Environ Sci Pollut Res DOI 10.1007/s11356-015-4083-9
RESEARCH ARTICLE
Methane and CO2 emissions from China’s hydroelectric reservoirs: a new quantitative synthesis Siyue Li & Quanfa Zhang & Richard T. Bush & Leigh A. Sullivan
Received: 25 May 2014 / Accepted: 4 January 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract Controversy surrounds the green credentials of hydroelectricity because of the potentially large emission of greenhouse gases (GHG) from associated reservoirs. However, limited and patchy data particularly for China is constraining the current global assessment of GHG releases from hydroelectric reservoirs. This study provides the first evaluation of the CO2 and CH4 emissions from China’s hydroelectric reservoirs by considering the reservoir water surface and drawdown areas, and downstream sources (including spillways and turbines, as well as river downstream). The total emission of 29.6 Tg CO2/year and 0.47 Tg CH4/year from hydroelectric reservoirs in China, expressed as CO2 equivalents (eq), corresponds to 45.6 Tg CO2eq/year, which is 2-fold higher than the current GHG emission (ca. 23 Tg CO2eq/year) from global temperate hydropower reservoirs. China’s average emission of 70 g CO2eq/kWh from hydropower amounts to 7 % of the emissions from coal-fired plant alternatives. China’s hydroelectric reservoirs thus currently mitigate GHG emission when compared to the main alternative source of electricity with potentially far great reductions in GHG emissions and benefits possible through relatively minor changes to reservoir management and design. On average, the sum of Responsible editor: Gerhard Lammel Electronic supplementary material The online version of this article (doi:10.1007/s11356-015-4083-9) contains supplementary material, which is available to authorized users. S. Li (*) : R. T. Bush : L. A. Sullivan Southern Cross GeoScience, Southern Cross University, Military Road, PO Box 157, Lismore, NSW 2480, Australia e-mail: [email protected] S. Li e-mail: [email protected] S. Li : Q. Zhang Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Garden, The Chinese Academy of Sciences, Wuhan 430074, China
drawdown and downstream emission including river reaches below dams and turbines, which is overlooked by most studies, represents the equivalent of 42 % of the CO2 and 92 % of CH4 that emit from hydroelectric reservoirs in China. Main drivers on GHG emission rates are summarized and highlight that water depth and stratification control CH4 flux, and CO2 flux shows significant negative relationships with pH, DO, and Chl-a. Based on our finding, a substantial revision of the global carbon emissions from hydroelectric reservoirs is warranted. Keywords Hydropower reservoirs . Greenhouse gas (GHG) . CO2 emission . Methane emission . Green energy . Carbon budget
Introduction Freshwaters, including lakes, rivers, and reservoirs, are substantial contributors to the global carbon budget (e.g., St. Louis et al. 2000; Richey et al. 2002; Lima et al. 2008; Aufdenkampe et al. 2011; Barros et al. 2011; Bastviken et al. 2011; Butman and Raymond 2011; Sarma et al. 2011; Li et al. 2012; Raymond et al. 2013; Li and Bush 2015). The combined aquatic carbon emission of 1.4–2.1 Pg C/year as carbon dioxide (CO2) surpasses the annual riverine carbon export of 1 Pg C/year to the oceans and, thus, represents a key component of global carbon biogeochemical cycling (Amiotte Suchet et al. 2003; Tranvik et al. 2009; Li et al. 2012; Lu et al. 2012; Raymond et al. 2013). The total carbon emission from inland waters, including methane (CH4) and CO2, balances the terrestrial carbon sink (cf. Bastviken et al. 2011). Of the major greenhouse gases, CH4 is particularly potent. The greenhouse effect of methane has progressively been revised upward to currently accepted level of 34 times
Environ Sci Pollut Res
greater than CO2 on an equivalent mass basis over a 100-year period horizon (cf. Shindell et al. 2009). Highly supersaturated aqueous CO2 concentrations relative to atmospheric equilibrium are contributable to upstream CO2 inputs and in situ mineralization of organic carbon with terrigenous origins in particular. CH4 emission at air-water interface is more complex. CH4 is generated in anoxic sediment by methanogens, while a large fraction of CH4 produced is likely oxidized by methanotrophs during the pelagic transport to epilimnion (Bastviken et al. 2008, 2011; Li and Zhang 2014). There are three pathways for GHG emissions, i.e., diffusion, ebullition, and degassing of waters passing through turbines and spillways and downstream river from hydropower reservoirs (Hertwich 2013). Prior studies showed that CO2 ebullitive emission is minor, and CH4 ebullition could be significant which largely depends on water depth. GHG emissions from waters passing through turbines and spillways are huge, which is clearly evident that intakes of turbines and spillways are located dozens of meters below the water surface for China’s hydropower reservoirs (Yang et al. 2014 and