1 SEPTEMBER 2014
ZHOU ET AL.
6591
Projected Changes in Temperature and Precipitation Extremes in China by the CMIP5 Multimodel Ensembles BOTAO ZHOU National Climate Center, China Meteorological Administration, Beijing, and Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science & Technology, Nanjing, China
QIUZI HAN WEN Climate Research Division, Environment Canada, Toronto, Ontario, Canada
YING XU AND LIANCHUN SONG National Climate Center, China Meteorological Administration, Beijing, China
XUEBIN ZHANG Climate Research Division, Environment Canada, Toronto, Ontario, Canada (Manuscript received 9 December 2013, in final form 1 May 2014) ABSTRACT This paper presents projected changes in temperature and precipitation extremes in China by the end of the twenty-first century based on the Coupled Model Intercomparison Project phase 5 (CMIP5) simulations. The temporal changes and their spatial patterns in the Expert Team on Climate Change Detection and Indices (ETCCDI) indices under the RCP4.5 and RCP8.5 emission scenarios are analyzed. Compared to the reference period 1986–2005, substantial changes are projected in temperature and precipitation extremes under both emission scenarios. These changes include a decrease in cold extremes, an increase in warm extremes, and an intensification of precipitation extremes. The intermodel spread in the projection increases with time, with wider spread under RCP8.5 than RCP4.5 for most indices, especially at the subregional scale. The difference in the projected changes under the two RCPs begins to emerge in the 2040s. Analyses based on the mixed-effects analysis of variance (ANOVA) model indicate that by the end of the twenty-first century, at the national scale, the dominant contributor to the projection uncertainty of most temperature-based indices, and some precipitation extremes [including maximum 1-day precipitation (RX1day) and maximum 5-day precipitation (RX5day), and total extremely wet day total amount (R95p)], is the difference in emission scenarios. By the end of the twenty-first century, model uncertainty is the dominant factor at the regional scale and for the other indices. Natural variability can also play very important role.
1. Introduction The Intergovernmental Panel on Climate Change (IPCC) in its Fifth Assessment Report (AR5) states that ‘‘warming of the climate system is unequivocal’’ and that ‘‘continued emissions of greenhouse gases will cause further warming and changes in all components of the climate system’’ (Stocker et al. 2013). The warming has
Corresponding author address: Botao Zhou, National Climate Center, China Meteorological Administration, No. 46 Zhongguancun Nandajie, Haidian, Beijing 100081, China. E-mail:
[email protected] DOI: 10.1175/JCLI-D-13-00761.1 Ó 2014 American Meteorological Society
exerted profound impacts over the globe on human society, ecosystems, and the environment across the globe, mainly through changes in climate extremes (Field et al. 2012). To better adapt to the changing climate, the public and policy makers now frequently demand information concerning future changes in climate extremes. The World Meteorological Organization (WMO) has initiated the Global Framework for Climate Services (GFCS) for managing climate risks (WMO 2009). China is particularly vulnerable to climate extremes such as droughts, floods, cold temperatures, and heat waves. Disasters caused by meteorological and climatic
6592
JOURNAL OF CLIMATE
VOLUME 27
TABLE 1. Information on the 24 climate models used in the present analysis. Atm. Resolution (lon 3 lat)
RCP4.5
RCP8.5
192 3 145
1
1
128 3 64
1
1
128 3 64
1
1
128 3 64
5
5
288 3 192
2
2
288 3 192
1
1
480 3 240
1
1
256 3 128
1
1
192 3 96
4
4
144 3 90
1
1
National Oceanic and Atmospheric Administration Geophysical Fluid Dynamics Laboratory/United States
144 3 90
1
1
Met Office Hadley Centre/United Kingdom
192 3 144
1
3
Met Office Hadley Centre/United Kingdom
192 3 144
4
4
Institute for Numerical Mathematics/Russia
180 3 120
Model
Institution/country
Australian Community Climate and Earth-System Simulator, version 1.0 (ACCESS1.0) Beijing Climate Center, Climate System Model, version 1.1 (BCC_CSM1.1) Beijing Normal University–Earth System Model (BNU-ESM) Second Generation Canadian Earth System Model (CanESM2) Community Climate System Model, version 4 (CCSM4) Community Earth System Model, version 1, bigeochemical (CESM1-BGC) Centro Euro-Mediterraneo per I Cambiamenti Climatici Climate Model (CMCC-CM) Centre National de Recherches Météorologiques Coupled Global Climate Model, version 5 (CNRM-CM5)
Commonwealth Scientific and Industrial Research Organization (CSIRO) and Bureau of Meteorology (BOM)/Australia Beijing Climate Center, China Meteorological Administration/China Beijing Normal University/China
Commonwealth Scientific and Industrial Research Organisation Mark, version 3.6.0 (CSIRO Mk3.6.0) Geophysical Fluid Dynamics Laboratory Earth System Model with Generalized Ocean Layer Dynamics (GOLD) component (GFDL-ESM2G) Geophysical Fluid Dynamics Laboratory Earth System Model with Modular Ocean Model 4 (MOM4) component (GFDL-ESM2M) Hadley Centre Global Environment Model, version 2–Carbon Cycle (HadGEM2-CC) Hadley Centre Global Environment Model, version 2–Earth System (HadGEM2-ES_ Institute of Numerical Mathematics Coupled Model, version 4.0 (INM-CM4) L’Institut Pierre-Simon Laplace Coupled Model, version 5A, low resolution (IPSL-CM5A-LR) L’Institut Pierre-Simon Laplace Coupled Model, version 5A, mid resolution (IPSL-CM5A-MR) L’Institut Pierre-Simon Laplace Coupled Model, version 5B, low resolution (IPSL-CM5B-LR) Model for Interdisciplinary Research on Climate, version 5 (MIROC5)
Model for Interdisciplinary Research on Climate, Earth System Model (MIROC-ESM)
Canadian Centre for Climate Modeling and Analysis/Canada National Center for Atmosphere Research/ United States National Science Foundation–Department of Energy–National Center for Atmosphere Research/United States Centro Euro-Mediterraneo per I Cambiamenti Climatici/Italy Centre National de Recherches Meteorologiques and Centre Europeen de Recherche et Formation Avancees en Calcul Scientifique/France Queensland Climate Change Centre of Excellence and Commonwealth Scientific and Industrial Research Organization/Australia National Oceanic and Atmospheric Administration (NOAA) Geophysical Fluid Dynamics Laboratory/United States
