Whether stable oxygen isotope (δ18O) in precipitation obeys the temperature effect and/or amount effect in the monsoon region has long been controversial. An intensive, precipitation event-based sampling project has been carried out at Guangzhou and Changsha stations in southeast China under the Asian monsoon influence. By dividing a year into summer and winter half years at respective station, we find prevalence of amount effect at both stations throughout the year. δ18O-temperature presents complex correlations, with the positive correlation significant at Guangzhou during the summer half year and at Changsha during the winter half year, but vague at either station during the rest of the year; the former attributable to a third mode of convection, while the latter indicative of the weakening monsoon influence accompanied by intensified local recycling. Our high-resolution data demonstrate a significant coexistence of temperature and amount effects of precipitation δ18O in the monsoon domain, conducive to climatic interpretation of δ18O in paleoclimate proxies in mid/low latitudes.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Since the initiating work of Dansgaard , stable oxygen isotope (δ18O) in precipitation has been widely studied in different regions in the world, revealing its relationship with local air temperature and precipitation amount, as well as regional atmospheric circulation. Theoretically, under Rayleigh distillation, δ18O in precipitation is positively correlated with local air temperature, thus demonstrating noticeable correlations with latitudes and altitudes [e.g., Dansgaard, 1964; Siegenthaler and Oeschger, 1980; Johnson and Ingram, 2004; Vuille et al., 2005a]. In China, precipitation δ18O is high in the overall southeast coastal China, and decreasing gradually to the northern and northeastern China under latitude effect [Araguás-Araguás et al., 1998; Liu et al., 2008] Due to the existence of Inter-Tropical Convergence Zone (ITCZ), however, the seasonal shift of the ITCZ and associated variation of atmospheric circulations usually give rise to dramatic changes in precipitation δ18O, thus complicating the correlation of δ18O with precipitation amount and temperature.
 Study of monthly data within the Global Network of Isotopes in Precipitation (GNIP) suggests that rain-out of moist, oceanic air masses moving inland associated with monsoon circulation could completely overshadow the dependence of δ18O on temperature, thus amount effect is proposed to outweigh the temperature effect in controlling δ18O variation in lower-latitudes [Araguás-Araguás et al., 1998]. Thompson et al. , Thompson , Yao and Thompson  and Yao et al.  interpreted ice core δ18O in the Himalayan region as a temperature proxy, which was questioned by Brown et al. , who, based on their modeling results, correlated precipitation δ18O with local precipitation amount instead of local or regional temperature.
 Controversies arise due to the deficiencies of climate model in considering such climate parameters as: 1) the complexity of dynamical and microphysical processes leading to the formation of individual precipitation event, and 2) the successive precipitation events at high variability being statistically smoothed by the observed isotopic monthly average [Jouzel et al., 1987]. Since previous studies are based on GNIP monthly data, thus likely to dismiss the daily or synoptic variations in the atmospheric circulation, we think it important to look into high-resolution data at event or daily time scale, in order to capture the accurate monsoon signal to the daily resolution and to clarify the climate significance of precipitation δ18O in the mid/low-latitudes. An intensive precipitation sampling project has therefore been carried out in East China influenced by the Asian monsoon, regulating the observations on the basis of precipitation event.
2. Study Area and Methods
 Our precipitation sampling was conducted at two stations, the Guangzhou Meteorological Satellite Ground Station (23°7′48″N, 113°19′12″E, 7 m a.s.l, later referred to as Guangzhou) nearby the South China Sea, and Changsha Meteorological Station (28°12′0″N, 113°4′0″E, 37 m a.s.l, later referred to as Changsha) to the south of the Yangtze River (Figure 1).