references therein). The dissolved GHGs in the hypolimnion would rapidly release into the atmosphere when the deep water with remarkable higher dissolved concentrations of GHGs passes through turbines and spillways due to the abrupt changes in pressure and temperature. Downstream rivers are also important sources of GHGs because of strong turbulence by water from dams, and current studies on tropical reservoirs demonstrate that a proportion of 2–32 % of CO2 and 9– 33 % of CH4 from reservoir surface is contributed by downstream rivers (Guerin et al. 2006; Kemenes et al. 2007, 2011). Although numerous studies have provided estimates of GHG emissions from freshwater environments, a lack of quantitative data has constrained accurate estimates of CH4 and CO2 contributions from major freshwaters such as artificial reservoirs, especially for Asia (cf. Barros et al. 2011). Understandably, with limited data to estimate carbon emissions, the reported values for CO2 and CH4 releases from reservoirs range widely (Table S1). The large range in published estimates reflects both the limitation of data and inherent differences in the various methods used for extrapolation (St. Louis et al. 2000; Lima et al. 2008; Barros et al. 2011). For example, for temperate reservoirs, the range is from 2.8 (Barros et al. 2011) to 12 mg CH4/m2/day (Lima et al. 2008), to 55 mg CH4/m2/day (Soumis et al. 2005). Further, the rates of CH4 release by global hydroelectric reservoirs range from 4 (Barros et al. 2011) to 70 Tg CH4/year (St. Louis et al. 2000), to 100 Tg CH4/year (Lima et al. 2008). Barros et al. (2011) first estimated CO2 and CH4 emissions from hydroelectric reservoirs using data from 85 globally distributed hydroelectric reservoirs. Their estimated annual carbon emission by hydroelectricity of 48 Tg C as CO2 and 27 Tg C-CO2 eq. as CH4 (3 Tg C-CH4) was far lower than an earlier estimate of 709 Tg C as CO2 equivalents per year with a
consideration of the matter of area difference (St. Louis et al. 2000), while a more recent estimate revised the data to be 76 Tg CO2-C/year and 7.3 Tg CH4-C/year (Hertwich 2013). The globally averaged CH4 and CO2 fluxes from hydroelectric reservoirs between 25° and 50° latitude were respectively 2.8 mg CH4/m2/day and 387 mg CO2/m2/day by Barros et al. (2011). These fluxes were remarkably lower than past observations (Table S1). For example, the CH4 flux was around 1/4 of that estimated by Lima et al. (2008) and only 1/20 that from Soumis et al. (2005), while the CO2 flux was between 1/4 to 1/2 that given in previous reports (St. Louis et al. 2000; Aufdenkampe et al. 2011). Moreover, the reservoirs examined in all of these studies were in the Americas and Northern Europe, and no data from Asia were included in these global estimates. This is clearly a major issue for China, where there is already a large number of hydroelectric reservoirs and many more planned for the future. Also, the drawdown area with typically much higher fluxes of CH4 emission (cf. Chen et al. 2009, 2013; Yang et al. 2012) has not been included in previous global and regional scenarios (St. Louis et al. 2000; Barros et al. 2011; Bastviken et al. 2011). Further, there is a lack of information on dam downstream carbon emissions such as spillways and turbines where a large share of GHGs especially methane emissions from reservoirs occurs (cf. Guerin et al. 2006; Roehm and Tremblay 2006; Kemenes et al. 2007; 2011; Yang et al. 2014), which further complicates the quantification of GHG emissions from inland waters. The clean energy credentials of hydroelectricity have come under increasing scrutiny due to the potential emissions of GHGs (Giles 2006). Some research based primarily on data from Brazil (e.g., Fearnside 1995, 2005; Fearnside and Pueyo 2012) indicate that GHGs, particularly CH4 emissions, discount the “green” credentials of hydropower. Likewise, previous studies of China’s hydropower dams also indicated high GHG emissions particularly in the seasonal drawdown area (Chen et al. 2009; Qiu 2009; Yang et al. 2012; Li and Zhang 2014). The carbon fluxes from reservoirs are relevant to reservoir age, size, latitude, and environmental factors (Barros et al. 2011), resulting in spatial and temporal heterogeneity in GHG fluxes (Table S2). In this study, we compile CO2 and CH4 emissions from Web of Science and Chinese Journals and quantify carbon emission from hydropower reservoirs in China by integrating reservoir surface area (water area), drawdown area, and reservoirs downstream (including spillways and turbines as well as rivers downstream the dam). Major controlling factors such as limnological and environmental variables of water temperature, pH, DO, Chl-a, and water depth on CH4 and CO2 fluxes are also identified. Our study ultimately will help guide global estimates and future trends in GHG emissions by making urgently needed Chinese data available for global modeling (most data are published in Chinese journals and are not available for international scholars).
