Stable Isotopic Variability of Daily Precipitation from Yungui Plateau of Southwest China: Insights from Moisture Source and Rainout Effect


 Long-term continuous monitoring of precipitation isotopes has great potential to advance our understanding of mechanisms that determine stable isotope variability in hydrological processes of monsoon regions. This study presents a 4-year daily data set of precipitation isotopes from Yungui Plateau of southwest China, influenced by southwest monsoon and East Asian monsoon. The local meteoric water line [LMWL, δ2H=8.12 (±0.04) δ18O+11.2(±0.4)] was established at the Tengchong (TC) site, which was similar to the global meteoric water line (GMWL, δ2H=8 δ18O+10) indicating little secondary sub-cloud evaporation in the falling rain. Precipitation δ18O exhibited significant inverse relationships with precipitation amount (r = -0.42) and air temperature (r = -0.43) throughout the entire period, which indicated that precipitation isotopic variability largely depended on the local meteorological conditions. Precipitation δ18O values are characterized by remarkably seasonal variability: In the summer monsoon period, moisture sources primarily originated from BoB source towards TC site experiencing local moisture recycling over land. The air masses were derived from the northern region of Africa and East Asia with the longest transporting distance and cyclonic activity over the source region in the Fall-winter period characterized with the depleted δ18O values. Precipitation δ18O at the TC site was estimated by a Rayleigh fractionation model considering rainout over BoB and land (Myanmar) during the vapor advection, and local recycling processes consistent with the observed precipitation δ18O values. These findings enhance our understanding of hydrological cycle in the Southwest monsoon regions and will potentially facilitate the interpretation of numerous isotopic proxy records from this region.


Introduction
The climate of Yungui Plateau in Southwest China is controlled by the westerlies and summer monsoon systems of southwest monsoon from Bay of Bengal (BoB) and East Asian monsoon from the South China Sea (SCS), characterized by seasonally strong rainfall and drought. The variability of the summer monsoon, its onset, duration, and failure have great impacts on regional societies and agriculture (Lin et al. 2015;Xu et al. 2002;Zhang et al. 2012). To better understand it, many researchers make huge attempts to reconstruct past climate and hydrology of the region by linking stable isotopes (δ 18 O and δ 2 H) records of paleoarchives (e.g., stalagmites and lacustrine sediments) to hydrological and meteorological conditions (Cheng et  Stable isotopes (δ 18 O and δ 2 H) in precipitation provides valuable information about moisture source and atmospheric circulation patterns that give rise to the rainfall on regional and global scales, and can be used to reconstruct past climate conditions. The so-called "temperature effect" (e.g., a positive correlation between δ 18 O and local air temperature) has been widely reported in the mid-to-high latitudes (Aggarwal et al. 2004;Dansgaard 1964).
According to this effect, the isotopic signals from paleoarchives have been widely considered as a key index of paleotemperature (Gao et al. 2015; Yao et al. 1996). By contrast, in lower latitudes characterized by the empirical "amount effect" (e.g., an anti-correlation between monthly precipitation amount and precipitation δ 18 O), historical changes in precipitation amount and local atmospheric humidity can be reconstructed in these regions (Pausata et al. 2011;Yadava et al. 2004). However, several literatures have showed the insigni cant or even non-existent amount effect at some low-latitude stations impacted by the Indian summer monsoon (Breitenbach et al. 2010;Sengupta;Sarkar 2006). Hence, this nding gives rise to several controversies regarding the amount effect for paleoclimatic reconstruction. Unlike the remarkable temperature effect in the high-latitude regions, potential factors controlling precipitation stable isotopes in the mid-low latitude monsoon regions are complex, it is de cient to unravel the potential mechanisms of precipitation stable isotopes to climatic and environmental forcing dependent solely on the statistical relationships with local meteorological parameters.
Using the observational isotope data, numerous studies examined a variety of hydrological processes determining the isotopic variability in precipitation such as rainout along transport paths, moisture sources, local microphysical processes within the cloud and precipitation type at both the daily and seasonal scales (Aggarwal et al. 2016;Wang et al. 2016). However, lack of precipitation isotope data has been restricted the systematic evaluation of these processes controlling isotopic variability of the southwestern monsoon regions available at monthly intervals from a few stations operating under the Global Network on Isotopes in Precipitation (GNIP) program. Recently, the Chinese Network of Isotopes in Precipitation (CHNIP) has begun to monitor precipitation isotopes at 40 sites across China at event-based resolution (Liu et al. 2014;Yao et al. 2013). Although these data sets are applied to determine spatial isotopic patterns and regional meteoric water line (Bowen et  Hence, the high frequency of event-based precipitation isotopes will provide the opportunity to delineate the climateisotope relationship and how moisture sources affect the isotopic compositions in precipitation, especially in the southwestern monsoon regions. Several studies, however, focused on either the in uence of large-scale convective activity, cloud formation mechanisms and condensation processes along air mass trajectories in determining changes in precipitation stable isotopes across various monsoon regions (Cai et  Previous studies were conducted to identify the local meteorological derivers determining the precipitation isotopic variability in this region (Li et al. 2017;Zhang et al. 2010). These studies have only depended on the synoptic timescale, which has a limited ability to explore the variability of stable isotopes in precipitation. In order to better ll the knowledge gaps, we present four years' daily precipitation isotopic data from the study site to explore atmospheric processes and moisture source that drive isotopic variability in this region. The objectives of this study are to: (1) characterize the variations of daily precipitation stable isotopes; (2) explore the relationship of precipitation stable isotopes with moisture sources and local meteorological parameters; (3) investigate the impact of moisture rainout processes during the monsoon periods via comparing observational data with model results from IsoGSM2 data.

