Which Countries' Clean Power Depends on the Weather
The countries whose clean power depends most on the weather are hydro countries: Portugal, Spain and Bosnia and Herzegovina swing the most of the 55 countries where hydro supplied at least 10% of electricity in their latest Ember year, with a typical year landing 27% to 41% away from its trend. Portugal's worst hydro year, 2017, came in 58.0% below trend, 8.16 TWh less than expected. Wind swings too: across 24 countries the median typical swing is 7.0%, less than hydro's 7.4%, and Costa Rica is the most exposed at 12%. Solar's median typical swing is 4.9% across 11 countries, the steadiest of the three.
Key Facts, Ember data to 2025
- 55 countries were scored for hydro: hydro supplied at least 10% of their electricity in their latest year, and the median typical swing against trend was 7.4% (2009 to 2024).
- Portugal had the largest typical hydro swing: 41.3% a year against trend over 15 scored years to 2024.
- The deepest hydro shortfall in any scored year was Portugal in 2017: 58.0% below trend, 8.16 TWh less than expected.
- The largest hydro shortfall in TWh was China in 2023: 102 TWh below trend (7.6%).
- In 2019, 14 of 55 scored hydro countries were at least 10% below trend at the same time, a likely dry year (Panama, Portugal, Spain, Kenya and Guatemala fell furthest).
- 24 countries were scored for wind: wind supplied at least 10% of their electricity in their latest year, and the median typical swing against trend was 7.0% (2010 to 2024).
- Costa Rica had the largest typical wind swing: 12.1% a year against trend over 15 scored years to 2024.
- The deepest wind shortfall in any scored year was Morocco in 2012: 28.8% below trend, 295 GWh less than expected.
- The largest wind shortfall in TWh was China in 2020: 64.0 TWh below trend (12.1%).
- In 2012, 5 of 17 scored wind countries were at least 10% below trend at the same time, a likely low wind year (Morocco, Romania, the United Kingdom, Sweden and Ireland fell furthest).
- 11 countries were scored for solar: solar supplied at least 10% of their electricity in their latest year, and the median typical swing against trend was 4.9% (2010 to 2024).
- Denmark had the largest typical solar swing: 11.5% a year against trend over 9 scored years to 2024.
- The deepest solar shortfall in any scored year was Israel in 2018: 33.5% below trend, 790 GWh less than expected.
- The largest solar shortfall in TWh was Germany in 2021: 5.56 TWh below trend (10.1%).
- 6 hydro countries are left out of the ranking because their hydro output changed by more than 2.5 times over the window: Cambodia, Laos, Malaysia, Ethiopia, Nepal and Angola.
Hydro: Most Exposed to Dry Years