Environ Sci Pollut Res
Materials and methods Data sources There are 19 hydropower reservoirs (66 study results) with data available on carbon emissions from ISI and Chinese journals (Fig. 1, Table S2). These reservoirs are located between 26° N and 46° N, spanning the temperate zone where China’s hydropower reservoirs are concentrated. Data on CO2 release in the river reach downstream of dam is also supplied. The global warming potential (GWP), i.e., 21–25, for the mass basis of CH4 relative to CO2 is recommended by IPCC (Intergovernmental Panel on Climate Change); however, a GWP of 34 is adopted in our study from the more recent work (Shindell et al. 2009). Thus, CO2 mass equivalent is calculated by summing the CO2 with CH4, multiplied by 34 to account for the specific long-lasting GWP of CH4 over a 100-year time horizon. Our collected data of CH4 emission fluxes are determined by floating chamber with gas chromatography (GC); CO2 fluxes in TGR (Three Gorges Reservoir), Nihe, and Shuibuya Reservoirs are determined by floating chamber with GC, while CO2 fluxes in Xin’anjiang, Wan’an, Hongfeng, Baihua, Xiufen, Hongyan, Hongjiadu, and Danjiangkou Reservoirs are calculated using a diffusive model with Henry’s constant (see SI context). Aquatic partial pressure of CO2 (pCO2) in the Wan’an and Xin’anjiang Reservoirs is measured using a continuous measurement system (equilibrator–non-dispersive infrared system), while pCO2 levels in the
Nihe Miyun
Danjiangkou Ertan
TGR Shuibuya Wan’an
Puding Yinzidu Dongfeng
Hongfeng
Suofengying
Baihua
Wujiangdu
Xiuwen
Wuqiangxi
Hongyan Hongjiadu
Fig. 1 Hydropower reservoirs with available GHG emission rates in China
other hydroelectric reservoirs are from pH and alkalinity. The limnological and environmental characteristics such as water temperature, pH, DO, and Chl-a are determined in site using the multi-water quality monitoring sondes after calibration (i.e., YSI 6600; Yellow Springs, USA). Water samples are collected in a depth of circa 50 cm below the water surface using previously acid-washed 5 L high-density polythene (HDPE) containers throughout the reservoir area. Alkalinity is titrated in triplicates using HCl (0.010–0.020 mol/L) in situ to pH 4.5 following the addition of a Methyl orange indicator (cf. Wang et al. 2011). Nutrients (N, P) are determined by a spectrophotometer and organic carbon by TOC analyzer. Major anions (Cl− and SO42−) are analyzed by ion chromatography, while major cations are analyzed by ICP-OES. The typical precisions of major ions are about ±5 %. The detailed information of datasets, limnological and environmental variables, sampling, and analyses are illustrated in Tables S2 and S3 (see citations in the supplementary document). Source apportionment of GHG emission GHG from classical reservoir surface GHG emission from reservoir surface was based on data collection from national and international literatures (see Table S2). GHG from reservoir downstream The immediate degassing of GHGs as water emerges from turbines and spillways are quantified using the total volume of water passing through turbines and the differences in CH4/ CO2 concentrations between the reservoir water at the turbine depth and the water below the dam (([CO2/CH4]above −[CO2/ CH4]below)×water outflow) (Roehm and Tremblay 2006). No measurement has been carried out on GHGs degassing from dams; here, the decrease in aquatic CO2 concentration (circa 30 μmol/L) through turbines is designated based on two cases (Fig. S1), and an estimated uncertainty of 15–45 μmol/L is accounted for extrapolation. A decrease of circa 30 % of CH4 concentration by water passing through turbines and spillways is due to immediate degassing (ca. 4 μmol/L) (Yang et al. 2009); this proportion is at the lower end of the immediate degassing ranges for tropical reservoirs (cf. Guerin et al. 2006). We also test errors using a possible variability of 50 %; thus, 2–6 μmol/L as immediate degassing is used for extrapolation. The estimates are comparable to source apportionment models from three well-studied reservoirs (Petit Saut, Balbina, and Samuel) (Tables 1 and 2). Area-specific fluxes of GHGs from river reaches below dams are much higher than upstream ones due to strong turbulence in the downstream rivers (Table S1). Very limited data and absences in downstream water areas affected by water
Environ Sci Pollut Res Table 1
CH4
Source proportions of CH4 and CO2 in the well-studied tropical hydroelectric reservoirs (Petit Saut, Balbina, and Samuel) Reservoir surfacea
Downstream Spillways and turbines
Reservoir surface/downstream=1/1b
80 % reservoir surface
20 % (9–33 %) reservoir surface
Guerin et al. 2006
100 % rivers downstreamc
15 % (7–25 %) reservoir surface
Guerin et al. 2006
CO2 a
Rivers downstream
Sources
Drawdown area is excluded from surface area
b
Kemenes et al. (2007), while Fearnside and Pueyo (2012) revise to 1/2
c
Kemenes et al. (2011)
releases from dams prevent our accurate estimates of GHG emissions from downstream rivers. Several studies on tropical reservoirs indicate that CO2 emission from downstream rivers accounts for 1.6–32 % of CO2 from reservoir surface, while downstream rivers make up 9–33 % of CH4 emission from reservoir water (Guerin et al. 2006; Kemenes et al. 2007, 2011). If these source apportionments are adopted in our calculations, downstream river emissions of GHGs comprise a small part of the total emissions; thus, the source models from tropical reservoirs show little effects on total quantifications and are considered as reliable. GHG diffusive flux in the reservoir drawdown area The littoral zones (drawdown area) in the reservoirs, which have been neglected in previous studies, are considered as notably high CH4 emitters (see Table S2). Water level fluctuations and land use determine CH4 fluxes in the drawdown area (cf. Yang et al. 2012). For example, from Table S2, a very high flux (119.8±42.2 mg CH4/m2/day) occurs in the rice
paddies and littoral marshes, whereas the other land use is characterized by a much lower mean flux (4.7±1.7 mg CH4/ m2/day). The inundated and drained periods for drawdown zones with varied elevation control the source or sink of GHGs; for example, dryland (cropland, fallow, and deforested land) in the drawdown area shows markedly higher CH4 emission flux in the inundated days, while rice paddy located in the upper zone of the drawdown area shows markedly higher CH4 emission flux in the drained days based on the previous observations in the TGR (Lu et al. 2011; Yang et al. 2012). For this study, these two significant land-use systems (rice paddy and dryland) for CH4 flux are categorized, and the total CH4 emission (FE) from the drawdown area is calculated as follows (Yang et al. 2012):
