Spatial variation of amount effect over peninsular ... - AGU Publications

PUBLICATIONS Geophysical Research Letters RESEARCH LETTER 10.1002/2015GL064517 Key Points: • Spatial variations of amount effect is driven by ratio of ISM to NEM rains • Significant amount effect is observed in NEM-dominated region 18 • More O depletion of NEM rain is likely due to the cyclonic storms over BoB

Supporting Information: • Tables S1 and S2 Correspondence to: P. R. Lekshmy, [email protected]

Citation: Lekshmy, P. R., M. Midhun, and R. Ramesh (2015), Spatial variation of amount effect over peninsular India and Sri Lanka: Role of seasonality, Geophys. Res. Lett., 42, 5500–5507, doi:10.1002/2015GL064517. Received 12 MAY 2015 Accepted 12 JUN 2015 Accepted article online 15 JUN 2015 Published online 4 JUL 2015

Spatial variation of amount effect over peninsular India and Sri Lanka: Role of seasonality P. R. Lekshmy1, M. Midhun1,2, and R. Ramesh1 1

Geosciences Division, Physical Research Laboratory, Ahmedabad, India, 2Department of Physics, Indian Institute of Technology Gandhinagar, Ahmedabad, India

Abstract The relationship between rain amount and rain δ18O of monsoon rain (amount effect) helps to reconstruct past monsoon variability from proxies (e.g., tree rings and speleothems). Analysis of new (and published) data of the δ18O of monsoon rains and vapor at nine stations shows that in regions of distinct seasonality in precipitation (e.g., peninsular India), the noise in such reconstructions can be minimized by a careful selection of sites. Peninsular India receives rain from both the Indian summer monsoon (ISM) and the northeast monsoon (NEM). Significant amount effect is observed only where the NEM rainfall is larger than or comparable to ISM rainfall. This is due to the higher quantity of NEM rain with more depleted 18O relative to ISM rain. NEM rain is more depleted in 18O because of cyclonic activity over Bay of Bengal, and the 18O depletion of Bay of Bengal surface waters due to post-ISM river runoff.

1. Introduction The monthly amount-weighted mean stable oxygen isotope ratio (δ18O) of tropical rain is negatively correlated with monthly rainfall, where the typical relationship is ~ 1.5‰ change in δ18O per 100 mm of rain [Dansgaard, 1964]. This so-called “amount effect” is exploited for the reconstruction of past monsoon rainfall from δ18O of cave calcites and tree rings from the tropics [e.g.,Yadava and Ramesh, 2005; Managave et al., 2011]. The δ18O variations of such proxies are also interpreted in terms of past monsoon intensity [e.g., Wang et al., 2008] or regional scale convection [e.g., Carolin et al., 2013]. The amount effect is known to exhibit a large spatial variation. Causes include complex cloud-rainout processes [Risi et al., 2008; Kurita, 2013; Moerman et al., 2013; Moore et al., 2014], seasonality of rain [Cobb et al., 2007; Yadava et al., 2007], and differences in the sources of moisture [Xie et al., 2011]. Peninsular India and Sri Lanka receive both the Indian summer monsoon (ISM, June–September) and northeast monsoon (winter monsoon or NEM, October–December). Over the east peninsular Indian coast and Sri Lanka, NEM is the dominant source of rain, while over the rest of peninsular India, ISM dominates (Figure 1a). During the ISM, driven by south westerly winds, the moisture source of rainfall lies in the Arabian Sea and southern Indian Ocean, while during the NEM, north easterly winds bring moisture mainly from the Bay of Bengal (BoB) [Wang, 2006]. In southwest India, the local rain amount plays a weaker role in determining δ18O of monsoon rain compared to moisture recycling associated with the large scale convective systems [Lekshmy et al., 2014]. The seasonal shift in the source of moisture and the associated change in δ18O of the incoming vapor further complicate the rainfall-δ18O correlation [Araguás-Araguás and Froehlich, 1998; Yadava et al., 2007; Feng et al., 2009; Warrier et al., 2010]. With an aim to optimally choose 18O-based monsoon proxy sites from this region, given the abundant teak trees and limestone caves that possibly house climate archives, we investigate here (i) the spatial variation of the amount effect and (ii) the causes for the seasonal difference in δ18O of rainfall in NEM and ISM in peninsular India and Sri Lanka that leads to a stronger amount effect.

2. Data and Methods

©2015. American Geophysical Union. All Rights Reserved.

