et al., 1999;Zhou & Yu, 2005).We instead highlight the importance of the Somali jet and South Asian routes in carrying atmospheric water towards East Asia, as with Cheng et al. (2021).Motivated by these findings, atmospheric water trajectories may best explain the interannual variability of these water source regions.This approach will reveal the relative influences, on variable atmospheric water transport, of thermodynamic and dynamic processes.An example future study of interannual variability in the EASM could be to identify and quantify its moisture sources and associated transport pathways for years when the East Asian monsoonal front is located at anomalous latitudes.

F
I G U R E 2 Net evaporation E (mm⋅day −1 ) responsible for East Asia precipitation P during June to September.This has been calculated by considering all the atmospheric water trajectories that reached the surface after backward tracing of precipitation from East Asia.The orange dashed box indicates the East Asia region from which the backward trajectories are initialised.Note, two different colour maps and scales are used to clearly show the contribution of local and remote evaporative sources responsible for East Asia rainfall.The 150 • E meridian indicated by the blue dotted line has been used for comparing the far-west Pacific and rest-of-Pacific contributions to the East Asia rainfall.(landgrid points between 21 • N-34.5 • N and 102 • E-124 • E) and the northern East Asia (land grid points between 34.5 • N-45 • N and 102 • E-144 • E) indicates that local recycling has a dominant role in the southern East Asia (20% in June to 29% in September) compared with its northern East Asia counterpart (17% in June to 19% in September).
Net precipitation P (mm⋅day −1 ) due to different atmospheric water source regions.The first row indicates the total net precipitation (including all the sources); subsequent rows indicate the contribution from East Asia (EA), the north Indian Ocean (NIO), the south Indian Ocean (SIO), the Pacific (PAC), and South Asia (SA).The domain of these panels is shown by the orange dashed box in Figure 2. The red dashed box indicates the southern EA region (only land grid points between 21 • N-34.5 • N and 102 • E-124 • E), and the remaining areas are termed as the northern EA (land grid points between 34.5 • N-45 • N and 102 • E-144 • E) and were used for the quantification.Two different scales have been used to distinctly visualise the total net precipitation and net precipitation due to the five evaporative sources.Note that the spatial structure of the total precipitation is determined by the non-local sources.E: evaporation.
Integrated precipitation over East Asia (EA) due to the five evaporative sources: units are in Sverdrups (1 Sv ≡ 10 9 kg⋅s −1 ) . The easterly trade winds over the Pacific Ocean are associated with the Schematic summary of the relative contribution of the south Indian Ocean (SIO) and the Pacific Ocean (PAC) evaporated waters to the East Asian summer rainfall.The red dashed box indicates the southern East Asia region (land grid points between 21 • N-34.5 • N and 102 • E-124 • E), and the remaining areas are termed as northern East Asia (land grid points between 34.5 • N-45 • N and 102 • E-144 • E).The orange and blue arrows indicate percentage contribution to the East Asian rainfall from the SIO and the PAC respectively.The arrows within the red dashed box show the contribution of specified evaporative sources to the southern East Asia summer rainfall, whereas arrows outside the box express the evaporative share to northern East Asia summer precipitation.The small orange arrows outlines the contribution of the SIO to the northern East Asia rainfall.Note, arrow shapes are subjectively drawn to mimic the "mean" water transport pathways.The thickness of the arrows corresponds to the amount of share (the thicker the arrows the higher the contribution).The relative contributions of the SIO and the PAC show the movement of atmospheric waters from southern to northern East Asia and back again, from these two basins, which is associated with seasonal migration of the monsoonal front.The shaded colour indicates net rainfall, including all sources, as Figure 3, row 1. WNPSH, and it would be interesting to study changes in the Pacific moisture source, subject to variable location of the WNPSH.A study by Volonté et al. (2022), using Lagrangian air trajectories, highlighted the importance of WNPSH location in East Asia precipitation.The state-of-the-art climate models under the Coupled Model Intercomparison Project are known to have significant biases in reproducing the climatological features of the EASM precipitation