Study area
The study site of Tengchong (TC), adjacent to Myanmar, is located in the southwestern region (SW) of Yunnan Province, China. The total area of the Yunnan province is approximately 3.9 ×10 5 km 2 and is an important component of the Yungui Plateau. The local climate is dominated by Indian monsoon, East Asian monsoon, and westerlies ( Fig. 1). Monthly average precipitation amount in TC ranges from 15.4 mm in December to 295.7 mm in July with total amount of 1532.4 mm (Fig. 2). The air temperature presents distinct seasonal variations with mean annual temperature of 15.4 °C (Fig. 2). The study site is characterized with evident dry (October to April) and wet (May to September) seasons. According to the historical meteorological data, the mean precipitation amount in the wet season accounts for 84% of the mean annual precipitation.

Field sampling and analysis
Daily precipitation samples were sampled at the TC meteorological station in Yunnan province during the observation period from 2009 to 2012. A total of 343 precipitation samples was measured for oxygen-18 and deuterium at the TC sampling site, respectively. All precipitation samples were sealed with screw caps, and wrapped with para lm to prevent sample loss and evaporation prior to analysis. In addition, the duration of each precipitation event, air temperature, relative humidity, and precipitation amount were recorded during the observation period. Deuterium excess (d-excess) is calculated as d= δ 2 H-8*δ 18 O to investigate non-equilibrium effects on the isotopic composition of precipitation (Dansgaard 1964).
In addition, monthly precipitation isotope composition data collected at the GNIP Kunming station were used in this study. The 15-year GNIP data were recorded from 1986 to 2003 (absent from 1993 to 1995) along with mean monthly temperature and precipitation amount at the Kunming station.