LEKSHMY ET AL.

We measured rainfall δ18O at nine stations (Figure 1, station numbers 7–15) over Kerala, southwestern India, during 2012 and 2013 with a high spatial resolution (average distance between stations ~80 km). The δ18O of atmospheric water vapor was also measured during a low-pressure storm event that occurred on the last week of April 2012 from two of these stations (7 and 14). The International Atomic Energy Agency protocol was used for sample collection (http://www-naweb.iaea.org/napc/ih/documents/other/gnip_manual_v2.02_en_hq.pdf).

AMOUNT EFFECT OVER INDIA AND SRI LANKA

5500

Geophysical Research Letters

10.1002/2015GL064517

Figure 1. (a) Spatial pattern of the ratio of ISM (June–Septemer, JJAS) to NEM (October–December, OND) rainfall. Locations (heights above the mean sea level) are as follows: (1) Mumbai (10 m), (2) Hyderabad (545 m), (3) Kakinada (8 m), (4) Belgaum (747 m), (5) Bangalore (897 m), (6) Mangalore (22 m), (7) Kozhikode-Kavil (30 m), (8) Wayanad (800 m), (9) Nilambur (35 m), (10) Kozhikode (20 m), (11) Palakkadu (26 m), (12) Thrissur (12 m), (13) Idukki (770 m), (14) Ernakulam (6 m), (15) Ponmudi (780 m), (16) Thiruvananthapuram (53 m), (17) Tirunelveli (4 m), (18) Mannar (5 m), (19) Anuradhapura (81 m), (20) Puttalam (2 m), (21) Maha Illuppallama (137 m), (22) Batticaloa (5 m), (23) Batalagoda (139 m), (24) Wellampittiya (5 m), (25) Colombo (7 m), (26) Nuwara Eliya (1880 m), (27) Tanamalwila (75 m), and (28) Hambantota (20 m). Wind vectors during ISM and NEM are schematically shown by arrows; circles: GNIP stations; triangles: our stations; square: Mangalore [Yadava et al., 2007]. 18 (b) Correlation coefficients between monthly rain and its δ O. Values significant at 0.05 levels are marked by a plus sign.

Daily rain water samples were collected using a carboy and funnel kept in an open space in order to collect unobstructed rain. Every day at 9:00 A.M. India Standard Time, the amount of rain was measured and a portion of the sample was transferred to leak proof bottles for further isotopic analysis. Care was taken at each stage of the sample collection to avoid evaporation. A Thermo Delta V Plus continuous flow mass spectrometer was used for isotopic analysis with an overall precision of 0.1‰ (1σ) for δ18O (for details, see Srivastava et al. [2015]). These data were complemented by rainfall δ18O data from other sites in the Global Network of Isotopes in Precipitation (GNIP) and δ18O data from Mangalore by Yadava et al., [2007]. The geographical coverage of GNIP stations is coarser in peninsular India, as only seven stations have at least five monthly rainfall δ18O observations. The spatial resolution of GNIP stations at Sri Lanka is higher, with 11 sampling sites across the island. Information on Tropical cyclones (http://www.rmcchennaieatlas.tn.nic.in/AboutEAtlas.aspx) from the India Meteorological Department (IMD) and CRU (Climatic Research Unit) precipitation data with 0.5° × 0.5° resolution is used for calculating the climatology (1981–2010 Common Era) of ISM and NEM rainfall.

3. The Amount Effect Within the region from ~6 to 19°N and 72.5 to 82.5°E, we observe a significant spatial correlation between monthly rainfall and δ18O at GNIP stations in peninsular India and Sri Lanka (Table 1). In peninsular India, observed correlations are not statistically significant (at P = 0.05 level), barring Kozhikode (GNIP with 92 monthly data) and Mangalore ([Yadava et al., 2007], with seven monthly data), which show significant positive correlations. Although Kozhikode-Kavil in Table 1 (distinct from GNIP station Kozhikode, ~30 km away) had only 15 months of data, it yields identical slopes and intercepts as GNIP data, albeit with a higher standard deviation. In Sri Lanka, 8 of the 11 stations show highly significant negative correlations (with slopes from 0.023 to 0.005‰ mm1), while at three GNIP stations the correlations are insignificant. The relative contributions of ISM and NEM rainfall in peninsular India and Sri Lanka exhibit high spatial variability (Figure 1). The ratio (r) of ISM (summer) to NEM (winter) rainfall was calculated using CRU

LEKSHMY ET AL.


