The impacts of climate change on water resources and agriculture in China

REVIEWS The impacts of climate change on water resources and agriculture in China Shilong Piao1, Philippe Ciais2, Yao Huang3, Zehao Shen1, Shushi Peng1, Junsheng Li4, Liping Zhou1, Hongyan Liu1, YuecunMa1,YihuiDing5,PierreFriedlingstein2,6,ChunzhenLiu7,KunTan1,YongqiangYu3,TianyiZhang3&JingyunFang1 China is the world’s most populous country and a major emitter of greenhouse gases. Consequently, much research has focused on China’s influence on climate change but somewhat less has been written about the impact of climate change on China. China experienced explosive economic growth in recent decades, but with only 7% of theworld’s arable land available to feed 22% of the world’s population, China’s economy may be vulnerable to climate change itself. We find, however, that notwithstanding the clearwarming that has occurred inChina in recent decades, current understanding does not allow a clear assessment of the impact of anthropogenic climate change on China’s water resources and agriculture and therefore China’s ability to feed its people. To reach a more definitive conclusion, future work must improve regional climate simulations— especially of precipitation—and develop a better understanding of the managed and unmanaged responses of crops to changes in climate, diseases, pests and atmospheric constituents. C limate change and its impacts on water resources and crop production is a major force with which China and the rest of the world will have to cope in the twenty-first century1,2. In China, despite the growing importance of industry, agri- culture has a central role in ensuring the food security and welfare of 1.3 billion people. At first glance, a map of China’s climate and eco- systems (Fig. 1) reveals an uneven distribution of water resources between the south, where water is abundant, and the drier north. Many regions lie in transitional zones where water resources, and hence agricultural production, could be affected positively or nega- tively by changes in climate. Over the past several decades, China has already experienced some devastating climate extremes2. For instance, the great flood of 1998 inundated213106hectares (21Mha)of landanddestroyed fivemillion houses in the Yangtze basin, causing an economic loss of over US$20 billion (ref. 3). Despite the enormous importance of the subject and the growing number of specific studies, multidisciplinary synthesis of the knowledge of climate impacts in China is scarce2. Our primary goal here is to review observations of climate, hydro- logy and agricultural production trends in China, and associate these observationswith likely future changes.We highlight themain areas of vulnerability and sources of uncertainty based on recent literature and publisheddata.Wepresent an analysis progressing fromwell-observed recent trends to more uncertain model projections and mechanisms. The first section deals with recent climate change observations and projections from climate models. Particular attention is given to drought and flood extremes. The second section addresses past and projected future trends in water resources, investigates whether the recent changes are unusual or within the bounds of normal climatic variability, and assesses the contribution of human withdrawals of water versus climate forcing. In this context, we review changes in glacier mass balance and their impact on hydrosystems. The last section integrates climate and atmospheric composition impacts on agricultural production, and the role of agricultural adaptation poten- tials, within a more conceptual and speculative framework. Through this analysis, we show that China’s climate has clearly warmed since 1960, with an increased frequency of heatwaves, and that glaciers are in retreat. But the geographic and interannual variability in water resources is so large, and the improvements of cropmanagement have been so important, that they prevent a clear conclusion on the net 1Department of Ecology, Department of Geography, College of Urban and Environmental Science, Key Laboratory for Earth Surface Processes of theMinistry of Education, and Center of Climate Research, Peking University, Beijing 100871, China. 2Laboratoire des Sciences du Climat et de l’Environnement, UMR CEA-CNRS-UVSQ, Batiment 709, CE L’Orme des Merisiers, Gif-sur-Yvette, F-91191, France. 3State Key Laboratory of Atmospheric Boundary Layer Physics and Atmospheric Chemistry, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China. 4Chinese ResearchAcademy of Environmental Sciences, Beijing 100012, China. 5Laboratory of Climate Studies, National Climate Center, ChinaMeteorological Administration, No. 46 ZhongguancunNaDa Jie, Beijing 100081, China. 