1. Introduction

Ice cores play an important role in paleoclimate studies, being unparalleled in resolution and types of climate and environmental information that can be extracted (Reference DansgaardDansgaard, 1997; Reference Kreutz, Mayewski, Meeker, Twickler, Whitlow and PittalwalaKreutz and others, 1997; Reference ThompsonThompson and others, 1998, Reference Thompson, Yao, Mosley-Thompson, Davis, Henderson and Lin2000; Reference Fischer, Wahlen, Smith, Mastroianni and DeckFischer and others, 1999; Reference PetitPetit and others, 1999). Because the Tibetan Plateau is sensitive to climate changes (Reference Yao, Thompson, Jiao, Mosley-Thompson and YangYao and others, 1995b; Reference Feng, Tang and WangFeng and others, 1998; Reference Liu and ChenLiu and Chen, 2000; Reference Gou, Chen, Jacoby and YangGou and others, 2006), ice-core studies there are particularly important for examining past climate and environment changes.

The Guliya ice cap, west Kunlun Shan, Qinghai–Tibetan Plateau, is the deepest polar-type ice cap in central Asia, with an area of 376 km² (Reference Yao, Thompson, Shi, Qin and JiaoYao and others, 1997). Three ice cores with respective lengths of 308.7, 93.2 and 34.5 m were successfully taken at 6200–6700ma.s.l. on the ice cap (35˚17^’ N, 81˚29^’ E) (Reference ThompsonThompson and others, 1995; Reference Yao, Jiao and YangYao and others, 1995a) and the longest one reached the bedrock. The 308.7 m ice core at 6200 ma.s.l. was drilled (Fig. 1) using an electromechanical drill in a dry borehole to 200 m and a thermal drill with an alcohol–water mixture from 200 m to bedrock (308.6 m). The ice temperatures in the borehole are –15.6˚C, –5.9˚C and –2.1˚C, corresponding to depths of 10 m, 200 m and the bottom, respectively. The core contained no hiatus, and the visible horizontal layers were evident throughout the core. Measurements of mass balance in 1990 and 1991 indicate that the ice cap receives ∼200 mmw.e.a⁻¹ of accumulation (Reference ThompsonThompson and others, 1997; Reference Yao, Thompson, Shi, Qin and JiaoYao and others, 1997). Reference ThompsonThompson and others (1995, Reference Thompson1997) and Reference YaoYao and others (1996) described in detail the field and laboratory methods used in climatic and environmental studies of the Guliya ice core. Numerous papers have described the Guliya ice core and the information extracted from it (e.g. Reference Yao, Jiao and LiYao and others, 1994a, Reference Yao, Jiao and Yang1995a, Reference Yao1996; Reference ThompsonThompson and others, 1995, Reference Thompson1997; Reference Yang, Yao, He and ThompsonYang and others, 2000, Reference Yang, Yao, Wang and Gou2006a,c). This paper examines the relationship between temperature and precipitation over the past 300 years, based on the upper part of the 308.7 m length ice-core record, and suggests mechanisms that could explain the observed relationship.

Fig. 1. The location of the Guliya (G) ice cap and the site where the 308.6 m Guliya ice core was drilled. Drill site elevation 6200m a.s.l. (after Reference ThompsonThompson and others, 1997).

2. Relationship Between Temperature and Precipitation Recorded in the Guliya Ice Core

Much work on the Qinghai-Tibetan Plateau has proved that there is a positive relationship between δ¹⁸O and temperature (Reference Yao, Jiao and YangYao and others, 1995a, Reference Yao, Thompson, Jiao, Mosley-Thompson and Yangb, Reference Yao1996). Variation of δ¹⁸O values in the Guliya ice core has been used as a proxy for historical temperature changes over the Qinghai-Tibetan Plateau. Net snow accumulation reflects total precipitation onto the ice cap and can be measured in the Guliya ice core. The core site lies high in the accumulation area of the glacier. For most glaciers, snow and ice thicknesses measured well into the accumulation zone closely mirror actual total precipitation. Because little snow melts high on the Guliya ice cap, we use the net accumulation as the total precipitation (Reference YaoYao and others, 1996). The accumulation on Guliya ice cap was measured by means of (1) accumulation stakes, (2) visible stratigraphy in pits and (3) insoluble particulate beta and tritium horizons (Reference ThompsonThompson and others, 1995). The precipitation on Guliya ice core is estimated from the net accumulation. As shown by Reference YaoYao and others (1996), net accumulation on this ice cap closely approximates actual precipitation.

