1. Introduction
Climate change will inevitably continue for the next few decades (IPCC 2013). Agriculture is extremely sensitive to climate change because the CO2 concentration and meteorological variables (i.e., temperature, precipitation, and solar radiation) determine resource availability and control fundamental processes involved in crop growth and development (IPCC 2014), emphasizing the need to formulate appropriate adaptive measures to mitigate the adverse impact of climate change on agriculture or take advantage of the possible opportunities associated with climate change (Lobell et al. 2008; Tao and Zhang 2010; IPCC 2014).
Rice (Oryza sativa L.) is one of the most important staple crops and is currently the most produced grain crop in China. Hunan Province is the largest rice cultivation area in China and produced 12.6% of the total rice production in 2013 [China Statistical Yearbook (National Bureau of Statistics 2013)] because it is naturally endowed with a typical humid subtropical climate favorable for rice production. Rice yield in this region thus has a significant influence on regional food security. It is generally acknowledged that the adverse impacts of climate change on agriculture will be more serious in tropical regions (Easterling et al. 2007) and developing countries (Lobell et al. 2008; IPCC 2014). Given its socioeconomic importance, it is of great necessity to assess the future climate change impacts on rice yield and the effectiveness of various adaptive measures.
Various methods have been used to explore the impacts of climate change on crop production. Among them, process-based crop models are advantageous in that they explicitly take into account interactions between genotype, environment, and management and quantify the relative effects of individual factors on crop development, growth, and final yield in a controlled manner (Martín et al. 2014; Rötter and Höhn 2015). During the past few decades, the application of crop models became a major data source for the assessments of the Intergovernmental Panel on Climate Change (IPCC) on agriculture (Gitay et al. 2001; Easterling et al. 2007; Porter et al. 2014).
The Crop Environment Resource Synthesis-Rice (CERES-Rice) model embedded in the Decision Support System for Agrotechnology Transfer (DSSAT) has proven to be a successful crop model (Jones et al. 2003). After several decades of development and refinement, it has been widely used all over the world (Ritchie et al. 1998; Timsina and Humphreys 2006a; Xiong et al. 2009; Phakamas et al. 2013). However, the reliability and accuracy of the model’s simulations depend not only on the robustness of the model but also on the availability of sufficient and high-quality empirical datasets, which are used to facilitate model calibration, validation, and evaluation (Hunt and Boote 1998; Timsina and Humphreys 2006b; Rötter et al. 2015).
China is a traditionally agricultural country with a long history of intensive cultivation and has been adopting various measures to collect data about rice production. One such measure is the establishment of a large number of agrometeorological experimental stations across most of the cultivation areas in China, which are responsible for collecting and recording data observed by well-trained agricultural technicians following standardized observation criteria and prescribed methods (China Meteorological Administration 1993). These high-quality data in combination with high-resolution daily weather data provide a valuable input to rigorously calibrate and validate the model, thus improving the model’s performance and facilitating its application in China.
With these data, several researchers have employed various versions of the CERES-Rice model to assess the potential response of rice yield to projected climate change in China (Jin et al. 1995; Lin et al. 2005; Yao et al. 2007; Xiong et al. 2009). Their findings provided useful understandings of the process and mechanism of the impacts of climate change and stressed the complexity of rice–climate relationships in China.
However, most of these previous studies were conducted either on extensive areas that had a limited number of representative stations (Jin et al. 1995; Yao et al. 2007) or on a scaled-up national 50 km × 50 km grid level (Lin et al. 2005; Xiong et al. 2009), which might not reflect the details of the situation in the study area. In addition, previous studies mainly focused on the prediction of rice production under future climatic conditions; few quantified the effectiveness of possible adaptive strategies, which has been drawing increasing attention from scientific communities such as the IPCC (Easterling et al. 2007; Porter et al. 2014) and other researchers (Howden et al. 2007; Lobell et al. 2008; Godfray et al. 2010; Rötter and Höhn 2015). Moreover, previous studies mainly used climate scenario data derived from the Special Report on Emissions Scenarios (SRES) in the third IPCC reports (Yao et al. 2007), which may not reflect the possible stabilization of greenhouse gas concentrations due to policy actions (Moss et al. 2010). Finally, uncertainty in climate change predictions presents a great challenge for assessing climate impacts and adaptation methods. However, only a few studies have adopted a multimodel ensemble approach, which has proven practical and effective at quantifying prediction uncertainty and thus provides better prediction quality than any single model (Kirtman et al. 2014).
In this study, basing on high-quality plot-scale agrometeorological experimental data, we used the recently developed CERES-Rice model, version 4.5 (v4.5), coupled with climate data generated by five general circulation models (GCMs) under the newly developed representative concentration pathways (RCPs) 4.5 and 8.5 climate scenarios to assess the potential climate change impacts and atmospheric CO2 fertilization effects on rice yield during three periods: the 2020s (2011–40), the 2050s (2041–70), and the 2080s (2071–99) relative to those during the baseline (1981–2010). We also evaluated the potential effectiveness of two adaptive measures widely practiced in Hunan Province and quantified the model uncertainty caused by climate models. The objective of this study was to gain insight into the impacts of climate change on future rice yield, and provide possible adaptive measures to cope with future climate change risks or capitalize on potential climatic opportunities.
2. Materials and methods
a. Study area and data-providing stations
Hunan Province (24°38′–30°08′N, 108°47′–114°15′E) covers an area of 211 800 km2 and has a population of 67 372 000. It is located in the middle reaches of the Yangtze River valley and has abundant surface and underground water because of its relative low altitude and abundant river systems. Its climate is dominated by a typical humid subtropical monsoon regime, with an annual mean temperature varying from 21° to 25°C and annual precipitation ranging between 1100 and 1500 mm, most of which falls from May to August. The soil in Hunan Province is mainly red loam, paddy soil, and yellow soil. The long growing season allows a double-cropping rice system; however, a single-cropping rice system is also practiced in some areas. With these naturally endowed resources and favorable agroclimatic conditions, as well as the convenient transportation allowed by its geographical centricity, Hunan Province has long been a primary rice cultivation area of China and plays a very important role in ensuring regional food security.
The stations used in this study refer to those agrometeorological experimental stations (AESs) that provide one of the key agricultural experimental data (including management practices, crop phenology, and crop yields) required to conduct CERES-Rice model simulations. However, not all stations can provide the required data. To enhance the model’s performance, we selected suitable stations based on the following criteria: they 1) must be located in a primary rice-production district; 2) must contain typical rice cultivars that had been cultivated for at least three years during which no diseases, insect pests, or extreme climate events occurred; 3) must have good records of management practices; and 4) must be located near weather stations. As a result, 15 cultivars in eight AESs for double-cropping rice (black dots) and 2 cultivars in two AESs for single-cropping rice (black triangles) were selected for use in this study (Fig. 1). These stations cover the majority of the rice-growing areas in Hunan Province. The general information on location, soil, and cultivars for the 10 selected stations is shown in Table 1.
