Low methane ( CH 4 ) emissions downstream of a monomictic subtropical hydroelectric reservoir ( Nam Theun 2 , Lao PDR )

Methane (CH4) emissions from hydroelectric reservoirs could represent a significant fraction of global CH4 emissions from inland waters and wetlands. Although CH4 emissions downstream of hydroelectric reservoirs are known to be potentially significant, these emissions are poorly documented in recent studies. We report the first quantification of emissions downstream of a subtropical monomictic reservoir. The Nam Theun 2 Reservoir (NT2R), located in the Lao People’s Democratic Republic, was flooded in 2008 and commissioned in April 2010. This reservoir is a trans-basin diversion reservoir which releases water into two downstream streams: the Nam Theun River below the dam and an artificial channel downstream of the powerhouse and a regulating pond that diverts the water from the Nam Theun watershed to the Xe Bangfai watershed. We quantified downstream emissions during the first 4 years after impoundment (2009–2012) on the basis of a high temporal (weekly to fortnightly) and spatial (23 stations) resolution of the monitoring of CH4 concentration. Before the commissioning of NT2R, downstream emissions were dominated by a very significant degassing at the dam site resulting from the occasional spillway discharge for controlling the water level in the reservoir. After the commissioning, downstream emissions were dominated by degassing which occurred mostly below the powerhouse. Overall, downstream emissions decreased from 10 GgCH4 yr −1 after the commissioning to 2 GgCH4 yr −1 4 years after impoundment. The downstream emissions contributed only 10 to 30 % of total CH4 emissions from the reservoir during the study. Most of the downstream emissions (80 %) occurred within 2–4 months during the transition between the warm dry season (WD) and the warm wet season (WW) when the CH4 concentration in hypolimnic water is maximum (up to 1000 μmol L) and downstream emissions are negligible for Published by Copernicus Publications on behalf of the European Geosciences Union. 1920 C. Deshmukh et al.: Low methane (CH4) emissions downstream the rest of the year. Emissions downstream of NT2R are also lower than expected because of the design of the water intake. A significant fraction of the CH4 that should have been transferred and emitted downstream of the powerhouse is emitted at the reservoir surface because of the artificial turbulence generated around the water intake. The positive counterpart of this artificial mixing is that it allows O2 diffusion down to the bottom of the water column, enhancing aerobic methane oxidation, and it subsequently lowered downstream emissions by at least 40 %.

In the present study, we quantified emissions downstream of the Nam Theun 2 Reservoir 81 (NT2R) located in Lao PDR on the basis of a high temporal (weekly to fortnightly) and spatial 82 (23 stations) resolution monitoring of CH 4 concentration. The significance of the aerobic CH 4 83 oxidation in the dynamics of CH 4 in the reservoir and the downstream rivers was also evaluated. 84 We characterized the seasonal patterns of downstream emissions and evaluated the contribution 85 of this pathway to CH 4 emissions by ebullition (Deshmukh et al., 2014) and diffusive fluxes at 86 the surface of the reservoir (Guérin et al., 2015). We finally discuss the contribution of 87 crimps and poisoned . A N 2 headspace was created and the vials were 160 vigorously shaken to ensure an equilibration between the liquid and gas phases prior to CH 4 161 concentration gas chromatography (GC) analysis. The concentration in the water was calculated 162 using the solubility coefficient of Yamamoto et al. (1976). 163