Number of runs
1
Institute Pierre-Simon Laplace/France
96 3 96
4
4
Institute Pierre-Simon Laplace/France
144 3 143
1
1
Institute Pierre-Simon Laplace/France
96 3 96
1
1
256 3 128
3
3
128 3 64
1
1
Atmosphere and Ocean Research Institute (The University of Tokyo), National Institute for Environmental Studies, and Japan Agency for Marine-Earth Science and Technology/Japan Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research Institute (The University of Tokyo), and National Institute for Environmental Studies/Japan
1 SEPTEMBER 2014
6593
ZHOU ET AL. TABLE 1. (Continued) Atm. Resolution (lon 3 lat)
RCP4.5
RCP8.5
Number of runs
Model
Institution/country
Model for Interdisciplinary Research on Climate, Earth System Model, Chemistry Coupled (MIROC-ESM-CHEM)
Japan Agency for Marine-Earth Science and Technology, Atmosphere and Ocean Research Institute (The University of Tokyo), and National Institute for Environmental Studies/Japan Max Planck Institute for Meteorology/Germany
128 3 64
1
1
192 3 96
3
3
Max Planck Institute for Meteorology/Germany
192 3 96
1
1
Meteorological Research Institute/Japan
320 3 160
1
1
Norwegian Climate Centre/Norway
144 3 96
1
1
Max Planck Institute Earth System Model, low resolution (MPI-ESM-LR) Max Planck Institute Earth System Model, medium resolution (MPI-ESM-MR) Meteorological Research Institute Coupled Atmosphere–Ocean General Circulation Model, version 3 (MRI-CGCM3) Norwegian Earth System Model, version 1 (intermediate resolution) (NorESM1-M)
events are estimated to have caused more than 200 billion Yuan per year in economic loss since 1990, accounting for 2.37% of the China’s gross domestic product (Wang and Zheng 2012). Thus, changes in climate extremes anticipated from global warming are of particular concern for Chinese policy makers. Scientifically sound projections of future changes in climate extremes are crucial for developing adaptation strategies to reduce climate risks and for a successful implementation of the GFCS in China. Several studies have investigated future changes in climate extremes anticipated for East Asia and China under different scenarios of greenhouse gas emissions (e.g., Ma et al. 2012; Chen et al. 2011; Gao et al. 2002, 2010; Jiang et al. 2004; Ding et al. 2007; Xu et al. 2009b; Sun et al. 2007; Chen et al. 2012; Jiang et al. 2012; Wang et al. 2012; Wu et al. 2012; Xu et al. 2013). The most recent of these results are based mainly on the global climate model (GCM) simulations available from the Coupled Model Intercomparison Project phase 3 (CMIP3). The results show increases in daily maximum and minimum temperatures with stronger warming at higher latitudes (Jiang et al. 2004). Xu et al. (2009b) also showed that CMIP3 models projected longer heat waves, accompanied with more frequent warm nights in the Yangtze River valley of China in the twenty-first century. More frequent and intense precipitation is projected for the middle and lower reaches of the Yangtze River valley, the southeast coastal region, the western part of northwest China, and the Tibetan Plateau (Jiang et al. 2012). Sun et al. (2010) showed that the frequency of intense snowfall events will decrease over southern China in the twenty-first century. In northern China, snowfall events are projected to increase initially and then subsequently decrease over the mid and late twenty-first century (Sun et al. 2010).
To facilitate research on climate extremes, the Joint WMO Commission for Climatology (CCL)–Climate Variability and Predictability (CLIVAR)–Joint Commission for Oceanography and Marine Meteorology (JCOMM) Expert Team on Climate Change Detection and Indices (ETCCDI) defined a set of climate change indices focusing on extremes that can be described from daily temperature and precipitation across different parts of the world (Frich et al. 2002; Klein Tank et al. 2009; Zhang et al. 2011). These indices have been widely used in detection, attribution, and projection of changes in climate extremes (e.g., Alexander et al. 2006; Min et al. 2011; Orlowsky and Seneviratne 2012; Donat et al. 2013; Sillmann et al. 2013a,b; Wen et al. 2013; Xu et al. 2013). Sillmann et al. (2013a) compared ETCCDI indices computed from observations and model simulations by the Coupled Model Intercomparison Project phase 5 (CMIP5; Taylor et al. 2012) and found that CMIP5 models are generally able to reproduce the historical trend patterns of these climate extreme indices. Compared with CMIP3, CMIP5 features substantial model improvements (Taylor et al. 2012) and adopts a new set of emission scenarios called representative concentration pathways (RCPs) (Moss et al. 2010; van Vuuren et al. 2011) for future climate simulations. Sillmann et al. (2013b) studied the projected future changes in climate extremes at the global, continental, and subcontinental scales. More detailed regional analysis is of particular value for regional policy making. However, detailed analysis of CMIP5 projected future changes in climate extremes for China is still lacking. This study aims at addressing this gap. We follow the global study of Sillmann et al. (2013b) and extend the methods to China and its subregions. Additionally, we attempt to provide some
6594
JOURNAL OF CLIMATE
understanding of the relative importance of various factors that contribute to the uncertainty in the future projections. The remainder of this paper is organized as follows. The data and methods are described in section 2. The projected spatiotemporal changes in temperature and precipitation indices are presented in sections 3 and 4, respectively. Uncertainties in the projection are quantified using a mixed-effects analysis of variance (ANOVA) model and discussed in section 5, followed by conclusions and discussion in section 6.
