Long-term climate variations in China and global warming signals

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 108, NO. D19, 4614, doi: /2003jd003651, 2003 Long-term climate variations in China and global warming signals Zeng-Zhen Hu Center for Ocean-Land-Atmosphere Studies, Calverton, Maryland, USA Song Yang Climate Prediction Center, NOAA, Camp Springs, Maryland, USA Renguang Wu Center for Ocean-Land-Atmosphere Studies, Calverton, Maryland, USA Received 1 April 2003; revised 18 June 2003; accepted 1 July 2003; published 11 October [1] In this work, the authors analyze the observed long-term variations of seasonal climate in China and then investigate the possible influence of increases in greenhouse gas concentrations on these variations by comparing the observations with the simulations of the second phase of the Coupled Model Intercomparison Project (CMIP2). The long-term variations of precipitation and temperature in China are highly seasonally dependent. The main characteristic of summer precipitation in China is a drying trend in the north and a wetting trend in the central part. The precipitation in winter shows an increasing trend in southern and eastern-central China. Interesting features have also been found in the transitional seasons. In spring, precipitation variations are almost opposite to those in summer. In autumn the precipitation decreases in almost the whole country except for the middle and lower reaches of the Yangtze River Valley. In addition, the seasonality of precipitation has become slightly weaker in recent decades in southern and eastern China. Pronounced warming is observed in the entire country in winter, spring, and autumn, particularly in the northern part of China. In summer a cooling trend in central China is particularly interesting, and cooling (warming) trends generally coexist with wetting (drying) trends. The correlativity between precipitation and temperature variations is weak in spring, autumn, and winter. It has also been found that the long-term climate variations in winter and summer in China may be connected to the warming trend in the sea surface temperature of the Indian Ocean. A comparison between the observed seasonal climate variations and the CMIP2 simulations of 16 models indicates that the observed long-term variations of winter, spring, and autumn temperature in China may be associated with increases in greenhouse gas concentrations. However, such a connection is not found for the summer temperature. The tremendous uncertainties among the models in precipitation simulations make it difficult to link the precipitation variations to global warming. INDEX TERMS: 1620 Global Change: Climate dynamics (3309); 3319 Meteorology and Atmospheric Dynamics: General circulation; 3309 Meteorology and Atmospheric Dynamics: Climatology (1620); 4215 Oceanography: General: Climate and interannual variability (3309); 1610 Global Change: Atmosphere (0315, 0325); KEYWORDS: global warming signals, Chinese climate, CMIP2 simulation Citation: Hu, Z.-Z., S. Yang, and R. Wu, Long-term climate variations in China and global warming signals, J. Geophys. Res., 108(D19), 4614, doi: /2003jd003651, Introduction [2] The climate variations in China are well known for their complexity due to the influence of many factors over a wide range of spatial and temporal scales. They are linked to global teleconnections and the El Niño-Southern Oscillation (ENSO) [Lau, 1992; Yang et al., 2002; Wu et al., 2003] and closely tied up with the variations of the Asian-Australian monsoon system. The climate in China differs significantly Copyright 2003 by the American Geophysical Union /03/2003JD from that in South Asia and Australia, although climate variations in these regions interact strongly with each other [Tao and Chen, 1957; Matsumoto, 1992; Ding, 1994; Lau et al., 2000; Wang et al., 2001]. The variability of Chinese climate is due largely to the activities of the summer and winter monsoons [Ding, 1994]. Precipitation and temperature anomalies, especially the summer and spring floods and droughts in China, are intimately related to the country s economy and people s lives. These floods and droughts have often been considered among the most severe natural disasters for the country. Thus the long-term climate variations in China and their possible connections to the increases in ACL 11-1

2 ACL 11-2 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE greenhouse gas concentrations are subjects of considerable scientific and practical interest. [3] There have already been a few observational studies on the long-term climate variations in China. Zhu and Wang [2002] found that an 80-year oscillation is an important component of the summer precipitation variations in East Asia. Hu [1997], Chang et al. [2000a, 2000b] and Gong and Ho [2002] demonstrated the existence of an interdecadal shift of the climate in China in Hu [1997] and Gong and Ho [2002] indicated that heating in the Indian Ocean may play a key role in the interdecadal variation of the summer climate in China through the changes in the Hadley cell and the subtropical high over the western Pacific. Chang et al. [2000a, 2000b] suggested that the interdecadal climate variations in China were related to interdecadal changes of the relationship between ENSO and the Chinese climate. Observational analyses of climate changes in East Asia during the last 100 years [Wang and Ye, 1993] showed a pronounced annual mean warming trend in northern China and a minor cooling trend in central China. Using stational data with a short interval, Chen et al. [1991] and Nitta and Hu [1996] found a similar trend pattern in Chinese summer climate. Wang and Gaffen [2001] show evidence of moisture