Skip to main content
NCBI home page
ACS AuthorChoice logoLink to ACS AuthorChoice
. 2024 Dec 6;58(50):22104–22113. doi: 10.1021/acs.est.4c06363

Immediate Effect of Floating Solar Energy Deployment on Greenhouse Gas Dynamics in Ponds

Nicholas E Ray †,‡,*, Meredith A Holgerson ‡, Steven M Grodsky §
PMCID: PMC11656716  PMID: 39642106

Abstract

graphic file with name es4c06363_0007.jpg

Floating photovoltaic (FPV) solar energy offers promise for renewable electricity production that spares land for other societal benefits. FPV deployment may alter greenhouse gas (GHG) production and emissions from waterbodies by changing physical, chemical, and biological processes, which can have implications for the carbon cost of energy production with FPV. Here, we use an ecosystem-scale experiment to assess how GHG dynamics in ponds respond to installation of operationally representative FPV. Following FPV deployments of 70% array coverage, daily whole-pond GHG emissions increased by 26.8% on a carbon dioxide-equivalent (CO2-eq) basis, and dissolved oxygen availability rapidly decreased. Despite increased emissions following FPV deployment, FPV-derived GHG emissions from waterbodies are likely lower than landscape GHG emissions associated with terrestrial solar and hydropower production on a CO2-eq kWh–1 basis. Adaptive management strategies like bubbler installation may reduce the magnitude of FPV impacts on GHG and dissolved oxygen dynamics.

Keywords: floating solar, photovoltaic, methane, carbon dioxide, ponds, biogeochemistry, dissolved oxygen

Short abstract

The ecosystem impacts of floating photovoltaic power plants are poorly constrained. In this study, we demonstrate an increase in greenhouse gas emissions from ponds following floating solar power plant deployment.

Introduction

Generating renewable energy to mitigate climate change while sparing terrestrial resources has driven the rapid and widespread deployment of floating photovoltaic solar energy (FPV).1−3 While there are potential cobenefits of generating electricity from FPV systems, such as enhanced energy production efficiency,4 land sparing,5 and reduced evaporation rates,6,7 FPV may theoretically affect waterbodies on which they are sited via reduction of dissolved oxygen,8−10 decreased light intensity,11,12 decreased temperatures,8−10 and changes in plankton and primary producer abundance and composition.7,10,13 An unexplored, potential sustainability trade-off of FPV is whether—and how—floating solar arrays influence greenhouse gas (GHG) emissions from recipient waterbodies.1,14 Knowing whether FPV leads to greater GHG emissions from waterscapes is key to accurate determination of the carbon footprint and savings of this burgeoning energy production system.1,15

Energy production technologies require land and can alter landscape GHG emissions,16−20 which may be particularly important when considering the carbon cost of renewable energy production from technologies touted as low carbon. For example, constructing hydropower reservoirs requires flooding land, typically resulting in enhanced emissions of carbon dioxide (CO2) and methane (CH4) to the atmosphere over decades to centuries.16 These emissions can be substantial. In Quebec, GHG emissions associated with flooded land contributed ∼70% of the carbon cost of electricity generation from hydropower in 2017.17 Similarly, land-use change-derived GHG emissions might contribute >50% of the total carbon cost of terrestrial PV electricity generation, depending upon PV efficiency and vegetation management scheme.18 Wind energy facilities might also alter ecosystem carbon-cycling processes by reducing plant growth and net ecosystem productivity in the nearby area.20 Accurate quantification of GHG emissions associated with FPV deployment is warranted to understand the sustainability trade-offs of this emerging renewable energy technology.

To date, most (>90%) FPV installations globally occur on reservoirs, lakes, and ponds that are less than 1 km2 in area.2,21 Small (<1 km2) waterbodies have high areal rates of CO2 and CH4 emissions and release globally significant quantities of CO2 and CH4 to the atmosphere (e.g., ∼6 or ∼15% of all CH4 emissions considering median or mean emission rates, respectively).22−24 Thus, understanding how FPV might influence small waterbody GHG cycling and exchange with the atmosphere is likely to be important both for estimating the CO2-equivalent performance of FPV installations relative to other energy production technologies and for inclusion in local and regional GHG budgets. Ponds are an excellent system to test whether FPV deployment might impact aquatic GHG dynamics and exchange with the atmosphere for several reasons: (1) most FPV installations have been deployed on small waterbodies;2 (2) ponds have high rates of CO2 and CH4 release to the atmosphere;22 and (3) pond GHG dynamics respond quickly to external perturbations.25,26

Production and consumption of CO2 and CH4 in ponds, lakes, and reservoirs are dependent on dissolved oxygen, temperature, and the balance between primary production and respiration—all conditions likely to be affected by FPV installation.14 Diffusive exchange of CO2 and CH4 between waterbodies and the atmosphere is dictated by several factors, including the difference in gas partial pressure between the waterbody and atmosphere, which, in turn, is influenced by temperature, wind speed immediately above the water surface, and convective cooling.27,28 CH4 also is emitted from waterbodies via ebullition, or the formation of bubbles containing CH4 in sediments that release CH4 to the atmosphere when they reach the water surface and burst; CH4 ebullition from small waterbodies is also controlled by factors such as temperature, dissolved oxygen, and organic matter availability that are likely affected by FPV installation.29,30

Here, we report results from the first two years of an ecosystem-scale experiment used to test the effect of FPV deployment on GHG dynamics and atmospheric GHG exchange in ponds (Figure 1).

Figure 1.

Figure 1

Open in a new tab

Aerial view of experimental ponds used in this study during floating photovoltaic array construction and deployment (right) and following completion of deployment (left). The two ponds behind the panel-covered ponds were sampled as controls for the experiment, with the third control and treatment pond, respectively, outside of the image view at the Cornell Experimental Ponds Facility. (Photo credit Jason Koski/Cornell University; reproduced with permission for commercial use by the author).

Methods

We deployed FPV arrays on constructed ponds at the Cornell Experimental Pond Facility in New York, USA in summer 2023 (Figure 1). Arrays were designed to maximize power production potential and thus also potential impacts (70% panel coverage). We measured water column temperature, dissolved oxygen saturation, and dissolved CO2 and CH4 concentrations in surface and bottom waters, quantified rates of CH4 ebullition, and determined treatment-specific air–water gas exchange rates (i.e., k600 values). Using these measurements, we calculated diffusive CO2 and CH4 emissions and compared total GHG emissions between ponds with and without FPV. We did not consider nitrous oxide dynamics in this study as it is unimportant in the greenhouse gas budget (<0.001% of the annual CO2-equivalent emissions budget) of ponds at the Cornell Experimental Pond Facility.31

Experimental Design

We sampled 16 ponds at the Cornell Experimental Ponds Facility in Ithaca, NY, USA, in summer 2022 to identify six ponds that were most similar based on the plant community, temperature, dissolved oxygen, pH, conductivity, dissolved nutrients, and dissolved GHG concentration. All ponds were constructed as 30 m × 30 m inverted-truncated pyramids with 2.4 m depth in 1958–1959. At the start of sampling in 2023, all ponds had a water depth of 1.85 m. The six most similar ponds used during the experiment were fishless and dominated by macrophytes (ponds 123, 124, 125, 128, 131, and 132).