6QUEST, Department of Earth Sciences, University of Bristol, Bristol, BS8 1RJ, UK. 7China Water Information Center (Hydrological Bureau), Lane 2 Baiguang Road, Beijing 100053, China. Decrease in PDSI Tarim River Yellow Yangtze Ri ve r Pearl River R iver Ri ve rLia oh e Increase in PDSI Deserts Grasslands Cultivated – 1 year 1 harvest Cultivated – 2 years 3 harvests or 1 year 2 harvests Forests and shrubs Cultivated – 1 year 3 harvests Figure 1 | Distribution of vegetation across China. The vegetation distribution reflects present-day climate gradients. The vast area covered by agriculture and regions with different crop rotations are given in green. The red dots represent the areas with a significant (P, 0.05) increase in drought expressed by the Palmer Drought Severity Index (PDSI; the higher the index the less drought) during the period 1960–2005 (see text). The green dots indicate the areas where a decrease in drought was observed. Annual PDSI at spatial resolution of 2.5u is from ref. 20. Inset, islands in area below map. Vol 467j2 September 2010jdoi:10.1038/nature09364 43 Macmillan Publishers Limited. All rights reserved©2010 zhangmk Highlight zhangmk Highlight zhangmk Highlight zhangmk Highlight zhangmk Highlight impact of historical climate changeon agriculture.Climate simulations point to serious potential vulnerabilities in China’s future agricultural security, but extensive uncertainties prevent a definitive conclusion. Evidence and projection of climate trends Stronger warming in the north. A strong warming4,5 of China over the past five decades is firmly supported by continuous measure- ments from 412 meteorological stations. The temperature has increased by 1.2 uC since 1960. The seven warmest years all occurred during the last decade (see Box 1).Winter warming (0.04 uCper year) is about four times the rate of summer warming (0.01 uC per year), and thus the temperature seasonal cycle amplitude has decreased by 0.03 uC per year (see Box 1). Moreover, northern China is warming faster than southern China4. Temperatures reconstructed over the past thousand years using tree-ringwidth confirm that the last century was the warmest period since 1600, although its temperatures are comparable to temperatures from the Medieval Warm Period2,6. As for future projections, IPCC global climate models tell us un- ambiguously that the warming trend will continue, but uncertainties about its extent and pace are large1 (see Box 1). China’s average tem- perature is estimated to increase further by 1–5 uCby2100 (ref. 1). This 4 uC range reflects not only uncertainty in IPCC greenhouse gas emis- sion economic scenarios7 (a range of 2 uC), but also the spread among climate models when forced by the same scenario1 (a range of 3 uC). Beyond mean annual values, impact studies need projections of seasonal temperature change. In looking at the output of 24 IPCC models1, we found a much stronger future warming rate in summer (from 0.0216 0.008 uC per year in the IPCC B1 scenario7 to 0.0496 0.009 uC per year in the IPCC A2 scenario7) than is currently observed1. Such a pronounced summer warming would inevitably enhance evapo-transpiration, increasing the risk of water shortage for agriculture. Increased rainfall contrast between northeastern and southern China. Precipitation in eastern China exhibits decadal-scale variability, forced by the East Asian and Indianmonsoons8.We analysed data from 355 rain gauge stations and observed no significant long-term trend in country-average precipitation since 1960 (see Box 1). However, there are significant regional precipitation trends (see Box 1). The drier regions of northeastern China (including North China and Northeast China) are receiving less and less precipitation in summer and autumn (a 12% decline since 1960). By contrast, the wetter region of southern China is experiencing more rainfall during both summer and winter. Similar regional summer precipitation trends are expected from the probableweakening of the summermonsoon since the late 1970s9,10. So far, the changes appear to fall within the bounds of normal decadal variability of rainfall (see Box 1). Future projections of precipitation by IPCC climate models1 are highly uncertain (see Box 1). For instance, in northern China, where a decrease in precipitation is observed today (see above) the models surprisingly project an increase in summer precipitation of 76 7% above2000–2006 levelsby2100 (ref. 1) (under the IPCCA1B scenario7). Models logically simulate a globally more intense hydrological cycle when forced by increasing greenhouse gases11, but over a region like northern China, they may not accurately reflect synoptic and oro- graphic rainfall processes12, nor regional climate forcing by dust and pollution aerosols4. In the light of this case study, one can appreciate that to reconcile the observed temperature and precipitation trends with future pro- jections for China remains a major scientific challenge. This can be addressed by using