55 countries, 2009 to 2024Rain and snowmelt decide how much water the dams have, so hydro output can swing widely from one year to the next.
| Country | Typical swing | Worst year | Below trend | Output lost | Share of power |
|---|---|---|---|---|---|
| 1. 馃嚨馃嚬 Portugal | 41.3% | 2017 | 58.0% | 8.16 TWh | 29.7% |
| 2. 馃嚜馃嚫 Spain | 29.5% | 2017 | 48.2% | 17.0 TWh | 11.4% |
| 3. 馃嚙馃嚘 Bosnia and Herzegovina | 27.0% | 2017 | 36.9% | 2.22 TWh | 36.3% |
| 4. 馃嚤馃嚢 Sri Lanka | 25.9% | 2012 | 42.9% | 2.48 TWh | 37.6% |
| 5. 馃嚟馃嚪 Croatia | 24.6% | 2012 | 29.1% | 1.98 TWh | 39.3% |
| 6. 馃嚭馃嚲 Uruguay | 21.5% | 2020 | 38.9% | 2.60 TWh | 27.3% |
| 7. 馃嚫馃嚠 Slovenia | 17.1% | 2022shared dry year | 35.3% | 1.72 TWh | 27.3% |
| 8. 馃嚬馃嚪 Turkey | 16.4% | 2014 | 35.8% | 22.6 TWh | 16.2% |
| 9. 馃嚪馃嚧 Romania | 15.3% | 2022shared dry year | 21.4% | 3.79 TWh | 24.2% |
| 10. 馃嚘馃嚳 Azerbaijan | 14.2% | 2023 | 28.7% | 660 GWh | 10.3% |
| 11. 馃嚝馃嚪 France | 13.9% | 2011 | 24.4% | 14.8 TWh | 10.3% |
| 12. 馃嚪馃嚫 Serbia | 13.8% | 2022shared dry year | 27.1% | 3.20 TWh | 22.3% |
| 13. 馃嚢馃嚜 Kenya | 13.1% | 2019shared dry year | 21.9% | 900 GWh | 20.9% |
| 14. 馃嚚馃嚠 Cote d'Ivoire | 13.1% | 2021 | 22.1% | 740 GWh | 27.8% |
| 15. 馃嚠馃嚬 Italy | 12.8% | 2022shared dry year | 33.9% | 14.6 TWh | 15.8% |
| 16. 馃嚫馃嚢 Slovakia | 12.2% | 2011 | 19.2% | 900 GWh | 10.9% |
| 17. 馃嚳馃嚰 Zimbabwe | 12.1% | 2016 | 33.5% | 1.50 TWh | 55.6% |
| 18. 馃嚨馃嚘 Panama | 11.6% | 2019shared dry year | 32.9% | 2.51 TWh | 53.1% |
| 19. 馃嚞馃嚬 Guatemala | 10.7% | 2009 | 21.9% | 815 GWh | 39.9% |
| 20. 馃嚝馃嚠 Finland | 10.4% | 2011 | 16.4% | 2.44 TWh | 15.2% |
| 21. 馃嚙馃嚧 Bolivia | 9.6% | 2016 | 26.9% | 625 GWh | 27.9% |
| 22. 馃嚮馃嚦 Vietnam | 9.5% | 2010 | 19.7% | 7.01 TWh | 33.0% |
| 23. 馃嚮馃嚜 Venezuela | 8.1% | 2014 | 18.8% | 14.9 TWh | 91.1% |
| 24. 馃嚢馃嚨 North Korea | 7.9% | 2015 | 22.5% | 2.87 TWh | 62.7% |
| 25. 馃嚜馃嚚 Ecuador | 7.7% | 2024 | 15.2% | 4.12 TWh | 77.9% |
| 26. 馃嚟馃嚦 Honduras | 7.6% | 2019shared dry year | 16.9% | 495 GWh | 30.5% |
| 27. 馃嚦馃嚞 Nigeria | 7.6% | 2024 | 18.3% | 2.05 TWh | 31.5% |
| 28. 馃嚫馃嚜 Sweden | 7.4% | 2013 | 14.0% | 9.99 TWh | 40.0% |
| 29. 馃嚚馃嚧 Colombia | 7.3% | 2024 | 16.2% | 9.48 TWh | 68.7% |
| 30. 馃嚢馃嚞 Kyrgyzstan | 7.2% | 2015 | 10.4% | 1.28 TWh | 94.7% |
| 31. 馃嚨馃嚟 Philippines | 7.1% | 2010 | 20.0% | 1.94 TWh | 10.3% |
| 32. 馃嚨馃嚲 Paraguay | 7.1% | 2024 | 13.9% | 6.78 TWh | 99.5% |
| 33. 馃嚚馃嚟 Switzerland | 6.9% | 2022shared dry year | 19.6% | 7.27 TWh | 52.3% |
| 34. 馃嚫馃嚛 Sudan | 6.7% | 2009 | 15.7% | 595 GWh | 77.9% |
| 35. 馃嚘馃嚬 Austria | 6.5% | 2011 | 16.7% | 6.87 TWh | 52.1% |
| 36. 馃嚞馃嚜 Georgia | 6.3% | 2020 | 13.7% | 1.30 TWh | 78.8% |
| 37. 馃嚘馃嚪 Argentina | 6.0% | 2021 | 15.4% | 3.57 TWh | 17.1% |