164
In the reservoir, water samples for AMO rate measurements were collected in the epilimnion and 165 in the metalimnion (at the oxicline). At RES9, the samples were taken in the middle of the water 166 column since the water column was well mixed. AMO was also performed at TRC1 167 (immediately downstream of the powerhouse). The water was collected in 1L HDPE bottles, 168 homogenized, oxygenated and redistributed to twelve serum vials (160 mL). Each vial contained 169 60 mL of water and 100 mL of air. Vials were covered with aluminium foil to avoid effect of 170 light on any bacterial activity and incubated in the dark (Dumestre et al., 1999;Murase and 171 Sugimoto, 2005) at 20°C to 30°C, depending on in situ temperatures. According to in situ 172 concentration of CH 4 in the water, different amounts of CH 4 were added by syringe while 173 withdrawing an equal volume of air from the headspace with a second syringe in order to obtain 174 concentrations of dissolved CH 4 in the incubated water ranging from in situ to four times in situ. 175 Incubations were performed with agitation to ensure continuous equilibrium between gas and 176 water phases. Total CH 4 concentrations in the vials were measured 5-times in a row at a 12 h 177 interval, and oxidation rates were calculated as the total loss of CH 4 in the vial (Guérin and 178 Abril, 2007). The oxidation rate for each concentration was the average value of the triplicates 179 with standard deviation (±SD). 180 The kinetics parameters of aerobic methane oxidation obtained from the experiment were 181 combined to the in situ CH 4 concentration profiles in order to calculate the integrated aerobic 182 methane oxidation in the oxic water column. As the aerobic methane oxidation rates we obtained 183 were potential, CH 4-ox were corrected for two limiting factors, the oxygen availability and the 184 light inhibition as described in Guerin and Abril (2007). The final equation to compute in situ 185 oxidation rates (CH 4-ox , mmol m -2 d -1 ) is: 186 CH 4-ox = C CH4 · S CH4-ox · C O2 / (C O2 + K m (O2) ) ·d· I(z) 187 with C CH4 , the CH 4 concentration; S CH4-ox , the specific CH 4-ox ; C O2 , the oxygen concentration; 188 K m(O2) , the K m of O 2 for CH 4 oxidation, d, depth of the water layer and I(z), the inhibition of 189 methanotrophic activity by light as defined by Dumestre et al. (1999) at the Petit Saut Reservoir. 190 Finally, the CH 4 oxidation rates were integrated in the oxic water column, from the water surface 191 to the limit of penetration of oxygen.
where F, the diffusive flux at water-air interface; k T , the gas transfer velocity at a given 212 temperature (T); ∆C = C w -C a , the concentration gradient between the water (C w ) and the 213 concentration at equilibrium with the overlying atmosphere (C a ). Afterward, the k T was 214 computed from k 600 with the following equation: 215 k T = k 600 × (600/Sc T ) n (2) 216 with Sc T , the Schmidt number of CH 4 at a given temperature (T) (Wanninkhof, 1992)  where C upstream is the CH 4 upstream of the site where degassing might occur and C downstream is the 245 CH 4 concentration in the water downstream of the degassing site. On each of these structures, the 246 degassing was calculated using the water discharges and the difference of CH 4 concentration 247 between the stations: (1) NTH3 located below the Nakai Dam and RES1, (2) TRC1 located 248 below the turbines and RES9, (3) NKT3 below the Regulating Dam and REG1, and (4) DCH3 249 below the Aeration Weir and DCH2 (Figure 1). In addition, degassing was calculated for the 250 occasional spillway releases from the Nakai Dam. 251 The estimation of the concentration upstream of the degassing sites was different for the four 252 sites. For the degassing below the turbines and below the regulating dam, the average of the 253 vertical profile of CH 4 concentrations at RES9 and REG1 were considered as concentrations 254 before degassing, respectively. Surface concentration at DCH2 was considered for the degassing 255 at the aeration weir. For the degassing below the Nakai Dam, since the continuous flow of 2 m 3 s -256 1 was released from the surface water layer, we considered the average CH 4 concentration in the 257 upper 3 m water layer at RES1 located ~100 m upstream of dam. For the spillway release of the 258 Nakai Dam, as the spillway gate is located at 12 m below the maximum reservoir water level, the 259 degassing due to spillway release was calculated using the average CH 4 concentration in the 260 upper 15 m water layer at RES1. 261 with surface O 2 concentrations ranging from 14 to 354 µmol L -1 (5 to 137% saturation) and the 271 hypolimnion was anoxic. In the CD season, the reservoir water column was poorly but entirely 272 oxygenated during a few weeks/month (127±93 µmol L -1 ). In the WD and WW seasons, the CH 4 273 concentrations ranged between 0.02 and 201.7 µmol L -1 in the epilimnion and 0.02 to 1000 µmol 274 L -1 in the hypolimnion. In the CD season, the CH 4 concentrations are only slightly higher in the 275 hypolimnion than in the epilimnion. After the starting of turbines, the hydrodynamics of the 276 water column at RES1 followed the same seasonal pattern as described before whereas the CH 4 277 vertical profiles of concentration at RES9 located upstream of the water intake were were occasionally observed when CH 4 -rich water was released from the spillway, especially in 296 2009. Ten kilometers downstream of the Nakai Dam, CH 4 concentration decreased down to 297 0.41±0.32 µmol L -1 at NTH4 and NTH5 without any clear seasonal pattern (Fig. 3 a). 298

Results
The concentrations observed below the Nakai Dam at the stations NTH4 and NTH5 were similar 299 to the CH 4 concentrations found in the pristine Nam Phao River (NPH1) in the watershed and 300 40% lower than the CH 4 concentrations at the station NTH7 located 50 km downstream of the 301 dam. They were 2 orders of magnitude lower than the concentrations observed downstream of

