Hydropower is an important renewable energy source in China, but it is
sensitive to climate change, because the changing climate may alter
hydrological conditions (e.g., river flow and reservoir storage). Future
changes and associated uncertainties in China's gross hydropower potential (GHP)
and developed hydropower potential (DHP) are projected using
simulations from eight global hydrological models (GHMs), including a large-scale
reservoir regulation model, forced by five general circulation models
(GCMs) with climate data under two representative concentration pathways
(RCP2.6 and RCP8.5). Results show that the estimation of the present GHP of
China is comparable to other studies; overall, the annual GHP is projected to
change by
Hydropower is a dominant renewable source of energy production and has received significant worldwide attention for further development (Ramachandra and Shruthi, 2007; Resch et al., 2008; Liu et al., 2011; Hamududu and Killingtveit, 2012; Stickler et al., 2013). China has a large gross potential, exceeded only by Russia (Zhou et al., 2015); hydropower provides about 17 % of China's total electricity production (all technologies), and accounts for more than 80 % of the nation's electricity energy from renewable sources in 2012 (CNREC, 2013). A survey of hydropower resources showed that the gross hydropower potential (GHP, the total energy from all natural runoff at stream gradient over the entire domain) and the technically exploitable installed capacity (maximum possible hydropower generation) in China are 694 and 542 GW, respectively (Yan et al., 2006). Hydropower development in China has been greatly impelled by increasing environmental issues and energy demands (Huang and Yan, 2009; Liu et al., 2011; Lu, 2004). China's installed hydropower capacity (IHC) has grown by 11 % per year during the past decade and reached 248 GW by 2012, which is about 46 % of the technically exploitable potential of China (Liu, 2013; WEC, 2013). China will further foster its hydropower development in the near future (IEA, 2014) by targeting a total IHC of 350 GW in 2020, and most of it will be from the hydropower stations in southwest China (GOSC, 2014).
Reduced hydropower generation has been reported to be associated with climate change (Qiu, 2010; Bahadori et al., 2013), and significant progress has been made in assessing the impacts of climate change on hydropower elsewhere in the world. For example, it was reported that a future decrease in climate-change-induced runoff would reduce energy generation and revenues of hydropower plants under current regulations in the Columbia River and California hydropower systems in the United States (Hamlet et al., 2010; Vicuña et al., 2011). By assessing the hydropower system in the Peribonka River basin (Quebec, Canada), Minville et al. (2009) suggested that annual hydropower would decrease by 1.8 % for the 2010–2039 period and then increase by 9.3 and 18.3 % for the 2040–2069 and 2070–2099 periods, respectively. Considerable impact of climate change on hydropower was reported in the Swiss and Italian Alps regions, but the impacts varied for different locations, hydropower systems, and projections of climate change (Schaefli et al., 2007; Gaudard et al., 2014; Maran et al., 2014). Most studies suggested that new adaptive management may mitigate projected losses of hydropower in the Alps regions (Majone et al., 2016).
China is a large country (about 9.6 million km
Runoff has experienced a significant decrease in the past decades and is likely to decrease more for many areas in China in the future (Ma et al., 2010; Tang et al., 2013; Han et al., 2014; Sun et al., 2014; Schewe et al., 2014; Leng et al., 2015), which may significantly affect the water availability and the hydropower potential of rivers and at the current hydropower facilities. So far, most related studies have focused on the environmental and ecological impacts of the dams in China (Fan et al., 2015) or hydropower (potential) variations at country continental level (Zhou et al., 2015; van Vliet et al., 2016), but the impacts of climate change on the hydropower of China are seldom reported as part of regional studies, i.e., including higher-quality data than what are available for global applications. It was partly due to the lack of continental hydrological simulations and necessary reservoir information at a large scale. The Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP) (Warszawski et al., 2014) provided multimodel hydrological projections over the world, making it possible to investigate the implications of continental water resource changes under climate change. Therefore, it is of great interest to determine the impacts of future climate change on hydropower potential with respect to the underlying prosperous development of hydropower in China.