Locations of selected rice stations in Hunan Province. Dots indicate double-cropping rice, and triangles indicate single-cropping rice.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
Information about rice stations, soils, and selected cultivars for model calibration and validation; the boldface year is the calibration year.
b. The CERES-Rice crop model
The CERES-Rice model used in this study is embedded in the DSSAT model suite (Jones et al. 2003). The DSSAT has developed a collection of independent programs that operate together and contains models for more than 27 crops (rice, maize, wheat, soybean, peanut, etc.); the CERES-Rice model is one of these. Since its first appearance, five versions of DSSAT have been released during the last 27 years (1989–2015), and the latest version—DSSAT v4.5—was distributed in 2012 (Hoogenboom et al. 2012). As compared with the previous version, the improvements in CERES-Rice v4.5 comprised bug fixes, an improved model structure, and new capabilities (Hoogenboom et al. 2012).
As an ecophysiological model, the CERES-Rice model was developed as a cultivar-specific and site-specific model operated on a daily time step, which could simulate rice growth (dry weight gain rate, leaf area index, grain filling rate, etc.), phenological development (germination, seedling emergence, flowering, physiological maturity, etc.), and yield based on plant physiological processes (Timsina and Humphreys 2006a). In the CERES-Rice model, there are additional adjustments for CO2 concentration, which exerts influence on variables like quantum efficiency and light-saturated photosynthesis rate (Jones et al. 2003). Because it is one of the most advanced and successful crop simulation models (Jones et al. 2003), the CERES-Rice v4.5 model was selected to conduct simulations in this study.
c. Input data for the CERES-Rice model
The CERES-Rice model requires high-quality input data to guarantee the accuracy of the simulation results. At a minimum, a dataset that describes cultivar genetic coefficients, weather and soil conditions, rice phenology, yield, and management practices is required for model operation (Jones et al. 1994; Hunt et al. 2001). Detailed information about these data is provided below.
1) Climate change scenarios
Climate scenarios are plausible descriptions of future climates (Moss et al. 2010), whose data include daily solar radiation, daily maximum and minimum temperature, and precipitation. Using a new coordinated parallel process rather than the traditional sequential process, scientists have developed a set of new scenarios called RCPs (Moss et al. 2010). The RCPs are named after the radiative forcing levels (W m−2) by the end of the twenty-first century. The Fifth Assessment Report of the IPCC selected four RCPs—RCP 2.6, RCP 4.5, RCP 6.0, and RCP 8.5—from the current literature to conduct climate change impact assessments (Fig. 2). In this study, RCP4.5 and RCP8.5 were selected as the future climate scenarios because they can cover both medium and extreme scenarios.
(a) Radiative forcing and (b) CO2 emissions under different RCPs.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
In this study, we selected five GCMs (HadGEM2-ES, GFDL-ESM2M, IPSL-CM5A-LR, MIROC-ESM-CHEM, and NorESM1-M) to generate climate data under RCP4.5 and RCP8.5. These five GCMs were adopted because they are the only GCMs from phase 5 of the Coupled Model Intercomparison Project (CMIP5) that were bias corrected and downscaled to the resolution of 0.5° × 0.5° by the Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP) to provide future climate data (Warszawski et al. 2014; Hempel et al. 2013). Table 2 shows the detailed information about these five GCMs. In addition, the future climate data for the 10 rice stations were extracted from the nearest grids for the application of the CERES-Rice model. The baseline spanned 30 yr (1981–2010), and three future periods (2020s, 2050s, and 2080s) were also studied.
Detailed information about the five GCMs used in this study. (Expansions of model acronyms are available at http://www.ametsoc.org/PubsAcronymList.)
2) Historical weather data
Daily measured weather data for the baseline period of 1981–2010 were obtained from the China Meteorological Data Sharing Service System (http://data.cma.cn/). They included daily sunshine hours, daily maximum and minimum air temperatures, and daily precipitation at each rice station. The daily solar radiation data required by the model were calculated using the Angstrom equation (Wang et al. 2008) based on the latitude and the daily sunshine hours from the empirical weather data. The historical weather data for six selected rice experimental stations (Changde, Nanxian, Yiyang, Wugang, Jingzhou, and Chenzhou) were obtained from local meteorological stations at the same locations, while the other four rice stations (Lixian, Loudi, Huaihua, and Lengshuitan) that did not have a local meteorological station were covered by data from the nearest meteorological station.
3) Soil data
Soil data are required by the CERES-Rice model to define the soil properties for each rice station. The required soil data consist of the soil type, color, slope, drainage condition, runoff potential, fertility factor, number of layers, and the characteristics of each layer. These data were derived from China Soil Scientific Database (http://www.soil.csdb.cn/). The general soil information for each rice station is shown in Table 1.
4) Agricultural experimental data
The agricultural experimental data for the baseline (1981–2010) were derived from local China agrometeorological experimental stations, as previously indicated, which are maintained by the China Meteorological Administration. These long-term high-quality data include detailed information about rice phenology, yields, yield components, and management practices and can be transformed into the required input data to CERES-Rice model such as the flowering duration (the number of days from transplanting date to flowering date) and the maturity duration (the number of days from transplanting date to maturity date). Records of management practices show that the rice was irrigated at all of the study stations; therefore this study simulated the climate change impacts on irrigated rice production.
5) Cultivar genetic coefficients
In DSSAT model, genetic coefficients are a set of parameters that characterize the nature of the genotypes for a given crop cultivar (Hunt et al. 1993). When combined, these genetic coefficients can define the development, vegetative growth, and reproductive growth of individual genotypes. They are used mathematically to quantify how a particular cultivar responds to external factors including climate change, variations in the CO2 concentration, and management practices (Román-Paoli et al. 2000). Different cultivars of the same crop have different coefficient values. Eight genetic coefficients are defined in the CERES-Rice model. These eight coefficients and their meanings are shown in Table 3.
Description of genetic coefficients in the CERES-Rice model.
d. Calibration, validation, and evaluation of the CERES-Rice model
Model calibration and validation were necessary to optimize the model input parameters and improve the reliability of the simulation results so that the model could be used in this study. Whether the model could be applied depends on its performance in simulating past processes occurring within an agricultural system. Model calibration is the adjustment of the crop cultivar genetic coefficients so that the simulated and observed growth duration and yield parameters are in close agreement; such a process is also used for model parameterization. Model validation compares simulated results with observed results using the calibrated crop cultivar genetic coefficients and other independent datasets. Model evaluation assesses the crop model’s performance or its simulating capacity, which is based on evaluating the degree of agreement between the simulated and the observed values of growth duration and yield parameters (Jones et al. 2003; Timsina and Humphreys 2006a).