5501


10.1002/2015GL064517

a

18

Table 1. Monthly Rainfall-δ O Correlations at Different Stations in Peninsular India and Sri Lanka Location (Station Number as in Figure 1) Thiruvananthapuram (16) Ponmudi (15) Idukki (13) Ernakulam (14) Thrissur (12) Palakkadu (10) Nilambur (9) Kozhikode-Kavil (7) Wayanad (8) Tirunelveli (17) Kozhikode (10) Mangalore (6) Bangalore (5) Belgaum (4) Kakinada (3) Hyderabad (2) Mumbai (1) Mannar (18) Anuradhapura (19) Maha Illuppalla (21) Puttalam (20) Batticaloa (22) Batalagoda (23) Colombo (25) Wellampittiya (24) Nuwara Eliya (26) Tanamalwila (27) Hambantota (28)

–1

Latitude (°N)

Longitude (°E)

Years

R

n

Slope (‰ mm )

Intercept

8.53 8.76 9.94 9.95 10.45 10.76 11.31 11.49 11.51 8.7 11.2 12.9 13 15.9 17 17.4 18.9 9 8.4 8.1 8 7.7 7.5 6.9 7 7 6.4 6.1

76.99 77.12 77.02 76.35 76.12 76.27 76.21 75.77 76.02 77.7 75.8 74.9 77.6 74.5 82.2 78.5 72.8 79.9 80.4 80.5 79.8 81.7 80.5 79.9 79.9 80.8 81.1 81.1

2012–2013 2012–2013 2012–2013 2012 2012–2013 2012–2013 2012–2013 2012–2013 2012 2003–2005 1997–2008 2000–2002 2003–2004 2003–2005 2003–2006 1997–2008 1960–1978 1983–1986 1983–1988 1992–1993 1992–1993 1983–1988 1992–1993 1983–1994 2009 1983–1993 1983–1986 1983–1993

0.07 0.09 0.31 0.54 0.28 0.35 0.57 0.47 0.01 0.25 0.29 0.85 0.31 0.18 0.35 0.24 0.16 0.48 0.75 0.85 0.64 0.39 0.63 0.06 0.74 0.14 0.45 0.45

22 17 15 9 17 14 12 15 7 12 92 7 13 12 24 55 50 19 37 15 15 49 16 53 12 61 10 39

0.001 ± 0.004 0.001 ± 0.001 0.002 ± 0.001 0.009 ± 0.005 0.001 ± 0.001 0.002 ± 0.001 0.003 ± 0.001 0.003 ± 0.002 0.004 ± 0.002 0.005 ± 0.006 0.003 ± 0.001 0.008 ± 0.002 0.01 ± 0.01 0.003 ± 0.005 0.006 ± 0.004 0.011 ± 0.005 0.001 ± 0.001 0.006 ± 0.003 0.015 ± 0.002 0.009 ± 0.002 0.012 ± 0.04 0.005 ± 0.002 0.012 ± 0.004 0.001 ± 0.002 0.023 ± 0.007 0.004 ± 0.004 0.010 ± 0.007 0.018 ± 0.006

3.9 ± 0.0.6 3.9 ± 0.6 4.2 ± 0.7 1.5 ± 1.4 3.5 ± 0.5 3.4 ± 0.6 3.6 ± 0.6 4.9 ± 0.8 3.1 ± 1.5 2.8 ± 0.5 4.9 ± 0.5 5.8 ± 0.9 2.5 ± 1.7 3.1 ± 1.2 3.9 ± 0.9 2.7 ± 0.8 2.0 ± 0.5 4.2 ± 0.6 2.1 ± 0.4 2.4 ± 0.3 2.6 ± 0.6 3.2 ± 0.39 2.6 ± 0.7 3.4 ± 0.5 0.87 ± 1.0 5.4 ± 0.7 3.5 ± 0.9 2.8 ± 0.5

a

Monthly rainfall below 5 mm is ignored. Linear correlation coefficients (R) significant at P = 0.05 level are highlighted (bold). GNIP stations (normal font); our stations and Mangalore [Yadava et al., 2007] (italics). The average value of statistically significant amount effect is–1.2‰ per 100 mm of rain (barring Kozhikode and Mangalore).