This study aims to present a regional overview of China's hydropower potential under future climate change by using the projections from eight global hydrological models (GHMs) under two future climate scenarios provided by the ISI-MIP. Changes in GHP are estimated to quantify the impacts of climate change on the total hydropower capacity, and changes in DHP (the developed hydropower potential of existing reservoirs) are estimated to examine the impacts on existing hydroelectric facilities through a hydropower scheme, which includes a reservoir operation module to regulate the simulated flow, similar to van Vliet et al. (2016). The study focuses on the hydropower potential but the changes in reservoir hydropower capacity caused by the development of hydroelectric facilities are not considered because changes in reservoir operations are to be optimized across multiple objectives – water supply and flood control in particular – and are prone to coordination between agencies and types of reservoir management (Tang et al., 2015). Nevertheless, this model-based analysis is expected to provide insight into future changes in current and additional potential hydropower generation of China, and to complement previous research studies at global scale (e.g., van Vliet et al., 2016). This study (1) assesses both gross and installed hydropower potential of China, and (2) provides an exhaustive uncertainty quantification with multimodel simulations, and thus (3) supports regional development of China by focusing on regional variability. The presented modeling framework is compatible with integrated assessment models (IAMs) which can combine socioeconomic analyses to further support the development of hydropower assets. This paper is organized as follows: Sect. 2 describes the method and data, Sect. 3 presents the results, Sect. 4 presents a discussion of the uncertainty associated with this study as well as the integration with socioeconomic analyses, and the last section presents the main conclusions.
Multimodel data are used to estimate GHP and DHP, and to address the
uncertainty in the simulations. Daily runoff and monthly discharge in China
are derived from eight GHMs: DBH (Tang et al., 2006, 2007b), H08 (Hanasaki
et al., 2008), Mac-PDM.09 (Gosling and Arnell, 2011), MATSIRO (Takata et
al., 2003), MPI-HM (Hagemann and Gates, 2003), PCR-GLOBWB (van Beek et al.,
2011), VIC (Liang et al., 1994), and WBMplus (Wisser et al., 2010) provided
by the ISI-MIP project (Warszawski et al., 2014). The model simulations are
driven by the same forcing data downscaled from CMIP5 climate projections of
the following GCMs: GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM,
and NorESM1-M (Hempel et al., 2013). Hydrological simulations were all
performed at a daily time step with a 0.5
GHP is defined as the total energy of natural runoff falling to the lowest
level (e.g., sea level) of a specific region. GHP is estimated from
discharge at each model grid cell: GHP
DHP is defined as the maximum possible hydropower generation at the existing hydroelectric facilities, which refers to all reservoirs in this study. Therefore, DHP is not the actual hydropower generation of the current hydropower plants since the latter is affected by many socioeconomic factors such as energy demand, electricity price, various water uses, etc. DHP is estimated using a hydropower scheme based on reservoir information, such as location, storage capacity, dam height, and IHC, but also the transient flow passing through the turbines, including flow release constraints (environmental, spill) and transient variations in the hydraulic head, as explained. Generally, reservoir regulation could fairly consider both socioeconomic factors and climate change in assessing the hydropower of a single reservoir or small regions (e.g., Madani and Lund, 2010; Pérez-Díaz and Wilhelmi, 2010; Wu and Chen, 2011; Gaudard et al., 2013). However, though it is possible to develop a large-scale hydrological model linking energy and water based on a conventional reservoir regulation approach, calibrating and validating the modeling for individual reservoirs and then implementing a comprehensive assessment of hydropower at a large scale remains challenging (Kao et al., 2015, van Vliet et al., 2016). Simplified, universal reservoir regulations were mostly adopted in large-scale hydrological modeling (Hanasaki et al., 2006; Döll et al., 2009; van Beek et al., 2011; Biemans et al., 2011) with limited regulation data and site-specific hydrological parameters.
In this study, socioeconomic factors (e.g., irrigation, energy price, and
demands) are generally not considered, although other human water use is
extensive in China (Tang et al., 2007a, 2008). The evaporation from water
surface of reservoirs is neglected for this analysis because it represents a
small fraction of the managed flow (Fekete et al., 2010; Liu et al., 2015).
We use the generic reservoir regulation rules from Hanasaki et al. (2006) to
derive the regulated flow at hydropower reservoirs. The reservoir operations
scheme has been extensively used under different development efforts in
large-scale studies (Biemans et al., 2011; Pokhrel et al., 2015; Döll et
al., 2009) and at a more regional scale (Voisin et al., 2013; Hejazi et al.,
2015). Regulation is set for flood control and then hydropower generation by
targeting monthly releases for the wet and dry seasons in a year. That is,
monthly release in a dry season is generally larger than monthly inflow of
reservoirs; it thus gradually reduces reservoir storage for flood control in
the coming wet season and provides more water for hydropower generation.
Monthly release (
Reservoir storage capacity and IHC at the provincial level in mainland China.