Because of the gene–environment interaction, the genetic coefficients of rice cultivars are station and cultivar specific. Therefore, the CERES-Rice model must be calibrated and validated for each rice station to obtain the specific genetic coefficients for each cultivar.
In this study, flowering duration, maturity duration, and grain yield were selected as growth duration and yield parameters. The calibration and validation processes were carried out as follows. First, the years during which the representative cultivars were planted and no diseases, insect pests, or extreme climate events occurred were selected as calibration and validation years. Among these years, the year with the best fertilization and irrigation managements was chosen for calibration, and another two years were chosen for validation (Table 1). Second, the CERES-Rice model was calibrated to estimate the genetic coefficients for each selected cultivar using the input data from the calibration year by employment of the generalized likelihood uncertainty estimation (GLUE) coefficient estimator module (He et al. 2010). Third, the CERES-Rice model was validated using the input data from the validation years.
e. Quantifying model uncertainty caused by climate models
Quantifying uncertainty caused by climate models plays an important role in selecting appropriate measures for adaptation and mitigation. Multimodel ensemble strategy is an effective approach to resolve uncertainty in climate change projections (Kirtman et al. 2014). Therefore, we adopted a multimodel ensemble approach to address uncertainty caused by climate models. The intermodel variability is commonly measured by intermodel standard deviations (Lobell et al. 2007). In this study, the standard deviations of simulated changes in yield from five GCMs were calculated to quantify model uncertainty.
3. Model simulation
After rigorous calibration, validation, and evaluation, the CERES-Rice model was operated to simulate the impacts of climate change on rice production and adaptive measures in Hunan Province under two RCPs. In this study, the simulation was conducted under the assumption that climate data, CO2 concentration, and adaptive measures are the only changing variables, while inputs such as cultivar and crop management practices during future periods remained the same as those during the baseline.
a. Simulation of the impacts of climate change on rice production
The impacts of climate change on rice phenology and yield were simulated by comparing the CERES-Rice model results during three future periods with those during the baseline period with climate data as the only variable and other inputs remaining the same as the baseline.
The effect of elevated CO2 concentration on rice yield in the 2080s was assessed by comparing the results from projected CO2 concentrations under two RCPs with results from a baseline CO2 concentration of 363 ppm, which was extracted from the NOAA Mauna Lua (Hawaii) CO2 database (http://www.co2.earth). Aside from the CO2 concentration, other inputs were held constant with the baseline.
b. Simulation of adaptive measures
Given data availability, the capacity of the CERES-Rice simulation, and the limited scope of this paper, we selected two of the most commonly employed adaptive measures: the use of high-temperature-tolerant rice cultivars and changing planting dates to evaluate their effectiveness in mitigating the adverse impacts of climate change on rice yield in the 2080s. These two measures represent an easily practiced household-level autonomous adaptation and have drawn most attention by the Fifth Assessment Report of the IPCC (Porter et al. 2014). The simulation was conducted on three cultivars with the largest amplitude of yield reduction: one from the early rice cultivars, one from the middle rice cultivars, and one from the late rice cultivars.
Regarding the use of a different rice cultivar, we simulated the effectiveness of selecting a cultivar that had the adaptive advantage of higher heat tolerance and/or higher yield as a substitute for the currently cultivated less adaptive and/or lower-yielding cultivar. The switch is conducted based on the principle of agrometeorological similarity, because similar temperature and solar radiation conditions between the place from which the cultivar is to be removed and the one to which it is to be transplanted guarantee a successful adaptation. For planting dates, we advanced and delayed the planting date by 15 days at 5-day intervals to assess the effect of this adaptive measure on mitigating the adverse effects of climate change.
4. Results
a. Climate change under RCP4.5 and RCP8.5
The projected CO2 concentrations under RCP4.5 and RCP8.5 and changes in ensemble mean of annual average temperature, precipitation, and solar radiation in the study area during the future periods (2020s, 2050s, and 2080s) under RCP4.5 and RCP8.5 relative to the baseline (1981~2010) are shown in Table 4. A detailed description of the ensemble-mean changes in climate variables during 2011–99 under two RCPs relative to the baseline at the 10 stations is shown in Fig. 3.
Projected CO2 concentration under RCP4.5 and RCP8.5 and changes in ensemble mean of annual average temperature, precipitation, and radiation under RCP4.5 and RCP8.5 in the study area relative to the baseline. The 
Changes in the ensemble mean at the 10 stations shown at the bottom of the figure (see Table 1 for names) of the monthly (a) maximum temperature, (b) minimum temperature, (c) precipitation, and (d) solar radiation from March to October during 2011–99 under (left) RCP4.5 and (right) RCP8.5 relative to the baseline.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
Table 4 shows that the ensemble mean of annual average temperature during the 2020s, 2050s, and 2080s is projected to increase by 0.9°, 2.0°, and 2.6°C, respectively, under RCP4.5, and 1.0°, 2.8°, and 4.8°C, respectively, under RCP8.5. Figures 3a and 3b illustrate that the ensemble-mean changes in monthly average maximum temperature at the 10 stations present a similar pattern with those of the minimum temperature. Generally, stations in the north exhibit larger increases in the maximum temperature than those in southern Hunan Province. Moreover, the increases in maximum and minimum temperature under the two RCPs mainly occur in March, August, September, and October.
As indicated in Table 4, in comparison with the baseline, the ensemble mean of annual average precipitation during the 2020s, 2050s, and 2080s is projected to increase by 1.4%, 5.1%, and 11.9%, respectively, under RCP4.5, and 1.5%, 4.0%, and 6.6%, respectively, under RCP8.5. Figure 3c indicates that the spatial distribution of the ensemble-mean changes in monthly average precipitation fluctuates greatly during different months, exhibiting an opposite trend from that of maximum temperature under two RCPs. The increases in precipitation are concentrated in April, June, and September.