| 38. 馃嚚馃嚤 Chile | 5.7% | 2021 | 11.0% | 2.34 TWh | 22.2% |
| 39. 馃嚞馃嚟 Ghana | 5.5% | 2015 | 16.1% | 1.12 TWh | 35.4% |
| 40. 馃嚙馃嚬 Bhutan | 5.5% | 2018 | 15.5% | 1.26 TWh | 100.0% |
| 41. 馃嚦馃嚧 Norway | 5.2% | 2019shared dry year | 10.8% | 15.1 TWh | 90.0% |
| 42. 馃嚨馃嚢 Pakistan | 4.9% | 2018 | 11.1% | 3.41 TWh | 23.1% |
| 43. 馃嚚馃嚪 Costa Rica | 4.8% | 2024 | 11.7% | 1.06 TWh | 82.2% |
| 44. 馃嚥馃嚥 Myanmar | 4.8% | 2020 | 12.1% | 1.21 TWh | 46.1% |
| 45. 馃嚳馃嚥 Zambia | 4.6% | 2016 | 13.7% | 1.75 TWh | 86.2% |
| 46. 馃嚚馃嚦 China | 4.2% | 2011 | 12.6% | 99.0 TWh | 13.3% |
| 47. 馃嚨馃嚜 Peru | 3.6% | 2016 | 8.4% | 2.22 TWh | 56.0% |
| 48. 馃嚦馃嚳 New Zealand | 3.5% | 2024 | 6.1% | 1.52 TWh | 54.8% |
| 49. 馃嚪馃嚭 Russia | 3.5% | 2015 | 6.2% | 11.0 TWh | 16.7% |
| 50. 馃嚙馃嚪 Brazil | 3.4% | 2021 | 11.9% | 48.9 TWh | 51.7% |
| 51. 馃嚥馃嚳 Mozambique | 3.3% | 2013 | 5.5% | 860 GWh | 82.4% |
| 52. 馃嚬馃嚡 Tajikistan | 3.0% | 2014 | 6.1% | 1.04 TWh | 94.4% |
| 53. 馃嚠馃嚫 Iceland | 2.4% | 2016 | 4.6% | 625 GWh | 71.2% |
| 54. 馃嚚馃嚘 Canada | 2.3% | 2010 | 5.6% | 20.8 TWh | 52.8% |
| 55. 馃嚚馃嚛 Democratic Republic of the Congo | 1.7% | 2012 | 5.4% | 430 GWh | 84.2% |
Typical swing is the average distance, in percent, between a year's output and its trend; below trend and output lost describe the single worst year. Share of power is hydro as a share of generation in the country's latest year, and countries must generate at least 10 TWh overall.
Left out of the hydro ranking because their hydro output changed by more than 2.5 times over the window, so new dams swamp any weather signal: 馃嚢馃嚟 Cambodia, 馃嚤馃嚘 Laos, 馃嚥馃嚲 Malaysia, 馃嚜馃嚬 Ethiopia, 馃嚦馃嚨 Nepal, 馃嚘馃嚧 Angola.
Years when many countries fell short together
- 2019: 14 of 55 scored countries were at least 10% below trend, a likely dry year. Furthest below: 馃嚨馃嚘 Panama, 馃嚨馃嚬 Portugal, 馃嚜馃嚫 Spain, 馃嚢馃嚜 Kenya, 馃嚞馃嚬 Guatemala.
- 2022: 14 of 55 scored countries were at least 10% below trend, a likely dry year. Furthest below: 馃嚨馃嚬 Portugal, 馃嚜馃嚫 Spain, 馃嚫馃嚠 Slovenia, 馃嚠馃嚬 Italy, 馃嚙馃嚘 Bosnia and Herzegovina.
Wind: Low Wind Years
24 countries, 2010 to 2024Wind fleets keep growing, so each year is compared with the midpoint of the years either side, which removes steady growth and leaves the weak and strong years.
| Country | Typical swing | Worst year | Below trend | Output lost | Share of power |
|---|---|---|---|---|---|
| 1. 馃嚚馃嚪 Costa Rica | 12.1% | 2013 | 22.8% | 145 GWh | 13.3% |
| 2. 馃嚫馃嚜 Sweden | 11.5% | 2014shared low wind year | 14.1% | 1.84 TWh | 22.8% |
| 3. 馃嚝馃嚠 Finland | 10.2% | 2021 | 16.1% | 1.63 TWh | 27.4% |
| 4. 馃嚠馃嚜 Ireland | 9.4% | 2010 | 23.2% | 865 GWh | 37.7% |
| 5. 馃嚞馃嚙 United Kingdom | 9.3% | 2010 | 18.5% | 2.33 TWh | 29.4% |