After identifying the six most similar ponds, we employed a before-after-control-impact (BACI) approach to test the short-term response of pond greenhouse gas dynamics to the installation of floating solar arrays. Floating solar arrays (Ciel et Terre International, France) were deployed on three ponds: the FPV array on pond 124 was constructed from June 15–29, 2023, pond 123 from June 29 to July 14, 2023, and pond 125 from September 18–28, 2023. During construction, we added one row of panels and then pushed the array further into the pond before beginning the next row. In this way, the pond slowly became more covered. In total, 70% of the surface area of each pond area was covered with solar panels, which, if operational, would have a capacity of approximately 52.25 kWp or 52,250 kWh pond–1 year–1 (assuming 1000 kWh per kWp; kWp values were provided by Ciel et Terre). The panels attached to the FPV array were recycled, decommissioned panels, and thus, no electricity was generated. We note that 70% coverage is slightly higher than the global median FPV coverage on 0.001 km2 waterbodies (∼58%) but is within the range of FPV coverage on ponds and small lakes2 and was the coverage suggested by Ciel et Terre as the high-end for these ponds during discussions about array size.

Water Column Temperature and Dissolved Oxygen

We characterized the temperature and dissolved oxygen concentrations of the water column in each pond using a thermistor and an optical dissolved oxygen sensor attached to a Manta +35 or a Manta +20 instrument (Eureka Water Probes, Austin, TX). Measurements were made at 25 cm depth intervals 9 times per pond between June 21 and October 25, 2023.

Dissolved Gas Sampling

We sampled for dissolved GHG concentrations in pond surface water on two occasions in 2022, and 14 occasions in 2023 using a headspace equilibration approach.26 Four of these sampling events were considered as before installation, two during FPV construction, and 10 after installation. On each sampling event, we collected triplicate water samples from the center of each pond at 5 cm depth in syringes and then created a headspace in the syringe with ambient air. After 5 min, the headspace was collected in evacuated exetainers for analysis of GHG concentration and determination of dissolved gas concentration. On 13 sampling occasions in 2023, we also sampled bottom water GHG concentrations by collecting water in a Van Dorn water sampler 25 cm from the pond bottom. We collected all water samples from the center of open water ponds using a canoe or kayak and from panel-covered ponds by walking on the array to near the pond center. Samples for the determination of atmospheric GHG concentrations were also collected during each sampling event. We determined sample GHG concentrations using a gas chromatograph equipped with a flame ionization detector and autosampler (Shimadzu GC 2014). We calculated dissolved gas concentrations using constants determined by Weiss32 and Wiesenburg and Guinasso.33

Ebullitive Flux Sampling

We deployed passive bubble trap samplers from May to October 2023 to measure rates of ebullitive CH4 flux.31,34 Bubble traps were constructed of inverted funnels (cross-sectional area of 0.059 m2) attached to graduated cylinders that were able to move freely up and down within a float attached to paracord strung across the pond. Three bubble traps were deployed in each pond by fully submerging the trap and ensuring that no bubbles were present. In this way, bubbles could accumulate in each trap for several days, until the volume of gas accumulated in each bubble trap was recorded. On average, there were 10 days between sampling events, though this time period was shorter in warm summer months when ebullition was more rapid and longer in cooler months when ebullition proceeded more slowly. In total, there were three sampling events considered as pre-installation, three during FPV construction, and nine sampling events post-installation.

One bubble trap in each pond was equipped with a stopcock, from which we collected gas samples for analysis of CH4 concentration on sampling events. We calculated ebullitive flux as

graphic file with name es4c06363_m001.jpg

where [CH4] is the concentration of CH4 in the trap (μL L–1) and Vm is the molar volume of gas at standard conditions (22.4 L mol–1). On 14 of 74 occasions, the volume of gas in the bubble traps was too small to sample. On these occasions, we used the mean concentration of CH4 from the preceding and following sampling events.31

Estimating k600 and Calculating Diffusive Flux

Diffusive exchange of dissolved gases between ponds and the atmosphere (mmol m–2 h–1) can be calculated from dissolved gas concentrations as35

graphic file with name es4c06363_m002.jpg

where Cwater and Cair indicate the gas concentration (μmol L–1) in the water and atmosphere, respectively, and the gas exchange coefficient kx (cm h–1) is estimated as

graphic file with name es4c06363_m003.jpg

where Sc is the temperature-dependent Schmidt number36 and x is equal to 0.66 when wind speed is ≤3 m s–1 and 0.5 when wind speed >3 m s–1. While there are several methods to estimate k600 mathematically, we suspected it differed between open and panel-covered ponds, and therefore, we measured k600 for CO2 and CH4 for the center and edge of each pond on November 2, 2023. We did this by measuring linear rates of CO2 and CH4 accumulation (or depletion) in a floating chamber (18.93 L; 0.071 m2 cross-sectional area) connected to a cavity-ringdown spectroscope (Los Gatos, Inc.) for 5 min and collecting surface water and air samples for analysis of CO2 and CH4 concentrations from the same location immediately after the 5 min incubation period as described previously. We then calculated k600 (cm h–1) as

graphic file with name es4c06363_m004.jpg

where “flux” is the areal rate of accumulation of CO2 or CH4 in the floating chamber (mmol m–2 h–1) and other variables are as described previously.

For all diffusive flux calculations, wind speed was determined by correcting wind speed data from the Ithaca-Tompkins weather station (NOAA Station WBAN-94761), which is located ∼2 km from the experimental ponds, using a previously established correction factor.31 In all cases, mean daily wind speed was <3 m s–1. We excluded measurements of k600 made from the edge of pond 125 and the center of pond 131 in statistical comparisons as the floating chamber was moved by wind across the pond surface during flux measurements.

Statistical Analysis

We compared GHG dynamics between ponds with and without FPV using a mixed model approach in R Statistical Software following a BACI approach.37 We constructed models using the lme4 package and then used least-squares mean tests to compare treatments via the emmeans package.38,39 Each model included treatment (i.e., ponds with and without FPV), before/after, the interaction between treatment and before/after, and day of year as fixed effects and pond number as a random effect. We considered samples collected from ponds before June 15, 2023, as “before” and samples collected after July 14, 2023, as “after” installation for these models. Data collected from June 15 to July 14 was excluded from statistical comparison as FPV construction and deployment for Pond 123 and 124 was taking place during this time. Pond 125 was classed as “No FPV” in statistical comparisons until FPV construction and deployment on the ponds began in September, at which point we removed it from our statistical analysis. We compared k600 values in the pond edge and center between ponds with and without FPV using one-tailed t tests.