In comparison with the baseline, the ensemble mean of annual average solar radiation during the 2020s, 2050s, and 2080s would change by −0.3%, 5.2%, and 6.3%, respectively, under RCP4.5 and −0.3%, 4.7%, and 7.7%, respectively, under RCP8.5 (Table 4). More specifically, Fig. 3d shows that the spatial distribution of the ensemble-mean changes in monthly average solar radiation is in good accordance with that of the maximum temperature, with the northern area showing larger increases than the southern part of Hunan Province under both RCPs. The increases in solar radiation are centered in May, June, and October.
b. The calibration and validation results for the CERES-Rice model
The calibration and validation results for the CERES-Rice model show that the observed and simulated flowering duration, maturity duration, and grain yields for the selected rice cultivars were in good agreement (Fig. 4). Most of the PD values were located within the ±15% lines, and the NRMSE values for the flowering duration, maturity duration, and grain yields were 8.4%, 6.0%, and 7.4%, respectively. Therefore, the CERES-Rice model had a good simulation performance at all the selected stations and could be used to conduct simulations. The calculated genetic coefficients for the selected cultivars are given in Table 5.
Comparisons of the simulated and observed (a) flowering durations, (b) maturity durations, and (c) yields at the 17 selected cultivars shown using the different symbols at the bottom of the figure (see Table 5). The black solid line is the 1:1 reference line, and the broken lines show the ±15% PD.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
Calculated genetic coefficients for the 17 selected rice cultivars.
c. Climate change impact simulation
1) Impacts on rice phenology
The modeling results of flowering duration of 17 rice cultivars are displayed in Fig. 5. For early rice cultivars (from No. 1 to No. 7), the ensemble-average flowering duration would decrease by 2.4–9.5 days under RCP4.5 and 2.3–14.9 days under RCP8.5. For middle rice cultivars (from No. 8 to No. 9), the ensemble-average flowering duration would be shortened by 0.42–6.4 days under RCP4.5; aside from No. 8 cultivar at Huaihua station whose ensemble-average flowering duration increased by 0.4 of a day in the 2020s and 3.2 days in the 2080s, other results showed a decreasing trend from 0.3 to 8.6 days under RCP8.5. However, the changes in the ensemble-average flowering duration of late rice cultivars (from No. 10 to No. 17) showed great diversity, ranging from −3.6 to 6.5 days under RCP4.5 and from −3.7 to 11.7 days under RCP8.5. Particularly, the ensemble-average flowering duration of late rice cultivar at Chenzhou station decreased 3.6 and 3.7 days, respectively, under RCP4.5 and RCP8.5, while that of late rice cultivar at Nanxian station increased 6.5 and 11.7 days, respectively, under RCP4.5 and RCP8.5, which was the most remarkable difference among all cultivars.
Change in flowering duration (days) in three future periods under (a) RCP4.5 and (b) RCP8.5 relative to the baseline. The numbers on the x axis represent the 17 rice cultivars (see Table 5). Box-and-whisker plots show the 5th, 25th, 50th, 75th, and 95th percentiles. The black dots within the bars indicate the multimodel ensemble means.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
Figure 6 shows the change in maturity duration for 17 rice cultivars. The ensemble-average maturity durations of most cultivars would decrease by 1.5–11.4 days under RCP4.5 and by 0.9–17.5 days under RCP8.5, except for those for the late rice cultivars at Nanxian and Yiyang stations, which changed by −2.9–3.0 days under RCP4.5 and −2.0–6.8 days under RCP8.5.
As in Fig. 5, but for change in maturity duration.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
In addition, the magnitude of the changes in flowering duration and maturity duration would increase while the model agreement would decrease from the 2020s to the 2080s under both RCPs.
Our findings are consistent with those of Challinor et al. (2007) and Devkota et al. (2013). Challinor et al. (2007) simulated that in India, when the optimal temperature for rice development was 28°C, the maturity duration in some regions would decrease by 10–35 days, while some other regions would experience a lengthening of the duration of up to 20 days during 2071–2100, relative to those during 1961–90 under the A2 scenario. Devkota et al. (2013) predicted that rice flowering durations during 2040–69 could be delayed by 4 days under the B1 scenario and 8 days under the A1F1 scenario relative to the historical data (1970–99) in central Asia.
2) Impacts on rice yield
The simulated changes of rice yield under RCP4.5 and RCP8.5 relative to the baseline, neglecting CO2 fertilization effects, are displayed in Fig. 7. In general, the rice yields for all rice cultivars showed an apparent decreasing trend under both RCPs. The decrease was relatively mild during the 2020s but intensified during the 2080s, and there was a notable decline in the model agreement from the 2020s to the 2080s. Specifically, the ensemble-average rice yields of all cultivars during the 2020s, 2050s, and 2080s would decrease by 12.3%, 17.2%, and 24.5% under RCP4.5 and 14.7%, 27.5%, and 47.1% under RCP8.5, respectively. For all cultivars, yields declined more acutely under RCP8.5 than RCP4.5 and the decreases in late rice cultivars were projected to be more significant than those of early and middle rice cultivars. The No. 10 late rice cultivar at Changde station would show the most severe yield reduction under RCP4.5, while the decrease in the No. 14 late rice cultivar at Lixian station would be the largest under RCP8.5. Our results are consistent with projections by previous studies that simulated climate change impacts alone without considering CO2 fertilization effects. Lin et al. (2005) found that the average yield of rainfed and irrigated rice would decrease by 1.1%–28.6% with large regional variability during 2010–79 under the A2 and B2 scenarios in China. Xiong et al. (2009) projected that changes in area-weighted rice yield varied between −26.2% and 6.3% from 2011 to 2100 under the A2 and B2 scenarios with a similar large range of regional variability in China.
As in Fig. 5, but for change in yield.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
If CO2 fertilization effects were considered, the ensemble-average yield decrease during the 2080s was mitigated by 8.6%–24.9% under RCP4.5 and 12.4%–58.1% under RCP8.5 (Fig. 8). For the early rice cultivars at Lixian and Nanxian stations, CO2 fertilization effects could even completely offset the adverse impacts of climate change on yield under both RCPs.
Change in yield in the 2080s under (a) RCP4.5 and (b) RCP8.5 relative to the baseline, without and with CO2 fertilization effects. The numbers on the x axis represent the 17 rice cultivars (see Table 5). Box-and-whisker plots show the 5th, 25th, 50th, 75th, and 95th percentiles The black dots within the bars indicate the indicate the multimodel ensemble means.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
Our results drew similar conclusions to most previous studies that showed an increasing trend in rice yield when CO2 fertilization effect was taken into account, when compared with those that neglected CO2 fertilization effect (Yao et al. 2007; Iizumi et al. 2008; Satapathy et al. 2014).
d. Adaptive measures simulation
1) Switching cultivars
As indicated in Fig. 9, switching from the No. 6 early rice cultivar at Wugang site to the late-maturing No. 3 early rice cultivar at Lengshuitan site would increase the ensemble-average rice yield in the 2080s by 16.8% under RCP4.5 and 16.5% under RCP8.5. Switching from the No. 8 middle rice cultivar at Huaihua site to the late-maturing No. 9 middle rice cultivar at Jingzhou site would increase the ensemble-average rice yield by 19.6% and 13.2% in the 2080s under RCP4.5 and RCP8.5, respectively, and switching from the No. 10 late rice cultivar at Changde site to the late-maturing No. 11 late rice cultivar at Chenzhou site would lead to an increased ensemble-average yield of 11.8% and 25.1% in the 2080s under RCP4.5 and RCP8.5, respectively.