| 6. 馃嚦馃嚤 Netherlands | 8.6% | 2010 | 17.6% | 850 GWh | 25.0% |
| 7. 馃嚚馃嚤 Chile | 8.4% | 2018shared low wind year | 14.9% | 625 GWh | 13.1% |
| 8. 馃嚛馃嚜 Germany | 8.2% | 2016shared low wind year | 14.2% | 13.2 TWh | 26.6% |
| 9. 馃嚘馃嚬 Austria | 7.5% | 2011 | 14.2% | 320 GWh | 11.4% |
| 10. 馃嚪馃嚧 Romania | 7.3% | 2018shared low wind year | 10.9% | 770 GWh | 12.0% |
| 11. 馃嚙馃嚜 Belgium | 7.2% | 2016shared low wind year | 10.3% | 625 GWh | 18.7% |
| 12. 馃嚨馃嚤 Poland | 7.1% | 2018shared low wind year | 14.7% | 2.21 TWh | 13.9% |
| 13. 馃嚥馃嚘 Morocco | 6.9% | 2012shared low wind year | 28.8% | 295 GWh | 15.2% |
| 14. 馃嚛馃嚢 Denmark | 6.4% | 2016shared low wind year | 11.6% | 1.68 TWh | 58.8% |
| 15. 馃嚭馃嚲 Uruguay | 6.3% | 2019 | 7.0% | 355 GWh | 40.6% |
| 16. 馃嚞馃嚪 Greece | 5.3% | 2014shared low wind year | 15.8% | 690 GWh | 20.9% |
| 17. 馃嚨馃嚬 Portugal | 5.2% | 2020 | 8.5% | 1.14 TWh | 26.8% |
| 18. 馃嚟馃嚪 Croatia | 4.9% | 2015 | 8.0% | 70.0 GWh | 21.2% |
| 19. 馃嚬馃嚪 Turkey | 4.4% | 2014shared low wind year | 11.3% | 1.09 TWh | 11.1% |
| 20. 馃嚙馃嚪 Brazil | 4.2% | 2020 | 11.1% | 7.09 TWh | 15.6% |
| 21. 馃嚘馃嚭 Australia | 3.8% | 2017 | 9.8% | 1.44 TWh | 13.7% |
| 22. 馃嚚馃嚦 China | 3.6% | 2020 | 12.1% | 64.0 TWh | 10.7% |
| 23. 馃嚜馃嚫 Spain | 3.5% | 2011 | 8.4% | 3.95 TWh | 20.4% |
| 24. 馃嚭馃嚫 United States | 3.0% | 2015 | 6.7% | 13.6 TWh | 10.3% |
Typical swing is the average distance, in percent, between a year's output and its trend; below trend and output lost describe the single worst year. Share of power is wind as a share of generation in the country's latest year, and countries must generate at least 10 TWh overall. 2 more countries meet the share bar but have fewer than 8 scorable years (the source is too new to measure), so they are not ranked.
Years when many countries fell short together
- 2012: 5 of 17 scored countries were at least 10% below trend, a likely low wind year. Furthest below: 馃嚥馃嚘 Morocco, 馃嚪馃嚧 Romania, 馃嚞馃嚙 United Kingdom, 馃嚫馃嚜 Sweden, 馃嚠馃嚜 Ireland.
- 2016: 7 of 24 scored countries were at least 10% below trend, a likely low wind year. Furthest below: 馃嚞馃嚙 United Kingdom, 馃嚛馃嚜 Germany, 馃嚝馃嚠 Finland, 馃嚚馃嚤 Chile, 馃嚠馃嚜 Ireland.
- 2014: 5 of 20 scored countries were at least 10% below trend, a likely low wind year. Furthest below: 馃嚞馃嚪 Greece, 馃嚫馃嚜 Sweden, 馃嚛馃嚜 Germany, 馃嚦馃嚤 Netherlands, 馃嚬馃嚪 Turkey.
- 2018: 6 of 24 scored countries were at least 10% below trend, a likely low wind year. Furthest below: 馃嚚馃嚤 Chile, 馃嚨馃嚤 Poland, 馃嚘馃嚬 Austria, 馃嚫馃嚜 Sweden, 馃嚪馃嚧 Romania.
Solar: Dull Years
11 countries, 2010 to 2024Solar fleets grow fast, so each year is compared with the midpoint of the years either side, and the swings here include years when new panels arrived unevenly as well as duller and sunnier years.
