A new modelling framework for assessing hydropower’s GHG footprint

Sara Mercier-Blais is a winner of the 2017 IHA Young Researcher of the Year award. In this article from her award submission, she outlines the rationale behind and methodology for the development of the G-res tool, a new framework for predicting greenhouse gas emissions from freshwater reservoirs.

Greenhouse gas (GHG) emissions from natural inland waters (i.e. streams, rivers, and lakes) are important sources of atmospheric carbon, on the same order of magnitude as carbon uptake by land and sea, and thus play a significant role in the global carbon budget.

Research over the last few decades has shown that freshwater reservoirs can also emit substantial amounts of GHGs. Estimates of GHG emissions are highly variable from one reservoir to another, with a few reports stating that in the first years after construction, some hydropower reservoirs emit as much carbon as would a coal-fired power plant producing similar amounts of energy.

Such statements logically challenge the ‘green’ energy aspect of hydroelectric power. However, considering that natural water bodies also emit carbon (such as the rivers dammed to create reservoirs), it is important to not only measure the emissions, but to also understand the pathways and factors controlling them.

This could help explaining the observed variability in reservoir GHG emissions. The carbonic GHGs, carbon dioxide (CO2) and methane (CH4), are produced from the decomposition of organic matter by bacteria in the sediments and water column of a water body.

The organic matter is sourced both from internal primary production and input from the terrestrial zone in natural and man‐made water bodies, with an extra source in reservoirs being the organic matter of surrounding flooded areas following reservoir construction. The resulting GHGs evade to the atmosphere through various pathways (Figure 1).

Figure 1: Schematic drawing showing the different carbon emissions pathway from the reservoir surface to the atmosphere.

Figure 1

Firstly, in all water bodies both gases will diffuse slowly from the sediment, up through the water column, and eventually be emitted. Secondly, because of the higher insolubility of CH4, it can also be liberated periodically as bubbles when CH4 production rates in sediments are high enough, a feature commonly found in dammed river systems and reservoirs.

A third emission pathway unique to reservoirs is the degassing of waters following transport through a dam into the downstream river. Many reservoirs exhibit a thermal stratification, where warmer surface water and colder deep water create a strong physical barrier, called the thermocline, which substantially slows the diffusion of gases from the deeper part of the reservoir to the surface.

GHGs thus accumulate in reservoir bottom waters up to quite high concentrations and if the dam intake is situated within this layer, then a significant amount of GHGs can be emitted at the exit of the outlet and further downstream.

Evidence suggests that the factors driving emissions vary between pathways, but fall into four major categories: climatic, geographic, edaphic (i.e. soil properties) and hydrologic. For example, it has been shown that diffusive emissions decrease with increasing age of the reservoir as the newly flooded soil is being degraded.

Factors controlling freshwater GHG emissions that appear in the natural inland water literature are also relevant to understand mechanisms in man-made water bodies. For example, it has been shown that CH4 bubbling predominantly occurs in the littoral zone of lakes.

The objective of the G‐res research project is to use our existing knowledge base to create a new modelling framework to assess the net GHG footprint from freshwater reservoirs, which excludes emissions occurring pre-dam construction.

In addition, this tool allows us to re‐evaluate the role of reservoir GHG emissions in the global carbon budget. Knowledge regarding the factors driving GHG emissions from reservoirs is distributed across scientific literature, but a comprehensive analysis of that literature had yet to be done.

As part of this project, an extensive literature review was conducted to summarise all past GHG estimates available in scientific literature.

In addition, >50 variables spanning the four potential driver categories were collected for 223 reservoirs in which >550 field measurements were made and used in our study. Using multiple regression analysis (elastic net), we found the best model to predict each of the four emission pathways (Table 1): CO2 diffusive emissions and CH4 diffusive, CH4 bubbling and CH4 degassing emissions (Figure 1).

Table 1: Predictive variables of CO2 and CH4 emission models and their effect on emissions. Red arrows represent negative effect on emissions and green arrows represent positive effect on emissions. Black x represents the variables that decides if CH4 degassing is included or not in the Net GHG footprint.

Table 1

The novelty of our model was to assess the change in emissions that is truly attributable to the transformation of a natural landscape into a reservoir, along with defining mechanistic models that predict carbon emissions from a reservoir over its lifetime (Figure 2).

Once we obtained the final models, we created a database of >5500 reservoirs containing all the variables needed (Table 1) in order to predict the net GHG footprint of each reservoir.

Figure 2: Example of change in annual CO2 emissions in time (years) for a reservoir. Black dot is representing the annual CO2 emissions for the actual age of a reservoir and the shaded area represents the integrated emission on 100 years without the pre‐existing emissions.

Figure 2

Ultimately, we present an updated estimate of the global carbon footprint of reservoirs equal to a range of 189–222 TgCO2e/yr, which is significantly lower than previous global assessments.

As the use of statistical models is not always user friendly, the G‐res tool was developed to simplify the understanding of the different elements pertaining to the use of the models and is available as a web interface.

The ultimate goal of this project was to develop a tool that would allow hydropower companies, consultants and other stakeholders to easily obtain the net GHG footprint for a specific reservoir (existing or future) in order to make well‐informed societal, economical, and environmental decisions.

Privacy Policy