Reconstructed droughts for the southeastern Tibetan Plateau over the past 568 years and its linkages to the Pacific and Atlantic Ocean climate variability Keyan Fang �� Xiaohua Gou �� Fahu Chen �� Jinbao Li �� Rosanne D���Arrigo �� Edward Cook �� Tao Yang �� Nicole Davi Received: 23 December 2008 / Accepted: 10 July 2009 / Published online: 6 August 2009 �� Springer-Verlag 2009 Abstract We present a Palmer Drought Severity Index reconstruction (r = 0.61, P \ 0.01) from 1440 to 2007 for the southeastern Tibetan Plateau, based on tree rings of the forest fir (Abies forrestii). Persistent decadal dry intervals were found in the 1440s���1460s, 1560s���1580s, 1700s, 1770s, 1810s, 1860s and 1980s, and the extreme wet epochs were the 1480s���1490s, 1510s���1520s, 1590s, 1610s���1630s, 1720s���1730s, 1800s, 1830s, 1870s, 1930s, 1950s and after the 1990s. Comparisons of our record with those identified in other moisture related reconstructions for nearby regions showed that our reconstructed droughts were relatively consistent with those found in other regions of Indochina, suggesting similar drought regimes. Spectral peaks of 2.3���5.5 years may be indicative of ENSO activity, as also suggested by negative correlations with SSTs in the eastern equatorial and southeastern Pacific Ocean. Signi- ficant multidecadal spectral peaks of 29.2���40.9 and 56.8��� 60.2 years were identified. As indicated by the spatial correlation patterns, the decadal-scale variability may be linked to SST variations in the northern Pacific and Atlantic Oceans. Keywords Drought Tree-ring PDSI SST Northeastern Tibetan Plateau 1 Introduction Investigating impacts of drought and flood extremes on human populations and the environment in monsoonal Asia provides great challenges. Although some modeling results predict an increase in the Asian summer monsoon rainfall due to global warming (Anderson et al. 2002 Meehl 1994), a decreased Indian summer monsoon was observed in the 1990s, coinciding with some of the warmest conditions over the past millennium (Kripalani et al. 2003). Current climate models show less success in predicting drought at the regional scale (Douville et al. 2006). One difficulty in understanding and predicting regional rainfall variability lies in the short climate records in monsoonal Asia (Bohner�� 2006). For example, over the Tibetan Plateau meteoro- logical data seldom extend back to the 1950s. This situa- tion could be greatly improved by using a paleoclimatic perspective, i.e. employing annual, exactly dated proxy records, such as tree-rings to reconstruct past climate (e.g. Buckley et al. 2007 D���Arrigo et al. 2005 Fan et al. 2008a Fang et al. 2009 Gou et al. 2007a, b Li et al. 2008 Sano et al. 2008 Shao et al. 2005 Wang et al. 2008 Zhang et al. 2003). The resulting paleoclimate records enable us to place recent climate conditions in a long-term context of climate change and allow us to evaluate the sensitivity of recent climate changes to natural or anthropogenic forcings (D���Arrigo and Wilson 2006). These long-term records can also allow for evaluation of decadal and multi-decadal variations and their statistical significances. Decadal to multi-decadal scale drought variations and characteristic modes of atmosphere���ocean circulation [e.g. Pacific Dec- adal Oscillation (PDO) and North Atlantic Oscillation (NAO)] can significantly impact climate of the globe (Buckley et al. 2007 D���Arrigo et al. 2005, D���Arrigo et al. 2006). However, physical mechanisms of these modes of K. Fang X. Gou (&) F. Chen T. Yang Key Laboratory of Western China���s Environmental Systems (MOE), Lanzhou University, 730000 Lanzhou, China e-mail: xhgou@lzu.edu.cn K. Fang e-mail: kf2278@columbia.edu K. Fang J. Li R. D���Arrigo E. Cook N. Davi Tree-Ring Laboratory, Lamont-Doherty Earth Observatory of Columbia University, New York, NY 10964, USA 123 Clim Dyn (2010) 35:577���585 DOI 10.1007/s00382-009-0636-2