2. Data and methods a. CMIP5 simulations and climate extremes indices The CMIP5 models produced simulations under some or all of the four RCPs: RCP2.6, RCP4.5, RCP6.0, and RCP8.5 (Taylor et al. 2012). We focus on RCP4.5 and RCP8.5, which are respectively a medium-low and high radiative forcing scenario. The two scenarios have the radiative forcing peaking at 4.5 and 8.5 W m22 by 2100 respectively. More details on the models and the forcing can be found at the CMIP5 website (http://cmip-pcmdi. llnl.gov/cmip5/availability.html, accessed March 2014). The ETCCDI indices from CMIP5 simulations were analyzed by Sillmann et al. (2013a,b) and are available online (at http://www.cccma.ec.gc.ca/data/climdex/index. shtml, accessed July 2013) for the twentieth and twentyfirst centuries. Index data computed from 24 models (see Table 1, which includes expansions for all model names) were available at the time of this analysis. These indices can be generally classified into four categories, namely absolute indices, threshold indices, duration indices, and percentile indices (Sillmann et al. 2013a). The absolute indices describe, for instance, the hottest or coldest day of a year, or the annual maximum 1-day or 5-day precipitation amounts. The threshold indices count the number of days when a fixed temperature or precipitation is exceeded. The duration indices describe the length of wet or dry spells, or warm and cold spells. The percentile indices depict the exceedance rates below or above a threshold, which is defined as the 10th or 90th percentiles from the 1961–90 base period. Detailed definitions of these indices are summarized in Table 2, which can also be found in Klein Tank et al. (2009), Sillmann et al. (2013a), and Zhang et al. (2011), or at the ETCCDI website (etccdi.pacificclimate.org/list_27_indices. shtml, accessed March 2014).
b. Data processing and analysis method The CMIP5 models have different spatial resolutions. Data from the different GCMs are converted to a common 18 3 18 grid using a bilinear interpolation scheme before
VOLUME 27
further processing. The projected change for a region is taken to be the median from the multimodel ensemble. Since the number of ensemble runs from different models differs, we use the model ensemble mean to represent the projections from a particular model if there is more than one run available. The analysis of projection uncertainty is based on individual runs. For this reason, we require at least two runs for a model to be included in this analysis in order to estimate contribution of internal variability to uncertainty. This leaves only 8 GCMs from the 24 models for the uncertainty analysis. We examine the spatial patterns of change by dividing China into eight regions (Fig. 1) used in China’s National Assessment Report on Climate Change (National Report Committee 2007). These regions were determined according to administrative boundaries and societal and geographical conditions. The projected changes in the annual and seasonal indices are summarized using box-and-whisker plots. These plots consist of the multimodel median, the interquartile model spread (the range between the 25th and 75th quantiles, box), and the full intermodel range (whiskers). The multimodel median is taken to be the projected change, while interquartile model spread and the intermodel range visualize uncertainties in the projection, and can also indicate model agreement or disagreement on the direction of the projected change. Uncertainty in the projection has three distinct sources: uncertainty due to different emission scenarios, uncertainty due to difference (both in structure and parameters) in GCMs, and uncertainty due to internal variability of the climate system. Hawkins and Sutton (2009, 2011) attempted to separate and quantify these sources. To understand the relative importance of the different sources to the uncertainty in the future projection, we use a mixed-effects ANOVA approach, which is similar to the one proposed in Li et al. (2012), to partition the variance of the projection to each source. We consider a two-factor model, with the two factors representing uncertainty caused by differences in scenarios and GCMs, respectively. More specifically, let Yijk 5 m 1 ai 1 bj 1 «ijk ,
(1)
where Yijk denotes a certain extreme index projected by the kth run of the jth model under the ith forcing scenario, m represents the grand mean of projection ensembles, ai represents effect on the projection associated with the ith (i 5 1, 2) forcing scenario, bj represents the effect on the projection associated with the jth ( j 5 1. . . 8) GCM simulation, and the noise term «ijk represents projection uncertainty caused by the difference among different runs from a GCM (i.e.,
1 SEPTEMBER 2014
6595
ZHOU ET AL. TABLE 2. Information of the climate extreme indices analyzed in the present analysis.
Label TNn TXx FD TR ID SU GSL
WSDI CSDI TN10p TN90p TX10p TX90p PRCPTOT SDII R95p RX5day RX1day R1mm CDD
Description
Category
Unit
Minimum of daily minimum temperature Maximum of daily maximum temperature Frost days, number of days when the daily minimum temperature below 08C Tropical nights, number of days when the daily minimum temperature above 208C Ice days, number of days when the daily maximum temperature below 08C Summer days, number of days when the daily maximum temperature above 258C Growing season length, number of days between the first occurrence of at least 6 consecutive days with the mean temperature above 58C and the first occurrence after 1 July (NH) or 1 January (SH) of at least 6 consecutive with the mean temperature below 58C Warm spell duration index, the number of days when the daily maximum temperature is above the 90th percentile for six consecutive days or longer Cold spell duration index, number of days when the daily minimum temperature is below the 10th percentile for six consecutive days or longer Cold nights, percentage of days when the daily minimum temperature is below the 10th percentile Warm nights, percentage of days when the daily minimum temperature is above the 90th percentile Cold days, percentage of days when the daily maximum temperature is below the 10th percentile Warm days, percentage of days when the daily maximum temperature is above the 90th percentile Total wet-day precipitation, annual total precipitation in wet days Simple daily precipitation intensity Very wet days, annual total precipitation when the daily precipitation exceeds the 95th percentile of the wet-day precipitation Maximum consecutive 5-day precipitation Maximum 1-day precipitation Number of wet days with at least 1 mm of precipitation Consecutive dry days, maximum number of consecutive days with less than 1 mm of precipitation
Absolute indices Absolute indices Threshold indices Threshold indices Threshold indices Threshold indices Threshold indices
8C 8C Days Days Days Days Days
Duration indices
Days
Duration indices
Days
Percentile indices
%
Percentile indices
%
Percentile indices
%
Percentile indices
%
Threshold indices
mm mm mm
Absolute indices Absolute indices Threshold indices Duration indices
mm mm Days Days
GCM generated internal variability). As in Li et al. (2012), we assume the effect of emission scenarios ai to be fixed because it is not possible to know the probability distribution of possible future emission, and the inter-GCM effect bj to be random. This setting allows us to estimate the effect of GCM differences, although we realize that the eight GCMs used here or even the larger set of CMIP5 models may not represent the plausible GCM population (Knutti et al. 2013). While the results by no means provide a precise partitioning of the contribution to uncertainty from different factors, they should nevertheless give us a general picture of the effects of various factors on future projections. The design is not orthogonal because different models may have a different number of runs available, and thus an explicit formula of its statistical inference is not available. Since our major objective is to quantify the relative contribution of each factor in the model, we employ the expected mean squares (EMS) for each factor as a rough indicator of its relative contribution to the total projection uncertainty. To account for the unbalanced nature of the problem, we use the type II weighted sum of squares (Langsrud 2003) instead of the
ordinary type I sum of squares in the analysis to calculate the mean squares.