The DHP of a reservoir is then estimated based on the monthly release
(including spilling water): DHP
A total of 447 reservoirs in China were selected from the Global Reservoir and Dam
(GRanD) database (Lehner et al., 2011) if the key information of reservoirs
was available. All the reservoirs were treated as hydropower plants and
operations are determined with Eq. (1). Run-of-the-river plants are not
considered in this study for lack of hydropower station types in the current
database. They presently do not represent a major capacity, although it could
change in the future for specific peak hour operations. Most reservoirs are
located in east China (EC) and south central China (SCC), and a few of them
are in north China. There are no IHC data associated with the GRanD
reservoirs. The IHCs of provinces of China (about 130 GW in total) reported
by CREEDI (2004) are then used to determine the IHC of each reservoir as
follows. Firstly, the provincial IHC is adjusted by the ratio of GRanD
storage capacity to the reported storage capacity (National Bureau of
Statistics of China,
Medians of the GHPs of China across the ensemble of all GCM-GHM
combinations over the historical period (1971–2000). The red rectangular
denotes a hotspot region (HS1). NC: north China, NEC: northeast China, EC:
east China, SCC: south central China, NWC: northwest China, SWC: southwest
China. Inner plot
GHP is determined from all GHM and GCM combinations over the historical period (1971–2000), and the ensemble median of gridded and regional annual mean GHPs are plotted in Fig. 2. GHP distribution is dominated by the terrain and water availability showing high values in southwest China and most south China, which is generally similar to previous studies at the global scale (Pokhrel et al., 2008; Zhou et al., 2015).
The analysis of GHP and DHP consists of two parts. In the first step, median estimates of the GHP and DHP over all GCM-GHM combinations are evaluated spatially and regionally over three periods: historical (1971–2000), near future (averages over the 2020–2050 period, referred to as 2035), and the end of the century (average of the 2070–2099 period and referred to as 2085, hereafter). A regional analysis is performed by evaluating changes between the two future periods and the benchmark historical period for the two RCPs. Emphasis is on regional change differences and uncertainties related to the two RCPs. The regions analyzed are north China (NC), northwest China (NWC), southeast China (SEC), southwest China (SWC), SCC, and EC, as shown in Fig. 2. The interquartile range (IQR, the difference between the 75th and 25th percentiles of the multimodel ensembles is calculated to address the uncertainty of hydropower estimation between GCM-GHM combinations, and is shown in parenthesis following the median estimates.
A second analysis consists of looking at the temporal changes in the DHP and GHP over all of China and evaluating the overall trends in changes with respect to uncertainties related to interannual variability, and uncertainties related to individual GCM and GHM model structures. This analysis evaluates the changes in the DHP and GHP as time series. Estimates are computed as a 31-year moving average from 1971 to 2099 and labeled with the center year. The labeled period is accordingly reduced to 1986–2084, and the labeled period of 2010–2084 is shown for a clear view.
Estimates of annual and seasonal GHP (GW) for regions and China over 1971–2000.
To further support planning and mitigation for China hydropower, we define hydropower hotspots as regions with large hydropower potential or currently high IHCs. Two hotspot regions are isolated to illustrate the projected impacts of climate change on currently developed hydropower and hydropower that is in planning phases or under construction. One hotspot (HS1; see Fig. 2) covers the areas with significant untapped hydropower potential in SWC, including most of the Jinsha River, Yalong River, Nu River, and Lancang River; many hydropower plants in this area are in the planning phases or under construction. The other hotspot region encompasses the Sichuan (including Chongqing) and Hubei provinces (HS2), and accounts for about 50 % of total adjusted IHC in China in this study. We perform an additional analysis of expected changes over those two regions, which overlay parts and combinations of the main regions addressed in the first part of this paper.
The ensemble median of the GHP of China is 644 GW (IQR: 200 GW), with a bias
of
Medians of relative changes in the GHP of China over the
2010–2084 period under RCP2.6
Medians of relative changes in the annual mean GHPs for
2020–2050
Figure 3 shows the ensemble medians of China's seasonal and annual relative GHP changes under RCP2.6 and RCP8.5 for the 2010–2084 period. Relative changes are estimated by subtracting the historical GHP (i.e., GHP in Fig. 2) from future periods, then dividing by the historical GHP. As shown in Fig. 3a, the annual GHP shows a small (< 2 %) decrease before 2020 and afterwards increases by nearly 3 % under RCP2.6. Spring and summer GHPs generally show larger decreases (before 2035) and smaller increases than the annual GHP; the winter GHP shows little changes before 2060 and relative changes similar to the annual one thereafter. The autumn GHP shows a larger increase (about 8 %) at the late 21st century than annual ones. The annual GHP under RCP8.5 would decrease by about 3 % in 2020 and then increase after 2040. Annual, summer, and autumn GHPs would increase by 6 to 8 % in the late 21st century, while winter and spring GHPs show small changes, especially after the 2060. Annual GHP estimates often show a large spread across the GHMs and GCMs; e.g., the IQRs are as large as twice the median of the annual GHP in the late 21st century for both RCPs (see Tables S4 and S5).