increase over most of China in , and Zhai and Ren [1999] demonstrated the changes of maximum and minimum surface temperature in China in In addition, the past 50-year trend of climate in western China is also described by Qin [2002]. [4] Recently, coupled general circulation model (CGCM) simulations have shown that increases in greenhouse gas concentrations intensify the South Asian summer monsoon and its variability [Meehl and Washington, 1993; Hu et al., 2000a] and diminish the Asian winter monsoon [Hu et al., 2000b]. After analyzing model simulations of climate change, Hulme et al. [1994] also pointed out the potential impact of global warming on the variations of temperature and precipitation in East Asia. The projection of climate in western China to the future 50 years is also made by Qin [2002]. These studies raise the questions of what the observed long-term variations of the seasonal climate in China are and whether these variations are connected to global warming. [5] Clearly, present CGCMs have serious defects in simulating regional climate. For example, none of the 10 models that participated in the Climate Variability and Predictability (CLIVAR)/Monsoon GCM Intercomparison Project can realistically reproduce the observed Mei-yu rainband [Kang et al., 2002]. The model deficiencies and the shortage of observational data make the detection and explanation of the long-term climate variations in China extremely difficult and uncertain. Nevertheless, the second phase of the Coupled Model Intercomparison Project (CMIP2) [Meehl et al., 2000] provides additional data to explore the problem. [6] To understand the long-term climate trends in China, it is necessary to analyze the trends for each season, because of the differences observed for each season. Furthermore, the potential impact of global warming on the seasonal climate variations in China may be understood by comparing the observed climate trends with those in the CMIP2 simulations. In this work, we analyze the observed long-term variations of the seasonal-mean precipitation and temperature in China using station data. We also explain the projected seasonal climate variations and their uncertainty using the CMIP2 results. In section 2 we describe the main features of observed and simulated data used in this study. The observed long-term seasonal climate variations and the responsible mechanisms are investigated in section 3. Section 4 provides discussions of the projected climate change. The uncertainty of model simulations is also discussed in section 4 by analyzing the differences between individual simulations and by examining the composite features for some selected models. Section 5 provides a summary and further discussions of the results. 2. Observed and Simulated Data 2.1. Observed Data [7] We analyze the monthly precipitation and temperature data of 160 meteorological stations in China for the period from January 1951 to February The data were collected and edited by the China Meteorological Administration and were relatively homogeneously distributed, especially in eastern China (see Figure 1a). The temperature data have not been adjusted for urban warming biases. In addition, the observed monthly mean sea surface temperature (SST) data are applied to study the relationship of ocean warming with the climate variations in China. The SST data are the reconstructed data with a resolution of 2 2 [Reynolds and Smith, 1994] Simulated Data [8] Simulations of the CMIP2 experiments by 16 models [Meehl et al., 2000] are used in the analysis (Table 1). Each model simulation consists of a control run with constant present-day atmospheric CO 2 and a greenhouse run with a standard gradual (1% year 1 compound) increase in CO 2. Each experiment was run for 80 years. Details about the individual models and experiments can be found at www-pcmdi.llnl.gov/cmip/ or in the work of Räisänen [2001, 2002] and Räisänen and Palmer [2001], and the references therein. Our analysis is focused on the comparison of the 20-year means for years centered at the CO 2 doubling with the 80-year means of the control run. We choose seasonal mean precipitation and surface air temperature as the prime variables. 3. Observed Long-Term Climate Variation 3.1. Rainfall [9] Figure 1a shows the summer (JJA, June, July, and August) mean precipitation in China. It can be seen that the precipitation amount decreases from southern (>700 mm) to northern (<300 mm) China. This is the most basic feature of the summer climate in China as described by Ding [1994]. Figure 1b depicts the linear trend over , which is calculated by linear regression after suppressing short timescale disturbances through a 5-year low-pass filter. It shows a wetting trend in central China, especially in the middle and lower reaches of the Yangtze River Valley. Figure 1 also shows a drying trend in parts of north and northeast China. A drying trend is also evident along the eastern edge of the Tibetan Plateau. This result is in line with the investigations of Nitta and Hu [1996], and Weng et al. [1999]. The