Simulated yields of Nos. 6, 8, and 10 cultivars during the 2080s under (a) RCP 4.5 and (b) RCP 8.5 with the current and replaced cultivar. Box-and-whisker plots show the 5th, 25th, 50th, 75th, and 95th percentiles. The black dots within the bars indicate the indicate the multimodel ensemble means.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
2) Changing planting dates
Simulation results for changing planting date to 15 days earlier and later than the current planting date varied with cultivars. Specifically, the ensemble-average yield for the No. 6 cultivar in the 2080s had a maximum increase of 14.5% under RCP4.5 and 22.7% under RCP8.5, and the ensemble-average yield of the No. 8 cultivar increased the most, by 21.6% and 14.2% in the 2080s under RCP4.5 and RCP8.5, respectively, when the planting dates were advanced by 15 days. Delaying the planting date of the No. 10 cultivar by 15 days led to the largest increase in ensemble-average yield, by 31.8% and 36.9% in the 2080s under RCP4.5 and RCP8.5, respectively (Fig. 10).
Simulated yields of Nos. 6, 8, and 10 cultivars with different planting dates in the 2080s under (a) RCP4.5 and (b) RCP8.5. Box-and-whisker plots show the 5th, 25th, 50th, 75th, and 95th percentiles. The black dots within the bars indicate the indicate the multimodel ensemble means.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
e. Model spread and uncertainty
The intermodel spread of changes in rice yield in three future periods under RCP4.5 and RCP8.5 is presented in Fig. 11. The standard deviations of multimodel ensembles for all cultivars increased from the 2020s to the 2080s under both RCPS. The spread arising from climate models was generally small (less than 20%), indicating that yields simulated from different climate models were in good consistency.
The intermodel spread of change in yields in three future periods under (a) RCP4.5 and (b) RCP8.5. The numbers on the x axis represent the 17 rice cultivars (see Table 5).
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
5. Discussion
In this study, we simulated climate change impacts and CO2 fertilization effects on rice phenology and yield, as well as the effectiveness of two adaptive measures in Hunan Province.
With respect to the changes of rice phenology, the ensemble-average flowering duration of the late rice cultivar at Nanxian station increased by 6.5 days in the 2080s under RCP4.5, which was the greatest increase for all of the simulations under RCP4.5. In contrast, the flowering duration of the late rice cultivar at Chenzhou station was projected to decrease by 3.6 days in the 2080s under RCP4.5. Among the results of five climate models in the 2080s under RCP4.5, the flowering durations of late rice cultivars at these two stations presented the most prominent difference with MIROC-ESM-CHEM climate data. To understand the underlying causes of such a notable difference, we calculated the monthly average minimum and maximum temperatures at these 10 stations during the growing season in MIROC-ESM-CHEM climate model in the 2080s under RCP4.5 (Fig. 12).
Monthly mean (a) maximum and (b) minimum temperature from March to October in the MIROC-ESM-CHEM climate model in the 2080s under RCP4.5.
Citation: Journal of Applied Meteorology and Climatology 55, 6; 10.1175/JAMC-D-15-0213.1
These phenological changes can be explained as follows: Rice is extremely sensitive to heat stress (≥35°C), and flowering is the most sensitive development stage to such stress (Prasad et al. 2006). In the CERES-Rice model, the daily thermal time (DTT) is used to describe different phases of rice development. A phenological phase is completed when DTT accumulation reaches a threshold defined by the genetic coefficients of each cultivar. When the daily average temperature is lower than the optimal temperature, the DTT increases with an increase in the average temperature and the phenological stage shortens. However, when the daily average temperature exceeds the optimal temperature, the DTT decreases with an increase in the average temperature and the phenological stage becomes longer. It is obvious that the monthly mean minimum and maximum temperature at Nanxian station from early July to late August are much higher than that at Chenzhou station (Fig. 12). Therefore, the flowering durations of late rice cultivars at Nanxian and Chenzhou stations would change in opposite directions. This mechanism also applies to the flowering duration changes for other early, middle, and late rice cultivars and maturity changes for all cultivars under both RCPs.
The simulation results showed that rice yields for all cultivars in Hunan Province would generally decrease under both RCPs. The largest yield reduction in the 2080s under RCP4.5 occurred in the late rice cultivar at Changde station. The reduction was most prominent with MIROC-ESM-CHEM climate data. To illustrate the relationship between yield reduction and future climate change in a more detailed way, we took MIROC-ESM-CHEM climate data at Changde station from 1981–2099 under RCP4.5 as an example and used a stepwise linear regression method to explore the possible effects of three climate variables (the daily average temperature, precipitation, and solar radiation) during the growing season on rice yield and the following equations were obtained.
These equations indicate that an increase in daily average temperature exerts the principal adverse effect on rice yield, while an increase in daily precipitation would also lead to a yield reduction, and an increase in daily solar radiation would increase the rice yield.
Three possible mechanisms might account for such a decrease in the rice yield associated with a temperature increase. First, higher temperatures that remain below the optimum temperature for rice growth and development generally cause the rice to grow faster, which results in less time for the grain itself to grow and mature, thus reducing the yield (Porter 2005; Hatfield et al. 2011; Krishnan et al. 2011). Second, a higher temperature that exceeds the optimum temperature would cause direct damage to the rice growth and reduce the yield (Porter 2005; Hatfield et al. 2011). When the daily average temperature is higher than 35°C, both the flowering and the filling stages would be adversely impacted (Porter 2005). During the flowering stage, high temperature affects anther dehiscence, reduces the pollen life span, and decreases the fertilization rate and thus leads to spikelet sterility (Prasad et al. 2006). The optimum daily average temperature for grain filling is between 22° and 28°C; if it exceeds 35°C, unfulfilled grains would undergo premature senescence, reducing the grain number (Prasad et al. 2006). The total number of days with a daily average temperature higher than 35°C at Changde station increases from 30 in the baseline to 264 in the 2020s, 552 in the 2050s, and 1077 in the 2080s in MIROC-ESM-CHEM climate model under RCP4.5, exerting a large adverse impact on rice yield. In addition, a higher minimum temperature would increase maintenance respiration and lead to a decrease in yield (Welch et al. 2010).