Scaling Emissions

To determine the relative impact of FPV installation on daily pond GHG emissions, we calculated diffusive fluxes and paired these with ebullitive fluxes for the 9 sampling events following the completion of panel installation. To calculate diffusive flux, we used the mean surface water CO2 or CH4 concentration for both treatments and calculated diffusive flux for pond centers and edges by treatment, as described previously. We used k600 values measured here for the pond centers and edges to calculate the diffusive flux for both pond sections. We made several assumptions as part of these calculations: that k600 values for each treatment and for the pond edge and center were constant throughout the sampling period, that pond edge and center dissolved gas concentrations were the same for each treatment on each sampling event, and that pond edge and center water temperatures were the same for each treatment on each sampling event.

Once we estimated edge and center diffusive flux, we calculated whole-pond diffusive flux, assuming edge area for both control and treatment ponds was 270 m2, the pond center surface area for control ponds was 630 m2, pond center surface area for treatment ponds was 270 m2 (this subtracts the total area of FPV array that is in physical contact with the water surface), and that fluxes were constant over a 24 h period. We then combined mean daily diffusive fluxes with the mean ebullitive flux from that sampling period (unless the bubble trap was reset on that sampling date, in which case we used the mean rate from the preceding bubble accumulation period), again assuming that ebullitive flux was constant over a 24 h period, that ebullitive flux was the same in the pond center and edge, and using the same areas described previously. For the final sampling date, bubble traps had been removed to avoid any possible damage if ponds froze, and we assumed ebullition for this date was negligible. We then converted all fluxes to CO2-equivalents, using a conversion rate of 1 kg CH4 to 27 kg CO2 as CH4 emissions from ponds are not derived from fossil fuel use.40

We then calculated the mean relative percent difference in total emissions when FPV is present. Combining this value with a previously published annual GHG budget from the Cornell Experimental Ponds facility (564.4 g CO2-eq m–2 year–1)31 allows us to estimate the increase in areal emissions associated with FPV in g CO2-eq m–2 year–1. We can then estimate relative environmental emissions per kWh produced by FPV multiplying annual enhancement in GHG emissions (151.3 g CO2-eq m–2 year–1) by pond area (900 m2), and then dividing by the total FPV power generation potential (52,250 kWh pond–1 year–1).

Results

Using a BACI approach, we demonstrate that FPV deployment with 70% coverage led to increased pond GHG emissions within days of deployment, and this effect lasted for weeks to months. Increased emissions were driven by greater CH4 ebullition which offset reduced diffusive CO2 and CH4 emissions in FPV-covered ponds.

Reduced Temperature and Dissolved Oxygen

Immediately following FPV deployment, ponds with FPV became colder than ponds without and tended to have more uniform temperatures throughout the water column (Figure S1). Similarly, the FPV installation had pronounced effects on dissolved oxygen availability; ponds with FPV had immediate reductions in dissolved oxygen saturation throughout the water column following FPV deployment (Figure 2).

Figure 2.

Figure 2

Open in a new tab

Depth profiles of dissolved oxygen saturation in experimental ponds with (123, 124, and 125) and without (128, 131, and 132) floating photovoltaic array installations. In ponds 123, 124, and 125, the period of floating photovoltaic array installation is shown in the light gray bar and the period of full array coverage is shown in the black bar. Each point indicates a dissolved oxygen measurement.

Increased Greenhouse Gas Concentrations

Prior to FPV installation, dissolved CO2 (Figure 3A,B) and CH4 (Figure 4A,B) concentrations in ponds with and without FPV tracked each other closely (p = 1.00; Table S2). Following FPV installation, dissolved CO2 concentration in the surface (121.1 ± 9.4 μmol CO2 L–1; mean ± SE) and bottom water (300.6 ± 50.7 μmol CO2 L–1) of ponds with FPV were both more than twice as high on average than dissolved CO2 concentrations in the surface (45.1 ± 5.3 μmol CO2 L–1; p < 0.001) and bottom water (117.3 ± 17.0 μmol CO2 L–1; p < 0.001) of ponds without FPV (Figure 3C,F). The observed increase in dissolved CO2 concentrations in FPV-covered ponds occurred within days of FPV installation, and the effect persisted over months up to the point that we ceased sampling in early November (Figure S2).

Figure 3.

Figure 3

Open in a new tab

Seasonal patterns of dissolved carbon dioxide concentrations in the (A, B) surface and (D, E) bottom water of experimental ponds with and without floating photovoltaic (FPV) arrays in 2022 and 2023, respectively, and comparison of dissolved carbon dioxide concentrations in (C) surface and (F) bottom water of ponds with and without FPV. In (A, B, D, E), each point indicates the mean gas concentration from pond centers on a given date by treatment group (i.e., open or 70% FPV coverage), and error bars indicate standard error. FPV installation is indicated by the gray bar, with the period of coverage indicated by the black bar. In panels (C, F), each point on the boxplot indicates a measured carbon dioxide concentration in an individual pond based on whether FPV is present (FPV) or not (open), and the p-value is the result of least-squares mean test of the treatment effect of the BACI mixed effect model (Tables S1 and S2).

Figure 4.

Figure 4

Open in a new tab

Seasonal patterns of dissolved methane concentration in the (A, B) surface and (D, E) bottom water of experimental ponds with and without floating photovoltaic (FPV) arrays in 2022 and 2023, respectively, and comparison of dissolved methane concentrations in (C) surface water and (F) bottom water of ponds with and without FPV. In (A–E), each point indicates the mean gas concentration from the pond centers on a given date by treatment group (i.e., open or 70% FPV coverage), and error bars indicate standard error. FPV installation is indicated by the gray bar, with the period of coverage indicated by the black bar. In panels (C, F) each point on the boxplot indicates a measured methane concentration in an individual pond based on whether FPV is present (FPV) or not (open), and the p-value is the result of least-squares mean test of the treatment effect of the BACI mixed effect model (Tables S1 and S2).

In both treatments, CH4 dynamics followed seasonal patterns, with the highest concentrations occurring during warm summer months. Like CO2, there was no difference in CH4 concentrations between control and FPV ponds prior to FPV installation (p > 0.90; Table S2). CH4 concentrations in surface water of ponds with FPV (8.4 ± 2.2 μmol of CH4 L–1) were higher than those without (4.2 ± 0.9 μmol of CH4 L–1) following FPV installation (p = 0.031; Figure 4C). Similarly, bottom water CH4 concentrations in FPV ponds (219.5 ± 69.0 μmol of CH4 L–1) were also higher than in ponds without FPV (47.8 ± 13.7 μmol of CH4 L–1) following FPV installation (p = 0.004; Figure 4F). The relative enhancement of dissolved CH4 concentrations in ponds with FPV relative to those in ponds without FPV oscillated over time following FPV installation (Figure S3).

Ebullitive Methane Fluxes

Ebullitive CH4 fluxes followed typical seasonal patterns in ponds both with and without FPV (Figure 5).25,31 Ebullitive CH4 emissions were on average nearly twice as high in ponds with FPV (0.21 ± 0.04 mmol CH4 m–2 h–1) compared to ponds without FPV (0.11 ± 0.02 mmol CH4 m–2 h–1) following FPV installation (p = 0.031; Figure 5). Rates of bubble accumulation were similar between pond types (p = 0.955; Figure S4A), so any changes in CH4 ebullition associated with FPV installation must have been driven by differences in bubble CH4 concentration—indeed, the CH4 concentration in bubble trap headspace in ponds with FPV (60.0 ± 4.70% CH4) was nearly twice as high as in ponds without FPV (34.4 ± 4.00% CH4; p < 0.001; Figure S4B).