3 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE ACL 11-3 summer (Figures 1b and 1c). An increase in winter precipitation is observed in some regions of south and southwest China and in the eastern part of central China, and a drying trend is evident in parts of north and northeast China (Figure 2b). The increase reaches 15 30% per 50 years, and the decrease is larger than 30% per 50 years (Figure 2c). The most significant drying trend is located in Inner Mongolia (Figure 2c), which has been a drought region for a long period. [11] Long-term variations are also pronounced in the transitional seasons. In spring (MAM, March, April, and May) the linear trend pattern is almost opposite to that of summer (compare Figures 3a and 1b), which means that seasonal partition of precipitation has been further shifting to summer. The drying trends appear mainly along the middle and lower reaches of the Yangtze River Valley, and the wetting trends are found along parts of the southern coast and parts of northeast and north China (Figure 3a). The amplitude of the decrease (increase) reaches 90 mm (15 mm) per 50 years. In autumn (SON, September, October, and November), a precipitation decrease is dominant for almost the whole country, except for central China and the western part of southwest China (Figure 3b). The major decrease in autumn precipitation extends from a broad area of central China to the southern part of northeast China with maximum values of mm per 50 years (Figure 3b). [12] The above analyses clearly demonstrate a strong seasonality in the long-term precipitation variations in China. The seasonality change is also an important element to measure climate change. Following the procedure of Walsh and Lawler [1981], we measure the long-term variation of the precipitation seasonality in China by a seasonality index (SI) expressed as: SI ¼ 1 P X 12 m¼1 jp m P 12 j; where P is the total precipitation in a year, and P m is the monthly precipitation for month m. According to the above equation, the value of SI in a year is the sum of the Figure 1. (a) Mean, (b) linear trend, and (c) ratio of the linear trend to the mean of summer precipitation in China during The contour intervals are 50 mm in Figure 1a, 50 mm per 50 years in Figure 1b, and 10% per 50 years in Figure 1c. Dark (light) shading marks the values greater (smaller) than 550 mm (450 mm) in Figure 1a and 10% per 50 years ( 10% per 50 years) in Figure 1c. The shading in Figure 1b represents significant linear trends at the significance level of 95% using the T-test. Solid dots in Figure 1a represent the station locations. increase in precipitation reaches 10 30% per 50 years in the middle and lower reaches of the Yangtze River Valley. On the contrary, a decrease of 10 20% is found in parts of north and northeast China (Figure 1c). [10] In winter (DJF, December, January, and February), the total precipitation amount (Figure 2a) is small for the whole country. The long-term trends (Figures 2b and 2c) have a relatively smaller spatial scale compared to those in Table 1. Sixteen Models Used in the Analyses Acronym Institute BMRC Bureau of Meteorology Research Center CCCma Canadian Centre for Climate Modeling and Analysis CCSR Center for Climate System Research CERFACS Centre Europeen de Recherche et de Formation Avancéen en Calcul Scientifique CSIRO Commonwealth Scientific and Industrial Research Organization ECHAM3 Max-Planck Institute for Meteorology ECHAM4 Max-Planck Institute for Meteorology GFDL Geophysical Fluid Dynamics Laboratory IAP Institute of Atmospheric Physics, Chinese Academy of Sciences INM Institute of Numerical Mathematics, Russian Academy of Sciences LMD Laboratoire de Metéorologie Dynamique, Institut Pierre Simon Laplace MRI Meteorological Research Institute NCAR National Center for Atmospheric Research PCM Department of Energy HadCM2 United Kingdom Met Office HadCM3 United Kingdom Met Office

4 ACL 11-4 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE The amplitudes of the decrease and increase are in the range of Temperature [13] Figure 5 shows the mean of summer temperature over (Figure 5a) and the linear trend over the period (Figure 5b). Mean temperature decreases from the southeastern coast to northeast China and the Tibetan Plateau. There are obvious negative trends in central China and positive ones elsewhere. The amplitude of cooling is larger than 0.3 C per 50 years in the middle reaches of the Yangtze River Valley, and the warming rate reaches 0.9 C per 50 years in part of north and south China. The cooling trend in central China, which contrasts sharply with the general warming trend of surface climate [Intergovernmental Panel on Climate Change, 2001], is particularly interesting. The cooling (warming) trends are generally colocated with the wetting (drying) trends in China shown in Figure 1b. This demonstrates the correlativity between the temperature and precipitation in summer, as pointed out by Nitta and Hu [1996]. [14] The long-term mean and variations of the winter temperature (Figure 6) show different features. First, the contrast between the mean temperature in northern and southern China is much larger in winter (Figure 6a) than in summer (Figure 5a). Second, a remarkable warming trend is dominant for almost the whole country, except for a minor cooling in a small part of southwest China Figure 2. Same as Figure 1 but for winter. The contour intervals are 30 mm in Figure 2a, 10 mm per 50 years in Figure 2b, and 15% per 50 years in Figure 2c. Dark (light) shading marks the values greater (smaller) than 120 mm (60mm) in Figure 2a and 15% per 50 years ( 15% per 50 years) in Figure 2c. absolute deviations of monthly mean precipitation from the overall mean divided by the total annual precipitation in that year. Figure 4 displays the SI averaged over (Figure 4a), (Figure 4b), and their differences (Figure 4c). The major feature in Figures 4a and 4b is that the mean seasonality increases from the south (<0.6) to the north (>1.1) of China. The differences in the mean SI between the first 20 years (Figure 4a) and the last 20 years (Figure 4b) show a slight decrease in the south and east of China (Figure 4c). Small increases exist in the western part of northeast China and Inner Mongolia. Figure 3. Same as Figure 1b but for (a) spring and (b) autumn precipitation. The contour interval is 30 mm per 50 years.