climatological data. Barring Sri Lanka and southeastern peninsular India, which receive more NEM rainfall, the rest of the stations receive rainfall mostly from ISM. Further, NEM rainfall shows stronger 18O depletion than ISM rainfall at most GNIP stations, which was reported as the cause for the response that is opposite to that expected by the amount effect [Yadava et al., 2007; Warrier et al., 2010], i.e., higher ISM rainfall with relatively enriched 18O and lower NEM rainfall with relatively depleted 18O. Our data for nine stations in Kerala for 2013, reconfirmed Lekshmy et al.’s [2014] earlier inference that no amount effect exists, as was the case in 2012. Thus, we infer that the spatial variation of the slope of the monthly rainfall-δ18O relation across peninsular India and Sri Lanka appears to be mainly determined by the spatial variation in the relative contributions of ISM and NEM. Significant negative correlations are observed only in regions where the NEM rainfall is higher than ISM rainfall. Weaker negative correlations are observed at stations where the ISM and NEM rainfalls are comparable to each other (e.g., Tirunelveli, Bangalore, and Kakinada). To check if this inference is robust and not an artifact of varying data lengths, we reanalyzed the data of all the stations with more than 2 years of data. Correlations for as many consecutive 2 year periods as allowed by the data length were computed. Figure 2 shows a plot of the computed correlation coefficients as a function of the number of data points. It is clear that for stations with less than 14 months of data, the correlation coefficient shows no consistent relationship between rainfall amount and δ18O. These stations have less data because they receive (a) only ISM rain for a few months; the rest of the year is dry (e.g., Mumbai; nine inverted triangles on the left region of Figure 2) or (b) mainly ISM rain, with occasional premonsoon showers enriched in 18O (e.g., Hyderabad; five triangles in the midregion of Figure 2). For stations with greater than 14 months of data, however, the significance or otherwise of the correlation appears to depend on r: the lower the ratio of ISM:NEM, the relationship displays a more consistent negative correlation. This is also shown in Table S1 in the supporting information: significant correlations also exist at Anuradhapura and Hambantota (stations 19 and 28). At Batticaloa (station 22), however, the correlation persisted only when r was 0.4. When r is lower, the correlation deteriorated. This is because the NEM rain

LEKSHMY ET AL.


5502


10.1002/2015GL064517

18

Figure 2. Linear correlation coefficient (R) between rainfall and δ O (calculated for every two consecutive years for GNIP stations with records longer than 2 years) plotted as a function of the number of monthly data points. The black lines connect correlation coefficients that are significant at 0.05 level. Points falling outside these lines are significant at a level smaller than 0.05 (i.e., higher significance). Different symbols denote different stations but with a common color coding based on the ratio (r) of ISM to NEM rainfall.

alone does not show an appreciable amount effect. The amount effect exists because of a lower quantity of weakly 18O-depleted ISM rain followed by higher quantity of strongly 18O-depleted NEM rain (there are more data points at Batticaloa because it receives small amounts of premonsoon and ISM rains). There are interannual variations in the significance of the amount effect as revealed by Table S1. While we believe that this is mainly due to the relative dominance or otherwise of NEM, the robustness of our conclusion needs further validation with longer data sets.

4. Controls on NEM Rainfall δ18O A stronger 18O depletion of NEM rain relative to ISM rain was previously explained by (i) seasonal runoff from the Ganga-Brahmaputra rivers post ISM, 18O depletion of the surface BoB, leading to 18O depleted NEM vapor and rain [Breitenbach et al., 2010; Singh et al., 2010]; or (ii) ISM rain in western India tends to be the first condensate (though the air masses travel a long distance over the Indian Ocean) and is therefore relatively enriched in 18O; whereas the NEM air mass travels a longer distance over land from the BoB [Warrier et al., 2010], further depleting the 18O of NEM rain over southwestern India due to rainfall occurring en route [Yadava et al., 2007]. But these two factors appear to have less control on the stronger 18O depletion of NEM than cyclonic activity over BoB, as discussed below. 4.1. The 18O Depletion in BoB Surface Water The δ18O of surface seawater across the BoB during and after ISM is 0.4 ± 0.6 (n = 141, September–October 2002 and August–September 1988 [Singh et al., 2010], September–December 2012–2013 [Achyuthan et al., 2013]). Calculation of the δ18O of the evaporating vapor flux from the ocean during each season was made using the Craig and Gordon [1965] model (entries 1 to 5 in Table 2): RE ¼ αkin

Rocean αeq

hRatm

1h

where, RE, Rocean, and Ratm are the δ18O of evaporative flux, sea surface water, and the ambient atmospheric vapor, respectively. The parameters αkin and αeq are the kinetic and equilibrium fractionation factors, respectively, and h, the relative humidity. Ratm and h were measured by our team, and some data were reported by Midhun et al. [2013] and Srivastava et al. [2015]. Constant sea surface temperature (28°C) and

LEKSHMY ET AL.