Figure 4 shows the future relative annual GHP changes under RCP8.5. The total GHP of China is projected to decrease by 1.7 % (IQR: 5.8 %) in 2035 and increase by 6.3 % (IQR: 13 %) in 2085. The annual GHP shows smaller changes for both periods under RCP2.6: GHP is projected to increase by 2 % (IQR: 5.7 %) and 3 % (IQR: 7.2 %) in 2035 and 2085, respectively (see Fig. S1, Tables S4, and S5 in the Supplement). Estimations under RCP2.6 generally show spatial patterns similar to those under RCP8.5 but different magnitudes. For simplicity, we prefer to show RCP2.6 results in the Supplement, because the interregional analysis patterns are similar to RCP8.5.
Under RCP8.5, annual GHP is projected to decrease by more than 10 % in eastern SWC and NWC, and northern SCC, while in the Tibet region and most of NC, it is projected to increase by more than 5 % in the near term (2035). The increase in annual GHP will dominate the relative changes in 2085, which shows a smaller decrease with respect to the historical period in SWC, SCC, and NWC, and larger increases in NC and western SWC. GHP changes might have significant implications in SWC, where more than half of China's hydropower structures are located and most hydropower stations are in planning phases or under construction (Wang et al., 2013). The projected annual GHP changes generally agree with the spatial patterns of discharge changes in China, i.e., they decrease in a large part of southern China and increase in most of NC (see Fig. S2). In addition to annual changes in GHP, seasonal changes are computed in both 2035 and 2085 (see Fig. S3 and Table S6). In the spring, the GHP will decrease on the southern edge of China, which is opposite to the trend of annual GHP at this location. Summer GHP changes resemble the annual ones, and relative changes in the autumn and winter GHPs seem to be larger than annual changes.
Medians of relative changes in the DHPs of present reservoirs in
China over the 2010–2084 period under RCP2.6
Medians of relative DHP changes for present reservoirs in China
across the ensemble of GCM-GHM combinations for 2020–2050
Relative monthly DHP changes in regions in China for 2020–2050
Percentiles of annual DHP changes (%) for regions in China over 2020–2050 (2035) and 2070–2099 (2085) across the ensemble of GCM-GHM combinations.
Figure 5 shows the estimation of DHP for existing reservoirs based on the present IHC of China (see Fig. 1). The projected DHP shows a slower decrease in the late 21st century than in the near future. The annual DHP will decrease by 2.2 % (RCP2.6) and 5.4 % (RCP8.5) in 2035, and decrease by 1 % (RCP2.6) and 3.6 % (RCP8.5) in 2085. Under RCP2.6, the annual DHP will decrease by up to 3 % by around 2025, and decrease by about 1 to 2 % after 2050. Seasonal DHP shows changes similar to the annual one except the winter DHP decreases by about 4 % in 2035. Under RCP8.5, the annual DHP will keep decreasing by more than 4 % for most years after 2020. Seasonal DHP will keep decreasing in the future and will resemble the annual DHP, wherein summer shows a relatively smaller decrease and winter shows a larger decrease. However, the winter DHP is relatively small in a year. Therefore, the larger decrease in winter contributes little to annual DHP changes. Projected changes in the annual DHP also show large spreads across the GHMs and GCMs; e.g., IQRs/medians are greater than 2 for most years, especially for RCP2.6 (see Tables S7 and S8).
Note that the GHP would increase after 2040 for both RCPs, while the DHP would decrease more or less. This is mainly due to the fact that GHP mostly increases in NC and west China, where a small number of installed capacity plants are located, while GHP significantly decreases in SC and central China, where a large number of installed capacity plants (expected large DHP) currently are located in this study (see Figs. 1, 4, and S1).