Figure 5.

Figure 5

Open in a new tab

(A) Seasonal pattern of ebullitive methane flux from ponds with and without floating photovoltaic (FPV) arrays in 2023. In panel (A), each point indicates the mean ebullitive flux on a given date by treatment group (i.e., open or 70% FPV coverage), and error bars indicate standard error. FPV panel installation is indicated by the gray bar, with the period of coverage indicated by the black bar. (B) Comparison of ebullitive methane flux from ponds with and without FPV arrays. Each point in panel (B) indicates a measured methane flux in an individual pond based on whether FPV panels are present (FPV) or not (open), and the p-value is the result of least-squares mean test of the treatment effect of the BACI mixed effect model (Tables S1 and S2).

Gas Transfer Velocities

Gas transfer velocities (i.e., k600 values) in the FPV-covered center of ponds were 4 times lower for CO2 (1.27 ± 0.18 cm h–1) and 3 times lower for CH4 (1.42 ± 0.59 cm h–1) than open pond centers in ponds without FPV (5.42 ± 2.23 and 4.25 ± 1.22 cm h–1 for CO2 and CH4 respectively; Table S3), though these differences were not statistically significant (p = 0.157 for k600CO2 and p = 0.107 for k600CH4), most likely due to the relatively small sample size. Ponds with and without FPV had similar k600 values for both CO2 and CH4 near the pond edge where no FPV was present (p > 0.39; Figure 1 and Table S3).

Pond GHG Emissions Post-FPV Deployment

Combining measured dissolved gas concentrations and k600 values to estimate diffusive CO2 and CH4 flux, we found that, on average, whole-pond diffusive CO2 emissions were 23.6 ± 7.50% lower and diffusive CH4 emissions were 17.5 ± 25.1% lower following FPV deployment (Figure 6A and Table S4). This reduction is associated with both reduced k600 values in the pond center, which minimizes the effect of enhanced GHG concentrations when calculating diffusive flux, and less total waterbody surface area, where GHGs can exchange with the atmosphere. We found that total ebullitive emissions in ponds with FPV were, on average, 57.2 ± 37.0% higher than those in ponds without FPV (Figure 6A). Considered together, enhanced ebullitive emissions in ponds with FPV offset reductions in diffusive emissions and ponds with FPV emitted 26.8 ± 20.2% more GHGs than ponds without FPV in terms of CO2-eq pond–1 day–1 (Figure 6A). However, enhancement was not consistent over time following panel installation (Figure 6B), with little effect of FPV on total GHG emissions in cooler months, when ebullition rates are naturally low.

Figure 6.

Figure 6

Open in a new tab

Relative percent difference in daily whole-pond greenhouse gas emissions following floating photovoltaic (FPV) deployment by the (A) emission category and (B) over time (Table S4). Positive values in (A) indicate higher emissions from ponds with FPV and negative values indicate reduced emissions in ponds with FPV, relative to ponds without FPV. Error bars indicate standard error (n = 9 for each category in panel (A), except n = 8 for ebullitive CH4 flux).

We can combine the average difference in GHG emissions in ponds with and without FPV determined here with a previously published annual GHG budget for the Cornell Experimental Ponds facility to estimate an annual pond emission enrichment of 151.3 g of CO2-eq m–2 year–1 for ponds with FPV. Combined with the estimated energy production capacity per FPV array, we estimate a water-use change carbon emission from FPV of 2.61 g of CO2-eq kWh–1.

Discussion

Our study demonstrates a clear effect of FPV deployment on aquatic biogeochemistry and GHG emission. We found immediate and sustained increases in dissolved CO2 and CH4 concentrations, reduced temperature, and near-anoxic conditions following FPV installation. Despite increased concentrations of dissolved CO2 and CH4, diffusive GHG emissions decreased following FPV installation due to a combination of reduced gas transfer velocity and pond area in contact with the atmosphere. Reduced diffusive emissions were offset by an increase in CH4 ebullition in FPV ponds, and overall, we estimate a 26.8% increase in greenhouse gas emissions following FPV installation using a carbon dioxide-equivalent basis. Along with reduced dissolved oxygen availability, we can consider these changes in the context of sustainability trade-offs associated with renewable energy production.

GHG emissions associated with water-use change during and after FPV installation require different considerations and accounting than for terrestrial PV. Proper estimation of GHG emissions associated with land-use change of terrestrial utility-scale solar power generation requires consideration of existing land cover, land management following installation, potential energy production of the solar facility, carbon lost during vegetation removal, the lost carbon sink potential, and changes in soil respiration as effects.18,41 The sum of GHG effects associated with FPV can be represented by air–water GHG exchange that integrates GHG dynamics taking place within the waterbody. Thus, quantification of direct GHG emissions associated with water-use change from FPV installation is more streamlined than that for terrestrial solar systems, but parsing the contribution of different processes to net emissions in ponds following FPV installation is more challenging. Currently, FPV installation entails no vegetation removal from waterbodies. However, plant dieback and decomposition can lead to enhanced production of CO2 and CH4 in the sediments and water column,31,42 which is likely to occur due to shading following FPV installation. Similarly, the lost carbon sink potential (i.e., reduced photosynthesis) can be reflected as either reduced CO2 uptake by the waterbody or greater CO2 emission to the atmosphere. Changes in sediment and water column respiration also will be reflected in water column GHG dynamics (e.g., increased concentrations of CO2 and CH4 following panel installation) and air–water GHG exchange.43,44 Reduced rates of gas exchange with the atmosphere might also lead to a buildup of dissolved gases and are likely driven by reduced turbulence at the air–water interface associated with the physical structure of the FPV array. All of these processes are likely contributing to our observation of enhanced CO2 and CH4 concentrations in surface waters of FPV ponds. We hypothesize that a biological mechanism drives the change in pond GHG dynamics following FPV installation, most likely a combination of reduced primary production and decomposition of macrophytes. Our observation of higher CH4 concentrations in ebullition samples provides some evidence for this hypothesis, indicative of potentially greater CH4 production in sediments and reduced production of oxygen-containing bubbles in the water column following FPV deployment. Further studies to better understand the mechanisms underlying our observations will improve our understanding of how FPV deployment may alter GHG dynamics across a range of aquatic environments.