Furthermore, precipitation might also be an important factor that impacts rice yield. Hunan Province is prone to flooding or waterlogging disasters from May to July because of its low terrain and abundant river systems (Huang et al. 2008). The projected notable increase in future precipitation in June would intensify the severity of flooding or waterlogging conditions (Fig. 3c), which might damage rice production in the future. The combination of these factors would produce adverse effects on rice yield in Hunan Province.
Third, variations in solar radiation, a driving variable in rice production, could also affect rice photosynthesis and thus impact rice yield (Welch et al. 2010). However, temperature and precipitation increases are more important and thus might cause more adverse impacts on yield. Therefore, the beneficial effects of increase in solar radiation could not offset the adverse impacts of increased temperature and precipitation, and rice yield would decrease in the future.
When CO2 fertilization effects were considered, yields of all rice cultivars would increase in different degrees. Elevated CO2 has two direct effects on C3 crops (e.g., rice, wheat, and soybean). On the one hand, elevated CO2 is beneficial to photosynthesis in C3 crops, increasing crop growth and grain yield (Kimball et al. 2002). On the other hand, elevated CO2 could decrease the conductivity of CO2 and water vapor through stomata, improving water-use efficiency and thus alleviating the drought damage (Ottman et al. 2001).
As for adaptive measures, rice yields would increase evidently when early or medium-maturing cultivars were replaced by late-maturing cultivars. This result can be attributed to the prolonged grain-filling duration, which contributes to the biomass accumulation (Lobell and Gourdji 2012). Besides, changing planting dates also proved to be an effective method to improve rice yields. This adaptive measure could change the amount and distribution of temperature, precipitation, and solar radiation during the growing season, thus avoiding the disadvantages and utilizing the advantages of climate change (Jalota et al. 2012). The warming trend in future periods would prolong the length of growing season, making both early and late planting possible (Zhang et al. 2004; White et al. 2009). In Hunan Province, the growing period for early, middle, and late rice cultivars is March–July, April–August, and June–October, respectively. Extreme high temperatures are most likely to occur in July–September. For early and middle rice cultivars, advancing the planting date could avoid extreme weather during the growing period and thus increase yield. Similarly, delaying the planting date for late rice cultivars could reduce the number of extreme weather events and benefit rice production.
6. Study uncertainties and limitations
Uncertainties are inherent in projections of the climate change impacts on rice production and can be attributed to different sources in modeling. This study is no exception. Uncertainties mainly arise from three sources. First, uncertainty in climate change projections by climate models is a significant contribution to the total projected uncertainty (Challinor et al. 2009). Weather and climate inputs are important in determining the predictive capability of agricultural models. Although we adopted five GCMs to address the uncertainties in climate models, it is impossible for these climate models to provide the equivalent of historical weather data for a future climate (Hallegatte 2009). The most important reason is that climate change itself is highly uncertain. The pathways of future greenhouse gas emissions cannot be projected accurately by the RCPs because they are greatly affected by political and socioeconomic factors. Additionally, there are inadequate projections of the climate processes in the climate models and a scale mismatch between the climate model outputs used in this study and the future climate data inputs needed by the CERES-Rice model. Both projecting the effects of emissions on climate and downscaling of the climate data lead to uncertainty in future climate projections (Asseng et al. 2013). Second, crop models also greatly contribute to the total projected uncertainty (Asseng et al. 2013). As a mathematical representation of a real-world system, a number of simplifications and limitations exist in the CERES-Rice model (Rosenzweig and Iglesias 1998). For example, the model assumes that environmental factors such as weeds, diseases, and insect pests are fully controlled and that soil problems such as salinity, acidity, or sodicity do not exist. Moreover, the CERES-Rice model could not perform well under extreme weather events such as floods and droughts because many simple, empirically derived relationships in this model may not hold under extreme climatic conditions (Timsina and Humphreys 2006b). Finally, because of a lack of related data and appropriate methods, this study was conducted under the assumption that, other than weather and CO2 concentration, changes in soil, management practices, and government policies and improvements in agricultural technology were not considered during the simulations, which is another source of total projected uncertainty.
7. Conclusions
In this study, the CERES-Rice model was calibrated and validated to simulate the impacts of climate change and adaptive measures. The simulation results indicate that increases in daily average temperature can either accelerate or slow down rice phenological development in Hunan Province under both RCPs, depending on whether the temperature exceeds the critical threshold defined in the CERES-Rice model. The yields of all cultivars would decrease in the future and the decreases were larger under RCP8.5 than RCP4.5. The uncertainly in simulated yield arising from climate models under both RCPs was generally small. With CO2 fertilization effects, the yield decreases would be alleviated during the 2080s under two RCPs. The adaptive measures evaluated in this study—replacing cultivars and changing planting dates—proved to be effective in mitigating the adverse effects of climate change on rice production in Hunan Province.
Although many uncertainties still exist, our results might increase our understanding of the mechanism and the degree of the impacts of climatic variables and CO2 fertilization on rice yield and help farmers and policymakers to develop effective and sustainable strategies for agriculture production in Hunan Province in the face of climate change.
Acknowledgments
This research is jointly funded by the National Basic Research Program of China (973 Program) (Grant 2012CB955403), the CAS Strategic Priority Research Program (Grant XDA05130701), and the Natural Science Foundation of China (Grant 41172154). We also express our thanks to the China Meteorological Administration and the Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP) of the Potsdam Institute for Climate Impact Research for providing climate data.
REFERENCES
Asseng, S., and Coauthors, 2013: Uncertainty in simulating wheat yields under climate change. Nat. Climate Change, 3, 827–832, doi:10.1038/nclimate1916.
Challinor, A. J., T. R. Wheeler, P. Q. Craufurd, C. A. T. Ferro, and D. B. Stephenson, 2007: Adaptation of crops to climate change through genotypic responses to mean and extreme temperatures. Agric. Ecosyst. Environ., 119, 190–204, doi:10.1016/j.agee.2006.07.009.
Challinor, A. J., T. R. Wheeler, D. Hemming, and H. D. Upadhyaya, 2009: Ensemble yield simulations: Crop and climate uncertainties, sensitivity to temperature and genotypic adaptation to climate change. Climate Res., 38, 117–127, doi:10.3354/cr00779.
China Meteorological Administration, 1993: Agro-Meteorological Observations Standard. China Meteorological Press, 230 pp.
Devkota, K. P., A. M. Manschadi, M. K. Devkota, O. Egamberdiev, and P. L. G. Vlek, 2013: Simulating the impact of climate change on rice phenology and grain yield in irrigated drylands of Central Asia. J. Appl. Meteor. Climatol., 52, 2033–2050, doi:10.1175/JAMC-D-12-0182.1.