 The sampling campaign was conducted from early June, 2006 till present. For comparative study, we use in this paper three-year data from June 2006 to May 2009. An open-mouthed bucket was set up in the meteorological observation field, with samples immediately taken after each precipitation event larger than 0.1 mm. Each collected precipitation sample was tightly sealed in a 15-ml polyethene bottle as full as possible, and stored cool in the refrigerators before δ18O measurement. Precipitation sampling was entrusted to local meteorological stations, with simultaneous meteorological parameters taken including air temperatures, precipitation amount, air pressure, wind speed, relative humidity, and start and end time of each event.
 The measurement of δ18O in precipitation was conducted on a Measurement and Analysis Technique (MAT) isotope ratio mass spectrometer 253 using water-gas equilibrium at the Key Laboratory of Tibetan Plateau Environment Changes and Land Surface Processes, Chinese Academy Sciences, and reported against Vienna Standard Mean Ocean Water (VSMOW) with a precision of ±0.05‰. The δ18O of each event was measured separately. For days witnessing over one event, δ18O is presented with amount-weighted mean of that daily precipitation. We especially noted samples of small amount in data treatment, and removed those lower than 0.3 mm corresponding to δ18O over 1.5‰ for possible evaporation during storage.
3.1. General Variation Features of δ18O in Precipitation
 At Guangzhou, the three-year data shows a range from −14.94‰ (August 10, 2007) to 5.89‰ (February 29, 2008) for daily precipitation δ18O. Amount-weighted δ18O in monthly precipitation shows rhythmic cycles, with May to October demonstrating obvious depletion while November to next April comparative enrichment (Figure 2a). The former six months witness high temperature (averaging around 28.5°C) and amount (averaging around 19.3 mm) per event, thus identified as the summer half year; while the rest months demonstrating generally low temperature (averaging around 18.6°C) and amount (averaging around 9.5 mm) per event are identified as the winter half year. δ18O in daily precipitation varies more dramatically in summers than winters (σ as 3.24 vs. 2.27) (Figure 2a).
 In comparison, three-year data of daily precipitation δ18O at Changsha varies from −19.77‰ (November 2, 2008) to 3.64‰ (April 1, 2007), with June to November witnessing general depletion while December to next May experiencing enrichment (Figure 2b). Correspondingly, air temperature is also much higher during June–November than that during December–May (23.3°C vs. 10.7°C), thus the former is identified as the summer half year, while the latter as the winter half year. Interestingly, the amount difference with seasons at Changsha is not as contrast as that at Guangzhou, averaging around 13 mm in summers as against 7 mm in winters at Changsha. In specific, unlike Guangzhou where little precipitation occurs in winters, Changsha experiences precipitation of high frequencies and amount. This may be attributed to the influence of the Asian Winter Monsoon, which influences mostly the mid-latitudes [Chang et al., 2006]. In general, both the winter and summer half years start later at Changsha than Guangzhou by one month.
 Despite different variation patterns of δ18O with time at the two stations, the daily and monthly values show noticeable depletion with significant increase in amount during summers, suggesting the tropical monsoon influence and its evolution processes [Thompson et al., 2000; Vuille et al., 2003, 2005b]. Guangzhou and Changsha stations show similar values in multiple-year amount-weighted δ18O average (−7.56‰ at Guangzhou vs. −7.91‰ at Changsha), suggesting significant continental recycling throughout the years in regions of interest. Furthermore, the comparative depletion at Changsha may be attributed to several factors, latitude and distance from the ocean being the most obvious causes.