5503


10.1002/2015GL064517

18

18

18

Table 2. The δ O of Vapor Flux From BoB Under Different Ambient Conditions of δ O of Sea Surface, δ O of Ambient Vapor, and Relative Humidity Calculated Using the Craig and Gordon [1965] Model With (1–5) and Without (6–9) the a Closure Assumption 18

18

18

δ O Sea Surface (‰)

δ O Ambient Vapor (‰)

h (%)

δ O Evaporating Vapor Flux (‰)

1-ISM 2 3 4 5-NEM

0 0 0.38 0.38 0.38

11.0 11.5 11.5 11.5 12.5

80 70 80 70 70

7.3 10.5 9.2 11.8 8.3

6 7-ISM 8-NEM 9

0 0 0.38 0.38

Under closure assumption ” ” ”

70 80 70 80

10.8 10.2 11.2 10.6

No.

a

18

Values in bold font pertain to observed ambient conditions. Data of δ O of ambient vapor and relative humidity are from Midhun et al. [2013] and Srivastava et al. [2015].

wind speed (5 m/s) for both the seasons are used, as their variations are minor, where the sea surface temperature climatology is from the National Oceanic and Atmospheric Administration optimum interpolated data [Reynolds et al., 2002]. National Centers for Environmental Prediction/National Center for Atmospheric Research reanalysis climatological average winds [Kalnay et al., 1996] during the seasons are less than 7 m/s and are considered as a smooth ocean surface condition [Merlivat and Jouzel, 1979]. The parameters αkin and αeq were calculated using the equations of Merlivat and Jouzel [1979]and Majoube [1971]. The calculation of δ18O values of the evaporating vapor flux was repeated assuming that the vapor over the BoB forms from the evaporating vapor flux alone, without interacting with advected vapor from elsewhere (local closure assumption, Ratm = RE; entries 6 to 9 in Table 2). The first five entries of Table 2 indicate that the δ18O of the evaporating vapor flux is not as sensitive to changes in the δ18O of the sea surface as it is to changes in h. For the same values of h and δ18O of vapor, a change of 0.38‰ in the δ18O of the sea surface causes only a 1.3‰ (h = 70%) to 1.9‰ (h = 80%) decrease in δ18O of the evaporating vapor flux. In contrast, the δ18O of the evaporating vapor flux decreases by 2.6‰ to 3.2‰ when the h decreases from 80 (ISM) to 70% (NEM) regardless of whether the δ18O of the sea surface 0‰ (ISM) or 0.38‰ (NEM). On the other hand, a decrease of 1.5‰ in the ambient water vapor δ18O (from 11.0 to 12.5‰) increases the δ18O of evaporating vapor flux by 3.5‰ (at 70% relative humidity and 0.38‰ sea surface δ18O). Observations and unpublished data of Midhun et al. [2013] and Srivastava et al. [2015] over BoB during ISM and NEM show that the 0.38‰ decrease in the δ18O of the sea surface is accompanied by reductions of h from 80 to 70% and of the δ18O of ambient water vapor from 11.0 to 12.5‰. Thus, the δ18O of evaporating vapor flux decreases only by a resultant of 1‰ (entries 1 and 5, Table 2). Repeating the calculation with the local closure assumption confirms this (entries 7 and 8, Table 2). But the observed mean 18O depletion of NEM rainfall is more than 3‰ relative to the ISM rainfall. Thus, the reason for the higher 18O depletion in NEM rain cannot be explained by the post-ISM 18O depletion of BoB surface waters by river runoff alone. 4.2. Continental Effect The second possibility is that the continuous rainout from a moist air mass originating from BoB traveling westward leads to a stronger 18O depletion of NEM rain in the western peninsular India (Figure 1). However, equally stronger 18O depletion is observed in NEM rain at eastern coastal peninsular India as well, where it travels a lesser distance than in the western portion of the peninsula. NEM shows an average δ18O of 5.9 ± 2.5‰ (Kozhikode and Mumbai, 1.9 ± 0.8‰ for ISM) in the western peninsula and 8.3 ± 1.2‰ (Kakinada and Hyderabad, 3.9 ± 1.4‰ for ISM) in the eastern peninsula (Figure 3a). Interestingly the first condensate on NEM shows stronger 18O depletion in the eastern peninsular India, and therefore the δ18O difference between ISM and NEM rain is not just due to ISM rain being the first condensate in the western peninsula as stated by Warrier et al., [2010]. On the contrary, NEM rain is relatively more enriched in 18O over the western peninsular India compared to the eastern side. This could be either due to the continental moisture recycling or the marine influence from the Arabian Sea, or both. Thus the continental effect also does not seem play a major role in the observed higher 18O depletion of NEM rainfall.