FE ¼
i¼b X
½Pi f in þ ð365−Pi Þ f dr Ai
i¼a
þ ½P0 f 0 in þ ð365−P0 Þ f 0 dr A0
ð1Þ
Table 2 Source proportion development of CH4 and CO2 from reservoir surfaces classically taken into account for reservoirs and downstream emissions in our study
CH4 CO2 a
Reservoir surfacea
Downstream Spillways and turbines
Rivers downstreame
Sources
Reservoir surface/downstream=1/9b
Schematic budgetc that is equal to 9-fold reservoir surfaceb Schematic budgetd
20 % reservoir surface 15 % reservoir surface
This study This study
Drawdown area is excluded from surface area
b
Upstream emission is proportional to the reservoir area, while the downstream emission is proportional to the streamflow and turbine depth (Fearnside and Pueyo 2012). Considering the very large surface area (∼3000 km2 ), small water outflow (577 m3 /s), and lower turbine depth (16 m) for Balbina, as for Tucurui with a similar surface area (2875 km2 ) as Balbina but 19 times (11,000 m3 /s) more water outflows and a turbine depth of 30 m, and the downstream CH4 emission including diffusive flux and bubbling at Tucurui represents a magnitude of nine times the CH4 emission from reservoir surface. This ratio is revised to 18 if the methodological factor is considered (Fearnside and Pueyo 2012). Valley-type Chinese reservoirs have relatively small surface area with larger water discharge and turbine depth greater than 30 m based on our data, thus a conservative ratio of 9 for downstream CH4 emissions to that from reservoir surface is designated. CH4 emission from turbines and river stretch below dam is proportional to increase c
Quantification using the differences between aquatic CH4 concentrations in turbine inflow and turbine outflow multiplying water volume passing through the turbines and spillways (Roehm and Tremblay 2006), which is comparable to the estimate from developed models through tropical reservoirs
d
CO2 emissions from turbines are calculated in the following: the differences between aquatic CO2 concentrations at turbine intake depth and in water discharge multiply water volume through turbine (Roehm and Tremblay 2006)
e
Ratios of CH4 and CO2 (GHG emissions from downstream river to reservoir surface) of 0.2 and 0.15 from tropical reservoirs is respectively adopted in our calculations. Because the emissions from downstream rivers are minor in comparison to total emissions from hydropower reservoirs, these uncertainties could be neglected. Schematic budget is the model of ([CO2/CH4]above −[CO2/CH4]below)×water outflow. This model is compared to source apportionments based on tropical reservoirs
Environ Sci Pollut Res
where i is the elevation from a to b (m), Pi is inundated days for dryland at varied elevation, and fin and fdr are the GHG fluxes of dryland in the inundated and drained durations, respectively. Ai is the submerged area at varied elevation. P′ is the inundated days of the rice paddy; f′in and f′dr are the GHG fluxes of the rice paddy during the inundated and drained seasons, respectively. A′ is the total area of the rice paddy. Here, we assume that areas with extremely high CH4 flux such as rice paddies and littoral marshes account for 10 % of the reservoir area, and the total drawdown area accounts for 30 % of the total reservoir area (cf. Chen et al. 2009). Accounting for the estimated uncertainty, a possible variability of 0–60 % is designated depending on seasonal hydrology. Due to similar geomorphological characteristics such as the locations of steep mountain areas for China’s hydropower reservoirs, our calculations are based on some simplifications with a combination of water level fluctuation and measurements in the TGR (cf. Lu et al. 2011; Yang et al. 2012). First, we neglect the dynamic duration of inundation at each elevation and designate an average inundated days for the dryland in the drawdown zone: Pi =200, fdr =0, and Ai is on average 20 % of the total reservoir area. Second, f′in =fin, P′=80. Eq. (1) can thus be simplified in the following:
shallow water; for example, bubbling CH4 emission occurs only where water depth is lower than 10 m (cf. Guerin et al. 2006). However, China’s valley reservoirs with gorge topography when coupled with removal of vegetation below the flooding lines effectively limit the ebullitive CH4 emission to the reservoir surface. In the drawdown area, soft vegetation and organic matter accumulation would facilitate CH4 production, resulting in the potential of bubbling flux. Thus, CH4 emission from China’s reservoir surfaces is confined largely to diffusion (Zhao et al. 2013), while bubbles are considered in the drawdown area (Yang et al. 2012; Table 3). The low ebullition flux (1.63 mg CH4/m2/day) of methane from TGR is considered for the drawdown dryland, while a much higher ebullitive flux (100 mg CH4/m2/day) is considered for littoral marshes with vegetation. This level is comparable to other temperate reservoirs with similar hydrological features (i.e., DelSontro et al. 2010) and the extremely high methane flux in the TGR marshes reported by Chen et al. (2009). With regard to CH4 emission quantifications in the drawdown zones, the inundated days are 200 and areal proportion of dryland and littoral marsh is respectively 10 and 20 % of the total reservoir based on Eq. (2) (simplifications are discussed earlier).
FE ¼ 200 f in 20%Atotal þ ð80 f in þ 140 f 0 dr Þ
Statistical analysis
10%Atotal
ð2Þ
Here, fin is the CH4 emission rate for drawdown area, Atotal is the total reservoir area, and f′dr is the CH4 emission rate of rice paddies and marshes (see Table S2). For CO2 emission calculations, we give a mean submergence time of 200 days and annual drawdown area of 30 % of the total reservoir area (cf. Chen et al. 2009; Yang et al. 2012). A CO2 flux of 2110 mg CO2/m2/day for temperate wetlands was adopted for areal outgassing from our reservoir drawdown zones (Aufdenkampe et al. 2011). This level is comparable to the lower limit of flux for marshes in China (Song et al. 2003; Yang et al. 2008). The total reservoir area used here is 28, 000 km2 with a storage of 600 km3, which were taken from China’s reservoirs registered in the ICOLD (International Commission on Large Dams; http://www.icold-cigb.net/).