c. Performance of CMIP5 models at regional scale Sillmann et al. (2013a) evaluated the performance of the CMIP5 models in simulating those extremes at the global scale based on the root-mean-square errors (RMSEs). Their results showed that the CMIP5 models are generally able to simulate global climate extremes. Following their approach, we assess model performance at regional for China. For this purpose, we compare climates of the indices from the models and from the observations. The observation data are based on gridded daily precipitation and temperature at 0.58 3 0.58 resolution constructed from 2416 observation stations over China by the National Climate Center, China Meteorological Administration (Xu et al. 2009a; Wu and Gao 2013). Indices were computed from this gridded datasets and then averaged to 18 3 18. The period for computing climatology is 1986–2005. The RMSE is calculated as qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (2) RMSE 5 h(X 2 Y)2 i ,
6596
JOURNAL OF CLIMATE
RMSE0 5
FIG. 1. Domains of eight subregions in China—NEC: northeast China; NC: north China; EC: east China; CC: central China; SC; south China; SWC1: southwest China, region 1; SWC2: southwest China, region 2; and NWC: northwest China. CN refers to China as a whole.
where X represents the model climatology of an index, Y is the corresponding climatology from the observational data, and the angular bracket denotes spatial averaging over China on the 18 3 18 grid. The collection of RMSEs for all individual models is used to derive the relative model error for each model, RMSE0 , defined as
VOLUME 27
RMSE 2 RMSEmedian , RMSEmedian
(3)
where RMSEmedian is the median of RMSEs for all individual models. RMSE0 provides a metric to measure a model’s performance relative to the other models, with respect to the observation. Negative RMSE0 indicates that the corresponding model performs better than the majority (50%) of models. Figure 2, similar to Fig. 10 of Sillmann et al. (2013a), summarizes the RMSE0 values for all models (rows) and all indices (columns) relative to the observation. This figure depicts the overall performance of individual models, with colder (warmer) colors indicating models that perform better (worse) than others on average. Despite the fact that the simulation ability of individual models is somehow different for different indices, most models generally perform reasonably well for most temperature indices, particularly CCSM4, CESM1-BGC, CMCC-CM, CSIRO-Mk-3.6.0, IPSLCM5A-MR, MPI-ESM-LR, and MPI-ESM-MR. Models that perform relatively well for all precipitation indices include CSIRO-Mk-3.6.0, IPSL-CM5A-LR, IPSL-CM5A_ MR, and MRI_CGCM3 (see Table 1 for information on models). Moreover, the ensemble median performs well
FIG. 2. The ‘‘portrait’’ diagram of relative RMSEs for the 1986–2005 climatologies of climate extreme indices simulated by the CMIP5 models with respect to the observation. The RMSEs are spatially averaged over China and the magnitude of the relative RMSEs is shown as colors.
1 SEPTEMBER 2014
ZHOU ET AL.
6597
FIG. 3. Temperature indices anomalies relative to the reference period 1986–2005 mean for RCP4.5 (blue) and RCP8.5 (red) over China. Solid lines indicate the ensemble medians and the shadings show the interquartile ensemble spread (25th and 75th quantiles). Time series are smoothed with a 20-yr running mean filter.
and outperforms individual models because some of the systematic errors in individual models are canceled out in the multimodel median. It shows better skill from the ensemble median, providing justification of the use of ensemble median for the future projection.
3. Changes in temperature indices a. Absolute and threshold indices 1) TEMPORAL EVOLUTION Figures 3a and 3b illustrate the temporal evolution of annual minimum daily minimum temperature (TNn) and annual maximum daily maximum temperature (TXx) under RCP4.5 and RCP8.5. Warming is observed in both
TNn and TXx, with larger increases under RCP8.5 than that under RCP4.5, and a slightly larger increase in TNn than in TXx under each emission scenario. Relative to the reference period 1986–2005, the projected ensemble median increases by the end of the twenty-first century in TNn and TXx are 2.98 and 2.78C under RCP4.5, and 5.88 and 5.58C under RCP8.5, respectively. Consistent changes are found in the temperature threshold indices; specifically, these indices are frost days (FD) and tropical nights (TR), which are derived from daily minimum temperature (Figs. 3c,d), and ice days (ID) and summer days (SU), which are derived from daily maximum temperature (Figs. 3e,f). Under both emission scenarios, decreases in FD and ID are projected whereas increases in TR and SU are
6598
JOURNAL OF CLIMATE
VOLUME 27
FIG. 4. The projected multimodel median change in (top) TNn and (bottom) TXx over the period 2081–2100 relative to the reference period 1986–2005 for (left) RCP4.5 and (right) RCP8.5. All changes are significant at the 95% significance level.
projected. By the end of the twenty-first century and under RCP4.5, the projected ensemble median of FD and ID will decrease by 21 days and 17 days while that of TR and SU will increase by 18 days and 25 days, respectively. Projected median changes under RCP8.5 are larger than those under RCP4.5 with a decrease of 43 days for FD and 32 days for ID, and an increase of 38 days for TR and 44 days for SU, respectively.
2) SPATIAL PATTERNS Figure 4 displays spatial distributions of the projected changes for annual TNn and TXx by the end of the twenty-first century. TNn and TXx are projected to increase in all subregions with stronger warming under RCP8.5. Projections under RCP8.5 are also associated with larger interquartile model spread (Fig. 5) due to different climate sensitivity in the models and/or positive regional land–atmosphere feedbacks. There are some spatial differences in the projected changes in TNn and TXx. For instance, under RCP8.5, the largest changes in TNn, exceeding 78C, occur at higher latitudes and higher
elevations including in northeast China, the northern part of northwest China, and the southern flank of southwest China (Fig. 4b), reflecting the influence of diminished snow cover. However, the most intense warming in TXx is mainly projected in eastern China, with a magnitude around 68C (Fig. 4d). The projected increases in TXx are also spatially more uniform than those for TNn. From a seasonal perspective, the projected median increase of TNn for RCP8.5 is generally greatest in DJF but is most spatially uniform across China in JJA (as is illustrated in Fig. 5). The projected increases of TXx for RCP8.5 in DJF are only slightly larger than those in JJA and are smaller than the projected TNn increase in DJF. A similar pattern of seasonal and regional changes in TNn and TXx is seen for RCP4.5, albeit less pronounced when compared to RCP8.5. FD and ID are both projected to decrease everywhere by the end of the twenty-first century under both RCP4.5 and RCP8.5, with strong model agreement (Figs. 6a,c), particularly in western China. The projections indicate that the SWC1 region will experience the
1 SEPTEMBER 2014
ZHOU ET AL.