LEKSHMY ET AL.


5504


10.1002/2015GL064517

18

Figure 3. (a) Monthly mean rainfall δ O at Kozhikode, Mumbai (western peninsular India), Hyderabad and Kakinada (eastern peninsular India). Seasonal averages (ISM-JJAS and NEM-OND) of stations in western and eastern peninsular 18 India are marked as blue and magenta lines in the figure, respectively. (b) GNIP time series of δ O of monthly rainfall at Kozhikode. Filled circles indicate the presence of cyclones in the BoB during NEM (shaded bars). (c) Variation of mean 18 18 rainfall (vertical bars), its δ O (filled circles) (means of eight stations at Kerala), and δ O of water vapor (mean of two stations, open circles) during a cyclonic storm that occurred during the last week of April 2012.

4.3. Effect of Cyclonic Disturbances in BoB The NEM season is characterized by increased cyclonic activity over the BoB. Cyclonic rains elsewhere in the tropics are also strongly depleted in 18O [e.g., Lawrence and Gedzelman, 1996; Fudeyasu et al., 2008]. In organized convective systems, processes such as increased post condensation isotopic equilibration of raindrops with the ambient water vapor and raindrop reevaporation occur. Such moisture recycling has been suggested to be responsible for stronger 18O depletion of rain as well as the water vapor in the subcloud layer [Kurita et al., 2011; Risi et al., 2012]. Tropical cyclones and low-pressure systems (depressions) in BoB during NEM, detected by the IMD, are marked in Figure 3b. The strongest 18O depletion of NEM rainfall at Kozhikode co-occurs with cyclonic activities over BoB (18 out of 23). Thus, the stronger 18O depletion of NEM rain could be due either to (i) the rain caused by the low-pressure systems or (ii) the lateral advection and mixing of 18O-depleted remnant vapor of the immediate prior low-pressure systems. Yet a few cyclones are observed during some months with no excess depletion in 18O (5 out of 23), possibly because the prevailing wind did not favor the advection of strongly 18O-depleted vapor to the west. In addition, there are a few months with stronger 18O depletion, which cannot be explained by depressions or cyclones detected by the IMD. There are common occurrences of weak low-pressure systems in the tropical Indian oceans during the pre-ISM (April and May) and NEM seasons, which do not intensify into depressions or cyclones. A cyclonic storm over the southern peninsular India during April 2012 showed (Figure 3c) 18O depletion in rain (average of eight stations) and in the ground level vapor (average of two stations). This typical, strong 18O depletion of rain caused by depressions was also recorded over the BoB [Midhun et al., 2013]. Thus, increased cyclonic activity over the BoB seems to play a significant additional role in the strong 18O depletion of NEM rain.

LEKSHMY ET AL.


5505


10.1002/2015GL064517

5. Conclusions Varying seasonal rain amounts in peninsular India and Sri Lanka lead to a large spatial variation in the slopes of the rainfall-δ18O relationships. The stronger 18O depletion of NEM rainfall is likely caused by increased cyclonic activity over the BoB, in addition to 18O depletion of its surface waters by river discharge. This leads to significant negative correlations between monthly rainfall and its δ18O chiefly in regions where the NEM contributes more than, or at least as much rain as, the ISM. The amount effect also vanishes when the ISM reduces significantly relative to the NEM. Interannual variations in the amount effect due to varying interannual contributions of ISM and NEM rain are also noted. Thus, a careful choice of sites for 18 O-based monsoon proxies can be made so as to minimize noise in the paleomonsoon signal that could arise at sites such as Kozhikode and Mangalore with inverse amount effects. Using proxies capable of providing annual resolution (e.g., fast-growing trees), past annual monsoon rainfall can be reconstructed at sites where the ratio of ISM to NEM rain continues to remain less than or comparable to unity, using the local amount effect.