We tested the data for normality using the KolmogorovSmirnov test before the statistical analyses. Variables that were non-normally distributed were natural logarithm (ln) transformed to satisfy normality assumptions. Analysis of variance (ANOVA) was performed to quantify the seasonal and spatial differences in the CO2 and CH4 fluxes (p
RESEARCH ARTICLE
Methane and CO2 emissions from China’s hydroelectric reservoirs: a new quantitative synthesis Siyue Li & Quanfa Zhang & Richard T. Bush & Leigh A. Sullivan
Received: 25 May 2014 / Accepted: 4 January 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract Controversy surrounds the green credentials of hydroelectricity because of the potentially large emission of greenhouse gases (GHG) from associated reservoirs. However, limited and patchy data particularly for China is constraining the current global assessment of GHG releases from hydroelectric reservoirs. This study provides the first evaluation of the CO2 and CH4 emissions from China’s hydroelectric reservoirs by considering the reservoir water surface and drawdown areas, and downstream sources (including spillways and turbines, as well as river downstream). The total emission of 29.6 Tg CO2/year and 0.47 Tg CH4/year from hydroelectric reservoirs in China, expressed as CO2 equivalents (eq), corresponds to 45.6 Tg CO2eq/year, which is 2-fold higher than the current GHG emission (ca. 23 Tg CO2eq/year) from global temperate hydropower reservoirs. China’s average emission of 70 g CO2eq/kWh from hydropower amounts to 7 % of the emissions from coal-fired plant alternatives. China’s hydroelectric reservoirs thus currently mitigate GHG emission when compared to the main alternative source of electricity with potentially far great reductions in GHG emissions and benefits possible through relatively minor changes to reservoir management and design. On average, the sum of Responsible editor: Gerhard Lammel Electronic supplementary material The online version of this article (doi:10.1007/s11356-015-4083-9) contains supplementary material, which is available to authorized users. S. Li (*) : R. T. Bush : L. A. Sullivan Southern Cross GeoScience, Southern Cross University, Military Road, PO Box 157, Lismore, NSW 2480, Australia e-mail: [email protected] S. Li e-mail: [email protected] S. Li : Q. Zhang Key Laboratory of Aquatic Botany and Watershed Ecology, Wuhan Botanical Garden, The Chinese Academy of Sciences, Wuhan 430074, China
drawdown and downstream emission including river reaches below dams and turbines, which is overlooked by most studies, represents the equivalent of 42 % of the CO2 and 92 % of CH4 that emit from hydroelectric reservoirs in China. Main drivers on GHG emission rates are summarized and highlight that water depth and stratification control CH4 flux, and CO2 flux shows significant negative relationships with pH, DO, and Chl-a. Based on our finding, a substantial revision of the global carbon emissions from hydroelectric reservoirs is warranted. Keywords Hydropower reservoirs . Greenhouse gas (GHG) . CO2 emission . Methane emission . Green energy . Carbon budget
Introduction Freshwaters, including lakes, rivers, and reservoirs, are substantial contributors to the global carbon budget (e.g., St. Louis et al. 2000; Richey et al. 2002; Lima et al. 2008; Aufdenkampe et al. 2011; Barros et al. 2011; Bastviken et al. 2011; Butman and Raymond 2011; Sarma et al. 2011; Li et al. 2012; Raymond et al. 2013; Li and Bush 2015). The combined aquatic carbon emission of 1.4–2.1 Pg C/year as carbon dioxide (CO2) surpasses the annual riverine carbon export of 1 Pg C/year to the oceans and, thus, represents a key component of global carbon biogeochemical cycling (Amiotte Suchet et al. 2003; Tranvik et al. 2009; Li et al. 2012; Lu et al. 2012; Raymond et al. 2013). The total carbon emission from inland waters, including methane (CH4) and CO2, balances the terrestrial carbon sink (cf. Bastviken et al. 2011). Of the major greenhouse gases, CH4 is particularly potent. The greenhouse effect of methane has progressively been revised upward to currently accepted level of 34 times
Environ Sci Pollut Res
greater than CO2 on an equivalent mass basis over a 100-year period horizon (cf. Shindell et al. 2009). Highly supersaturated aqueous CO2 concentrations relative to atmospheric equilibrium are contributable to upstream CO2 inputs and in situ mineralization of organic carbon with terrigenous origins in particular. CH4 emission at air-water interface is more complex. CH4 is generated in anoxic sediment by methanogens, while a large fraction of CH4 produced is likely oxidized by methanotrophs during the pelagic transport to epilimnion (Bastviken et al. 2008, 2011; Li and Zhang 2014). There are three pathways for GHG emissions, i.e., diffusion, ebullition, and degassing of waters passing through turbines and spillways and downstream river from hydropower reservoirs (Hertwich 2013). Prior studies showed that CO2 ebullitive emission is minor, and CH4 ebullition could be significant which largely depends on water depth. GHG emissions from waters passing through turbines and spillways are huge, which is clearly evident