Aerobic CH 4 oxidation in the reservoir and downstream of the powerhouse 404 and the Nakai Dam 405
In the reservoir, the potential AMO rates increased linearly with the CH 4 concentration ( Figure  406 5a,b,c) in both epilimnetic and metalimnic waters at the stations RES1, RES3 and RES7. The 407 AMO rates in the middle of the well-mixed water column at the station RES9 were not 408 statistically different from the AMO rates in the metalimnion at the other stations of the 409 reservoirs. Therefore, the AMO rates from RES9 were plotted versus the initial CH 4 410 concentration together with AMO rates from the metalimnion. The slope of the linear 411 correlation, or the so-called specific oxidation rate (SOR, d -1 ) in the metalimnion was similar for 412 the CD and WD seasons (SOR = 0.88 ± 0.03 d -1 ) (Figure 5a). In the epilimnion the SOR was 413 twice higher in the WD season (5.28 ± 0.43 d -1 ) than in the CD season (2.24±0.41 d -1 ) ( Figure  414 5b,c). Overall, the SOR in the epilimnion was two to fourfold higher than the SOR in the 415 metalimnion. Downstream of the powerhouse, the SOR was 1.47 ± 0.07 d -1 , that is intermediate and 2012, the average integrated oxidation rate at RES9 is 122 mmol m -2 d -1 that is more than 423 three times higher than the average integrated oxidation rate at RES1 (35 mmol m -2 d -1 ). Since 424 oxidation occurs from the surface to the bottom of the water column at RES9 and mostly around 425 the oxicline at RES1, the depth-integrated oxidation rates were 5-20 times higher at RES9 than at 426 RES1 during the WD season and no clear tendency can be drawn for the WW and CD seasons 427 (Table 1). At RES9, the total amount of oxidized CH 4 decreased from 5 to 1 Gg(CH 4 ) y -1 428 between 2010 and 2012 whereas it ranged between 0.4 and 0.7 Gg(CH 4 ) y -1 without clear trend at 429 RES1 (Table 1). 430   Table 2 (Guérin et al., 2015). The emissions at RES9 correspond to 20 to 40% of the total 535 downstream emissions (Table 2). Therefore, a very significant amount of CH 4 that could be 536 emitted downstream is emitted at the reservoir surface and this contributes to lower downstream 537 emissions. 538 However, the mixing at the water intake has a strong impact on aerobic CH 4 oxidation. The 539 vertical mixing allows O 2 to penetrate down to the bottom in the vicinity the water intake and 540 enhances both oxidation at the water intake and downstream of the powerhouse. On average, 541 depth-integrated CH 4 oxidation at RES9 upstream of the water intake is one order of magnitude 542 higher than at the station RES1 upstream of the Nakai Dam where the water column is thermally 543 stratified. Over the 3-km 2 -area representative for RES9 between 2010 and 2012, aerobic CH 4 544 oxidation consumed an amount of CH 4 that is equivalent to 50% of total CH 4 downstream 545 emissions (Table 1 and 2). In absence of artificial mixing, aerobic CH 4 oxidation would only 546 remove an amount of CH 4 that is equivalent to the amount of CH 4 removed by oxidation at RES1 547 that is on average, that is 11% of total downstream emissions over the three years of monitoring 548 (Table 1 and 2). Total downstream emissions were therefore lowered by 20% due to the 549 enhancement of aerobic CH 4 oxidation at RES9 if we compare total downstream emissions to 550 total downstream emissions plus the amount of CH 4 that would not be oxidized in absence of 551 mixing (oxidation at RES9 minus oxidation at RES1). In addition, aerobic methane oxidation in 552 the downstream channel might be enhanced too since water from RES9 being transferred to the 553 artificial downstream channel is better oxygenated than it would be in absence of artificial 554 mixing. 555

Contribution of downstream emissions to CH 4 gross emissions 485
Overall, the design of the water intake that mixes the whole water column decreases virtually 556 downstream emissions since part of the CH 4 is outgassed at the reservoir surface instead of being 557 transported and emitted downstream. The very positive consequence of this artificial mixing at 558 the water intake is that the mixing allows O 2 to penetrate down to the bottom of the water 559 column enhancing aerobic methane oxidation both at the water intake and in the river/channel 560 downstream of the powerhouse. Roughly, CH 4 emissions from NT2 Reservoir are lowered by 561 40% or more due to the artificial mixing of the water column at the water intake. Emissions downstream of the Nam Theun 2 Reservoir have a low contribution to total emissions 573 also because a very significant amount of CH 4 that could be emitted downstream of the reservoir 574 is (1) emitted upstream of the water intake and (2) is oxidized in the vicinity of the water intake 575 because of the artificial mixing it generates. This artificial mixing contributes to improve the 576 water quality downstream of the turbines since the water that passes through is well oxygenated 577 (70% saturation). The other positive consequence is that it generates a hotspot of aerobic 578 methane oxidation that contributes to the oxidation of 20% of the CH 4 that would potentially be 579 emitted at the water intake or downstream of the turbines. This study shows that downstream 580 emissions from future or existing reservoirs could be significantly mitigated by the adoption of 581 water intake-design or the installation of devices enhancing artificial water column 582 destratification and oxygenation upstream of the turbines. 583 On the basis of these results, different from those previously published, we recommend that 584 estimates at the global scale of emissions below dams take into account the mixing status of 585 reservoirs, the water residence time and depth of the water intake and its impact on the 586 oxygenation of the water column immediately upstream of the turbines. 587 Acknowledgements 588 The authors thank everyone who contributed to the NT2 monitoring programme, especially the 589   Total (Mg(CH 4 ) month -1 ) J a n -0 9 A p r -0 9 J u l -0 9 O c t -0 9 J a n -1 0 A p r -1 0 J u l -1 0 O c t -1 0 J a n -1 1 A p r -1 1 J u l -1 1 O c t -1 1 J a n -1 2