HYSPLIT backward trajectories analysis
To determine the origin of air masses and transport paths, air mass back trajectory was calculated for each precipitating day using the Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model version 4.0 (Stein et al. 2015). Meteorological data at 1°×1° and 3-hourly resolution from the Global Data Assimilation System (GDAS) were used to drive the HYSPLIT model. Backward trajectory was set for a time period of 10 days and started at 1500 m above ground level. The selection of 1500 m height corresponding to the 850 mb pressure level was considered the major condensation and strong horizontal moisture transport in this region (Breitenbach et al. 2010). Air parcel positions and speci c humidity were calculated in the HYSPLIT model. Trajectory frequency passing through a certain 1°×1° box was calculated to visualize the spatial pattern of precipitating air mass transport.
For selected typical precipitation events, ten-day backward trajectories representing daily rainfall samples were calculated as trajectory ensembles. Each including twenty-seven ensemble members released at 12:00 LT on the day of precipitating sample collection. HYSPLIT ensembles analysis was helpful for reducing the uncertainty of a single run for trajectory analysis.

Models and data sources
To assess the effect of air rainout processes from main moisture source of BoB on the precipitation isotopes variability during monsoon period, three boxes including ocean box (Box 3), land box (Box 2) and site box (Box 1) were designed during the monsoon periods (Fig. S1). The region of BoB was the main moisture source fed the southwest China during the monsoon periods (Seen from backward trajectory discussion later). The moisture suffered fractionation by rainout during advection over the moisture source of BoB (Ocean box) and the land Box (Myanmar) and is further changed by recycling before rainout in the TC study site (Site box). Rayleigh distillation model was considered to explore the step-wise changes in the vapor isotopic composition and then the determination of precipitation isotopes in the TC study site. The detailed description of modeling the isotopic composition in precipitation was listed in the supplementary material.
In this model, some meteorological parameters such as wind speed, relative humidity, speci c humidity and the total column water vapor in the atmosphere across BoB and the study region. These variables were produced from the ERA-Interim data set from the European Centre for Medium Weather Forecast with 0.75° × 0.75° spatial resolution (http://www.ecmwf.int/en/research/climate-reanalysis/era-interim). The SST data with 1° × 1° spatial resolution was available from National Oceanic and Atmospheric Administration (NOAA) optimum interpolation SST version 2 (V2) (https://www.ncdc.noaa.gov/oisst/data-access).
Due to the lack of the isotopic compositions of water vapor, the data set of Isotope Global Spectral Model version 2 (IsoGSM2) was used to estimate the rainout fraction, which is compared with the calculated stable isotopes in water vapor from Rayleigh model during the transporting processes. The IsoGSM2 is a water isotope-enabled general circulation model, developed by (Yoshimura et al. 2008). IsoGSM2 has a horizontal resolution of 1.8° × 1.8° and 28 vertical levels, and a time resolution of six hours. IsoGSM2 simulations were used for obtaining estimates of vapor isotopic compositions (Wei et al. 2018).

Precipitation isotopic compositions: daily and seasonality
The wide range of precipitation δ 18 O and δ 2 H values are observed from 2009 to 2012 ( Fig. 3 and Table 1). The full dataset is presented in Table S1. Results show that daily δ 18 O values vary from -26.8 ‰ to 4.1 ‰ with an amountweighted average of -10.1 ‰ and δ 2 H values range from -200.9 ‰ to 36.9 ‰ with an amount-weighted average of -70.5 ‰ (Fig. 3 and Table 1 (Fig. S2), which is associated with moisture sources from Indian Ocean region with high speci c humidity and westerlies from the Mediterranean regions with low speci c humidity.
The correlations of daily precipitation δ 18 O with precipitation amount and air temperature are shown during the entire period (Table 1). Results show that a signi cant negative correlation of daily precipitation δ 18 O with precipitation amount and air temperature is observed, which is consistent to the monthly variations (Fig. S3). By contrast, a signi cant correlation of air temperature with precipitation δ 18 O is found during the Fall-winter period but not for the pre-monsoon and monsoon periods at the TC site.