FIG. 5. Projected changes in annual (ANN), summer (JJA), and winter (DJF) TNn and TXx over the period 2081–2100 relative to the reference period 1986–2005 under RCP4.5 (blue) and RCP8.5 (red). Boxes indicate the interquartile model spread (25th and 75th quantiles) with the horizontal line indicating the ensemble median and the whiskers showing the extreme range of the CMIP5 ensemble.
6599
6600
JOURNAL OF CLIMATE
VOLUME 27
FIG. 6. As in Fig. 5, but for changes in annual FD, TR, ID, SU, and GSL.
greatest decrease of FD and ID, with FD decreasing by 55 days and ID by 51 days under RCP8.5. Decreases for the NWC region under RCP8.5 are also substantial with 49 days decrease in FD and 43 days
decrease in ID. The smallest projected change of FD occurs in SC (14 days for RCP8.5) where the fewest frost days were observed during the reference period 1986–2005.
1 SEPTEMBER 2014
ZHOU ET AL.
6601
FIG. 7. As in Fig. 4, but for (top) WSDI and (bottom) CSDI. All changes are significant at the 95% significance level.
TR and SU are projected to increase by more than 16 and 21 days under RCP4.5 and 38 and 42 days under RCP8.5 in all subregions respectively except SWC1, which includes the Tibetan Plateau (Figs. 6b,d). The largest projected change occurs in SWC2, with an increase of TR and SU by 41 and 50 days under RCP4.5 and by 89 and 93 days under RCP8.5, respectively. This means that if the future emissions follow the path of RCP8.5, Yunnan province located in the SWC2 could experience nighttime temperatures above 208C and daytime temperatures above 258C during the entire summer season. In addition, projected increases of TR (62 days) and SU (57days) for SC under RCP8.5 are larger, even though projected changes of TNn and TXx in SC are actually less than those for other subregions. This may be due to the high local temperature climatology in SC. Small increase in temperature extremes in SC can cause significant impacts on SU and TR in that region. The growing season length (GSL) is projected to be much longer in the future. Both RCP4.5 and RCP8.5 will result in an increase in GSL across China (Fig. 6e). The largest increase under RCP8.5 occurs in SWC1 (67 days)
followed by CC (50 days). The smallest increase of GSL occurs in SC (10 days), because this region is comparatively warm and the GSL itself is already very large during the reference period.
b. Duration and percentile indices The projected changes in temperature duration indices are consistent with what would be expected from warming and changes in absolute and threshold indices. That is, cold spell duration index (CSDI) is projected to decrease and warm spell duration index (WSDI) to increase (Fig. 7). The projected magnitude of increase in WSDI is much larger under RCP8.5 at 136 days than RCP4.5 at 49 days. The projected amplitude of decrease in CSDI is more comparable under RCP4.5 and RCP8.5, with the median projected decrease being 3 days and 4 days, respectively. The asymmetry in the projected change of CSDI is due to the fact that the projected CSDI under the two RCPs would be mostly zero by the end of the twenty-first century. Under RCP8.5, the largest increases in WSDI, which exceed 150 days, are mainly projected in western China, while projected
6602
JOURNAL OF CLIMATE
VOLUME 27
FIG. 8. As in Fig. 3, but for TN10p, TN90p, TX10p, and TX90p. Changes are displayed as absolute exceedance rate.
decreases in CSDI that exceed 4 days are concentrated in Xinjiang, Tibet, and southern China where the climatological CSDI is larger in the reference period 1986–2005. Projected changes of cold nights (TN10p) and cold days (TX10p), warm nights (TN90p), and warm days (TX90p) are shown in Fig. 8. Consistent with warming and projections of the absolute and threshold temperature indices, a decrease in TN10p and TX10p and an increase in TN90p and TX90p are projected. Changes in nighttime temperature indices (TN10p and TN90p) are projected to be stronger than those in daytime temperature indices (TX10p and TX90p). TN10p (TX10p) decreases from about 10% in 1961–90 to 1.7% (2.6%) by the end of this century under RCP4.5 and to 0.4% (0.9%) by the end of this century under RCP8.5. These changes indicate that cold temperatures that occurred once every 10 days in the late twentieth century are projected to occur less than once in 200 days for TN10p and 100 days for TX10p by the end of the twenty-first century under RCP8.5. Projections from different models converge as the projection approaches the zero exceedance rate (i.e., all models project fewer and fewer cold nights and days toward the end of the twenty-first century). For TN90p and TX90p, the median projected increases are from about 10% in the base period to 41% and 36% under RCP4.5, and 67% and 59% under RCP8.5 by the end of the twenty-first century,
respectively. That is, hot temperatures that occurred once every 10 days in the late twentieth century would become everyday weather by the end of the twenty-first century under RCP8.5. Figure 9 shows the spatial distribution of projected changes in TN10p, TN90p, TX10p, and TX90p, which are rather uniform across all subregions in China. Under RCP8.5, TN10p will be in the range from 0.1% (NEC) to 0.5% (SC), TN90p from 61% (NEC) to 75% (SWC2), TX10p from 0.4% (NEC) to 1.6% (SWC2), and TX90p from 51% (NEC) to 68% (SWC1).
4. Changes in precipitation indices In the following, changes in precipitation indices except the number of wet days (R1mm) and consecutive days (CDD) are expressed as percentage change relative to the reference period 1986–2005.
a. Temporal evolution Figure 10 shows the temporal evolution of projected changes in precipitation indices. Overall, precipitation amount, including wet day annual total amount (PRCPTOT), averaged wet day daily amount (SDII), and total extremely wet day total amount (R95p) are all projected to increase in the twenty-first century. Under RCP8.5 (RCP4.5), PRCPTOT and SDII are
1 SEPTEMBER 2014
ZHOU ET AL.