regional models fitted with aerosol and chemistry effects on climate and improved description of land–atmosphere feedback processes, to enable improved impact studies and to design cost-effective adaptation measures4. A country of drought and floods. China is at risk from heavy rainfalls, heatwaves and drought5,13,14. Heatwaves have occurredmore frequently during the past 50 years, except over central China15. A significant reduction of cold days in winter has also been observed16. Trends in heavy rainfall events causing floods show high spatial heterogeneity13,14. These extreme events seem tobecomemore frequent over northwestern China and the mid- to lower reaches of the Yangtze River, but less frequent in northeastern China and the northwestern Yangtze River13. Meanwhile, a general decrease in the number of rainy days has been observed across the entire country (see Box 1). According to regional climatemodels, the frequency of heatwaves and rainfall extremes in the future may increase over most of the country17. Drought is one of the most severe manifestations of climate vari- ability in China. It is a source of concern for agriculture and human life, given that the country is already quite dry18 (3.323 106 km2 of drylands). Over the past six decades, very severe droughts hit China in the 1960s, in the late 1970s and early 1980s, and in the late 1990s19. Recently, northeastern China has suffered particularly from drought20,21 while, surprisingly, arid regions of northwestern China have enjoyed less-severe droughts (Fig. 1), as indicated by rising lake levels and increased vegetation cover on desert margins22. A key question is whether northeastern China will continue to suffer from drought in the future. Here again, results from climate models and scenarios indicate a large range of uncertainty. Under the IPCC A1B scenario7, it is predicted that the recently observed dipole of drier northeast China and wetter northwest China may further intensify23. In contrast, under the IPCC B1 scenario7, a decrease of drought in northeast China is projected24. To understand and project drought occurrences better, we need to pay more attention to the effects of soil moisture feedbacks on climate25. Analysing the big droughts of the twentieth century26 should further help researchers to identify key drought traits such as duration, intensity and extent, the processes that affect ecosystems and water resources most adversely, and the differences in regional responses. a b 600 800 1,000 1,200 1960 1970 1980 1990 2000 0 20 40 60 80 100 Yangtze River Yellow River A nn ua l s tr ea m flo w (k m 3 yr –1 ) Figure 2 | Observed inter-annual variation in annual runoff in two major Chinese rivers. a, Observed inter-annual variation in the Yangtze River annual runoff at the Datong station (red dot on China map in inset; grey shading indicates the area of the YangtzeRiver basin) from1960 to 2000. The red dotted line is the fit to the Datong data: y5 2.05x2 3,172 (R2 5 0.05, P5 0.16). b, Observed inter-annual variation in the Yellow River annual runoff at the Lanzhou (upper basin), Huayuankou (lower basin), and Gaocun (lower basin) stations (green, blue and red dots on China map in inset; grey shading indicates the area of the Yellow River basin) from 1960 to 2000. The green dotted line is the fit to the Lanzhou data: y5 20.27x1 560 (R2 5 0.19, P, 0.01). The blue dotted line is the fit to the Huayuankou data: y5 20.7x1 1,434 (R2 5 0.31, P, 0.01). The red dotted line is the fit to the Gaocun data: y5 20.87x1 1,764 (R2 5 0.44, P, 0.01). The decreasing trend in Yellow River annual runoff is at least partially induced by climate change (see text). A linear regression t-test was conducted to determine whether the slope of the regression line differed significantly from zero. REVIEWS NATUREjVol 467j2 September 2010 44 Macmillan Publishers Limited. All rights reserved©2010 zhangmk Highlight zhangmk Highlight zhangmk Highlight zhangmk Highlight Box 1 jClimate change in China Although China’s overall mean annual temperature has significantly increased over the past five decades (Box 1 Figure a), there are remarkable regional contrasts (Box 1 Figure c). The largest warming is found in northeast China, with a trend of 0.36 uC per decade, and Inner Mongolia, with 0.4 uC per decade. The smallest warming trend is found over southwest China with a trend of 0.15 uC per decade (possibly related to the cooling effects of increasing aerosol content100). Annual precipitation trends in China for the period 1960–2006 show strong differences between northeastern (decrease), northwestern (increase) and southeasternChina (increase). The northeastern decrease ismostly causedby the decrease in summer andautumnprecipitation,while the southeastern increasemainly results froman increase in summer andwinter precipitation (Box 1 Figure d). Seasonal changes