3.2. Temperature and Precipitation Amount Dependency of δ18O in Precipitation
 Any formation of precipitation is caused by some kind of cooling process, thus Rayleigh distillation regulates the positive control of temperature over isotope contents of precipitation [Dansgaard, 1964]. The observed composition of individual rain is, however, a function of several parameters. The variability of daily precipitation δ18O depends not only on temperature, but also on other atmospheric or geographical parameters [Joussaume et al., 1984]. The acquisition of meteorological parameters simultaneous to precipitation events allows for a comprehensive analysis of atmospheric factors controlling δ18O variation. Bivariate correlation analysis of monitored meteorological parameters shows temperature and amount as bearing close correlations with other meteorological factors; we therefore simply focus on the control of temperature and amount over daily δ18O variation. Stepwise linear regression is adopted to pick out the more important factor. In practice, factors with probability surpassing 95% confidence level enter the correlation model, while those with probability smaller than 90% confidence level are excluded (Table 1). Precipitation amount turns out to be more significant at Guangzhou throughout the year and at Changsha in the summer half year, while temperature outstands amount as more significant at Changsha in the winter half year. Besides, temperature exerts a secondary effect on precipitation δ18O variation at Guangzhou in the summer half year, but it fails the stepwise regression at Guangzhou in the winter half year and at Changsha in the summer half year.
Table 1. Statistics of Stepwise Linear Regression of δ18O in Precipitation With Synchronous Temperature and Amounta
Models shown in the table all surpass the 95% confidence level.
 Considering the prevalence of amount controls over daily precipitation δ18O at both stations regardless of seasons, we used a modified two-step regression model to distinguish the independent control of temperature and amount over precipitation δ18O [Bowen and Wilkinson, 2002]. First, the relationship between precipitation δ18O and amount is described by a linear regression, which is used to produce an amount-dependent δ18O value. The value is then deducted from the original δ18O value to produce a residual. The δ18O residual is thus correlated with simultaneous temperature to study the amount-independent temperature effect on δ18O. As shown in Figures 2a and 2b, precipitation occurs more frequently at Guangzhou during summers than winters, while otherwise at Changsha, where winter precipitation contributes nearly half of annual precipitation (∼43%). This is attributed to different dominative atmospheric processes at respective sites. Negative correlation between δ18O and amount surpasses 99% confidence level at both stations throughout the year, and both stations show a higher δ18O-P slope during winter than summer half years (Figures 2c and 2e). The correlation between the δ18O residual and temperature, though all positive, only demonstrates reliable significance at Guangzhou during summers and at Changsha during winters (Figures 2d and 2f).
 To assess the co-existence of temperature and amount effects in precipitation δ18O at both stations, we adopted a multiple linear regression model within each season as
whereby T, for simultaneous temperature, and P, for daily amount, are treated equally in producing δ18O values. As shown in Table 2, the multiple linear regressions of δ18O with temperature and amount all reach 99% significance level with different correlation coefficients, suggesting the co-existence of temperature and amount effects at both stations throughout the year. The slopes for δ18O-P and δ18O-T correlations at Guangzhou during winters are, respectively, similar with those at Changsha during summers (Table 2), though the positive δ18O-T correlation in both cases fails to pass the 99% confidence level. This suggests some commonality in precipitation formation processes at both sites, and possible vying among various atmospheric circulation systems and relevant fractionation/condensation processes. During the rest seasons at respective station, temperature shows a larger covariate than amount with δ18O (Table 2). This is especially the case with summer precipitation at Guangzhou, highlighting a larger control and more important role of surface temperature than amount over δ18O.
Table 2. Results for the Multiple Linear Regressiona
Here “a” refers to the δ18O-T slope, “b” refers to δ18O-P slope, “c” refers to the constant in the multiple linear regression, and R stands for the correlation coefficient. Numeric figures in parentheses are the standard errors, and bold values indicate that they failed to pass the 99% confidence level.
4. Discussion and Conclusion
 Event-based precipitation sampled at Guangzhou and Changsha in southeast China from June, 2006, to May, 2009, is presented in this study, revealing dramatic depletion of precipitation δ18O during summer monsoons. By dividing the data into winter and summer half years according to their different seasonal precipitation patterns, we find significant coexistence of temperature and amount effects at both sites influenced by the Asian monsoon (summer monsoon and/or winter monsoon). Three-year data also show predominant amount effect throughout the year. The temperature effect demonstrates a more complex coexistence, with positive δ18O-T correlation significant at Guangzhou during summers and at Changsha during winters, when both sites are dominated by a single atmospheric circulation system [Chang et al., 2006; Ding and Sikka, 2006], while positive δ18O-T correlation vague during the rest season of the year at each site.