Relationships of meteorological parameters with isotopic compositions
In light of the time-varying isotope values in the SW regions described above, the potential relationships between precipitation isotopes and local meteorological parameters were explored at the daily (Table 1) and monthly scale (Fig. S3). Precipitation δ 18 O exhibited more signi cant inverse relationships with precipitation amount and air temperature at the monthly and daily scales, which indicated that air temperature and precipitation amount play important roles in determining the isotopic variability at the TC site. Relatively higher precipitation amount pattern was found from May to October (Fig. 2), which indicates that amount effect largely contribute to isotopic variability (Zhang et al. 2010). The amount effect derived from monthly data suggests that the lighter rainfall with smaller raindrops was more inclined to kinetic fractionation resulting in enriched heavy isotope ratios below the cloud in the low-latitude regions (Dansgaard 1964;Risi et al. 2008). The Pearson correlation coe cients of amount effect is remarkably higher at the monthly (r= -0.74, p < 0.001, Fig. S3) scale than at the daily (r= -0.42, p < 0.001, Table 1) scale. The weak correlation between precipitation δ 18 O and daily precipitation amount further indicates that local The signi cant anti-temperature effect (negative correlation between precipitation δ 18 O and air temperature) was also observed (Table 1 and Fig. S3). The correlation between precipitation δ 18 O and air temperature is also found in other regions of Yunnan province such as Kunming and Mengzi site (Li et al., 2017). However, a signi cant negative between precipitation δ 18 O and air temperature was found during the Fall-winter period but not for the pre-monsoon and monsoon periods ( Table 1). The contrasting levels of correlation for the different seasons may indicate that the mechanism of the temperature effect is complex in this region. Some complex local (e.g., vapor recycling or subcloud evaporation) and regional atmospheric processes (e.g., convective activity and precipitating rainout) weaken the effect of air temperature on precipitation δ 18

Impacts of moisture sources on precipitation isotopic variability
The time-series of precipitation δ 18 O and δ 2 H exhibit a clear trend from the enriched isotopic values in the premonsoon period to depleted isotopic values in the summer monsoon and Fall-winter period ( Fig. 3 and Table 1), which is attributed to changes in moisture source across seasons (Araguás-Araguás et al. 2000; Cai; Tian 2020). One interesting depletion phenomenon is that the most depleted precipitation isotopic values were found in the Fall-winter period (δ 18 O= -11.9 ‰, δ 2 H= -83.4 ‰, Table 1), which is consistent to the isotopic results from South Asian regions  Figure 5. In the premonsoon period, there are two branches of trajectories: one branch from the interior of Arabian Peninsula-Iranian Plateau and another from BoB (Fig. 5a). The higher temperatures and salinities observed in the Red Sea and Arabian Sea with high evaporation rates, where it feeds water vapor to the downstream regions with the enriched isotopic compositions (Ganssen; Kroon 1991). In the summer monsoon period, air trajectories mainly originate from the northern Indian Ocean and crossing the BoB towards the TC site (Fig. 5b). These water vapor experience the rainout distillation processes along the trajectory and then result in the depleted isotopic compositions in precipitation at a given site.
However, in the Fall-winter period, the length of trajectories prolongs, and they split into two branches from the west and east of the study site (Fig. 5c)