Acknowledgments We acknowledge ISRO-GBP, Government of India for funding. We thank the two anonymous reviewers and Kim Cobb for comments and suggestions that significantly improved this manuscript. Jud Partin is thanked for suggesting improvements in presentation. The 18 new monthly rainfall-δ O data are made available through the supporting information. CRU-TS3.21 precipitation is available at http://badc.nerc.ac.uk/ data/cru/. GNIP data are available at http://www.univie.ac.at/cartography/ project/wiser/. The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.

LEKSHMY ET AL.

References Achyuthan, H., R. D. Deshpande, M. S. Rao, B. Kumar, T. Nallathambi, K. Shashi Kumar, R. Ramesh, P. Ramachandran, A. S. Maurya, and S. K. Gupta (2013), Stable isotopes and salinity in the surface waters of the Bay of Bengal: Implications for water dynamics and palaeoclimate, Mar. Chem., 149, 51–62, doi:10.1016/j.marchem.2012.12.006. Araguás-Araguás, L., and K. Froehlich (1998), Stable isotope composition of precipitation over southeast Asia, J. Geophys. Res., 103, 28,721–28,742, doi:10.1029/98JD02582. Breitenbach, S. F. M., J. F. Adkins, H. Meyer, N. Marwan, K. K. Kumar, and G. H. Haug (2010), Strong influence of water vapor source dynamics on stable isotopes in precipitation observed in Southern Meghalaya, NE India, Earth Planet. Sci. Lett., 292, 212–220, doi:10.1016/ j.epsl.2010.01.038. Carolin, S. A., K. M. Cobb, J. F. Adkins, B. Clark, J. L. Conroy, S. Lejau, J. Malang, and A. A. Tuen (2013), Varied response of western Pacific hydrology to climate forcing over the last glacial period, Science, 340(6140), 1564–1566, doi:10.1126/science.1233797. Cobb, K. M., J. F. Adkins, J. W. Partin, and B. Clark (2007), Regional-scale climate influences on temporal variations of rainwater and cave dripwater oxygen isotopes in northern Borneo, Earth Planet. Sci. Lett., 263, 207–220, doi:10.1016/j.epsl.2007.08.024. Craig, H., and L. Gordon (1965), Deuterium and oxygen 18 variations in the ocean and marine atmosphere, in Stable Isotopes in Oceanographic Studies and Paleotemperatures, edited by E. Tongiori, pp. 9–130, Consiglio Nazionale Delle Richerche Laboratorio Di Gelogica Nucleare, Pisa, Italy. Dansgaard, W. (1964), Stable isotopes in precipitation, Tellus, 16(4), 436–468. Feng, X., A. M. Faiia, and E. S. Posmentier (2009), Seasonality of isotopes in precipitation: A global perspective, J. Geophys. Res., 114, D0811, doi:10.1029/2008JD011279. Fudeyasu, H., K. Ichiyanagi, A. Sugimoto, K. Yoshimura, A. Ueta, M. Yamanaka, and K. Ozawa (2008), Isotope ratios of precipitation and water vapor observed in Typhoon Shanshan, J. Geophys. Res., 113, D12113, doi:10.1029/2007JD009313. Kalnay, E., et al. (1996), The NCEP/NCAR 40-year reanalysis project, Bull. Am. Meteorol. Soc., 77, 437–470. Kurita, N. (2013), Water isotopic variability in response to mesoscale convective system over the tropical ocean, J. Geophys. Res. Atmos., 118, 10,376–10,390, doi:10.1002/jgrd.50754. Kurita, N., D. Noone, C. Risi, G. A. Schmidt, H. Yamada, and K. Yoneyama (2011), Intraseasonal isotopic variation associated with the Madden-Julian Oscillation, J. Geophys. Res., 116, D24101, doi:10.1029/2010JD015209. Lawrence, R. J., and D. S. Gedzelman (1996), Low stable isotope ratios of tropical cyclone rains, Geophys. Res. Lett., 23, 527–530, doi:10.1029/ 96GL00425. 18 Lekshmy, P. R., M. Midhun, R. Ramesh, and R. A. Jani (2014), O depletion in monsoon rain relates to large scale organized convection rather than the amount of rainfall, Sci. Rep., 4, 1–5, doi:10.1038/srep05661. Majoube, M. (1971), Fractionation of oxygen18 and deuterium between water and its vapour, J. Chem. Phys., 68, 1423–1436. 18 Managave, S. R., M. S. Sheshshayee, A. Bhattacharyya, and R. Ramesh (2011), Intra-annual variations of teak cellulose δ O in Kerala, India: implications to the reconstruction of past summer and winter monsoon rains, Clim. Dyn., 37(3–4), 555–567, doi:10.1007/s00382-010-0917-9. Merlivat, L., and J. Jouzel (1979), Global climatic interpretation of the deuterium-oxygen-18 relationship for precipitation, J. Geophys. Res., 84(9), 5029–5033, doi:10.1029/JC084iC08p05029. Midhun, M., P. R. Lekshmy, and R. Ramesh (2013), Hydrogen and oxygen isotopic compositions of water vapor over the Bay of Bengal during monsoon, Geophys. Res. Lett., 40, 6324–6328, doi:10.1002/2013GL058181. 18 Moerman, J. W., K. M. Cobb, J. F. Adkins, H. Sodemann, B. Clark, and A. A. Tuen (2013), Diurnal to interannual rainfall δ O variations in northern Borneo driven by regional hydrology, Earth Planet. Sci. Lett., 369–370, 108–119, doi:10.1016/j.epsl.2013.03.014. Moore, M., Z. Kuang, and P. N. Blossey (2014), A moisture budget perspective of the amount effect, Geophys. Res. Lett., 41, 1329–1335, doi:10.1002/2013GL058302. Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, and W. Wang (2002), An improved in situ and satellite SST analysis for climate, J. Clim., 15, 1609–1625, doi:10.1175/1520-0442(2002)0152.0.CO;2. 18 Risi, C., S. Bony, and F. Vimeux (2008), Influence of convective processes on the isotopic composition (δ O and δD) of precipitation and water vapor in the tropics: 2. Physical interpretation of the amount effect, J. Geophys. Res., 113, D19306, doi:10.1029/2008JD009943. Risi, C., et al. (2012) Process-evaluation of tropospheric humidity simulated by general circulation models using water vapor isotopic observations: 2. Using isotopic diagnostics to understand the mid and upper tropospheric moist bias in the tropics and subtropics, J. Geophys. Res., 117, D05304, doi:10.1029/2011JD016623. 18 Singh, A., R. A. Jani, and R. Ramesh (2010), Spatiotemporal variations of the δ O–salinity relation in the northern Indian Ocean, Deep Sea Res., Part I, 57(11), 1422–1431, doi:10.1016/j.dsr.2010.08.002. Srivastava, R., R. Ramesh, N. Gandhi, R. A. Jani, and A. K. Singh (2015), Monsoon onset signal in the stable oxygen and hydrogen isotope ratios of monsoon vapor, Atmos. Environ., 108, 117–124, doi:10.1016/j.atmosenv.2015.02.062.