Our initial analysis suggests that the carbon cost of water-use change associated with FPV deployment (2.61 g CO2-eq kWh–1) is low compared to environmental emissions associated with hydropower (∼10 g CO2-eq kWh–1)17 and terrestrial solar (0–50 g CO2-eq kWh–1).18 Addition of environmental emissions measured here to GHG footprint estimates of FPV determined using life cycle assessments that do not include changes to waterbody GHG exchange with the atmosphere (73.3 and 40 g CO2-eq kWh–1)45,46 increases estimated emissions per kWh by 4.0–6.5%. We can also combine FPV environmental emissions to provide a rough estimate of current and potential GHG emissions from FPV globally. In 2023, the reported global installed capacity of FPV was 2.55 GWp and practical potential electricity generation from FPV might be able to produce up to 9434 TWh year–1 with 30% array coverage on >100,000 reservoirs.2,6 Using our measured emissions per kWh, we can estimate that at present, FPV-derived GHG emissions from waterbodies are 6.7 Gg CO2-eq year–1 (assuming ∼1000 kWh kWp–1). At modeled practical potential generation of 9434 TWh year–1, FPV-derived waterbody GHG emissions may increase to 24.6 Tg CO2-eq year–1. Importantly, our study addresses the effects of the upper range of possible FPV deployment array coverage on GHG dynamics, so the values presented here likely represent a high-end estimate of the environmental carbon emissions from FPV.

Our estimates provide a first glimpse at the interface between FPV development and GHG emissions, opening the door for future studies on sustainable FPV deployment considerations, several of which we describe here. We measured the effect of the FPV installation on pond GHG emissions in just the first year of deployment. GHG response to FPV deployment may be temporary, persist, or follow nonlinear dynamics as pond ecosystems potentially adapt to biophysical effects of FPV installation. For example, GHG emissions from hydropower reservoirs diminish over time as the reservoir ages.16,47 The ponds used in this study have almost no watershed and are not connected to a stream or river, thus limiting inputs of organic matter and nutrients from outside the ecosystem. GHG emissions from ponds, lakes, and reservoirs with high rates of organic matter and nutrient loading typically have higher GHG emissions than those with lower rates of organic matter and nutrient loading from the surrounding watershed,48,49 and it is unclear how FPV installation might affect GHG cycling and emissions in these more nutrient-rich systems. It is important to determine if GHG cycling and emissions from waterbodies dominated by phytoplankton will be similarly impacted by FPV installation as macrophyte systems and whether FPV installation alters primary producer communities. Experimental evidence suggests shading similar to that expected from FPV may lead to increased phytoplankton biomass and reduced macrophyte biomass,50 though this remains to be tested. Shifts in primary producer abundance and dominance following FPV installation might influence GHG cycling in several ways, including by altering rates of photosynthesis and CO2 uptake,51,52 loading of organic matter to the sediment,29 and mixing and stratification dynamics.26,31 The waterbodies used in this study are on the small end of those currently used for FPV power generation,2 though using these systems allowed for experimental comparison and provided insight into the effects of FPV installation on larger systems. Areal rates of GHG emission are usually inversely related with waterbody area,22 so we might expect a reduction in emissions associated with FPV deployment on larger waterbodies. Large hydropower reservoirs have been suggested as particularly appealing for FPV deployment and cogeneration of electricity, so it is important to understand the consequences of FPV deployment on larger and deeper systems.1,53 Finally, the total area of pond used for electricity generation in this study (70%) is on the higher end of typical panel coverage (global mean FPV coverage ∼34%, regression estimated mean FPV coverage for 0.001 km2 ponds is ∼58%).2 Whether a reduction in percent coverage leads to a reduced impact on GHG dynamics and emissions or simply reduces the area of the waterbody impacted remains unclear.

Adaptive strategies could be employed to minimize the negative biogeochemical impacts of FPV deployment. Reducing the environmental impacts of FPV on waterbodies may influence public and legal acceptance of large-scale adoption of this emerging renewable energy technology.3,54 We suggest three options that may reduce the effect of FPV on GHG emissions and dissolved oxygen availability. (1) A threshold for percent cover of FPV arrays can be identified to generate renewable electricity from FPV while managing GHG emissions. (2) Engineering considerations such as the tilt of FPV panels and the distance between FPV panels and the water surface can be made for biogeochemical benefits (i.e., reduced shading), as opposed to designs determined to maximize energy generation and minimize potential wind damage to the FPV system exclusively. Moving panels further from or making them more perpendicular to the water surface will likely allow for more interaction between the atmosphere and water surface, allowing for more gas exchange, including reoxygenation, and potentially venting of GHGs. (3) Incorporation of aeration devices that hang beneath the FPV array could relieve oxygen stress and likely reduce CH4 production. Use of aerators is rapidly gaining traction as a method to both improve dissolved oxygen conditions and reduce CH4 emissions in aquaculture settings55 and could be similarly employed in FPV systems. Each of these potential adaptive strategies may have trade-offs, such as reduced power generation, increased costs, or alteration of other ecosystem processes, which can be reconciled concurrently to inform a sustainable energy transition.3

Sustainability trade-offs of FPV in aquatic ecosystems are driven by a range of interactive factors, and the goal of net-zero emissions is affected by biogeochemical processes on land and in water.15 An improved understanding of the biogeochemical and ecological mechanisms linked to changes in GHG pools and fluxes will allow for not only better prediction of the GHG impact of FPV deployment but also possible consequences for biodiversity and sustainability writ large. For example, reduced dissolved oxygen is likely to have important consequences for freshwater biodiversity, which is experiencing rapid decline globally.56,57 Persistently low dissolved oxygen concentrations might also be associated with a change in microbial community structure or primary producer abundance, that, in turn, may feedback to GHG dynamics.58 A holistic understanding of the ecological and biogeochemical impacts of FPV deployment is needed, and these impacts should be considered not only for the waterbody in which FPV is deployed but also in the broader context of trade-offs of shifting energy production from land to water.

Acknowledgments

Funding for this work was provided by an Academic Venture Fund award to S.M.G. and M.A.H. from the Cornell Atkinson Center for Sustainability. We thank Benj Sterret, Ash Canino, Jera Jansen, Caitlin Davis, Mônica Antunes-Ulysséa, Sheena Dwyer-McNulty, Trifosa Simamora, Tim Boycott, Kathy Stehnjem, Daniel Consolvo, Lee Fitzgerald, Wulfgar Ramsey, and Dave Grodsky for assistance with panel installation and sampling. Floats for two ponds were donated by Ciele et Terre, and recycled solar panels were donated by Green Clean Solar. Matt Thomas at the Cornell Statistical Consulting Unit provided helpful conversation regarding statistical analysis. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Data Availability Statement

The data that support the findings of this study are available for download online via the Figshare Repository (10.6084/m9.figshare.25674810.v1). The R script used to conduct statistical analysis and generate the figures presented here is available in the GitHub repository as an R Markdown script (https://github.com/nray17/Floating_Solar_GHGs_2022_2023).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.4c06363.

  • Additional information regarding pond water temperatures, enhancement of dissolved CO2 and CH4 concentrations over time following FPV installation, rates of bubble accumulation and the CH4 concentrations of accumulated headspace in bubble traps, enhancement of ebullitive CH4 flux over time, summaries of statistical models, summaries of least-squares mean tests, calculated gas transfer velocities (i.e., k600), and estimated daily GHG fluxes and emissions from ponds (PDF)

The authors declare no competing financial interest.