5 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE ACL 11-5 and 1.5 C per 50 years for autumn (Figure 7b) in northern China. Small cooling trends (> 0.5 C) are observed in part of southwest China. The above analyses show that the opposite trends of temperature and precipitation (the correlativity between temperature and precipitation) in China occur in summer but not in spring, autumn, and winter. Also, the warming trends are the largest in winter, and the smallest in summer Possible Connection With the Indian Ocean Sea Surface Temperature [16] Ocean temperature is an important factor in affecting the low-frequency variations of climate. Hu [1997] found that the interdecadal variability of the summer precipitation and temperature in East Asia is largely influenced by the changes of SST and convective activity over the tropical Indian Ocean and western Pacific. The convective activity is usually enhanced when the SST is warmer than normal, so the subtropical high over East Asia is intensified through the enhancement of the Hadley cell. As a result, subtropical East Asia, including south and southwest China and the Southwest Islands of Japan, is under the control of a positive height anomaly at 500 hpa that causes above normal temperature and below normal precipitation. The Figure 4. Seasonality index (SI) for monthly mean precipitation averaged over (a) , (b) , and (c) the difference between the two periods (Figures 4b and 4a). The contour intervals are 0.1 in Figures 4a and 4b and 0.02 in Figure 4c. Dark (light) shading marks the values greater (smaller) than 0.8 (0.6) in Figures 4a and 4b. Shading in Figure 4c denotes significant differences at the level of 95% using the T-test. See the text for the definition of SI. (Figure 6b). The warming rates reach values larger than 3.5 C per 50 years in some regions of northern China and <1.0 C per 50 years (or even minor cooling) in southwest China (Figure 6b). [15] In spring and autumn the long-term variations of temperature are dominated by obvious warming in most parts of the country (Figure 7), similar to the warming in winter (Figure 6b). The intensity of the warming trend in spring (Figure 7a) is larger than that in autumn (Figure 7b) but smaller than that in winter (Figure 6b). The warming rates are larger than 2.5 C per 50 years for spring (Figure 7a) Figure 5. Summer mean temperature in China (a) over and (b) the linear trends. The contour intervals are 2 C in Figure 5a and 0.3 C per 50 years in Figure 5b. Dark (light) shading marks the values greater (smaller) than 26 (22) C in Figure 5a. The shaded regions in Figure 5b denote significant linear trends at the significance level of 95% using the T-test.

6 ACL 11-6 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE Figure 6. Same as Figure 5 but for winter. The contour intervals are 2 C in Figure 6a and 0.5 C per 50 years in Figure 6b. Dark (light) shading marks the values greater (smaller) than 2 ( 2) C in Figure 6a. Figure 7. Same as Figure 6b but for (a) spring and (b) autumn. following analysis is devoted to discussions of a possible connection between the seasonal climate variations in China and the SST changes in the Indian Ocean. [17] Figure 8 shows the time series and the corresponding spatial patterns of the leading modes of empirical orthogonal function (EOF) analysis of the summer and winter mean SST in the Indian Ocean. Positive values are dominant in the spatial pattern with the maximum center in the subtropical part of the South Indian Ocean (Figures 8b and 8c). The evolution of the time series is consistent in summer and winter, and the warming trends become pronounced after the mid-1970s (Figure 8a). The time evolution and spatial patterns in spring and autumn (not shown) are similar to that in summer and winter. [18] The connection between the summer and winter climate variations and the Indian Ocean SST is demonstrated by the simultaneous correlations between the EOF time series and the precipitation and temperature in China (see Figure 9). The correlation is calculated using detrended and smoothed (a 5-year filter) data. Interestingly, the patterns of SSTprecipitation correlation (Figure 9) are generally similar to those of precipitation and temperature trends, particularly for the precipitation in East China (Figures 1b and 2b), the summer temperature in central and south China (Figure 5b), and winter temperature in north and central China (Figure 6b). These features suggest a connection between the SST anomaly in the Indian Ocean and the Figure 8. (a) Time series and (b) (c) the corresponding spatial patterns of the leading empirical orthogonal function (EOF) of summer and winter mean SST in the Indian Ocean. The contour intervals are 0.1 and shading marks negative values in Figures 8b and 8c (courtesy of Soo-Hyun Yoo).

7 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE ACL 11-7 a further understanding of the results by analyzing the individual simulations by selected models. We focus on the possible connection between the long-term climate variations in China and increases in greenhouse gas concentrations. [20] Figure 10 shows the 16-model means of precipitation and temperature for summer and winter, from the control runs. The models capture the general features observed reasonably well (compare Figure 10 and Figures 1a, 2a, 5a, and 6a). However, discrepancies exist in regional scale Figure 9. Simultaneous correlations between the time series of the leading EOF modes of the SST in the Indian Ocean and the precipitation and temperature in China over for (a) summer precipitation and SST, (b) winter precipitation and SST, (c) summer temperature and SST, and (d) winter temperature and SST. The contour intervals are 0.2. The shaded regions indicate significant correlations at the confidence level of 95%. low-frequency variations of Chinese climate as indicated by Hu [1997]. However, the statistical correlations between spring/autumn precipitation/temperature and the SST are not significant. The reason of the seasonal dependence in the link is not clear. 4. Simulated Climate Change [19] In this section, we first analyze the mean characteristics of the seasonal precipitation and temperature changes in the CMIP2 simulations by the 16 models. Then, we aim at Figure 10. The 16-model mean precipitation in (a) summer and (b) winter and mean temperature in (c) summer and (d) winter in the control runs. The contour intervals are 50 mm in Figure 10a, 30 mm in Figure 10b, 2 C in Figure 10c, and 4 C in Figure 10d. The dark (light) shading is used for values larger (less) than 550 mm (450 mm) in Figure 10a, 120 mm (60 mm) in Figure 10b, 38 C (34 C) in Figure 10c, and 12 C (4 C) in Figure 10d.