Easterling, W. E., and Coauthors, 2007: Food, fibre and forest products. Climate Change 2007: Impacts, Adaptation and Vulnerability, M. L. Parry, et al., Eds., Cambridge University Press, 273–313.
Gitay, H., S. Brown, W. Easterling, and B. Jallow, 2001: Ecosystems and their goods and services. Climate Change 2001: Impacts, Adaptation, and Vulnerability, J. J. McCarthy et al., Eds., Cambridge University Press, 237–342.
Godfray, H. C. J., and Coauthors, 2010: Food security: The challenge of feeding 9 billion people. Science, 327, 812–818, doi:10.1126/science.1185383.
Hallegatte, S., 2009: Strategies to adapt to an uncertain climate change. Global Environ. Change, 19, 240–247, doi:10.1016/j.gloenvcha.2008.12.003.
Hatfield, J. L., K. J. Boote, B. A. Kimball, L. H. Ziska, R. C. Izaurralde, D. Ort, A. M. Thomson, and D. Wolfe, 2011: Climate impacts on agriculture: Implications for crop production. Agron. J., 103, 351–370, doi:10.2134/agronj2010.0303.
He, J., J. W. Jones, W. D. Graham, and M. D. Dukes, 2010: Influence of likelihood function choice for estimating crop model parameters using the generalized likelihood uncertainty estimation method. Agric. Syst., 103, 256–264, doi:10.1016/j.agsy.2010.01.006.
Hempel, S., K. Frieler, L. Warszawski, J. Schewe, and F. Piontek, 2013: A trend-preserving bias correction—The ISI-MIP approach. Earth Syst. Dyn., 4, 219–236, doi:10.5194/esd-4-219-2013.
Hoogenboom, G., and Coauthors, 2012: Decision Support System for Agrotechnology Transfer (DSSAT) version 4.5.1.023. University of Hawaii, CD-ROM.
Howden, S. M., J. F. Soussana, F. N. Tubiello, N. Chhetri, M. Dunlop, and H. Meinke, 2007: Adapting agriculture to climate change. Proc. Natl. Acad. Sci. USA, 104, 19 691–19 696, doi:10.1073/pnas.0701890104.
Huang, X., and Coauthors, 2008: Flood hazard in Hunan province of China: An economic loss analysis. Nat. Hazards, 47, 65–73, doi:10.1007/s11069-007-9197-z.
Hunt, L. A., and K. J. Boote, 1998: Data for model operation, calibration and evaluation. Understanding Options for Agricultural Production, G. Y. Tsuji, G. Hoogenboom, and P. K. Thornton, Eds., Systems Approaches for Sustainable Agricultural Development, Vol. 7, Kluwer Academic Publishers, 9–39, doi:10.1007/978-94-017-3624-4_2.
Hunt, L. A., J. W. Pararajasingham, J. W. Jones, G. Hoogenboom, D. T. Imamura, and R. M. Ogoshi, 1993: GENCALC: Software to facilitate the use of crop models for analyzing field experiments. Agron. J., 85, 1090–1094, doi:10.2134/agronj1993.00021962008500050025x.
Hunt, L. A., J. W. White, and G. Hoogenboom, 2001: Agronomic data: Advances in documentation and protocols for exchange and use. Agric. Syst., 70, 477–492, doi:10.1016/S0308-521X(01)00056-7.
Iizumi, T., M. Yokozawa, Y. Hayashi, and F. Kimura, 2008: Climate change impact on rice insurance payouts in Japan. J. Appl. Meteor. Climatol., 47, 2265–2278, doi:10.1175/2008JAMC1800.1.
IPCC, 2013: Climate Change 2013: The Physical Science Basis. Cambridge University Press, 1535 pp., doi:10.1017/CBO9781107415324.
IPCC, 2014: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part A: Global and Sectoral Aspects. C. B. Field et al., Eds., Cambridge University Press, 1132 pp. [Available online at http://www.ipcc.ch/pdf/assessment-report/ar5/wg2/WGIIAR5-PartA_FINAL.pdf.]
Jalota, S. K., H. Kaur, S. S. Ray, R. Tripathi, B. B. Vashisht, and S. K. Bal, 2012: Mitigating future climate change effects by shifting planting dates of crops in rice–wheat cropping system. Reg. Environ. Change, 12, 913–922, doi:10.1007/s10113-012-0300-y.
Jin, Z., D. Ge, H. Chen, and J. Fang, 1995: Effects of climate change on rice production and strategies for adaptation in Southern China. Climate Change and Agriculture: Analysis of Potential International Impacts, C. Rosenzweig et al., Eds., American Society of Agronomy, 307–323.
Jones, J. W., and Coauthors, 1994: Input and output files. Decision Support System for Agrotechnology Transfer (DSSAT) Version 3, Vol. 2, G. Y. Tsuji, G. Uehara, and S. Balas, Eds., University of Hawaii, 1–94.
Jones, J. W., and Coauthors, 2003: The DSSAT cropping system model. Eur. J. Agron., 18, 235–265, doi:10.1016/S1161-0301(02)00107-7.
Kimball, B. A., K. Kobayashi, and M. Bindi, 2002: Responses of agricultural crops to free-air CO2 enrichment. Adv. Agron., 77, 293–368, doi:10.1016/S0065-2113(02)77017-X.
Kirtman, B. P., and Coauthors, 2014: The North American Multimodel Ensemble: Phase-1 seasonal-to-interannual prediction; phase-2 toward developing intraseasonal prediction. Bull. Amer. Meteor. Soc., 95, 585–601, doi:10.1175/BAMS-D-12-00050.1.
Krishnan, P., B. Ramakrishnan, K. Raja Reddy, and V. R. Reddy, 2011: High-temperature effects on rice growth, yield, and grain quality. Adv. Agron., 111, 87–206, doi:10.1016/B978-0-12-387689-8.00004-7.
Lin, E., W. Xiong, H. Ju, Y. Xu, Y. Li, L. Bai, and L. Xie, 2005: Climate change impacts on crop yield and quality with CO2 fertilization in China. Philos. Trans. Roy. Soc., 360B, 2149–2154, doi:10.1098/rstb.2005.1743.
Lobell, D. B., and S. M. Gourdji, 2012: The influence of climate change on global crop productivity. Plant Physiol., 160, 1686–1697, doi:10.1104/pp.112.208298.
Lobell, D. B., C. Bonfils, and P. B. Duffy, 2007: Climate change uncertainty for daily minimum and maximum temperatures: A model inter-comparison. Geophys. Res. Lett., 34, L05715, doi:10.1029/2006GL028726.