Figure 2 shows the yearly δ¹⁸O and net accumulation recorded in the Guliya ice core for the past 300 years. In order to examine the relationship between temperature and precipitation, Reference Yang, Yao, He and ThompsonYang and others (2000) applied the cumulative anomaly method, plotting the running total of the yearly differences between each year’s value and the long-term mean. This method clearly displays long-term trends. Applied to climatic time series, the cumulative anomaly indicates periods where negative anomalies predominate (downward slopes on the cumulative anomaly curve) and periods where positive anomalies predominate (upward slopes on the cumulative anomaly curve). Thus, cumulative anomaly curves indicate changing trends over time. The general slope of the cumulative anomaly, rather than its position relative to the zero, indicates the trend. Figure 3 shows the cumulative anomalies of δ¹⁸O and net accumulation recorded in Guliya ice core from AD1690 to 1991. Figure 3 demonstrates that over the period of this study the climate in the Guliya region exhibits only two dominant phases: warm when conditions are wet, and cold when conditions are dry. Since 300 years BP, the climate in the Guliya region can be separated into three periods: warm and wet from AD 1690 to the end of the 18th century; cold and dry from the 19th century to the 1930s; and warm and wet again from the 1940s to the present (Fig. 3).

Fig. 2. Yearly δ¹⁸O and net accumulation values recorded in the Guliya ice core for the past 300 years.

Fig. 3. The cumulative anomalies (the running total of deviations from the mean) of δ¹⁸O and net accumulation recorded in the Guliya ice core during AD1690–1991, displayed with displaced time axes (28 years) to illustrate the phase lag between similar oscillations.

Adjusting for a lag time of about 30 years, Figure 3 also highlights the similarity in variation of the trends of cumulative anomalies between temperature and precipitation. The correlation coefficient reaches around 0.8 (p < 0.001) when the variation of the cumulative anomaly of the net accumulation lags behind that of the temperature by about 20–40 years. When we take into account the lag in precipitation relative to the temperature, even small fluctuations in the secular changes appear to be very consistent between temperature and precipitation. The cross-correlation analysis of the temperature and precipitation yields the highest correlation coefficient (R = 0.97, p < 0.001) when precipitation lags by 28 years.

Our study also used the 11 year moving average of the temperature and precipitation to illustrate the relationship between temperature and precipitation (Fig. 4), and we calculated the lag correlation coefficients for the 11 year averages (Fig. 5). Figure 5 again shows the best correlation when the precipitation lags behind temperature by about 20–40 years. The highest correlation coefficient was attained (R = 0.57, p < 0.001) when precipitation lagged temperature by 29 years.

Fig. 4. The 11 year moving average of δ¹⁸O and net accumulation recorded in the Guliya ice core over the past 300 years, displayed with displaced time axes (29 years) to illustrate the phase lag between similar oscillations.

Fig. 5. The lag correlation coefficient between the net accumulation and the temperature. Data used for calculation are 11 year moving average from AD 1690 to 1990 (Fig. 4). Each lag step represents 1 year and the peak in the correlation coefficient (0.57) corresponding to the 29 year lag step.

These data pose the question why temperature and precipitation should be correlated in this manner. In other words, why are variations of precipitation so sensitive to the temperature changes and what mechanisms can explain the 20–40 year time lag of precipitation behind temperature?

3. Discussion

Reference YaoYao and others (1996) considered various mechanisms involving evaporation and transport of moisture from the ocean to the plateau. They recognized that the lag effect between variations of precipitation and temperature could relate to the large specific heat of the ocean, preventing a rapid response to external heating or cooling. This effect causes ocean temperatures to lag changes in the air temperature. Because ocean evaporation intensity depends on the ocean temperature, and precipitation depends on evaporation (and transport), precipitation will also lag behind the temperature changes.