8 ACL 11-8 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE Figure 11. The (a) 16-model mean, (b) spread, and (c) ratios (mean/spread) of summer precipitation differences between 2 CO 2 (years 61 80) and the corresponding control runs (80 years). The contour intervals are 8 mm in Figures 11a and 11b and 0.1 in Figure 11c. The dark (light) shading is used for values larger (less) than 24 mm (8 mm) in Figure 11a, 56 mm (40 mm) in Figure 11b, and 0.5 (0.3) in Figure 11c. features. For example, the observed summer rainband on the southern side of the Yangtze River (Figure 1a) is not well simulated by the model mean (Figure 10a). This is a common problem in GCM modeling [Kang et al., 2002]. Another problem is that the model mean simulation gives a much warmer climatology (Figures 10c and 10d) compared with the observations (Figures 5a and 6a) The 16-Model Mean of Rainfall Change [21] Figure 11a shows the 16-model mean of the differences in summer precipitation between the global warming scenario runs in years and the corresponding control runs. The differences are referred as (2 CO 2 - control) in the following discussion. The major characteristics of the simulated influence of the increase in greenhouse gas concentrations include a wetting trend over almost all of China, except for western parts of Inner Mongolia. The wetting trend is even more pronounced in southwest and northeast China. This contrasts with the observed longterm variations of the summer precipitation shown in Figure 1b. Thus the observed summer precipitation variations in China might not directly link with increases in greenhouse gas concentrations. However, the enormous spread among the simulations (Figures 11b and 11c) makes the results of (2 CO 2 - control) very uncertain. The ratios of the 16-model mean to the spread are less than 1.0 in most regions, except for part of north and northeast China (Figure 11c). [22] Correspondingly, we also analyzed the precipitation simulations for winter, spring, and autumn (figures not shown). In the global warming scenario the winter precipitation increases in northern-central China and decreases along the southern and southeastern coast. Those features are different from those of the observed long-term precipitation change (See Figure 2b). The spread of the model simulation is also large, meaning a large uncertainty in the winter precipitation simulations in the global warming scenario. For spring, precipitation increases in the global warming scenario in almost all of the country, except for a small part of the southeastern coast. The maximum increase is in western-central China with a value of 36 mm, and the largest spread is in the middle and lower reaches of the Yangtze River Valley and the southeastern coast. In autumn, major wetting trends occur in the southern and eastern coast, and drying or small wetting trends appear from the middle reaches of the Yangtze River Valley to northeast China. The spreads are large in the middle and lower reaches of the Yangtze River Valley and the southern coast. In both spring and autumn, the simulations show tremendous disagreement with the observed long-term precipitation variations (Figure 3). However, the reduction of the seasonality in the global warming scenario is consistent with the observations shown in Figure 4. [23] Assuming that the mean represents the signal and the spread is the noise, it can be claimed that the ratios of signal to noise are very small in the seasonal precipitation change of ensemble simulations of the CMIP2 runs. The ratios are <0.5 in many places, demonstrating a tremendous uncertainty and thus a challenge in simulating the seasonal precipitation changes over China that may result from increases in greenhouse gas concentrations The 16-Model Mean of Temperature Change [24] The common features of seasonal temperature changes are characterized by warming trends over the whole country. The warming lies in the range of C in summer (Figure 12a), C in winter (Figure 13a), C in spring (not shown), and C in autumn (not shown). Interestingly, the patterns of spread are similar between summer (Figure 12b) and autumn and between winter (Figure 13b) and spring. The uncertainty increases from the southeastern coast to Inner

9 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE ACL 11-9 However, such a similarity cannot be found between the simulations and the observations for summer Simulations by Selected Models [26] The above analysis of the mean global warming simulations of the 16 models has indicated a similarity in the long-term change between the observed and simulated temperature in winter, suggesting a possible connection between increases in greenhouse gas concentrations and the observed changes in Chinese winter temperature. A similar but weaker feature has also been found for the temperature in spring and autumn. However, the analysis of the temperature in summer and precipitation in various seasons encounters difficulties in explaining the long-term Figure 12. Same as Figure 11 but for summer temperature. The contour intervals are 0.2 C in Figure 12a, 0.1 C in Figure 12b, and 0.2 in Figure 12c. The dark (light) shading is used for values larger (less) than 2.4 C (2.0 C) in Figure 12a, 1.0 C (0.8 C) in Figure 12b, and 3.0 (2.6) in Figure 12c. Mongolia in summer (Figure 12c) and from northeast to southwest China in winter (Figure 13c). Nevertheless, the ratios of the mean to the spread are in the range of , larger than those for the precipitation simulations discussed above. This implies that the simulations of the long-term temperature changes may be more reliable than those simulations for precipitation. [25] The simulated temperature change patterns in winter (Figure 13), spring, and autumn are similar to those observed (Figures 6b and 7), indicating that increases in greenhouse gas concentrations may play a role in the observed temperature variations in China in these seasons. Figure 13. Same as Figure 11, but for winter temperature. The contour intervals are 0.2 C in Figure 13a, 0.1 C in Figure 13b, and 0.2 in Figure 13c. The dark (light) shading is used for values larger (less) than 2.4 C(2.0 C) in Figure 13a, 1.0 C (0.8 C) in Figure 13b, and 3.0 (2.6) in Figure 13c.