Supplementary Material

es4c06363_si_001.pdf (560.6KB, pdf)

References

  1. Almeida R.; Schmitt R.; Grodsky S.; Flecker A.; Gomes C. P.; Zhao L.; Liu H.; Barros N.; Kelman R.; McIntyre P. B. Floating Solar Power: Evaluate Trade-Offs. Nature 2022, 606, 246–249. 10.1038/d41586-022-01525-1. [DOI] [PubMed] [Google Scholar]
  2. Nobre R.; Rocha S. M.; Healing S.; Ji Q.; Boulêtreau S.; Armstrong A.; Cucherousset J. A Global Study of Freshwater Coverage by Floating Photovoltaics. Sol. Energy 2024, 267, 112244 10.1016/j.solener.2023.112244. [DOI] [Google Scholar]
  3. Grodsky S. M. Matching Renewable Energy and Conservation Targets for a Sustainable Future. One Earth 2021, 4 (7), 924–926. 10.1016/j.oneear.2021.07.001. [DOI] [Google Scholar]
  4. Dörenkämper M.; Wahed A.; Kumar A.; de Jong M.; Kroon J.; Reindl T. The Cooling Effect of Floating PV in Two Different Climate Zones: A Comparison of Field Test Data from the Netherlands and Singapore. Sol. Energy 2021, 219, 15–23. 10.1016/j.solener.2021.03.051. [DOI] [Google Scholar]
  5. Goswami A.; Sadhu P.; Goswami U.; Sadhu P. K. Floating Solar Power Plant for Sustainable Development: A Techno-Economic Analysis. Environ. Prog. Sustainable Energy 2019, 38 (6), e13268 10.1002/ep.13268. [DOI] [Google Scholar]
  6. Jin Y.; Hu S.; Ziegler A. D.; Gibson L.; Campbell J. E.; Xu R.; Chen D.; Zhu K.; Zheng Y.; Ye B.; Ye F.; Zeng Z. Energy Production and Water Savings from Floating Solar Photovoltaics on Global Reservoirs. Nat. Sustainability 2023, 6 (7), 865–874. 10.1038/s41893-023-01089-6. [DOI] [Google Scholar]
  7. Abdelal Q. Floating PV; An Assessment of Water Quality and Evaporation Reduction in Semi-Arid Regions. Int. J. Low-Carbon Technol. 2021, 16 (3), 732–739. 10.1093/ijlct/ctab001. [DOI] [Google Scholar]
  8. de Lima R. L. P.; Paxinou K.; Boogaard F. C.; Akkerman O.; Lin F. Y. In-situ Water Quality Observations under a Large-scale Floating Solar Farm Using Sensors and Underwater Drones. Sustainability 2021, 13 (11), 6421 10.3390/su13116421. [DOI] [Google Scholar]
  9. Andini S.; Suwartha N.; Setiawan E. A.; Ma’arif S. Analysis of Biological, Chemical, and Physical Parameters to Evaluate the Effect of Floating Solar PV in Mahoni Lake, Depok, Indonesia: Mesocosm Experiment Study. J. Ecol. Eng. 2022, 23 (4), 201–207. 10.12911/22998993/146385. [DOI] [Google Scholar]
  10. Ziar H.; Prudon B.; Lin F. Y.; Roeffen B.; Heijkoop D.; Stark T.; Teurlincx S.; de Senerpont Domis L.; Goma E. G.; Extebarria J. G.; Alavez I. N.; van Tilborg D.; van Laar H.; Santbergen R.; Isabella O. Innovative Floating Bifacial Photovoltaic Solutions for Inland Water Areas. Prog. Photovoltaics 2021, 29 (7), 725–743. 10.1002/pip.3367. [DOI] [Google Scholar]
  11. Bax V.; van de Lageweg W. I.; Hoosemans R.; van den Berg B. Floating Photovoltaic Pilot Project at the Oostvoornse Lake: Assessment of the Water Quality Effects of Three Different System Designs. Energy Rep. 2023, 9, 1415–1425. 10.1016/j.egyr.2022.12.080. [DOI] [Google Scholar]
  12. Ilgen K.; Schindler D.; Wieland S.; Lange J. The Impact of Floating Photovoltaic Power Plants on Lake Water Temperature and Stratification. Sci. Rep. 2023, 13 (1), 7932 10.1038/s41598-023-34751-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Li W.; Wang Y.; Wang G.; Liang Y.; Li C.; Svenning J. C. How Do Rotifer Communities Respond to Floating Photovoltaic Systems in the Subsidence Wetlands Created by Underground Coal Mining in China?. J. Environ. Manage. 2023, 339, 117816 10.1016/j.jenvman.2023.117816. [DOI] [PubMed] [Google Scholar]
  14. Nobre R.; Boulêtreau S.; Colas F.; Azemar F.; Tudesque L.; Parthuisot N.; Favriou P.; Cucherousset J. Potential Ecological Impacts of Floating Photovoltaics on Lake Biodiversity and Ecosystem Functioning. Renewable Sustainable Energy Rev. 2023, 188, 113852 10.1016/j.rser.2023.113852. [DOI] [Google Scholar]
  15. Zickfeld K.; MacIsaac A. J.; Canadell J. G.; Fuss S.; Jackson R. B.; Jones C. D.; Lohila A.; Matthews H. D.; Peters G. P.; Rogelj J.; Zaehle S. Net-Zero Approaches Must Consider Earth System Impacts to Achieve Climate Goals. Nat. Clim. Change 2023, 13 (12), 1298–1305. 10.1038/s41558-023-01862-7. [DOI] [Google Scholar]
  16. Soued C.; Harrison J. A.; Mercier-Blais S.; Prairie Y. T. Reservoir CO2 and CH4 Emissions and Their Climate Impact over the Period 1900–2060. Nat. Geosci. 2022, 15, 700–705. 10.1038/s41561-022-01004-2. [DOI] [Google Scholar]
  17. Levasseur A.; Mercier-Blais S.; Prairie Y. T.; Tremblay A.; Turpin C. Improving the Accuracy of Electricity Carbon Footprint: Estimation of Hydroelectric Reservoir Greenhouse Gas Emissions. Renewable Sustainable Energy Rev. 2021, 136, 110433 10.1016/j.rser.2020.110433. [DOI] [Google Scholar]
  18. van de Ven D. J.; Capellan-Peréz I.; Arto I.; Cazcarro I.; de Castro C.; Patel P.; Gonzalez-Eguino M. The Potential Land Requirements and Related Land Use Change Emissions of Solar Energy. Sci. Rep. 2021, 11 (1), 2907 10.1038/s41598-021-82042-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Lovering J.; Swain M.; Blomqvist L.; Hernandez R. R. Land-Use Intensity of Electricity Production and Tomorrow’s Energy Landscape. PLoS One 2022, 17, e0270155 10.1371/journal.pone.0270155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Wu D.; Grodsky S. M.; Xu W.; Liu N.; Almeida R. M.; Zhou L.; Miller L. M.; Roy S. B.; Xia G.; Agrawal A. A.; Houlton B. Z.; Flecker A. S.; Xu X. Observed Impacts of Large Wind Farms on Grassland Carbon Cycling. Sci. Bull. 2023, 68, 2889–2892. 10.1016/j.scib.2023.10.016. [DOI] [PubMed] [Google Scholar]