that intakes of turbines and spillways are located dozens of meters below the water surface for China’s hydropower reservoirs (Yang et al. 2014 and references therein). The dissolved GHGs in the hypolimnion would rapidly release into the atmosphere when the deep water with remarkable higher dissolved concentrations of GHGs passes through turbines and spillways due to the abrupt changes in pressure and temperature. Downstream rivers are also important sources of GHGs because of strong turbulence by water from dams, and current studies on tropical reservoirs demonstrate that a proportion of 2–32 % of CO2 and 9– 33 % of CH4 from reservoir surface is contributed by downstream rivers (Guerin et al. 2006; Kemenes et al. 2007, 2011). Although numerous studies have provided estimates of GHG emissions from freshwater environments, a lack of quantitative data has constrained accurate estimates of CH4 and CO2 contributions from major freshwaters such as artificial reservoirs, especially for Asia (cf. Barros et al. 2011). Understandably, with limited data to estimate carbon emissions, the reported values for CO2 and CH4 releases from reservoirs range widely (Table S1). The large range in published estimates reflects both the limitation of data and inherent differences in the various methods used for extrapolation (St. Louis et al. 2000; Lima et al. 2008; Barros et al. 2011). For example, for temperate reservoirs, the range is from 2.8 (Barros et al. 2011) to 12 mg CH4/m2/day (Lima et al. 2008), to 55 mg CH4/m2/day (Soumis et al. 2005). Further, the rates of CH4 release by global hydroelectric reservoirs range from 4 (Barros et al. 2011) to 70 Tg CH4/year (St. Louis et al. 2000), to 100 Tg CH4/year (Lima et al. 2008). Barros et al. (2011) first estimated CO2 and CH4 emissions from hydroelectric reservoirs using data from 85 globally distributed hydroelectric reservoirs. Their estimated annual carbon emission by hydroelectricity of 48 Tg C as CO2 and 27 Tg C-CO2 eq. as CH4 (3 Tg C-CH4) was far lower than an earlier estimate of 709 Tg C as CO2 equivalents per year with a
consideration of the matter of area difference (St. Louis et al. 2000), while a more recent estimate revised the data to be 76 Tg CO2-C/year and 7.3 Tg CH4-C/year (Hertwich 2013). The globally averaged CH4 and CO2 fluxes from hydroelectric reservoirs between 25° and 50° latitude were respectively 2.8 mg CH4/m2/day and 387 mg CO2/m2/day by Barros et al. (2011). These fluxes were remarkably lower than past observations (Table S1). For example, the CH4 flux was around 1/4 of that estimated by Lima et al. (2008) and only 1/20 that from Soumis et al. (2005), while the CO2 flux was between 1/4 to 1/2 that given in previous reports (St. Louis et al. 2000; Aufdenkampe et al. 2011). Moreover, the reservoirs examined in all of these studies were in the Americas and Northern Europe, and no data from Asia were included in these global estimates. This is clearly a major issue for China, where there is already a large number of hydroelectric reservoirs and many more planned for the future. Also, the drawdown area with typically much higher fluxes of CH4 emission (cf. Chen et al. 2009, 2013; Yang et al. 2012) has not been included in previous global and regional scenarios (St. Louis et al. 2000; Barros et al. 2011; Bastviken et al. 2011). Further, there is a lack of information on dam downstream carbon emissions such as spillways and turbines where a large share of GHGs especially methane emissions from reservoirs occurs (cf. Guerin et al. 2006; Roehm and Tremblay 2006; Kemenes et al. 2007; 2011; Yang et al. 2014), which further complicates the quantification of GHG emissions from inland waters. The clean energy credentials of hydroelectricity have come under increasing scrutiny due to the potential emissions of GHGs (Giles 2006). Some research based primarily on data from Brazil (e.g., Fearnside 1995, 2005; Fearnside and Pueyo 2012) indicate that GHGs, particularly CH4 emissions, discount the “green” credentials of hydropower. Likewise, previous studies of China’s hydropower dams also indicated high GHG emissions particularly in the seasonal drawdown area (Chen et al. 2009; Qiu 2009; Yang et al. 2012; Li and Zhang 2014). The carbon fluxes from reservoirs are relevant to reservoir age, size, latitude, and environmental factors (Barros et al. 2011), resulting in spatial and temporal heterogeneity in GHG fluxes (Table S2). In this study, we compile CO2 and CH4 emissions from Web of Science and Chinese Journals and quantify carbon emission from hydropower reservoirs in China by integrating reservoir surface area (water area), drawdown area, and reservoirs downstream (including spillways and turbines as well as rivers downstream the dam). Major controlling factors such as limnological and environmental variables of water temperature, pH, DO, Chl-a, and water depth on CH4 and CO2 fluxes are also identified. Our study ultimately will help guide global estimates and future trends in GHG emissions by making urgently needed Chinese data available for global modeling (most data are published in Chinese journals and are not available for international scholars).
