Hydroelectricity Releases CO2

We estimate:

• 0.1 pounds of CO2e per kilowatt hour from wind and solar, mostly manufacture and placement
  
• 0.3 from nuclear, mostly fuel processing
• 1.0 from hydroelectricity, mostly vegetation in reservoirs decaying to methane (3 to 11 for reservoirs in forests)
  
• 1.8 average from US grid (or see "Nations" tab for over 100 other countries)
• Spreadsheet shows the steps of the calculations

Hydroelectric reservoirs convert some of the carbon in the area they flood to methane. Methane has 27-83 times as much global warming effect as CO2, so hydroelectricity which converts carbon to methane can have major global warming effects.

Methane forms when plant matter decays in the absence of oxygen (anaerobic or anoxic). This decay often happens in reservoirs. The bottom of tropical reservoirs is often low in oxygen because surface waters are warm, stay on top, and mix little with the bottom. The bottom of northern reservoirs is often low in oxygen when the surface is frozen. Oxygen is low in any climate when a reservoir suffers eutrophication: blooms of algae, whose decay uses up the oxygen in the water.

One source of methane is the decay of the original plants and carbon in the soil, which were flooded when a reservoir was created.

A second source of methane is that reservoirs convert to methane some of the carbon continuously arriving from detritus, branches and leaves falling in tributaries.

Third, every year plants grow on the sides of reservoirs when water levels drop, and then are covered and decay when water levels rise again.

Any of this annual carbon which is released as CO2 does not contribute to global warming, since it was recently captured from an equal amount of CO2 in the atmosphere, so there is no net effect. But any which is converted to methane has a large global warming effect, since methane is 81-83 times as powerful as CO2 in the first 20 years, and 27-30 times as powerful over 100 years.

Introductions to the issues are in New Scientist, Scitizen, and International Rivers. The best detailed overview is by Farrer. She explains clearly where the CO2 and methane come from and go, citing a range of studies. Larger wood of many species does not decay under water; 42% does decay. The highest greenhouse gas emissions per kilowatt hour of hydroelectricity are from shallow reservoirs which flood and decay and drain large areas relative to the power generated.

The studies treat methane as causing 21 or 25 times as much as an equal weight of CO2, which are estimates for the relative effect over 100 years. Over 20 years, EPA says the multiplier would be 81-83, since most of methane's effect is immediate. If one is concerned about the next 20 years, the larger multiplier would be appropriate and hydroelectricity would be seen as much worse.

Hydroelectric emissions in the CO2List footprint calculator are based on the average of nine reservoirs reported by Farrer in Brazil in various ecosystems: savanna, Atlantic forest and rainforest. The data for the nine reservoirs did not include emissions at the spillway and turbines, for which we had a choice of two multipliers, 1.9 from the Petit Saut dam in French Guiana rainforest, which has been more intensively studied, or 15 from a study of Tucurui, a large dam in Brazilian rainforest. We chose the smaller adjustment. The individual reservoirs would have estimates of 0.03 to 11.3 pounds CO2e per kilowatt hour (excluding distribution losses). The average of the various ecosystems, with more or less plant matter to decay, is 1 pounds per kilowatt hour. The calculator increases this for transmission losses in each country.

A study by the University of California-Berkeley had an overlapping range of estimated emissions, from 5.9 pounds/kwh in tropical forests to 4.5 in temperate forests, 2.9 in boreal forests, and 0.8 in woodland/shrubland, 0.2 in grassland and 0.1 in desert scrub, all omitting the decay from annual detritus arriving. They cite IPCC's 2003 Guidance for the original carbon content of the flooded areas, including soils.

The 1 pound/kwh which the calculator uses, is at the low end, since it represents an average of ecosystems. When hydroelectric reservoirs are known to be in forests, the Berkeley estimates would be better. Though incomplete, they are internally consistent, they are comparable to more specific local studies, and they reflect the large amount of plant matter in forests.

The Tucurui dam in Brazil was estimated to have total emissions of 2.3-3.4 pounds/kwh. Petit Saut in French Guiana is estimated at 5.9 pounds/kwh, total. A study of two large reservoirs in boreal forests in Finland yields a figure of 5.0 pounds/kwh, omitting emissions at the turbines, spillway and downstream, as well as methane bubbles in the reservoir from soil decay. A study by Hydro Quebec and the University of Quebec measured only emissions from the reservoir surface, not the turbines, spillway, or downstream, and compared them to natural lakes, rather than to pre-existing land use.