Effects of rainout processes during the monsoon periods
As discussed above, moisture originated from BoB regions largely determine the isotopic compositions of most precipitating events based on HYSPLIT backward analysis during the pre-monsoon and summer monsoon periods (Fig. 5). Air masses from ocean to the terminal site experience rainout, advection and local recycling processes, and thus alter the isotopic compositions in water vapor and precipitation. To assess the in uence of these processes on isotopic compositions in precipitation, we design three boxes (ocean, land, and site, Fig. S1) to model in terms of processes including four states: evaporation at the source region, transport of moisture and rainout over ocean and land, moisture recycling via transpiration and evaporation from surface waters in the land, and nally the Rayleigh rainout in the terminal TC sampling site (Fig. 6). The detailed description of modelling isotopic compositions of precipitation at TC is listed in the supplementary materials.
The TC precipitation isotopic compositions are estimated by Rayleigh condensation of the emerging water vapor from the ocean and land boxes. A box of 1°×1° around TC is considered to estimate the rainout fraction and precipitation δ 18 O values. In the Rayleigh model for the precipitation isotope, water vapor isotopic composition is considered as an important parameter, which is derived from the calculation of classic Rayleigh fractionation formula and IsoGSM2 data. The spatial pattern of water vapor δ 18 O values from IsoGSM2 is shown in Figure S5. The modeled predicted precipitation δ 18 O values at TC are more enriched than the observed precipitation δ 18 O values (Fig. 7), which may be related with moisture recycling during vapor advection over BoB and land (Myanmar). There is overall similarity in the pattern of variations whereas the modeling results by IsoGSM2 (r=0.53) are better matched with Rayleigh model (r=0.38) (Fig. 7cd). This discrepancy could be resulted from different uncertainties in the model which are not addressed. Our model calculates the isotopic composition in precipitation based on a set of simple Rayleigh fractionation processes occurring over the ocean and land. To that end, many input parameters are derived from satellite measurements of meteorological variables which are not continuous measurements entailing some uncertainties. According to the model results, a summary picture can be listed as follows (Fig. 6). The daily rainout fraction over the ocean ranges from 0 to 38% whereas over the land varies between 0 to 62 %. The mean evaporated vapor δ 18 O values over the ocean is approximately -10 ± 0.07 ‰, which is similar to the previous results by (

Conclusions
In this study, daily observations of isotopic compositions (δ 18 O, δ 2 H, and d-excess) from 2009 to 2012 in TC sampling site were presented. The relationships between variations of precipitation δ 18 O with local meteorological variables (e.g., precipitation amount and air temperature) were investigated to assess the response of precipitation δ 18 O to climate changes. In addition, the in uence of moisture sources and moisture rainout processes on precipitation variability was identi ed from the HYSPLIT trajectory data and Rayleigh distillation models. Results showed that precipitation δ 18 O exhibited remarkably seasonal variability, with higher values in pre-monsoon period and lower values in the Fall-winter period. The signi cant negative relationships of precipitation amount and air temperature with precipitation δ 18 O were found, indicating that the variability of precipitation δ 18 O greatly depended on local climatic conditions. Compared with precipitation processes, the contrasting moisture sources and its transporting distance largely contribute to the seasonal variability of precipitation isotopic compositions. In the summer monsoon period, moisture sources primarily originated from BoB source regions while air masses were derived from the northern region of Africa and East Asia with the longest transporting distance in the Fall-winter period characterized with the depleted δ 18 O values. In addition, the precipitating air masses from BoB transporting towards TC sampling site experience precipitation rainout processes during summer monsoon. The rainout effect during the moisture advection and the moisture recycling processes over land box (Myanmar) are two important factors in determining the isotopic composition of TC precipitation. The ndings of this study enhance our understanding of the temporal variations of precipitation stable isotopes at the TC site, and shed a new light on the study of climatic and environmental controls on precipitation stable isotope composition in SW monsoon regions. Figure 1 Geographical distribution of study region and precipitation sampling site of Tengchong (TC). The dominant circulation systems (arrows with upper case letter) of the Indian monsoon (A), East Asian monsoon (B), and Westerlies (C). Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.   Temporal patterns of daily precipitation δ18O, δ2H, and d-excess values at TC sampling site. Yellow, blue, and gray bars represent the Pre-monsoon, monsoon, and Fall-winter periods, respectively.

Figure 4
Plots of δ18O and δ2H in precipitation from TC sampling site. KM represents Kunming of the GNIP site. Spatial patterns of 10-day trajectory frequency during pre-monsoon, summer monsoon and Fall-winter periods. Stars indicate the location of TC sampling site. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 6
A schematic to illustrate the water vapor stemmed from ocean (Box 3) and its advection towards the land (Box 2). The isotopic compositions of the water vapor are changed by rainout and recycling from transpiration and evaporation from surface waters. This modi ed water vapor produced precipitation over the TC sampling site (Box 1). The values in brackets represent the mean rainout fraction and δ18O values in water vapor and precipitation, respectively.