10 ACL HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE changes of these fields under the global warming scenario. It is possible that these changes are more strongly linked to other factors, such as the warming trend of the Indian Ocean, as discussed in section 3, and the Pacific Oceans, which themselves may be part of the global warming. On the other hand, the above analyses have also revealed a large uncertainty in the simulations of Chinese precipitation and summer temperature by the 16 models. [27] It is well known that the performance in climate modeling by even the state-of-the-art GCMs differs significantly from one model to another. While one model simulates a specific field for a particular season most successfully, other models may be better in simulating other fields during other seasons. Therefore it is important to analyze the precipitation and temperature simulations for the various seasons by each individual model and discuss the salient features for selected models. It is important to know which models are better in simulating regional climate and its change, even if the reasons for the differences are not fully clear. [28] Figure 14 shows the changes in summer precipitation simulated by CCSR (Figure 14a), ECHAM4 (Figure 14b), HadCM2 (Figure 14c), and the mean of the three models Figure 14. Summer precipitation differences between 2 CO 2 (years 61 80) and the corresponding control runs (80 years) of (a) CCSR, (b) ECHAM4, (c) HadCM2, and (d) the mean of the three models. The contour intervals are 0.2 mm. The dark (light) shading is used for values larger (less) than 0.4 mm (0.0 mm). Figure 15. Same as Figure 14 but for winter precipitation of (a) ECHAM3, (b) IAP, (c) HadCM3, and (d) the mean of the three models.

11 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE ACL minor wetting trend in the north, which are the major features of the observations (Figures 2b and 2c), are reasonably captured by these models. However, the detailed regional features are not well simulated by the models. This is not surprising since these three models are quite different in many features, including resolutions, physical processes, land surface schemes, and ocean-atmosphere coupling strategy. [30] ECHAM3, NCAR, and PCM are the three models (Figure 16) closest to simulating the observed winter temperature variations (Figure 6b). They all suggest that increases in greenhouse gas concentrations may be an important contributor to the observed winter temperature variations. However, no model can reasonably simulate the observed change in summer temperature including the cooling in central China shown in Figure 5b. Figure 16. Winter temperature differences between 2 CO 2 (years 61 80) and the corresponding control runs (80 years) of (a) ECHAM3, (b) NCAR, (c) PCM, and (d) mean of the three models. The contour intervals are 0.3 C. The dark (light) shading is used for values larger (less) than 2.7 C (2.1 C) in Figure 16a and 2.1 C (1.5 C) in Figures 16b, 16c, and 16d. (Figure 14d). These models generally reproduce the major features of the observed precipitation (Figures 1b and 1c), although large differences exist. Model documentations show that one of the common features of these models is the use of water and heat flux adjustments. [29] Figure 15 shows the winter precipitation changes simulated by ECHAM3 (Figure 15a), IAP (Figure 15b), HadCM3 (Figure 15c) models, and the mean (Figure 15d). The wetting trend in southern China and the drying or 5. Summary and Discussion [31] In this study, we have analyzed the observed longterm variations of the seasonal climate in China. We have also investigated the possible linkage of these variations to increases in greenhouse gas concentrations by analyzing the CMIP2 simulations by the 16 models. [32] The long-term climate variations in China have a strong seasonality. The main characteristic of summer precipitation variation is an opposite trend between central and northern China, including a drying trend in the north and a wetting trend in central China. In winter, precipitation increases in south and southwest China and in the southern parts of north and northeast China. On the other hand, a drying trend is evident in the northern parts of north and northeast China. In spring the precipitation variations are almost opposite to the summer patterns. In autumn the Chinese precipitation decreases over almost the whole country, except for the middle and lower reaches of the Yangtze River Valley. The seasonality has become weaker in southern and eastern China in recent decades. Pronounced warming dominates in the whole country in winter, spring, and autumn, particularly in northern China. In summer the cooling trend in central China, which contrasts sharply with the general warming trend of surface climate in spring, autumn, and winter, is particularly interesting. In summer, cooling (warming) trends generally coexist with the wetting (drying) trends. However, the correlativity between precipitation and temperature variations is weak in spring, autumn, and winter. The long-term climate variations in summer and winter in China may be connected to the warming trend in the SST of the Indian Ocean. The SST anomaly in the Indian Ocean affects the climate and general circulation through changes in the Hadley circulation. The detailed physical processes and responsible mechanisms need to be better understood through future studies. [33] A comparison of the observed seasonal climate variations with the ensemble mean of CMIP2 simulations from 16 models indicates that the observed variations of winter temperature in China could be largely affected by increases in greenhouse gas concentrations. However, such a relationship cannot be found in summer. The tremendous uncertainties among the models in the precipitation simulations make the explanation of the observed variations difficult. The lack