Figure 6 shows the relative DHP changes of each reservoir in 2035 and 2085 under RCP8.5. More than 60 % of the reservoirs show an expected decrease in DHP in 2035 (Fig. 6a). Most DHP increases are less than 5 % in China, and only several reservoirs show more than a 10 % increase in NC. Most reservoirs in the southern EC and SCC show small change (less than 5 %) in DHP, while important reservoirs that decrease by more than 10 % are found in northern SCC and eastern SWC. Large decreases (more than 20 %) in DHP mainly occur in northern SCC and EC, as well as in some reservoirs in SWC and NWC. The relative DHP changes in 2085 (Fig. 6b) show a pattern similar to that in 2035, except that fewer reservoirs show large decreases in NWC and northern SCC, but more decreases are expected in the southern EC and SCC. Under RCP2.6, relative DHP changes generally resemble RCP8.5 patterns with larger increases in DHP in southern China and smaller decreases in DHP overall (see Fig. S5). It seems that hydropower is more sensitive to discharge reduction for large reservoirs; e.g., DHP shows a relatively large decrease at most reservoirs with large storage capacities, but only small changes at many small reservoirs in SCC and SWC (Fig. 6).
Figure 7 shows the regional relative changes in the aggregated monthly DHP
in 2035 and 2085 under RCP8.5 (numbers are shown in Table 2). In 2035
(Fig. 7a), the monthly DHP will decrease for most regions except for NC,
where it will increase slightly (less than 5 %). The monthly DHP in NEC
shows a very small increase from April to July, and decreases by about
In 2085 (Fig. 7b), the changes in monthly DHP are generally similar to
those in 2035, except for a larger increase (4.8 to 10 %) in NC,
larger decreases in May (
The monthly DHP changes in 2035 and 2085 under RCP2.6 (see Fig. S6 and Table 2) are smaller than those under RCP8.5. Monthly DHP changes are mostly less than 5 % and show more increases in NC, NEC, NWC, and SWC, especially in 2085; the annual DHP will decrease by 2.2 and 1.3 %, accounting for 0.7 and 0.4 % of total IHC, in 2035 and 2085, respectively.
Relative changes in the monthly GHPs (DHPs) and discharges (reservoir inflow) in the hotspot regions for 2020–2050 (2035) and 2070–2099 (2085) under RCP8.5. HS1: the hotspot region in southwest China (see Fig. 2); HS2: Sichuan and Hubei provinces (see Fig. 6). Grey areas denote the IQRs across the ensemble of GCM-GHM combinations.
We now analyze the temporal trends over the two identified hydropower hotspots (HS1 and HS2) as defined in the experimental approach.
Figure 8a shows the relative changes in the monthly GHP and discharge of HS1
in 2035 and 2085. The annual GHP in HS1 is projected to increase by about
2 % (IQR: 12 %) in 2035, during which relatively large increases will
occur in March, June, and from September to December, and considerable
decreases will occur in April (
For HS2, the monthly DHP will significantly decrease by 3.3 % (April) to 7.8 % (August) for most months, but increase slightly in May (0.9 %) and June (0.5 %) in 2035 (Fig. 8c). The monthly DHP in May and June will also increase by about 1 % in 2085 (Fig. 8d) and decrease by 1.2 % (July) to 8.6 % (November) in other months. The annual DHP will decrease by about 5.7 % (IQR: 5.4 %) and 5 % (IQR: 10.7 %), accounting for 1 and 0.9 % of total IHC, in 2035 and 2085, respectively. Changes in the monthly mean inflow are smaller than or relatively close to those for DHP in 2085. This indicates that reservoir regulation may offset the impact of discharge changes on hydropower to some degree. Under RCP2.6, the DHP in HS2 shows similar but smaller changes than those under RCP8.5 (see Fig. S7c, d). The monthly DHP will decrease by 0.5 to 5 % in 2035 and by less than 2 % for most months in 2085. The annual DHP will decrease by about 2.6 and 0.8 %, accounting for 0.46 and 0.13 % of total IHC, for 2035 and 2085, respectively.
For both hotspot regions, the uncertainties related to different GCM-GHM combinations are as large as the expected changes, but they are consistent in their monthly patterns and the direction of the annual changes for HS1 by the end of the century and HS2 for both the near term and the end of century.
A selection of hydropower potentials (GHP, DHP) in this study was estimated using multimodel simulations of runoff and discharge under different climate change scenarios. An ensemble of hydropower potential is generated that represents uncertainties due to model structure (multimodel) and emissions (RCPs). The combination of multiple potential estimates and the understanding of their regional diversity and associated uncertainties should provide support for integrated analysis from the generation perspective. A socioeconomic analysis, discussed hereafter, would complement the decision support analysis from the water and energy demand perspective. In this section, we further discuss the uncertainties, how the results relate to the modeling framework and the socioeconomic perspective.