One study of nitrous oxide from reservoirs finds its net effect is less than 10% of other gases, so it has not been included.

Greenhouse gases are also released from quarrying, earthmoving, concrete manufacture, and building the turbines, though the only study in the spreadsheet shows construction is not as significant as greenhouse gases from reservoirs.

Hydropower Studies Summarized by IPCC. IPCC's Special Report in 2011 summarizes some studies about greenhouse gas emissions from hydroelectricity. However they excluded from their summary results any estimate of land use change (Annex 2 Methodology, p.14), which is one of the two biggest sources of hydroelectric emissions, along with leaves and twigs arriving from tributaries to decompose into methane in the bottom of reservoirs.

They included studies which had numbers on any two stages in the life cycle (p.13), and treated the results as if each study were a total of all stages of the life cycle. Some of the older studies omitted emissions from the reservoir, looking mostly at construction We tried to avoid this incompleteness problem by combining, from different studies, the strongest results for each stage of the life cycle.

A surprising number of studies of reservoirs show tonnes of emissions, but not electricity generated, so one cannot calculate emissions per kilowatt hour; IPCC did not seek electricity figures from public records and did not use those studies.

The following are the studies used for the IPCC Special Report's quantitative estimates. Special Report on Renewable Energy Sources and Climate Change Mitigation (Annex 2 Methodology, pp.25-6) May 2011 http://srren.ipcc-wg3.de/report/

Barnthouse, L.W., G.F. Cada, M.-D. Cheng, C.E. Easterly, R.L. Kroodsma, R. Lee, D.S. Shriner, V.R. Tolbert, and R.S. Turner (1994). Estimating Externalities of the Hydro Fuel Cycles. Report 6. Oak Ridge National Laboratory, Oak Ridge, TN, USA, 205 pp. http://www.osti.gov/energycitations/servlets/purl/757384-22LKCY/webviewable/757384.pdf

Omits all emissions from decay of the original flooded carbon, annual detritus, annual shoreline growth, reservoir surface, turbines, spillway and downstream. It was done in 1994 before those gaps were well-researched.

Denholm, P., and G.L. Kulcinski (2004). Life cycle energy requirements and greenhouse gas emissions from large scale energy storage systems. Energy Conversion and Management, 45(13-14), pp. 2153-2172.

Dones, R., T. Heck, C. Bauer, S. Hirschberg, P. Bickel, P. Preiss, L.I. Panis, and I. De Vlieger (2005). Externalities of Energy: Extension of Accounting Framework and Policy Applications: New Energy Technologies. ENG1-CT-2002-00609, Paul Scherrer Institute (PSI), Villigen, Switzerland, 76 pp. http://gabe.web.psi.ch/pdfs/externe_pol/WP6_Technical_Report_Release_2.pdf

Summarizes an early version of the data in the 2007 report from the same principal author (below), and less complete than the 2007 study. No data given on emissions from decay of the original flooded carbon, annual detritus, annual shoreline growth, reservoirs, turbines, spillways and downstream, and less explanation given in general than the limited amount in the 2007 report.

Dones, R., C. Bauer, R. Bolliger, B. Burger, T. Heck, A. Roder, M.F. Emenegger, R. Frischknecht, N. Jungbluth, and M. Tuchschmid (2007). Life Cycle Inventories of Energy Systems: Results for Current Systems in Switzerland and Other UCTE Countries. Ecoinvent Report No. 5, Paul Scherrer Institute, Swiss Centre for Life Cycle Inventories, Villigen, Switzerland, 185 pp. Available at: www.ecolo.org/documents/documents_in_english/Life-cycle-analysis-PSI-05.pdf.

Omits emissions from turbines, spillway and downstream. Emissions from reservoir surface largely based on natural lakes (p.101) without adjustment, though natural lakes do not have recently flooded carbon decomposing under the water. Furthermore, reservoirs are designed to have more outflow than most natural lakes, to drive turbines, so reservoirs have more inflow and thus more detritus arriving and decaying into methane in the reservoirs.

The report is dated 2007, but all data on reservoir emissions come from 2000 or before except a 2002 study from the Swedish power company, which also omits most reservoir emissions. They are aware of the newer Brazilian and Finnish studies, which measure reservoirs rather than lakes, but say they are not finished yet.