12 ACL HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE of the influences from changes of snow cover, soil moisture, land use, solar radiation, volcano activity, and so on may also be a reason for the divergence among the model simulations. The urban heating island effect is also an important factor in affecting the detection of possible global warming signal from the observed temperature change in China [Zhao, 1991; Portman, 1993; Hulme et al., 1994]. [34] It should be pointed out that the linear trend of the climate does not necessarily equal the global warming signal. First, the increase of greenhouse gas concentrations was not linear within the twentieth century. Second, the interdecadal variability of the climate is pronounced in China, especially for precipitation variations [Wang et al., 2000]. The amplitudes of the trends shown in Figures 1 3 and 5 7 depend on the time interval examined [Wang et al., 1998, 2000]. In addition, since there may be a lowfrequency variation on the timescale of years in Chinese climate [Schlesinger and Ramankntty, 1994], the trends in temperature found for may be in part attributed to the natural variability. Therefore there still exist many challenges in identifying appropriately the global warming signal from observed climate records. Despite the shortcomings of the models, comparison between observed climate change in China and the simulations of the climate in the twentieth century (C20C) will hopefully help in understanding the observed climate change in China better as the models improve their representations of C20C. [35] For the variations of the summer precipitation and temperature in central China, there have been a few works providing reasonable explanations, in which links between the climate variations and the changes of aerosols were established [Qian and Giorgi, 1999; Menon et al., 2002; Kaiser and Qian, 2002]. For example, by analyzing climate model simulations, Menon et al. [2002] found that a large amount of black carbon (soot) particles and other pollutants caused changes in the precipitation and temperature over China and may be partially responsible for the tendency toward the increasing floods in the south, droughts in the north, and the cooling in central China during the last several decades. The mechanism suggested is that the black carbon can affect regional climate by absorbing sunlight, heating the air, and thereby altering the large-scale atmospheric circulation and the hydrologic cycle. [36] Acknowledgments. The authors thank editor A. Robock and anonymous reviewers for their comments to significantly improve the manuscript. The authors are indebted to J. Meehl, D. Straus, E. Schneider, L. Bengtsson, and R. Stouffer for their suggestions, to C. Covey for his help in getting the CMIP data, and to Jinhong Zhu and Dao-Yi Gong for providing the station data in China. This work was supported by grants from the National Science Foundation (ATM ) and the U.S. Department of Energy (De-FG02-01ER63256). All CMIP2 modeling groups are acknowledged for making the simulations available. CMIP is supported and the model data are distributed by the Program for Climate Model Diagnosis and Intercomparison (PCMDI) at the Lawrence Livermore National Laboratory (LLNL). References Chang, C.-P., Y. Zhang, and T. Li, Interannual and interdecadal variations of the East Asian summer monsoon and tropical Pacific SSTs, part 1: Roles of the subtropical ridge, J. Clim., 13, , 2000a. Chang, C.-P., Y. Zhang, and T. Li, Interannual and interdecadal variations of the East Asian summer monsoon and tropical Pacific SSTs, part 2: Meridional structure of the monsoon, J. Clim., 13, , 2000b. Chen, L., et al., Tentative analysis of climate change in China in last 40 years (in Chinese), Quat. J. Appl. Meteorol., 2, , Ding, Y., Monsoons Over China, 419 pp., Kluwer Acad., Norwell, Mass., Gong, D.-Y., and C.-H. Ho, Shift in the summer rainfall over the Yangtze River valley in the late 1970s, Geophys. Res. Lett., 29(10), 1436, doi: /2001gl014523, Hu, Z.-Z., Interdecadal variability of summer climate over East Asia and its association with 500 hpa height and global sea surface temperature, J. Geophys. Res., 102, 19,403 19,412, Hu, Z.-Z., M. Latif, E. Roeckner, and L. Bengtsson, Intensified Asian summer monsoon and its variability in a coupled model forced by increasing greenhouse gas concentrations, Geophys. Res. Lett., 27, , 2000a. Hu, Z.