The monsoon provides a linkage. The Indian monsoon strength is related to the temperature difference between ocean surface and the Tibetan Plateau (Reference Duan, Wang and PuDuan and others, 2002). When the climate warms, temperatures on the land and ocean surface increase and evaporation from the ocean surface strengthens. Transport of water vapor to the Tibetan Plateau intensifies, increasing precipitation over the plateau. When the climate cools, temperatures on the land and ocean surface decrease and evaporation from the ocean surface weakens. Water-vapor transport to the Tibetan Plateau diminishes and precipitation on the plateau decreases. Even for moisture from local convective clouds, climate warming increases precipitation because local convective activity intensifies, and climate cooling has the opposite effect (Reference Yao, Yang and LuYao and others, 1994b).

Additional factors may contribute to the lag of precipitation variations behind temperature changes. The following feedback processes should also be considered. When climate starts warming, the temperature on land will increase first, due to the difference in heat capacity between land and ocean. Because of its geographical position and sensitivity to climate change, Tibetan Plateau temperatures should increase particularly rapidly. Then the heat difference between land and ocean increases and the monsoon intensifies, which in turn intensifies water-vapor transport from ocean to the Tibetan Plateau. Previous studies (e.g. Reference Yao, Jiao and YangYao and others, 1995a) have noted that moisture sources in the Bay of Bengal, the Arabian Sea and local convective water vapor all influence precipitation on glaciers from the Guliya ice cap to the Tanggula mountains. These three moisture sources are closely related to the strength of the monsoon. When climate warms, precipitation on the Tibetan Plateau increases. As precipitation increases, the surface soil moisture also increases. The summertime climate on the Tibetan Plateau is relatively warm. Such warm/wet climate conditions benefit vegetation growth and water storage capacity, and decrease the runoff. Considering the water balance, enhanced land surface moisture along with high temperatures should increase evaporation and the local convective precipitation. In fact, because of the plateau’s geographical position, local convective activities are very strong in summer and local convective precipitation occurs frequently. The local convection recycles water vapor from the ocean continuously, and thus local convective precipitation contributes a large percentage (46.86% at least in the central Tibetan Plateau) of the total precipitation (Reference Yang, Yao, Wang, Tian and GouYang and others, 2006b).

When the climate cools, temperatures on the plateau cool first (due to the difference in heat capacity between land and ocean), the heat difference between land and ocean decreases and the monsoon weakens. Moisture transport from the ocean decreases and the precipitation on the plateau decreases. This causes the surface moisture to decrease. When it is cold over the Tibetan Plateau, the cold/ dry conditions cause the vegetation to diminish. The water storage capacity of the Tibetan Plateau declines and runoff increases. The water balance changes, with reduced surface moisture and lower temperatures decreasing evaporation and lowering the percentage of the local convective precipitation. Precipitation on the Tibetan Plateau then further decreases through this positive feedback process. The expansion or diminution of vegetation in response to changing temperature and precipitation conditions is not instantaneous and requires some time for the vegetative response. This response time, coupled with associated temperature and precipitation processes, may contribute to lag of precipitation behind temperature by 20–40 years.

4. Conclusions

The Guliya ice cap is an ideal site for acquiring ice cores for climate-change studies. The analysis of net accumulation and temperature proxy (δ¹⁸O) data covering the past 300 years archived in Guliya ice cores shows that precipitation correlates well with temperature. Over a long time period, positive precipitation anomalies correspond to positive temperature anomalies, and vice versa. However, changes in the precipitation regime lag behind changes in temperature by about 20–40 years. Likely reasons for this lag include differences between the specific heat of the oceans and the land, but other mechanisms also appear to be involved. A lag in the precipitation response following changes in temperature may arise from the slow warming of the ocean surface and its effect on evaporation. However, at the same time a slow vegetation response to temperature and precipitation affects retention of soil moisture, which in turn feeds back by affecting precipitation. Because variations in temperature precede those of precipitation, future trends in precipitation might be predictable based on current observations of variations of temperature trends.