Though the ensemble mean of projected GHP of China for the historical period is relatively close to the reported data, there is large discrepancy among GHMs. During the historical period, discrepancy in hydropower potential is much smaller among GCMs, because the GCM climate data are bias corrected to a historical reference. It implies that validation or bias correction may be helpful to reduce the uncertainty in the projections of GHMs. However, currently most GHMs, except for a few such as WaterGAP, are not calibrated against historical observations, and thus often show a large uncertainty in streamflow projections (Schewe et al., 2014). For annual estimates, it should be more effective and important to enhance the middle- and long-term hydrological prediction in order to fine tune the estimates of DHP and GHP. Therefore, validation and calibration of GHMs with hydrological observations (as with the WaterGAP model) are necessary in future studies which are effective in narrowing the differences among GHMs (Müller Schmied et al., 2014; Döll et al., 2016).
As stated, the uncertainty in the streamflow projections certainly propagates to the estimation of DHP. Though a universal reservoir regulation is applied to all modeled discharge, there is still a large spread across GCM-GHM combinations. The large uncertainty in DHP should be mainly due to the large discrepancy of GCM climate data since the reservoirs used in this study are mostly located in areas with low model agreements in future discharge projections (see Fig. 1 in Schewe et al., 2014). This also partly explains why the total DHP (Fig. 5) shows somewhat larger spread than the total GHP (Fig. 3) of China.
The projections of hydropower potential in this study are generally consistent with previous studies (e.g., Zhou et al., 2016; van Vliet et al., 2016), but this analysis is extended for different potential development scenarios (gross and installed hydropower potential) and focuses on regional variability of China in the future. The uncertainty sources in the GHP and DHP estimates are further discussed as follows, with respect to climate forcing, discrepancy among GHMs, reservoir regulation rules, and other sources like lack of consideration of socioeconomic factors in the hydropower estimation.
The GHP estimates are representative of the effect of climate change on hydropower if all of the natural runoff could be captured. The effect of climate change on DHP is an intermediate estimate that takes into consideration the regulation of reservoirs as if they were operated for hydropower only. The combined direct and indirect impacts of climate change on hydropower can be very complicated, e.g., the more frequent extreme heat and drought may reduce power generation capacity in the future (Bartos and Chester, 2015) which may affect the electricity supply from the state grid and therefore change the demand of hydropower generation. Nevertheless, the strong linkage between climate–streamflow–hydropower potential is one of the main ways that climate change affects hydropower. Climate change can directly modulate regional water availability, such as the increased temperature and depressed precipitation may give rise to drought events (Dai et al., 2004), while more intensive and spatially concentrated rainfall may result in more floods (Wasko et al., 2016). The consequent streamflow variations will then directly affect GHP and the reservoir storage which is associated with DHP. Besides the temperature and precipitation, several climate variables can alter river streamflow (e.g., Tang et al., 2013; Liu et al., 2014) and then affect hydropower potential. In this study, streamflow changes under changing climate conditions are projected by GHMs, and most of them include several climate variables, but two models use only temperature and precipitation as input (see Table S1). Therefore, this study presents the compound effects of the changes of multiple climate variables on hydropower potential. It is beyond the scope of this analysis to identify the contributions of all climate variables to the changes of hydropower potential. We limit the current analysis of meteorological forcing to a discussion on changes in temperature and precipitation, which are main drivers for streamflow projections.
As an example, the relative annual temperature and precipitation changes in
2035 and 2085 compared to the historical period are presented in Figs. S8 and S9,
respectively. Temperature is projected to increase by more than
1.5
The estimation of GHP is subject to large spread across the GCM-GHM ensemble, and most regions show poor agreement between models in the signs of the GHP changes (see Fig. S4). The large uncertainty in climate projections (Knutti and Sedlacek, 2013) certainly will propagate to the hydropower estimates as GHMs are sensitive to climate forcing (Müller Schmied et al., 2014). The GCM uncertainty predominates in GHP estimates in this study. For both future periods, the spread owing to differences between GCMs dominates the total ensemble spread in most of southern China for RCP8.5 and RCP2.6 and in some of NEC for RCP2.6 (Fig. S10). The ranges of GHP and DHP estimates across GHMs and GCMs are further summarized in Tables S4–S8, in terms of the 25th, 50th, and 75th percentiles of the ensemble of GCMs and GHMs, respectively.