-Z., L. Bengtsson, and K. Arpe, Impact of global warming on the Asian winter monsoon in a coupled GCM, J. Geophys. Res., 105, , 2000b. Hulme, M., Z. C. Zhao, and T. Jiang, Recent and future climate change in East Asia, Int. J. Climatol., 14, , Intergovernmental Panel on Climate Change, Climate Change 2001: The Scientific Basis, Contribution of Working Group 1 to the Third Assessment Report of the Intergovernment Panel on Climate Change, edited by J. T. Houghton et al., 881 pp., Cambridge Univ. Press, New York, Kaiser, D. P., and Y. Qian, Decreasing trends in sunshine duration over China for : Indication of increased haze pollution?, Geophys. Res. Lett., 29(21), 2042, doi: /2002gl016057, Kang, I.-S., et al., Intercomparison of the climatological variations of Asian summer monsoon precipitation simulated by 10 GCMs, Clim. Dyn., 19, , Lau, K.-M., East asian summer monsoon rainfall variability and climate teleconnection, J. Meteorol. Soc. Jpn., 70, , Lau, K.-M., K. M. Kim, and S. Yang, Dynamical and boundary forcing characteristics of regional components of the Asian summer monsoon, J. Clim., 13, , Matsumoto, J., The seasonal changes in Asian and Australian monsoon regions, J. Meteorol. Soc. Jpn., 70, , Meehl, G. A., and W. M. Washington, South Asian summer monsoon variability in a model with doubled atmospheric carbon dioxide concentration, Science, 260, , Meehl, G. A., G. J. Boer, C. Covey, M. Latif, and R. J. Stouffer, The Coupled Model Intercomparison Project (CMIP), Bull. Am. Meteorol. Soc., 81, , Menon, S., J. Hansen, L. Nazarenko, and Y. Luo, Climate effects of black carbon aerosols in China and India, Sciences, 297, , Nitta, T., and Z.-Z. Hu, Summer climate variability in China and its association with 500 hpa height and tropical convection, J. Meteorol. Soc. Jpn., 74, , Portman, D. A., Identifying and correcting urban bias in regional time series: Surface temperature in China s northern plains, J. Clim., 6, , Qian, Y., and F. Giorgi, Interactive coupling of regional climate and sulfate aerosol models over eastern Asia, J. Geophys. Res., 104, , Qin, D., (Ed.), Evaluation of the Environmental Changes in western China (in Chinese), pp. 1 80, China Sci. Press, Beijing, Räisänen, J., CO 2 -induced climate change in CMIP2 experiments: Quantification of agreement and role of internal variability, J. Clim., 14, , Räisänen, J., CO 2 -induced changes in interannual temperature and precipitation variability in 19 CMIP2 experiments, J. Clim., 15, , Räisänen, J., and T. N. Palmer, A probability and decision-model analysis of multimodel ensemble of climate change simulations, J. Climate, 14, , Reynolds, R. W., and T. M. Smith, Improved global sea surface temperature analyses using optimum interpolation, J. Clim., 7, , Schlesinger, M. E., and N. Ramankntty, An oscillation in the global climate system of period years, Nature, 367, , Tao, S., and L. Chen, The structure of the summer circulation over Asia, Acta Meteorol. Sin., 28, , Walsh, P. D., and D. N. Lawler, Rainfall seasonality: Description, spatial patterns and change through time, Weather, 36, , Wang, B., R. Wu, and K.-M. Lau, Interannual variability of the Asian summer monsoon: Contrasts between the Indian and the western North Pacific-East Asian monsoons, J. Clim., 14, , Wang, S. W., and D. Z. Ye, Analysis of the global warming during the last one hundred years, in Climate Variability, edited by Ye et al., pp , China Meteorol. Press, Beijing, Wang, S. W., J. Ye, D.-Y. Gong, J. Zhu, and T. Yao, Construction of annual mean temperature series for the last one hundred years in China (in Chinese), Quat. J. Appl. Meteorol., 9(4), , 1998.

13 HU ET AL.: GLOBAL WARMING IN CHINESE CLIMATE ACL Wang, S. W., D. Y. Gong, J. Ye, and Z. Chen, Seasonal precipitation series of eastern China since 1880 and variability (in Chinese), Acta Geogr. Sin., 55(3), , Wang, X. L., and D. J. Gaffen, Late twentieth century climatology and trends of surface humidity and temperature in China, J. Clim., 14, , Weng, H.-Y., K.-M. Lau, and Y.-K. Xue, Multi-scale summer rainfall variability over China and its long-term link to global sea surface temperature variability, J. Meteorol. Soc. Jpn., 77, , Wu, R., Z.-Z. Hu, and B. P. Kirtman, Evolution of ENSO-related rainfall anomalies in East Asia and the processes, J. Clim, in press, Yang, S., K.-M. Lau, and K.-M. Kim, Variations of the East Asian jet stream and Asian-Pacific-American winter climate anomalies, J. Clim., 15, , Zhai, P., and F. Ren, On changes of China s maximum and minimum temperature in , Acta Meteorol. Sin., 13, , Zhao, Z., Temperature changes and urbanization in China for the last 39 years (in Chinese), Meteorol. Month., 17(4), 14 17, Zhu, J., and S. Wang, 80 yr oscillation of summer rainfall over north China and East Asian summer monsoon, Geophys. Res. Lett., 29(14), 1672, doi: /2001gl013997, Z.-Z. Hu and R. Wu, Center for Ocean-Land-Atmosphere Studies, 4041 Powder Mill Road, Suite 302, Calverton, MD 20705, USA. (hu@cola. iges.org) S. Yang, Climate Prediction Center, NOAA, 5200 Auth Road, Camp Springs, MD 20746, USA. (song.yang@noaa.gov)