Environ Sci Pollut Res
Materials and methods Data sources There are 19 hydropower reservoirs (66 study results) with data available on carbon emissions from ISI and Chinese journals (Fig. 1, Table S2). These reservoirs are located between 26° N and 46° N, spanning the temperate zone where China’s hydropower reservoirs are concentrated. Data on CO2 release in the river reach downstream of dam is also supplied. The global warming potential (GWP), i.e., 21–25, for the mass basis of CH4 relative to CO2 is recommended by IPCC (Intergovernmental Panel on Climate Change); however, a GWP of 34 is adopted in our study from the more recent work (Shindell et al. 2009). Thus, CO2 mass equivalent is calculated by summing the CO2 with CH4, multiplied by 34 to account for the specific long-lasting GWP of CH4 over a 100-year time horizon. Our collected data of CH4 emission fluxes are determined by floating chamber with gas chromatography (GC); CO2 fluxes in TGR (Three Gorges Reservoir), Nihe, and Shuibuya Reservoirs are determined by floating chamber with GC, while CO2 fluxes in Xin’anjiang, Wan’an, Hongfeng, Baihua, Xiufen, Hongyan, Hongjiadu, and Danjiangkou Reservoirs are calculated using a diffusive model with Henry’s constant (see SI context). Aquatic partial pressure of CO2 (pCO2) in the Wan’an and Xin’anjiang Reservoirs is measured using a continuous measurement system (equilibrator–non-dispersive infrared system), while pCO2 levels in the
Nihe Miyun
Danjiangkou Ertan
TGR Shuibuya Wan’an
Puding Yinzidu Dongfeng
Hongfeng
Suofengying
Baihua
Wujiangdu
Xiuwen
Wuqiangxi
Hongyan Hongjiadu
Fig. 1 Hydropower reservoirs with available GHG emission rates in China
other hydroelectric reservoirs are from pH and alkalinity. The limnological and environmental characteristics such as water temperature, pH, DO, and Chl-a are determined in site using the multi-water quality monitoring sondes after calibration (i.e., YSI 6600; Yellow Springs, USA). Water samples are collected in a depth of circa 50 cm below the water surface using previously acid-washed 5 L high-density polythene (HDPE) containers throughout the reservoir area. Alkalinity is titrated in triplicates using HCl (0.010–0.020 mol/L) in situ to pH 4.5 following the addition of a Methyl orange indicator (cf. Wang et al. 2011). Nutrients (N, P) are determined by a spectrophotometer and organic carbon by TOC analyzer. Major anions (Cl− and SO42−) are analyzed by ion chromatography, while major cations are analyzed by ICP-OES. The typical precisions of major ions are about ±5 %. The detailed information of datasets, limnological and environmental variables, sampling, and analyses are illustrated in Tables S2 and S3 (see citations in the supplementary document). Source apportionment of GHG emission GHG from classical reservoir surface GHG emission from reservoir surface was based on data collection from national and international literatures (see Table S2). GHG from reservoir downstream The immediate degassing of GHGs as water emerges from turbines and spillways are quantified using the total volume of water passing through turbines and the differences in CH4/ CO2 concentrations between the reservoir water at the turbine depth and the water below the dam (([CO2/CH4]above −[CO2/ CH4]below)×water outflow) (Roehm and Tremblay 2006). No measurement has been carried out on GHGs degassing from dams; here, the decrease in aquatic CO2 concentration (circa 30 μmol/L) through turbines is designated based on two cases (Fig. S1), and an estimated uncertainty of 15–45 μmol/L is accounted for extrapolation. A decrease of circa 30 % of CH4 concentration by water passing through turbines and spillways is due to immediate degassing (ca. 4 μmol/L) (Yang et al. 2009); this proportion is at the lower end of the immediate degassing ranges for tropical reservoirs (cf. Guerin et al. 2006). We also test errors using a possible variability of 50 %; thus, 2–6 μmol/L as immediate degassing is used for extrapolation. The estimates are comparable to source apportionment models from three well-studied reservoirs (Petit Saut, Balbina, and Samuel) (Tables 1 and 2). Area-specific fluxes of GHGs from river reaches below dams are much higher than upstream ones due to strong turbulence in the downstream rivers (Table S1). Very limited data and absences in downstream water areas affected by water
Environ Sci Pollut Res Table 1
CH4
Source proportions of CH4 and CO2 in the well-studied tropical hydroelectric reservoirs (Petit Saut, Balbina, and Samuel) Reservoir surfacea
Downstream Spillways and turbines
Reservoir surface/downstream=1/1b
80 % reservoir surface
20 % (9–33 %) reservoir surface
Guerin et al. 2006
100 % rivers downstreamc
15 % (7–25 %) reservoir surface
Guerin et al. 2006
CO2 a
Rivers downstream
Sources
Drawdown area is excluded from surface area
b
Kemenes et al. (2007), while Fearnside and Pueyo (2012) revise to 1/2
c
Kemenes et al. (2011)
releases from dams prevent our accurate estimates of GHG emissions from downstream rivers. Several studies on tropical reservoirs indicate that CO2 emission from downstream rivers accounts for 1.6–32 % of CO2 from reservoir surface, while downstream rivers make up 9–33 % of CH4 emission from reservoir water (Guerin et al. 2006; Kemenes et al. 2007, 2011). If these source apportionments are adopted in our calculations, downstream river emissions of GHGs comprise a small part of the total emissions; thus, the source models from tropical reservoirs show little effects on total quantifications and are considered as reliable. GHG diffusive flux in the reservoir drawdown area The littoral zones (drawdown area) in the reservoirs, which have been neglected in previous studies, are considered as notably high CH4 emitters (see Table S2). Water level fluctuations and land use determine CH4 fluxes in the drawdown area (cf. Yang et al. 2012). For example, from Table S2, a very high flux (119.8±42.2 mg CH4/m2/day) occurs in the rice
paddies and littoral marshes, whereas the other land use is characterized by a much lower mean flux (4.7±1.7 mg CH4/ m2/day). The inundated and drained periods for drawdown zones with varied elevation control the source or sink of GHGs; for example, dryland (cropland, fallow, and deforested land) in the drawdown area shows markedly higher CH4 emission flux in the inundated days, while rice paddy located in the upper zone of the drawdown area shows markedly higher CH4 emission flux in the drained days based on the previous observations in the TGR (Lu et al. 2011; Yang et al. 2012). For this study, these two significant land-use systems (rice paddy and dryland) for CH4 flux are categorized, and the total CH4 emission (FE) from the drawdown area is calculated as follows (Yang et al. 2012):