 Different influences of atmospheric circulations and relevant hydrological cycle may contribute to those differences. During the Asian winter monsoon, the extratropical region is prevailed by low-level north-easterlies associated with the anticyclonic circulations of the surface Siberian-Mongolian High (SMH), which forms a strong cold dome in the northeast [Chang et al., 2006]. Changsha is subjected to the strong cold-core high pressure in winters, witnessing local convection and intense advection of air masses bearing contrast heat features. Both lead to general enrichment of δ18O at high temperature with intensified local evaporation and/or exchanges of falling raindrops with ambient vapor, whereas general depletion of δ18O at low temperature with the cooling of condensation temperature; Thus a good demonstration of temperature effect of δ18O in precipitation. As the SMH moves southeastward toward coastal China, it gradually renders its cold front to the warm marine moisture loaded by Walker Circulation extending across the Pacific Ocean[Webster et al., 1998], resulting in convergence of air masses of differential heating. Precipitation at Guangzhou in winters is thus from frequent kinetic fractionation and advective cooling, disturbing the temperature control over δ18O.
 During the summer monsoon, Guangzhou at the monsoon front demonstrates significant temperature effect, which is noteworthy as intense convection featured in mid/low-latitudes is regarded as the major cause for δ18O disobeying temperature effect [Araguás-Araguás et al., 1998; Brown et al., 2006]. It thus highlights a unique atmospheric process in the monsoonal domain, a third important mode of convection apart from deep and shallow convections—cumulus congestus clouds terminating near the 0°C isotherm level [Johnson et al., 1999]. The height of the 0°C isotherm level is much higher in the tropical and subtropical regions than that in temperate latitudes (i.e., above 5 km vs. below 2 km) [Mason, 1982]. At Guangzhou during summers, surface temperature increases lead to enhancement of land-ocean pressure gradient and uplift of the 0°C isotherm level, resulting in condensation at the cloud base entirely beneath the 0°C isothermal level besides intense convection. Fluctuations of the 0°C isothermal level with surface temperature contributes to the significant positive correlation between δ18O and temperature at Guangzhou during summer monsoon dominance. With the summer monsoon advancing northward and losing its intensity en route, its leading zone and the monsoon rainy belt correspondingly moves from low to mid and high latitudes [Ding and Sikka, 2006]. Changsha then is subject to multiple atmospheric processes, including convection from the summer monsoon, advection due to complex land-ocean terrain, and local recycling associated with evaporation and evapo-transpiration in summers. Such complicated atmospheric circulation processes weakens the temperature control over δ18O in precipitation.
 Our study reveals the coexistence of temperature and amount effects on precipitation δ18O in the Asian monsoon region. The existence of significant temperature effect during intensified monsoon, as at Guangzhou during summer monsoon and at Changsha during winter monsoon, is attributed to the seasonal shift of the 0°C isothermal level, and in association, the temperature vertical profile regulated by the thermodynamic principles, while lack thereof attributed to the interplays of the weakening Asian (summer or winter) monsoon and strengthening local recycling (as at Guangzhou during winters and at Changsha during summers). The study therefore implies that obvious temperature effect exists when a single atmospheric circulation dominates.
 This work was supported by the National Natural Science Foundation of China (grants 40830638 and 41101021), Chinese Academy of Sciences Innovation Group (KZCX2-YW-T11) and the Third Pole Environment program (GJHZ0960). The authors were also indebted to Guangzhou Meteorological Satellite Ground Station and Changsha Meteorological Bureau for providing necessary data presented in the paper. Finally, we would like to thank Xiaoyun Huang, Xiaoli Zhao, Xinmei Huang, Zuhong Luo, Weilin Weng and others for precipitation sampling and meteorological note-taking.
 The Editor thanks the two anonymous reviewers.