6603
FIG. 9. As in Fig. 5, but for TN10p, TN90p, TX10p, and TX90p. Changes are displayed as absolute exceedance rate.
projected to increase by 14% (8%) and 15% (8%), respectively. The percentage changes are very similar for PRCPTOT and SDII, indicating that increases in total precipitation are projected to be mainly due to increases in precipitation intensity across the precipitation distribution rather than changes in precipitation frequency. The projected percentage increases in maximum 1-day precipitation (RX1day) and maximum 5-day precipitation (RX5day), indicators of the extreme aspect of precipitation, are larger than those for PRCPTOT and SDII. Under RCP8.5 (RCP4.5), the increase is projected to be 21% (11%) for RX5day and 26% (13%) for RX1day by 2100, suggesting larger increases in more extreme daily precipitation. The number of wet days (i.e., R1mm) is projected to change little, and the number of consecutive
dry days is projected to decrease during the second half of the twenty-first century. Time series of precipitation indices under RCP8.5 start to deviate from those under RCP4.5 in the early 2040s, indicating that the difference in the impacts of emission scenarios on precipitation indices is very small before that time. This finding is consistent with changes in temperature indices.
b. Spatial patterns Figure 11 presents the spatial distribution of projected median changes in PRCPTOT and R95p under RCP4.5 and RCP8.5. By the end of the twenty-first century, PRCPTOT is projected to increase uniformly across the country (Figs. 11a,b). The projected PRCPTOT changes have large spread among models, but the majority
6604
JOURNAL OF CLIMATE
VOLUME 27
FIG. 10. As in Fig. 3, but for precipitation indices.
of models agree on the sign of change in all subregions (Fig. 14a). The projected percent increase is larger in northern China than in southern China. The largest increase is projected for western China, due to the drier climate toward the north and northwest. The spatial distribution of R95p changes is similar to that of PRCPTOT, especially in western and northern
China (Figs. 11c,d). The percentage increase in R95p is larger than that of PRCPTOT, implying a disproportionally larger contribution to the total precipitation change from an increase in precipitation falling on very wet days. Additionally, models have better agreement in the sign of projected changes in R95p than in PRCPTOT (Fig. 14).
1 SEPTEMBER 2014
ZHOU ET AL.
6605
FIG. 11. As in Fig. 4, but for (top) PRCPTOT and (bottom) R95p. Black solid dots indicate grid points with changes that are not significant at the 95% significance level.
Figure 12 summarizes changes in annual and seasonal RX1day and RX5day in the eight subregions. The projected changes for these two indices are similar in space with slightly wider spread for RX1day in most cases. For annual indices, the median RX5day projection shows increases from 19% (CC and SC) to 28% (NEC) under RCP8.5 and RX1day increases from 22% (SC) to 32% (NC and SWC2). In northern China (i.e., NEC, NC, and NWC), the median projected increases of RX1day and RX5day are more pronounced in December–February (DJF) than in June–August (JJA). For example, the strongest increase of RX1day (RX5day) in DJF is 53% (51%) compared with 27% (19%) increase in JJA in NC. The opposite seasonal pattern is noticed in southern China such as in SC, SWC1, and SWC2, where the projected increases in JJA are larger than those in DJF. This may not necessarily reflect the absolute change in the extreme values because of seasonal variability in precipitation climatology. There do not appear to be seasonal differences in the projected changes in extreme precipitation in other subregions (e.g., EC and CC).
Figure 13 displays projected changes in the number of wet days (R1mm) and consecutive dry days (CDD). The number of wet days increases north of ;308N and decreases in the south. The maximum number of consecutive dry days is projected to decrease in the north including NC, NWC, and NEC (Fig. 14e) but increase in other regions.
5. Relative importance of different factors contributing to the uncertainty in the projections The main results of uncertainty assessment are summarized in Fig. 15. For temperature-based indices, forcing uncertainty is a dominant factor for China as a whole; however, internal variability and differences between models also play important roles at the regional scale. For example, for the relatively small region SWC1, the effect of model differences is as strong as that of the difference in scenarios for indices such as TR, ID, SU, and GSL. Another example is that internal variability has the largest contribution to the uncertainty in projected changes of ID in CC. For the precipitation-based indices,
6606
JOURNAL OF CLIMATE
FIG. 12. As in Fig. 5, but for (left) RX5day and (right) RX1day.
VOLUME 27
1 SEPTEMBER 2014
ZHOU ET AL.
6607
FIG. 13. As in Fig. 4, but for (top) R1mm and (bottom) CDD. Black solid dots indicate grid points with changes that are not significant at the 95% significance level.
the major sources of uncertainties are diverse for different indices at both national and regional scales. Differences in both models and forcing scenarios have a large influence on projected changes in PRCPTOT for the country, but the main contributors vary among subregions. For example, it appears that the dominant contributors are forcing scenarios in NEC and SWC1, internal variability in EC, and differences in GCMs in CC and SWC2. The differences between forcing scenarios appear to be the most important contributor to projection uncertainty in R95P, SDII, RX1day, and RX5day, similar to those of temperature indices. This is consistent with other studies that show that precipitation rate and extreme precipitation in particular are strongly influenced by the availability of atmospheric moisture (Kharin et al. 2013; Zhang et al. 2013). For example, over Northern Hemispheric land, projected changes in RX1day and RX5day scale well with projected temperature increases (Zhang et al. 2013), and this is also the case for projected changes in 20-yr return values of annual maximum daily precipitation (Kharin et al. 2013).
6. Summary and concluding remarks In this study, we examined projected changes in ETCCDI extreme indices that are based on daily temperature and precipitation data in China as simulated by the CMIP5 multimodel ensemble under RCP4.5 and RCP8.5 forcing scenarios. Results generally show decreases in cold extremes, increases in warm extremes, and an intensification of precipitation extremes in the future warmer world. They are in agreement with previous CMIP3based studies (e.g., Chen et al. 2012; Jiang et al. 2012; Wang et al. 2012; Xu et al. 2013). The larger changes in the climate extreme indices corresponding to stronger radiative forcing changes are also in agreement with previous findings. However, compared with earlier studies, this study is more comprehensive, providing more detailed information on the future changes in climate extremes for China and its subregions based on multiple aspects, new scenarios, and the new generation of climate models. We also quantified the relative contribution of various factors to the uncertainty in the projections. The main findings are summarized below:
6608
JOURNAL OF CLIMATE
FIG. 14. As in Fig. 5, but for annual PRCPTOT, SDII, R95p, R1mm, and CDD, respectively.
VOLUME 27
1 SEPTEMBER 2014
ZHOU ET AL.
6609
FIG. 15. Relative importance (in percentage) of factors contributing to the projection uncertainties for China and each of the eight subregions. Factors include climate models (red), RCPs (yellow), and internal variability (blue).