FE ¼
i¼b X
½Pi f in þ ð365−Pi Þ f dr Ai
i¼a
þ ½P0 f 0 in þ ð365−P0 Þ f 0 dr A0
ð1Þ
Table 2 Source proportion development of CH4 and CO2 from reservoir surfaces classically taken into account for reservoirs and downstream emissions in our study
CH4 CO2 a
Reservoir surfacea
Downstream Spillways and turbines
Rivers downstreame
Sources
Reservoir surface/downstream=1/9b
Schematic budgetc that is equal to 9-fold reservoir surfaceb Schematic budgetd
20 % reservoir surface 15 % reservoir surface
This study This study
Drawdown area is excluded from surface area
b
Upstream emission is proportional to the reservoir area, while the downstream emission is proportional to the streamflow and turbine depth (Fearnside and Pueyo 2012). Considering the very large surface area (∼3000 km2 ), small water outflow (577 m3 /s), and lower turbine depth (16 m) for Balbina, as for Tucurui with a similar surface area (2875 km2 ) as Balbina but 19 times (11,000 m3 /s) more water outflows and a turbine depth of 30 m, and the downstream CH4 emission including diffusive flux and bubbling at Tucurui represents a magnitude of nine times the CH4 emission from reservoir surface. This ratio is revised to 18 if the methodological factor is considered (Fearnside and Pueyo 2012). Valley-type Chinese reservoirs have relatively small surface area with larger water discharge and turbine depth greater than 30 m based on our data, thus a conservative ratio of 9 for downstream CH4 emissions to that from reservoir surface is designated. CH4 emission from turbines and river stretch below dam is proportional to increase c
Quantification using the differences between aquatic CH4 concentrations in turbine inflow and turbine outflow multiplying water volume passing through the turbines and spillways (Roehm and Tremblay 2006), which is comparable to the estimate from developed models through tropical reservoirs
d
CO2 emissions from turbines are calculated in the following: the differences between aquatic CO2 concentrations at turbine intake depth and in water discharge multiply water volume through turbine (Roehm and Tremblay 2006)
e
Ratios of CH4 and CO2 (GHG emissions from downstream river to reservoir surface) of 0.2 and 0.15 from tropical reservoirs is respectively adopted in our calculations. Because the emissions from downstream rivers are minor in comparison to total emissions from hydropower reservoirs, these uncertainties could be neglected. Schematic budget is the model of ([CO2/CH4]above −[CO2/CH4]below)×water outflow. This model is compared to source apportionments based on tropical reservoirs
Environ Sci Pollut Res
where i is the elevation from a to b (m), Pi is inundated days for dryland at varied elevation, and fin and fdr are the GHG fluxes of dryland in the inundated and drained durations, respectively. Ai is the submerged area at varied elevation. P′ is the inundated days of the rice paddy; f′in and f′dr are the GHG fluxes of the rice paddy during the inundated and drained seasons, respectively. A′ is the total area of the rice paddy. Here, we assume that areas with extremely high CH4 flux such as rice paddies and littoral marshes account for 10 % of the reservoir area, and the total drawdown area accounts for 30 % of the total reservoir area (cf. Chen et al. 2009). Accounting for the estimated uncertainty, a possible variability of 0–60 % is designated depending on seasonal hydrology. Due to similar geomorphological characteristics such as the locations of steep mountain areas for China’s hydropower reservoirs, our calculations are based on some simplifications with a combination of water level fluctuation and measurements in the TGR (cf. Lu et al. 2011; Yang et al. 2012). First, we neglect the dynamic duration of inundation at each elevation and designate an average inundated days for the dryland in the drawdown zone: Pi =200, fdr =0, and Ai is on average 20 % of the total reservoir area. Second, f′in =fin, P′=80. Eq. (1) can thus be simplified in the following:
shallow water; for example, bubbling CH4 emission occurs only where water depth is lower than 10 m (cf. Guerin et al. 2006). However, China’s valley reservoirs with gorge topography when coupled with removal of vegetation below the flooding lines effectively limit the ebullitive CH4 emission to the reservoir surface. In the drawdown area, soft vegetation and organic matter accumulation would facilitate CH4 production, resulting in the potential of bubbling flux. Thus, CH4 emission from China’s reservoir surfaces is confined largely to diffusion (Zhao et al. 2013), while bubbles are considered in the drawdown area (Yang et al. 2012; Table 3). The low ebullition flux (1.63 mg CH4/m2/day) of methane from TGR is considered for the drawdown dryland, while a much higher ebullitive flux (100 mg CH4/m2/day) is considered for littoral marshes with vegetation. This level is comparable to other temperate reservoirs with similar hydrological features (i.e., DelSontro et al. 2010) and the extremely high methane flux in the TGR marshes reported by Chen et al. (2009). With regard to CH4 emission quantifications in the drawdown zones, the inundated days are 200 and areal proportion of dryland and littoral marsh is respectively 10 and 20 % of the total reservoir based on Eq. (2) (simplifications are discussed earlier).
FE ¼ 200 f in 20%Atotal þ ð80 f in þ 140 f 0 dr Þ
Statistical analysis
10%Atotal
ð2Þ
Here, fin is the CH4 emission rate for drawdown area, Atotal is the total reservoir area, and f′dr is the CH4 emission rate of rice paddies and marshes (see Table S2). For CO2 emission calculations, we give a mean submergence time of 200 days and annual drawdown area of 30 % of the total reservoir area (cf. Chen et al. 2009; Yang et al. 2012). A CO2 flux of 2110 mg CO2/m2/day for temperate wetlands was adopted for areal outgassing from our reservoir drawdown zones (Aufdenkampe et al. 2011). This level is comparable to the lower limit of flux for marshes in China (Song et al. 2003; Yang et al. 2008). The total reservoir area used here is 28, 000 km2 with a storage of 600 km3, which were taken from China’s reservoirs registered in the ICOLD (International Commission on Large Dams; http://www.icold-cigb.net/).
We tested the data for normality using the KolmogorovSmirnov test before the statistical analyses. Variables that were non-normally distributed were natural logarithm (ln) transformed to satisfy normality assumptions. Analysis of variance (ANOVA) was performed to quantify the seasonal and spatial differences in the CO2 and CH4 fluxes (p