Lobell, D. B., M. B. Burke, C. Tebaldi, M. D. Mastrandrea, W. P. Falcon, and R. L. Naylor, 2008: Prioritizing climate change adaptation needs for food security in 2030. Science, 319, 607–610, doi:10.1126/science.1152339.
Martín, M. M., J. E. Olesen, and J. R. Porter, 2014: A genotype, environment and management (GxExM) analysis of adaptation in winter wheat to climate change in Denmark. Agric. For. Meteor., 187, 1–13, doi:10.1016/j.agrformet.2013.11.009.
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 Bureau of Statistics, 2013: China Statistical Yearbook. China Statistics Press, 967 pp.
Ottman, M. J., and Coauthors, 2001: Elevated CO2 increases sorghum biomass under drought conditions. New Phytol., 150, 261–273, doi:10.1046/j.1469-8137.2001.00110.x.
Phakamas, N., A. Jintrawet, A. Patanothai, P. Sringam, and G. Hoogenboom, 2013: Estimation of solar radiation based on air temperature and application with the DSSAT v4.5 peanut and rice simulation models in Thailand. Agric. For. Meteor., 180, 182–193, doi:10.1016/j.agrformet.2013.06.015.
Porter, J. R., 2005: Rising temperatures are likely to reduce crop yields. Nature, 436, 174, doi:10.1038/436174b.
Porter, J. R., A. J. Challinor, K. Cochrane, S. M. Howden, M. M. Iqbal, D. B. Lobell, and M. I. Travasso, 2014: Food security and food production systems. Climate Change 2014: Impacts, Adaptation, and Vulnerability, C. B. Field et al., Eds., Cambridge University Press, 485–533.
Prasad, P. V. V., K. J. Boote, L. H. Allen, J. E. Sheehy, and J. M. G. Thomas, 2006: Species, ecotype and cultivar differences in spikelet fertility and harvest index of rice in response to high temperature stress. Field Crops Res., 95, 398–411, doi:10.1016/j.fcr.2005.04.008.
Rinaldi, M., N. Losavio, and Z. Flagella, 2003: Evaluation and application of the OILCROP-SUN model for sunflower in southern Italy. Agric. Syst., 78, 17–30, doi:10.1016/S0308-521X(03)00030-1.
Ritchie, J. T., U. Singh, D. C. Godwin, and W. T. Bowen, 1998: Cereal growth, development and yield. Understanding Options for Agricultural Production, G. Y. Tsuji, G. Hoogenboom, and P. K. Thornton, Eds., Systems Approaches for Sustainable Agricultural Development, Vol. 7, Kluwer Academic Publishers, 79–98.
Román-Paoli, E., S. M. Welch, and R. L. Vanderlip, 2000: Comparing genetic coefficient methods using the CERES-Maize model. Agric. Syst., 65, 29–41, doi:10.1016/S0308-521X(00)00024-X.
Rosenzweig, C., and A. Iglesias, 1998: The use of crop models for international climate change impact assessment. Understanding Options for Agricultural Production, G. Y. Tsuji, G. Hoogenboom, and P. K. Thornton, Eds., Systems Approaches for Sustainable Agricultural Development, Vol. 7, Kluwer Academic Publishers, 267–292.
Rötter, R. P., and J. Höhn, 2015: An overview of climate change impact on crop production and its variability in Europe, related uncertainties and research challenges. Climate Change and Food Systems: Global Assessments and Implications for Food Security and Trade, A. Elbehri, Ed., FAO, 108–145.
Rötter, R. P., F. Tao, J. G. Höhn, and T. Palosuo, 2015: Use of crop simulation modelling to aid ideotype design of future cereal cultivars. J. Exp. Bot., 66, 3463–3476, doi:10.1093/jxb/erv098.
Satapathy, S. S., D. K. Swain, and S. Herath, 2014: Field experiments and simulation to evaluate rice cultivar adaptation to elevated carbon dioxide and temperature in sub-tropical India. Eur. J. Agron., 54, 21–33, doi:10.1016/j.eja.2013.11.010.
Tao, F. L., and Z. Zhang, 2010: Adaptation of maize production to climate change in North China Plain: Quantify the relative contributions of adaptation options. Eur. J. Agron., 33, 103–116, doi:10.1016/j.eja.2010.04.002.
Timsina, J., and E. Humphreys, 2006a: Performance of CERES-Rice and CERES-Wheat models in rice–wheat systems: A review. Agric. Syst., 90, 5–31, doi:10.1016/j.agsy.2005.11.007.
Timsina, J., and E. Humphreys, 2006b: Applications of CERES-Rice and CERES-Wheat in research, policy and climate change studies in Asia: A review. Int. J. Agric. Res., 1, 202–225, doi:10.3923/ijar.2006.202.225.
Wang, E., Q. Yu, D. Wu, and J. Xia, 2008: Climate, agricultural production and hydrological balance in the North China Plain. Int. J. Climatol., 28, 1959–1970, doi:10.1002/joc.1677.
Warszawski, L., K. Frieler, V. Huber, F. Piontek, O. Serdeczny, and J. Schewe, 2014: The Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP): Project framework. Proc. Natl. Acad. Sci. USA, 111, 3228–3232, doi:10.1073/pnas.1312330110.
Welch, J. R., J. R. Vincent, M. Auffhammer, P. F. Moya, A. Dobermann, and D. Dawe, 2010: Rice yields in tropical/subtropical Asia exhibit large but opposing sensitivities to minimum and maximum temperatures. Proc. Natl. Acad. Sci. USA, 107, 14 562–14 567, doi:10.1073/pnas.1001222107.
White, M. A., and Coauthors, 2009: Intercomparison, interpretation, and assessment of spring phenology in North America estimated from remote sensing for 1982–2006. Global Change Biol., 15, 2335–2359, doi:10.1111/j.1365-2486.2009.01910.x.
Xiong, W., D. Conway, E. Lin, and I. Holman, 2009: Potential impacts of climate change and climate variability on China’s rice yield and production. Climate Res., 40, 23–35, doi:10.3354/cr00802.
Yao, F., Y. Xu, E. Lin, M. Yokozawa, and J. Zhang, 2007: Assessing the impacts of climate change on rice yields in the main rice areas of China. Climatic Change, 80, 395–409, doi:10.1007/s10584-006-9122-6.
Zhang, X., M. A. Friedl, C. B. Schaaf, and A. H. Strahler, 2004: Climate controls on vegetation phenological patterns in northern mid- and high latitudes inferred from MODIS data. Global Change Biol., 10, 1133–1145, doi:10.1111/j.1529-8817.2003.00784.x.