inprecipitationpatterns are also apparent. In autumn, most regions except the Qinghai–Xizang Plateau show a decrease in precipitation. In stark contrast, winter has experienced an increase in precipitation across China, particularly in northwestern China (16% per decade) and Qinghai–Xizang Plateau (14% per decade). We have analysed past trends in heatwaves from long-term meteorological observations. We defined heatwaves as June–August days with temperatures exceeding the 90th percentile with respect to the 1960–2006 reference period15. An increase in frequency of heatwave events occurred acrossmost of China, except in central China. TheQinghai–Xizang Plateau and coastal regions of southern China show the largest increase in frequency of heatwave events (over two days per decade). Most of China has experienced a decrease in the annual number of raindays, particularly in the southwest and northeastern part. In southeastern China, the decrease in the annual number of raindays is coincident with an increase in annual precipitation, implying an increase in rainfall intensity13. 1960 1970 1980 1990 2000 –100 0 100 200 300 400 500 D iff er en ce fr om 1 96 0– 20 06 (m m ) –1 –0.5 0 0.5 1 Tr en d (m m p er y ea r) Tr en d (° C p er y ea r) * * 80 90 100 110 120 130 20 30 40 50 * * * * –1 0 1 2 –1 0 1 2 –1 0 1 2 –1 0 1 2 –1 0 1 2 –1 0 1 2 –1 0 1 2 –1 0 1 2 –1 0 1 2 * * * * Autumn Winter Spring Summer Annual 80 90 100 110 120 130 20 30 40 50 –12 –8 –4 –1 0 1 4 8 12 1960 1970 1980 1990 2000 –1 0 1 2 3 4 5 6 7 a b c d e f D iff er en ce fr om 1 96 0– 20 06 (° C ) Year Year Sp rin g Su m m er Au tu m n W int er An nu al Sp rin g Su mm er Au tum n W int er An nu al 0 0.01 0.02 0.03 0.04 0.05 B1 B1 A1B A1B A2 A2 * * * * * 80 90 100 110 120 130 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 0 0.02 0.04 0.06 0.08 * * ** * * * ** * * ** * * * ** * * * ** * * ** * * ** * * ** * * ** * 20 30 40 50 * * ** * * * ** * * ** * * * ** * * * ** * * ** * * ** * * ** * * ** * Autumn Winter Spring Summer Annual 80 90 100 110 120 130 Longitude (°E) Longitude (°E) 20 30 40 50 4 3 2 1 0 –1 –2 –3 –4 La tit ud e (° N ) La tit ud e (° N ) Trend in temperature (°C per year) Trend in precipitation (percentage per year) Trend in freq uency of heat w ave ep isod es (d ays p er d ecad e) Trend in num b er of p recip itation d ays (d ays p er d ecad e) Box 1 Figure | Observed trends and future projections of climate in China. a, Observed mean annual temperature variations between 1960 and 2006 across the country, expressed as deviation from the mean during that period (blue line). The blue dotted line is a fit to the data: y5 0.0263x2 52.13 (R2 5 0.54, P5 0.001). The inset shows trends in seasonal temperature (uC per year) during the period 1960–2006. The data come from the climate records of 412 meteorological stations. The three coloured bars on the right-hand side show the projected temperature range by 2100 for the three IPCC marker scenarios A1B, A2 and B1. Model output comes from ref. 1 and uses an ensemble of 24models. b, As for a but for precipitation variations, but the data come from climate records at 355 meteorological stations where all daily precipitation data are available during the period of 1960–2006. The blue dotted line is a fit to the data: y5 0.0454x2 90.03 (R2 5 0.00, P5 0.93). c, Spatial patterns of the trend in seasonal temperature (uC per year, shown as bar graphs) from 1960 to 2006. d, Spatial patterns of the trend in seasonal precipitation (percentage per year, shown as bar graphs) from 1960 to 2006. e, Spatial patterns of the trend in frequency of summer heatwave episodes (days per decade, shown as colour scale) from 1960 to 2006. Heatwave episodes were defined as hot summer (June–August) days with temperatures exceeding the 90th percentile with respect to the reference period (1960–2006). f, Spatial patterns of the trend in rainfall days with precipitation exceeding 0mm (days per decade, shown as colour scale) from 1960 to 2006. A linear regression t-test was conducted to determine whether the slope of the regression line differed significantly from zero. Asterisks and black-edged circles indicates that the trend is statistically significant (P, 0.05). NATUREjVol 467j2 September 2010 REVIEWS 45 Macmillan Publishers Limited. All rights reserved©2010 zhangmk Highlight River runoff and water resources As the demand by agriculture, industry and households for water increases, its availability is becoming a key factor in China’s develop- ment27,28. China’s total fresh water volume is 2.813 1012m3, with 2.73 1012m3 of surface water and 0.833 1012m3 of groundwater29. Although this water resource is large in absolute value, ranking sixth in the world, the per capita water resource is only 25% of the world average30. Moreover, the distribution of water resources is spatially and seasonally uneven. The north of the country, similar in lan