The uncertainties in the DHP estimates, partially due to reservoir
operations, are evaluated by looking at empirical parameters (
In addition to the uncertainties in climate projections, GHMs, and regulation-related parameters, this study is also subject to considerable uncertainty in the estimates of DHP due to the simplification of reservoir regulation rules and limitations of the data. The different assumptions in the reservoir regulation are also a possible cause of the discrepancies compared to other studies (e.g., Fekete et al., 2010). The simplification of reservoir geometry for the computation of the hydraulic head may affect the DHP estimates in this study. More complex expression of the reservoir geometry may provide better approximation depending on the river geomography (Fekete et al., 2010) and then produce different DHP estimates. We further address here two main assumptions for estimating DHP: the effect of simple reservoir regulations and the lack of water withdrawals in the system for socioeconomic water demand.
Hydropower operations differ based on the reservoir storage and inflow characteristics, and if the plant is operated for general generation (daily load) or for capacity of generation during peak hour load. Furthermore, most reservoirs operated have multiple objectives to be combined with hydropower, implying even more complexities in hydropower operations. Large-scale water resources management modules have been efficient at evaluating the state of managed water resources over continents (Biemans et al., 2011). Although not operational models, the research models mimic reasonably well the impact of impoundment (regulation and withdrawals) on flow over large areas. Usually, two types of release rules are used. (1) The flood control/hydropower rule mostly releases water uniformly over the year with interannual variability in the release and minimizes spilling. With a quasi-constant release target over one whole operational year, the storage buffers the seasonality in flow (for larger reservoirs), which affects DHP estimates. (2) An irrigation rule tops off the reservoir storage as much as possible before the start of the irrigation season, and then releases water with a monthly pattern following the monthly demand anomalies. In our regulation scheme, we only used the flood control/hydropower rule with no water or energy demand information. The estimates of DHPs in this paper are therefore an upper bound of hydropower generation and consistent with the concept of potential with respect to an operational context with more complex water management for competitive water uses and uncertainties in water demand.
Adaptation of reservoir operations to the impact of climate change can be complex as they will need to be adjusted for both changes in water resources and increased competition between water uses (Vicuna et al., 2008, Vicuña et al., 2011; Finger et al., 2012; Jamali et al., 2013), and potentially to changes in energy demand and energy infrastructure as well. In our analysis where the reservoir operations depend on the historical long-term mean annual inflow, an increase (decrease) in mean annual inflow into reservoirs as well as the change in seasonality affect the estimation of DHPs. As release targets are maintained, the change in storage head for large reservoir storage and reservoir spilling/drying are driving the estimates of DHP changes. No adaptation measure was applied to our reservoir operations, as it would add another level of uncertainty to be further quantified and evaluated with respect to the competing water uses as addressed next.
The projected changes in hydropower potential indicate the impacts of
climate change, but they do not represent future prospects because
socioeconomic and technical evolutions are not considered. Therefore, more
uncertainty may arise:
Anthropogenic water use is expected to increase
along with the increase in temperature and population (Kendy et al., 2007;
Elliott et al., 2014; Leng and Tang, 2014); however, it is not considered.
Socioeconomic water demand has been the focus of recent research in global
integrated assessment and is taken into consideration in association with the
changes in population, industrial development, policy choices with respect
to carbon emissions, economy between countries, and energy demands (Hejazi
et al., 2014). Changes in infrastructure should also be taken into
consideration when estimating DHP. The South–North Water Diversion project,
which was designed to divert about 44.8 billion m Hydropower generation is affected by its potential
generation and by its integration into the electrical grid. The use of
pumped-storage hydroelectricity (Huang and Yan, 2009), potential changes in
seasonal energy demand, and the electricity price of the power grid may
affect the actual hydropower generation of a region or a reservoir. Variation in climate change-induced energy demand may also affect actual
hydropower generation; e.g., increasing temperature may lead to more energy
demand in summer and less in winter (Pereira-Cardenal et al., 2014). These
cumulative effects on hydropower merit further study and associated
uncertainty quantification.
The development of hydropower assets is driven by the potential production of the plants/region as well as by its economic value, which in turns depends on the energy demand, distance between the demand and the generation, and prices of other electricity generation technologies (natural gas and coal for example). Other factors such as policy, technology development, electricity market, city expansion, and industry location can affect the economic value of hydropower. This analysis supports the first step for developing hydropower assets (potential generation) and is driven by the natural resources conditions (climate and hydrology). It also provides an extensive uncertainty quantification. We discussed some sensitivity analyses and adaptation approaches. Beyond mitigation and adaptation in reservoir operations, a lower DHP would most likely need to be balanced by energy production from other sources, likely from costlier technologies, implying regional economic impact, which would need to be taken into consideration in the socioeconomic analyses. The socioeconomic analysis is beyond the scope of this paper. However, the regional assessment performed in this study provides the necessary information for integration into the regional version of IAMs for national sustainable mitigation and adaptation policy making. It also provides a first assessment for regional developers to create case studies in specific sites to determine the feasibility from a natural resource and economical perspective.