They do document how much cement, concrete and steel are in the dams, but no figures on how much CO2e they allocate to these sources (proprietary).

They use a 150-year time frame for dams, with no evidence, rather than the normal 100-year IPCC planning period. The main effect of the 150 years is to allocate a third of the construction emissions out of the IPCC time frame, even though they are emitted at the beginning, not the end.

They also have useful information on distribution losses for 26 European countries, ranging from 3.8% stated for Finland to 18.2% in Croatia (p.168)

Horvath, A. (2005). Decision-making in Electricity Generation Based on Global Warming Potential and Life-cycle Assessment for Climate Change. University of California Energy Institute, Berkeley, CA, USA, 16 pp. Available at: repositories.cdlib.org/ucei/devtech/EDT-006 http://escholarship.org/uc/item/8jh5x7z4

Report on hydroelectricity from Lake Powell on the Colorado River in the Utah desert.

Omits CO2e released immediately from arriving detritus as it decomposes in the water, not trapped in sediment.

Otherwise the lake is a relatively low emitter of CO2e, since not much biomass was flooded, and limited amounts arrive each year from the sparsely vegetated lower basin. However detritus does arrive from the upper basin of the Colorado River, which includes the western and southern slopes of the Rocky Mountains in Colorado, Wyoming and Utah.

Even omitting all detritus arriving, the report estimates 49 grams of CO2e per kilowatt hour for this site in desert scrub (0.1 pounds/kwh), and estimates emissions for the same type of dam (p.9) would be 2,696 g/kwh in tropical forests (5.9 pounds/kwh), 2,053 in temperate forests (4.5) and 1,296 in boreal forests (2.9).

The IPCC Special Report does not include those higher estimates.

In addition to these numbers the report says 4.6 g CO2e per kilowatt hour accumulate in bottom sediments, to be released when the dam is decommissioned (p.10).

For dam construction, the report uses the most complete input-output analysis (p.2), which includes indirect emissions from all the suppliers to the dam.

It also counts foregone carbon capture as an emission (p.9).

Another aspect of desert reservoirs is evaporation of water vapor, which is also a greenhouse gas. It is not clear if the atmosphere holds more water because of Lake Powell than it would if the river continued unhindered to the Sea of Cortez and Pacific.

IEA (1998). Benign Energy? The Environmental Implications of Renewables. International Energy Agency, Paris, France, 128 pp.

Pacca, S. (2007). Impacts from decommissioning of hydroelectric dams: A life cycle perspective. Climatic Change, 84(3-4), pp. 281-294.

Rhodes, S., J. Wazlaw, C. Chaffee, F. Kommonen, S. Apfelbaum, and L. Brown (2000). A Study of the Lake Chelan Hydroelectric Project Based on Life-cycle Stressor-effects Assessment. Final Report. Scientific Certification Systems, Oakland, CA, USA, 193 pp. Rhodes http://www.chelanpud.org/relicense/study/refer/4841_1.PDF

Report done to support relicensing, omits influx of detritus, which is now the main source of emissions, since the dam was built in 1926. The reservoir is bordered by wooded hills in Washington State, about 100 miles from Seattle and Puget Sound. The report also does not explain how it allocates carbon between methane and CO2.

Ribeiro, F.d.M., and G.A. da Silva (2009). Life-cycle inventory for hydroelectric generation: a Brazilian case study. Journal of Cleaner Production, 18(1), pp. 44-54.

Vattenfall (2008). Vattenfall AB Generation Nordic Certified Environmental Product Declaration EPD® of Electricity from Vattenfall´s Nordic Hydropower. Report No. S-P-00088, Vattenfall, Stockholm, Sweden, 50 pp. http://www.environdec.com/reg/epd88.pdf and http://www.environdec.com/reg/088/dokument/08_waterEPD.pdf

Swedish electric company reporting on their own operations. They omit ongoing arrival of detritus, and most details on their calculations, saying the information is proprietary. They say their land was cleared before flooding, removing emissions from the reservoir, but do not address the decay of the removed vegetation or the remaining leaf litter decaying under water and they identify a quarter as much soil carbon as given in the source they say they used. Further discussion.

Zhang, Q., B. Karney, H.L. MacLean, and J. Feng (2007). Life-Cycle Inventory of Energy Use and Greenhouse Gas Emissions for Two Hydropower Projects in China. Journal of Infrastructure Systems, 13(4), pp. 271-279.