1) TNn and TXx are projected to have larger increase and interquartile model spread under RCP8.5 than RCP4.5. The spatial pattern of changes in TNn and TXx differ slightly with a larger increase in TNn than in TXx in general. The greatest projected TNn change occurs in the higher latitudes and higher elevation regions of northeast China, the northern part of northwest China, and the southern flank of southwest China. The largest warming in TXx is projected for eastern China and the northern part of northwest China. Projected changes in other temperature extremes indices are consistent with those of TNn and TXx. The number of frost days and ice days are expected to decrease with the largest decrease projected in SWC1 and NWC. More tropical nights and summer days are expected toward the end of the twenty-first century, especially in SWC2 and SC. The growing season length (GSL) is also projected to become longer in China, particularly in SWC1. Cold spell duration index (CSDI) is projected to decrease and warm spell duration index (WSDI) to increase. The strongest increases of WSDI are mainly projected for western China, and the decreases of CSDI are mainly projected in Xinjiang, Tibet, and southern China. Decreases in TN10p and TX10p and increases in TN90p and TX90p are also predominant features. Moreover, changes in warm
and cold nights are more remarkable than warm and cold days. For example, under the RCP8.5 scenario, cold nights or days that are defined based on 1961–1900 climate would be very rare by the end of the twentyfirst century. 2) Annual total precipitation amount (PROPTOT), average daily precipitation rate (SDII), the proportion of heavy precipitation in total annual precipitation (R95p), and extreme precipitation including both RX1day and RX5day are all projected to increase. Percent changes in extreme precipitation RX1day and RX5day are larger than projected for other aspects of precipitation examined, indicating larger disproportionately increases in extremes as compared to other aspects of precipitation. The percent increase in precipitation extremes is more pronounced in DJF than in JJA in northern China (e.g., NEC, NC, and NWC) whereas the seasonal difference is opposite in southern China (e.g., SC, SWC1, and SWC2). This difference may be related to the difference in the seasonal amount of precipitation rather than an indication of larger increase in the magnitude of extreme precipitation. It may also reflect possible changes in circulations. The number of wet days appears to increase in the north (north of 308N) and decrease in the south (south of 308N). However, the changes are small in general and model agreement on the sign of projected change is poor.
6610
JOURNAL OF CLIMATE
3) Uncertainties in the projection are larger under RCP8.5 than under RCP4.5 at a given future time (except TN10p and TX10p) and increase over time under both forcing scenarios. In addition, differences in projected change under the two forcing scenarios start to emerge only in the 2040s. Two factors including the committed warming from emissions already made and the time that it takes for the RCPs to diverge are at play. Three factors—natural climate variability, difference in forcing scenarios, and uncertainty in climate models—contribute to the uncertainty in the projections. Their relative contributions differ depending on indices, spatial scale, and time period. At the national scale and by the end of the twenty-first century, differences in the forcing scenarios play the dominant role. This is also the case for extreme precipitation indices, notably RX1day, RX5day, and R95p, which are more temperature sensitive than other indices such as total precipitation. At the regional scale, differences among climate models become another important contributor for most indices even at the end of the twenty-first century. However, it is unclear if the elevated importance in model uncertainty compared with forcing uncertainty is due to larger variability at smaller spatial scales or is an indication of a lack of ability in models at simulating climate at regional scales. Acknowledgments. We acknowledge the World Climate Research Program’s Working Group on Coupled Modeling, which is responsible for CMIP, and thank climate modeling groups for producing and making available their model output. For CMIP, the U.S. Department of Energy’s Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. This research was conducted while B. Zhou was visiting Climate Research Division of Environment Canada under a collaborative project of the Joint Working Group XIII between the Meteorological Service Canada and the China Meteorological Administration. This research received support from the National Natural Science Foundation of China (41275078), the National Basic Research Program of China (2012CB955902) and the Climate Change Specific Fund (CCSF201329) of China.
REFERENCES Alexander, L., and Coauthors, 2006: Global observed changes in daily climate extremes of temperature and precipitation. J. Geophys. Res., 111, D05109, doi:10.1029/2005JD006290.
VOLUME 27
Chen, H., J. Sun, X. Chen, and W. Zhou, 2012: GCM projections of heavy rainfall events in China. Int. J. Climatol., 32, 441–450, doi:10.1002/joc.2278. Chen, W., Z. Jiang, and L. Li, 2011: Probabilistic projections of climate change over China under the SRES A1B scenario using 28 AOGCMs. J. Climate, 24, 4741–4756, doi:10.1175/ 2011JCLI4102.1. Ding, Y., G. Ren, Z. Zhao, Y. Xu, Y. Luo, Q. Li, and J. Zhang, 2007: Detection, causes and projection of climate change over China: An overview of recent progress. Adv. Atmos. Sci., 24, 954–971, doi:10.1007/s00376-007-0954-4. Donat, M. G., and Coauthors, 2013: Updated analyses of temperature and precipitation extreme indices since the beginning of the twentieth century: The HadEX2 dataset. J. Geophys. Res., 118, 2098–2118, doi:10.1002/jgrd.50150. Field, C. B., and Coauthors, Eds., 2012: Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. Cambridge University Press, 582 pp. Frich, P., L. V. Alexander, P. Della-Marta, B. Gleason, M. Haylock, A. M. G. Klein Tank, and T. Peterson, 2002: Observed coherent changes in climatic extremes during the second half of the twentieth century. Climate Res., 19, 193–212, doi:10.3354/ cr019193. Gao, X., Z. Zhao, and F. Giorgi, 2002: Changes of extreme events in regional climate simulations over East Asia. Adv. Atmos. Sci., 19, 927–942, doi:10.1007/s00376-002-0056-2. ——, Y. Shi, and F. Giorgi, 2010: A high resolution simulation of climate change over China. Sci. China, 54D, 462–472. Hawkins, E., and R. Sutton, 2009: The potential to narrow uncertainty in regional climate predictions. Bull. Amer. Meteor. Soc., 90, 1095–1107, doi:10.1175/2009BAMS2607.1. ——, and ——, 2011: The potential to narrow uncertainty in regional precipitation change. Climate Dyn., 37, 407–418, doi:10.1007/ s00382-010-0810-6. Jiang, D.-B., H.-J. Wang, and X.-M. Lang, 2004: Multimodel ensemble prediction for climate change trend of China under SRES A2 scenario. Chin. J. Geophys., 47, 878–886, doi:10.1002/cjg2.564. Jiang, Z., J. Song, L. Li, W. Chen, Z. Wang, and J. Wang, 2012: Extreme climate events in China: IPCC-AR4 model evaluation and projection. Climatic Change, 110, 385–401, doi:10.1007/ s10584-011-0090-0. Kharin, V. V., F. W. Zwiers, X. Zhang, and M. Wehner, 2013: Changes in temperature and precipitation extremes in the CMIP5 ensemble. Climatic Change, 119, 345–357, doi:10.1007/ s10584-013-0705-8. Klein Tank, A. M. G., F. W. Zwiers, and X. Zhang, 2009: Guidelines on analysis of extremes in a changing climate in support of informed decisions for adaptation. WCDMP-No. 72, WMO-TD No. 1500, 56 pp. [Available online at www.wcrpclimate.org/documents/WCDMP_TD_1500.pdf.] Knutti, R., D. Masson, and A. Gettelman, 2013: Climate model genealogy: Generation CMIP5 and how we got there. Geophys. Res. Lett., 40, 1194–1199, doi:10.1002/grl.50256. Langsrud, Ø., 2003: ANOVA for unbalanced data: Use type II instead of type III sums of squares. Stat. Comput., 13, 163–167, doi:10.1023/A:1023260610025. Li, G., X. Zhang, F. Zwiers, and Q. H. Wen, 2012: Quantification of uncertainty in high-resolution temperature scenarios for North America. J. Climate, 25, 3373–3389, doi:10.1175/ JCLI-D-11-00217.1. Ma, T., Z. Wu, and Z. Jiang, 2012: How does cold wave frequency in China respond to a warming climate? Climate Dyn., 39, 2487–2496, doi:10.1007/s00382-012-1354-8.