5506


10.1002/2015GL064517

Wang, B. (2006), The Asian Monsoon, Springer, Berlin. Wang, Y., H. Cheng, R. L. Edwards, X. Kong, X. Shao, S. Chen, J. Wu, X. Jiang, X. Wang, and Z. An (2008), Millennial- and orbital-scale changes in the East Asian monsoon over the past 224,000 years, Nature, 451(7182), 1090–1093, doi:10.1038/nature06692. Warrier, C. U., M. P. Babu, P. Manjula, K. Velayudhan, A. S. Hameed, and K. Vasu (2010), Isotopic characterization of dual monsoon precipitation-evidence from Kerala, India, Curr. Sci., 98, 1487–1495. 18 Xie, L., G. Wei, W. Deng, and X. Zhao (2011), Daily δ O and δD of precipitations from 2007 to 2009 in Guangzhou, South China: Implications for changes of moisture sources, J. Hydrol., 400, 477–489, doi:10.1016/j.jhydrol.2011.02.002. Yadava, M. G., and R. Ramesh (2005), Monsoon reconstruction from radiocarbon dated tropical Indian speleothems, Holocene, 15(1), 48–59, doi:10.1191/0959683605h1783rp. Yadava, M. G., R. Ramesh, and K. Pandarinath (2007), A positive “amount effect” in the Sahayadri (Western Ghats) rainfall, Curr. Sci., 93, 560–564.

LEKSHMY ET AL.


5507