  21. Nobre R.; Boulêtreau S.; Cucherousset J. Small Lakes at Risk from Extensive Solar-Panel Coverage. Nature 2022, 607, 239 10.1038/d41586-022-01891-w. [DOI] [PubMed] [Google Scholar]
  22. Holgerson M. A.; Raymond P. A. Large Contribution to Inland Water CO2 and CH4 Emissions from Very Small Ponds. Nat. Geosci. 2016, 9 (3), 222–226. 10.1038/ngeo2654. [DOI] [Google Scholar]
  23. Rosentreter J. A.; Borges A. V.; Deemer B. R.; Holgerson M. A.; Liu S.; Song C.; Melack J.; Raymond P. A.; Duarte C. M.; Allen G. H.; Olefeldt D.; Poulter B.; Battin T. I.; Eyre B. D. Half of Global Methane Emissions Come from Highly Variable Aquatic Ecosystem Sources. Nat. Geosci. 2021, 14 (4), 225–230. 10.1038/s41561-021-00715-2. [DOI] [Google Scholar]
  24. Raymond P. A.; Hartmann J.; Lauerwald R.; Sobek S.; McDonald C.; Hoover M.; Butman D.; Striegl R.; Mayorga E.; Humborg C.; Kortelainen P.; Dürr H.; Meybeck M.; Ciais P.; Guth P. Global Carbon Dioxide Emissions from Inland Waters. Nature 2013, 503 (7476), 355–359. 10.1038/nature12760. [DOI] [PubMed] [Google Scholar]
  25. van Bergen T. J. H. M.; Barros N.; Mendonça R.; Aben R. C. H.; Althuizen I. H. J.; Huszar V.; Lamers L. P. M.; Lürling M.; Roland F.; Kosten S. Seasonal and Diel Variation in Greenhouse Gas Emissions from an Urban Pond and Its Major Drivers. Limnol. Oceanogr. 2019, 64 (5), 2129–2139. 10.1002/lno.11173. [DOI] [Google Scholar]
  26. Ray N. E.; Holgerson M. A.; Andersen M. R.; Bikše J.; Bortolotti L. E.; Futter M.; Kokori̅te I.; Law A.; McDonald C.; Mesman J. P.; Peacock M.; Richardson D. C.; Arsenault J.; Bansal S.; Cawley K.; Kuhn M.; Shahabinia A. R.; Smufer F. Spatial and Temporal Variability in Summertime Dissolved Carbon Dioxide and Methane in Temperate Ponds and Shallow Lakes. Limnol. Oceanogr. 2023, 68 (7), 1530–1545. 10.1002/lno.12362. [DOI] [Google Scholar]
  27. Read J. S.; Hamilton D. P.; Desai A. R.; Rose K. C.; MacIntyre S.; Lenters J. D.; Smyth R. L.; Hanson P. C.; Cole J. J.; Staehr P. A.; Rusak J. A.; Pierson D. C.; Brookes J. D.; Laas A.; Wu C. H.. Lake-Size Dependency of Wind Shear and Convection as Controls on Gas Exchange Geophys. Res. Lett. 2012, 39 (9), 10.1029/2012GL051886. [DOI]
  28. Wanninkhof R.; Asher W. E.; Ho D. T.; Sweeney C.; McGillis W. R. Advances in Quantifying Air-Sea Gas Exchange and Environmental Forcing. Annual Rev. Mar. Sci. 2009, 1, 213–244. 10.1146/annurev.marine.010908.163742. [DOI] [PubMed] [Google Scholar]
  29. Grasset C.; Moras S.; Isidorova A.; Couture R. M.; Linkhorst A.; Sobek S. An Empirical Model to Predict Methane Production in Inland Water Sediment from Particular Organic Matter Supply and Reactivity. Limnol. Oceanogr. 2021, 66 (10), 3643–3655. 10.1002/lno.11905. [DOI] [Google Scholar]
  30. Aben R. C. H.; Barros N.; Van Donk E.; Frenken T.; Hilt S.; Kazanjian G.; Lamers L. P. M.; Peeters E. T. H. M.; Roelofs J. G. M.; De Senerpont Domis L. N.; Stephan S.; Velthuis M.; Van De Waal D. B.; Wik M.; Thornton B. F.; Wilkinson J.; Delsontro T.; Kosten S. Cross Continental Increase in Methane Ebullition under Climate Change. Nat. Commun. 2017, 8 (1), 1682 10.1038/s41467-017-01535-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Ray N. E.; Holgerson M. A. High Intra-Seasonal Variability in Greenhouse Gas Emissions from Temperate Constructed Ponds. Geophys. Res. Lett. 2023, 50, e2023GL104235 10.1029/2023GL104235. [DOI] [Google Scholar]
  32. Weiss R. F. Carbon Dioxide in Water and Seawater: The Solubility of a Non-Ideal Gas. Mar. Chem. 1974, 2, 203–215. 10.1016/0304-4203(74)90015-2. [DOI] [Google Scholar]
  33. Wiesenburg D. A.; Guinasso N. L. Equilibrium Solubilities of Methane, Carbon Monoxide, and Hydrogen in Water and Sea Water. J. Chem. Eng. Data 1979, 24 (4), 356–360. 10.1021/je60083a006. [DOI] [Google Scholar]
  34. Wik M.; Crill P. M.; Varner R. K.; Bastviken D. Multiyear Measurements of Ebullitive Methane Flux from Three Subarctic Lakes. J. Geophys. Res.: Biogeosci. 2013, 118 (3), 1307–1321. 10.1002/jgrg.20103. [DOI] [Google Scholar]
  35. MacIntyre S.Trace Gas Exchange across the Air-Sea Interface in Freshwater and Coastal Marine Environments. In Biogenic Trace Gases: Measuring Emissions from Soil and Water; Matson P.; Harriss R., Eds.; Blackwell Science, 1995; pp 52–97. [Google Scholar]
  36. Wanninkhof R. Relationship between Wind Speed and Gas Exchange over the Ocean. J. Geophys. Res.: Oceans 1992, 97 (C5), 7373–7382. 10.1029/92JC00188. [DOI] [Google Scholar]
  37. Lenth R. V.R: A Language and Environment for Statistical Computing; CRAN: Contributed Packages: Vienna, Austria, 2014.
  38. Bates D.; Maechler M.; Bolker B.; Walker S. Fitting Linear Mixed-Effects Models Using Lme4. J. Stat. Software 2015, 67, 1–48. 10.18637/jss.v067.i01. [DOI] [Google Scholar]
  39. Lenth R. V.emmeans: Estimated Marginal Means, aka Least-Squares Means; CRAN: Contributed Packages: 2018.
  40. Forster P.; Storelvmo T.; Armour K.; Collins W.; Dufresen J.; Frame D.; Lunt D.; Mauritsen T.; Palmer M.; Watanabe M.; Wild M.; Zhange H.. The Earth’s Energy Budget, Climate Feedbacks and Climate Sensitivity. In Climate Change 2021—The Physical Science Basis; Masson-Delmotte V.; Zhai P.; Pirani A.; Connors S.; Pean C.; Berger S.; Caud N.; Chen Y.; Goldfarb L.; Gomis M.; Huang M.; Leitzell K.; Lonnoy E.; Matthews J.; Maycock T.; Waterfield T.; Yelekci O.; Yu R.; Zhou B., Eds.; Cambridge University Press: Cambridge, United Kingdom and New York, NY, 2021; pp 923–1054. [Google Scholar]