variation are still not well understood, especially for the Asian continent (D���Arrigo et al. 2005, 2006 Hurrell 1995 Mantua et al. 1997). As a result of it vast size and elevation (average height more 4,000 m above sea level), the Tibetan Plateau (hereafter TP) strongly impacts the Asian summer mon- soon circulation by heating the troposphere (Bohner �� 2006 Duan et al. 2006 Webster et al. 1998). Numerous dendroclimatological studies have been conducted in the northeastern TP (Gou et al. 2007a, b, 2008 Li et al. 2008 Liu et al. 2006 Shao et al. 2005 Sheppard et al. 2004 Wang et al. 2008 Zhang et al. 2003), however, very limited research has been performed on the southern TP (Brauning �� 2001 Brauning �� and Mantwill 2004 Fan et al. 2008a, b). An improved tree-ring data network with dense spatial coverage and a long time-span for the southern TP is therefore needed to capture monsoon variability and to provide insights for the prediction of the Asian monsoon. Fan et al. (2008a) described a 350-year spring drought reconstruction for the southeastern TP using a Residual tree-ring chronology that highlights the high-frequency variations. In this study, we present a new 660-year ring- width ARSTAN chronology, which is better at retaining the low-frequency signals, from three forest fir (Abies forrestii) sites to the south of previous study areas on the southeastern TP. We herein use the most reliable period of this chronology to reconstruct annual drought variations since AD 1440. We use this reconstructed 568-year drought history to investigate the variability of regional drought and compare our reconstruction with others for Indochina, northern and northeastern TP. We also explore the drought variability in relation to Pacific and Atlantic climate vari- ability. In particular, we focus on decadal to multi-decal climate signals in our tree-ring drought reconstruction for the southeastern TP. 2 Materials and methods 2.1 Tree-ring data The three sampling tree-ring sites are located in the Hengduan Mountains, southeastern TP. The Hengduan Mountains appear to block moisture advecting from the southwest, resulting in sharply increased rainfall on the windward side and a rain shadow effect on the leeward side of the mountain ranges (Fig. 1). As shown in Fig. 1, from the southwestern lowlands in Myanmar to the TP the averaged January���December precipitation for 1951���2007 ranges from 50 to 350 mm per month based on the gridded GPCC full data reanalysis (v4, 0.5�� resolution). Our tree- ring sites are located along the leeward regions, where precipitation is around 75 mm/month (Fig. 1). According to the Deqin meteorological station during the period 1954���2000 (Table 1), maximum monthly precipitation is found in July (132 mm), and minimum precipitation in December (7 mm). The monthly averaged temperature for the Deqin station in this region is 6.4��C. The maximum monthly temperature is seen in July (13.7��C), and the minimum monthly temperature in January (-1.5��C). The seasonality of temperature and rainfall exhibits a mon- soonal-type pattern, with peaks of both temperature and rainfall occurring from June through August. Precipitation during the monsoonal season (June���August) accounts for 49.5% of the total annual precipitation. Tree-ring samples were collected at three sites (WXI, WEX and PTG) from both living and subfossil wood specimens of A. forrestii (Fig. 1), an endemic tree species in southwestern China (Farjon 1990). Ring-width cores were taken at breast height (1.3 m). In total, 43 cores from 24 trees at WXI, 41 cores from 25 trees at WEX, and 29 cores from 15 trees at PTG were collected (Fig. 1 Table 1). Core samples were mounted, air dried, and polished with fine abrasive papers. Exact calendar years of each growth ring were assigned by visual crossdating (Fritts 1976). The visually crossdated tree rings were then mea- sured and checked by the program COFECHA for quality control (Holmes 1983). A cubic smoothing spline function with a 50% cutoff of 200 year was fitted to all the raw ring widths to remove biological trends, which is around 67% of the mean segment length and allows preservation of low- frequency signals (Cook and Kairiukstis 1990). The tree- ring indices were calculated as residuals after performing an adaptive power transformation, which can stabilize the Fig. 1 Map of locations of the sampling sites and associated climate records, and contours for the mean January���December averaged GPCC v4 0.5 precipitation (mm/month) from 1951 to 2007 578 K. Fang et al.: Drought reconstruction for southeastern TP 123