An overview of projected future changes in the GHP and DHP of China was
presented using hydrological simulations derived from multiple GHMs and
GCMs. A reservoir regulation scheme was incorporated to estimate the DHP
using current infrastructure. Historical GHP simulation was evaluated at the
regional scale (overall bias
The GHP of China is projected to change by
The annual DHP in China will decrease by about 2.2 to 5.4 % (0.7–1.7 % of total IHC) and 1.3 % to 4 % (0.4–1.3 % of total IHC) from 2020 to 2050 and from 2070 to 2099, respectively. These changes are mostly contributed by the large decrease in SCC and EC, where most reservoirs and large IHCs are located currently. The DHP will decrease with some regional disparities as well. It will mainly decrease in southern China (e.g., EC, SCC, and part of SWC), and will increase considerably in NC and the region of Tibet. China's DHP also shows a small decrease in late spring and early summer and a relatively large decrease in other months.
The impact of climate change on hydropower is particularly of concern in two identified hotspot regions that have rich hydropower potential. One hotspot located in SWC shows increases of nearly 2 to 6 % and 4 to 11 % in the annual GHP from 2020 to 2050 and from 2070 to 2099, respectively. This region has the most hydropower plants currently in planning phases or under construction, and will be the most important region for targeting hydropower development in China in the near future (GOSC, 2014). The result herein suggests the necessity of considering climate change for future hydropower development. In another hotspot region – the Sichuan and Hubei provinces – which holds nearly half of China's total IHC and is closer to urban centers, the DHP will decrease by 2.6 to 5.7 % (0.46–0.97 % of total IHC) and 0.8 to 5 % (0.13–0.91 % of total IHC) from 2020 to 2050 and 2070 to 2099, respectively. Though the DHP seasonality would be optimized (e.g., retain a high water level or increase release) to reduce the effects of monthly inflow decrease for certain years, it is mostly subject to the streamflow seasonality during a long-term period according to the parameterization of reservoir regulation. In this hotspot region, relatively small changes of monthly DHP will occur in late spring and early summer, while large decreases will occur in other months. If actual hydropower changes proportionally to the DHP under climate change, the reservoirs in this region might be unable to provide as much hydropower generation as in the present day. The significant DHP decrease in dry season (e.g., in winter) will further increase challenges to managing the competitive water uses and regulation of reservoirs.
The projected effects of climate change on GHPs and DHPs of China are related but in the opposite direction of change because most areas with high IHC show decrease in DHP while most areas with high GHP show increase in GHP. Even though GHPs are generally projected to increase by the second half of the 21st century, DHPs given the current infrastructure will not be able to mitigate the hydrological changes and thus will decrease without future update of regulation rules. Those trends tend to be consistent even under the range of uncertainty captured by an ensemble of global climate models, hydrological models, and two bounding climate change scenarios.
Hydropower is a relatively cheap and clean energy, which also facilitates the penetration of other renewable energy (solar and wind) into the grid. Both the numerous exploitable potential and the well-advanced technology of hydropower would greatly help China reduce its greenhouse gas emissions and environmental pollution while developing its economy (Kaygusuz, 2004; Chang et al., 2010; Hu and Cheng, 2013; Liu et al., 2013). It is necessary to involve climate risk assessment in hydropower development because the projected decrease in flow under climate change conditions will lead to reduced potential hydropower generation with high regional disparities. China has shown strong motivation to develop large hydroelectric facilities in the future. This motivation is not only driven by hydropower energy but also by flood control or irrigation demands. This research provides a preliminary regional assessment of climate change impacts on hydropower potential, which could guide the development of hydropower technology, e.g., pumped-storage hydroelectricity, interbasin transfer, and joint reservoir operations, in order to mitigate the impact of climate change on renewable electricity generation in China.
All the discharge and runoff data used to estimate the hydropower potential
are publicly available from the Inter-Sectoral Impact Model Intercomparison
Project (ISI-MIP,
Q. Tang and X. Liu designed the research, N. Voisin processed the reservoir data, X. Liu conducted analyses, Q. Tang, X. Liu, N. Voisin, and H. Cui wrote the manuscript.
We acknowledge the modeling groups (listed in Table S1 of this paper) and the ISI-MIP coordination team for providing the model data. This research is supported by the National Natural Science Foundation of China (41425002 and 41201201) and the National Youth Top-notch Talent Support Program in China. Edited by: T. Kjeldsen Reviewed by: two anonymous referees