  41. Turney D.; Fthenakis V. Environmental Impacts from the Installation and Operation of Large-Scale Solar Power Plants. Renewable Sustainable Energy Rev. 2011, 15, 3261–3270. 10.1016/j.rser.2011.04.023. [DOI] [Google Scholar]
  42. Theus M. E.; Ray N. E.; Bansal S.; Holgerson M. A. Submersed Macrophyte Density Regulates Aquatic Greenhouse Gas Emissions. J. Geophys. Res.: Biogeosci. 2023, 128 (10), e2023JG007758 10.1029/2023JG007758. [DOI] [Google Scholar]
  43. Bastviken D.; Cole J.; Pace M.; Tranvik L.. Methane Emissions from Lakes: Dependence of Lake Characteristics, Two Regional Assessments, and a Global Estimate Global Biogeochem. Cycles 2004, 18 (4), 10.1029/2004GB002238. [DOI]
  44. Bastviken D.; Cole J. J.; Pace M. L.; Van de-Bogert M. C.. Fates of Methane from Different Lake Habitats: Connecting Whole-Lake Budgets and CH4 Emissions J. Geophys. Res.: Biogeosci. 2008, 113 (2), 10.1029/2007JG000608. [DOI]
  45. Islam M. I.; Jadin M. S.; Mansur A. A.; Kamari N. A. M.; Jamal T.; Lipu M. S. H.; Azlan M. N. M.; Sarker M. R.; Shihavuddin A. S. M. Techno-Economic and Carbon Emission Assessment of a Large-Scale Floating Solar PV System for Sustainable Energy Generation in Support of Malaysia’s Renewable Energy Roadmap. Energies 2023, 16 (10), 4034 10.3390/en16104034. [DOI] [Google Scholar]
  46. Clemons S. K. C.; Salloum C. R.; Herdegen K. G.; Kamens R. M.; Gheewala S. H. Life Cycle Assessment of a Floating Photovoltaic System and Feasibility for Application in Thailand. Renewable Energy 2021, 168, 448–462. 10.1016/j.renene.2020.12.082. [DOI] [Google Scholar]
  47. Prairie Y. T.; Alm J.; Beaulieu J.; Barros N.; Battin T.; Cole J.; del Giorgio P.; DelSontro T.; Guérin F.; Harby A.; Harrison J.; Mercier-Blais S.; Serça D.; Sobek S.; Vachon D. Greenhouse Gas Emissions from Freshwater Reservoirs: What Does the Atmosphere See?. Ecosystems 2018, 21 (5), 1058–1071. 10.1007/s10021-017-0198-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Beaulieu J. J.; DelSontro T.; Downing J. A. Eutrophication Will Increase Methane Emissions from Lakes and Impoundments during the 21st Century. Nat. Commun. 2019, 10, 1375 10.1038/s41467-019-09100-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Davidson T. A.; Audet J.; Jeppesen E.; Landkildehus F.; Lauridsen T. L.; Søndergaard M.; Syväranta J. Synergy between Nutrients and Warming Enhances Methane Ebullition from Experimental Lakes. Nat. Clim. Change 2018, 8 (2), 156–160. 10.1038/s41558-017-0063-z. [DOI] [Google Scholar]
  50. Yamamichi M.; Kazama T.; Tokita K.; Katano I.; Doi H.; Yoshida T.; Hairston N. G.; Urabe J. A Shady Phytoplankton Paradox: When Phytoplankton Increases under Low Light. Proc. R. Soc. B 2018, 285 (1882), 20181067 10.1098/rspb.2018.1067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Brothers S. M.; Hilt S.; Meyer S.; Köhler J. Plant Community Structure Determines Primary Productivity in Shallow, Eutrophic Lakes. Freshwater Biol. 2013, 58 (11), 2264–2276. 10.1111/fwb.12207. [DOI] [Google Scholar]
  52. Schindler D. E.; Carpenter S. R.; Cole J. J.; Kitchell J. F.; Pace M. L. Influence of Food Web Structure on Carbon Exchange between Lakes and the Atmosphere. Science 1997, 277 (5323), 248–251. 10.1126/science.277.5323.248. [DOI] [Google Scholar]
  53. Lee N.; Grunwald U.; Rosenlieb E.; Mirletz H.; Aznar A.; Spencer R.; Cox S. Hybrid Floating Solar Photovoltaics-Hydropower Systems: Benefits and Global Assessment of Technical Potential. Renewable Energy 2020, 162, 1415–1427. 10.1016/j.renene.2020.08.080. [DOI] [Google Scholar]
  54. Nilson R. S.; Stedman R. C. Reacting to the Rural Burden: Understanding Opposition to Utility-Scale Solar Development in Upstate New York. Rural Sociol. 2023, 88 (2), 578–605. 10.1111/ruso.12486. [DOI] [Google Scholar]
  55. Fang X.; Wang C.; Zhang T.; Zheng F.; Zhao J.; Wu S.; Barthel M.; Six J.; Zou J.; Liu S. Ebullitive CH4 Flux and Its Mitigation Potential by Aeration in Freshwater Aquaculture: Measurements and Global Data Synthesis. Agric., Ecosyst. Environ. 2022, 335, 108016 10.1016/j.agee.2022.108016. [DOI] [Google Scholar]
  56. Living Planet Report 2018: Aiming Higher; WWF: Gland, Switzerland, 2018.
  57. Croijmans L.; de Jong J. F.; Prins H. H. T. Oxygen Is a Better Predictor of Macroinvertebrate Richness than Temperature - A Systematic Review. Environ. Res. Lett. 2021, 16, 023002 10.1088/1748-9326/ab9b42. [DOI] [Google Scholar]
  58. Bush T.; Diao M.; Allen R. J.; Sinnige R.; Muyzer G.; Huisman J. Oxic-Anoxic Regime Shifts Mediated by Feedbacks between Biogeochemical Processes and Microbial Community Dynamics. Nat. Commun. 2017, 8 (1), 789 10.1038/s41467-017-00912-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

es4c06363_si_001.pdf (560.6KB, pdf)

Data Availability Statement

The data that support the findings of this study are available for download online via the Figshare Repository (10.6084/m9.figshare.25674810.v1). The R script used to conduct statistical analysis and generate the figures presented here is available in the GitHub repository as an R Markdown script (https://github.com/nray17/Floating_Solar_GHGs_2022_2023).

Articles from Environmental Science & Technology are provided here courtesy of American Chemical Society

RESOURCES