1 SEPTEMBER 2014
ZHOU ET AL.
Min, S.-K., X. Zhang, F. W. Zwiers, and G. C. Hegerl, 2011: Human contribution to more intense precipitation extremes. Nature, 470, 378–381, doi:10.1038/nature09763. Moss, R. H., and Coauthors, 2010: The next generation of scenarios for climate change research and assessment. Nature, 463, 747– 756, doi:10.1038/nature08823. National Report Committee, 2007: China’s National Assessment Report on Climate Change (in Chinese). Science Press, 148 pp. Orlowsky, B., and S. I. Seneviratne, 2012: Global changes in extreme events: Regional and seasonal dimension. Climatic Change, 110, 669–696, doi:10.1007/s10584-011-0122-9. Sillmann, J., V. V. Kharin, X. Zhang, F. W. Zwiers, and D. Bronaugh, 2013a: Climate extremes indices in the CMIP5 multimodel ensemble: Part 1. Model evaluation in the present climate. J. Geophys. Res., 118, 1716–1733, doi:10.1002/jgrd.50203. ——, ——, F. W. Zwiers, X. Zhang, and D. Bronaugh, 2013b: Climate extremes indices in the CMIP5 multimodel ensemble: Part 2. Future climate projections. J. Geophys. Res., 118, 2473– 2493, doi:10.1002/jgrd.50188. Stocker, T. F., and Coauthors, Eds., 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp. Sun, J., H. Wang, W. Yuan, and H. Chen, 2010: Spatial-temporal features of intense snowfall events in China and their possible change. J. Geophys. Res., 115, D16110, doi:10.1029/ 2009JD013541. Sun, Y., S. Solomon, A. Dai, and R. W. Portmann, 2007: How often will it rain? J. Climate, 20, 4801–4818, doi:10.1175/JCLI4263.1. Taylor, K. E., B. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, doi:10.1175/BAMS-D-11-00094.1. van Vuuren, D. P., and Coauthors, 2011: The representative concentration pathways: An overview. Climatic Change, 109, 5– 31, doi:10.1007/s10584-011-0148-z. Wang, H.-J., and Coauthors, 2012: Extreme climate in China: Facts, simulation and projection. Meteor. Z., 21, 279–304, doi:10.1127/0941-2948/2012/0330. Wang, W. G., and G. G. Zheng, 2012: Annual Report on Actions to Address Climate Change (2012): Climate Finance and Low
6611
Carbon Development (in Chinese). Social Sciences Academic Press, 306 pp. Wen, Q. H., X. Zhang, Y. Xu, and B. Wang, 2013: Detecting human influence on extreme temperatures in China. Geophys. Res. Lett., 40, 1171–1176, doi:10.1002/grl.50285. WMO, 2009: Report of the World Climate Conference-3. World Meteorological Organization, 80 pp. [Available online at http://library.wmo.int/opac/index.php?lvl=notice_display& id=4043#.U9f9NfldVuI.] Wu, J., and X. J. Gao, 2013: A gridded daily observation dataset over China region and comparison with the other datasets. Chin. J. Geophys., 56, 1102–1111. Wu, Z., Z. Jiang, J. Li, S. Zhong, and L. Wang, 2012: Possible association of the western Tibetan Plateau snow cover with the decadal to interdecadal variations of northern China heatwave frequency. Climate Dyn., 39, 2393–2402, doi:10.1007/ s00382-012-1439-4. Xu, J., Y. Shi, X. Gao, and F. Giorgi, 2013: Projected changes in climate extremes over China in the 21st century from a high resolution regional climate model (RegCM3). Chin. Sci. Bull., 58, 1443–1452, doi:10.1007/s11434-012-5548-6. Xu, Y., X. Gao, Y. Shen, C. Xu, Y. Shi, and F. Giorgi, 2009a: A daily temperature dataset over China and its application in validating a RCM simulation. Adv. Atmos. Sci., 26, 763–772, doi:10.1007/s00376-009-9029-z. ——, C. Xu, X. Gao, and Y. Luo, 2009b: Projected changes in temperature and precipitation extremes over the Yangtze River Basin of China in the 21st century. Quat. Int., 208, 44–52, doi:10.1016/j.quaint.2008.12.020. Zhang, X., L. Alexander, G. C. Hegerl, P. Jones, A. Klein Tank, T. C. Peterson, B. Trewin, and F. W. Zwiers, 2011: Indices for monitoring changes in extremes based on daily temperature and precipitation data. Wiley Interdiscip. Rev.: Climate Change, 2, 851–870, doi:10.1002/wcc.147. ——, H. Wan, F. W. Zwiers, G. C. Hegerl, and S.-K. Min, 2013: Attributing intensification of precipitation extremes to human influence. Geophys. Res. Lett., 40, 5252–5257, doi:10.1002/ grl.51010.