| Country | Typical swing | Worst year | Below trend | Output lost | Share of power |
|---|---|---|---|---|---|
| 1. 馃嚛馃嚢 Denmark | 11.5% | 2021 | 26.4% | 470 GWh | 13.2% |
| 2. 馃嚠馃嚤 Israel | 7.9% | 2018 | 33.5% | 790 GWh | 16.2% |
| 3. 馃嚞馃嚪 Greece | 7.4% | 2018 | 10.0% | 420 GWh | 19.7% |
| 4. 馃嚙馃嚞 Bulgaria | 6.3% | 2021 | 17.4% | 310 GWh | 18.2% |
| 5. 馃嚜馃嚫 Spain | 5.0% | 2018 | 13.6% | 2.01 TWh | 21.9% |
| 6. 馃嚠馃嚬 Italy | 4.9% | 2016 | 6.6% | 1.56 TWh | 16.8% |
| 7. 馃嚚馃嚟 Switzerland | 4.9% | 2021 | 12.1% | 390 GWh | 12.1% |
| 8. 馃嚛馃嚜 Germany | 4.7% | 2021 | 10.1% | 5.56 TWh | 18.1% |
| 9. 馃嚚馃嚤 Chile | 4.1% | 2020 | 10.7% | 900 GWh | 25.1% |
| 10. 馃嚙馃嚜 Belgium | 3.8% | 2021 | 6.3% | 375 GWh | 14.5% |
| 11. 馃嚘馃嚭 Australia | 3.2% | 2017 | 9.8% | 965 GWh | 19.7% |
Typical swing is the average distance, in percent, between a year's output and its trend; below trend and output lost describe the single worst year. Share of power is solar as a share of generation in the country's latest year, and countries must generate at least 10 TWh overall. 14 more countries meet the share bar but have fewer than 8 scorable years (the source is too new to measure), so they are not ranked.
Years when many countries fell short together
No year met the bar: at least 5 countries, and at least 25% of those scored that year, 10% or more below trend.
The Five Most Exposed Hydro Countries
Indexed to each country's own average (100)| Year | Portugal | Spain | Bosnia and Herzegovina | Sri Lanka | Croatia |
|---|---|---|---|---|---|
| 2010 | 138 | 141 | 142 | 101 | 136 |
| 2011 | 99 | 102 | 77 | 83 | 75 |
| 2012 | 48 | 68 | 73 | 59 | 72 |
| 2013 | 118 | 123 | 128 | 125 | 128 |
| 2014 | 133 | 130 | 105 | 82 | 134 |
| 2015 | 74 | 94 | 98 | 108 | 95 |
| 2016 | 135 | 121 | 99 | 76 | 102 |
| 2017 | 51 | 61 | 67 | 72 | 79 |
| 2018 | 106 | 114 | 113 | 115 | 114 |
| 2019 | 76 | 82 | 108 | 87 | 87 |
| 2020 | 103 | 101 | 81 | 90 | 84 |
| 2021 | 102 | 99 | 118 | 130 | 106 |
| 2022 | 56 | 58 | 83 | 122 | 81 |
| 2023 | 103 | 83 | 119 | 108 | 121 |
| 2024 | 128 | 114 | 96 | 126 | 100 |
| 2025 | 130 | 109 | 93 | 116 | 86 |
How This Is Measured, and Caveats
- Who is included: a country is scored for a source when that source supplied at least 10% of its generation in its latest Ember year and the country generated at least 10 TWh in total (the floor the records pages use), so a single dam or wind farm does not headline.
- How a year is scored: each of the 15 years before a country's latest year is compared with the midpoint of the year before and the year after. Steady growth or decline, which wind and solar have and hydro sometimes does, sits on that line and is not counted as a swing; a weak or strong year stands out. The deviation is (actual minus midpoint) over the midpoint. A year is only scored while the source supplied at least 2% of generation in it and in both neighbouring years, and a country needs at least 8 scored years.
- The latest year of each series is not scored, because it has no following year to compare with, so the newest figures on each country's page are not in these tables. The baseline also carries a little noise of its own: a very weak year pulls the baseline of the years either side down, so they can look slightly better than they were.
- Typical swing is the mean absolute deviation. The worst year is the most negative deviation, and output lost is the midpoint minus the actual figure in that year, in TWh.
- A shared year needs at least 5 countries and at least 25% of those scored that year to be 10% or more below trend. That points to a common cause such as a drought or a calm spell across a region, but annual data cannot prove it, and some countries in a shared year are far apart.
- Annual data cannot separate weather from plant outages, new capacity switching on part way through a year, curtailment, or changes in how Ember or a national agency reports a source. Hydro is also left out where output changed by more than 2.5 times over the window.
- This is variability against trend in the past. It is not a forecast, and a country with a small swing can still have a bad year. Each country is scored over its own years: the last scored year is 2024 for countries with 2025 data and a year earlier for the rest.
Weather Exposure FAQ
Which countries' hydropower varies most from year to year?
Of 55 countries where hydro supplied at least 10% of electricity in their latest Ember year, Portugal (41%), Spain (29%), Bosnia and Herzegovina (27%), Sri Lanka (26%) and Croatia (25%) have the largest typical swing against trend over 2009 to 2024. The figure is the average distance, in percent, between each year's output and the midpoint of the years either side.
How much does wind output swing from year to year?
Across 24 countries where wind supplied at least 10% of electricity, the median typical swing against trend was 7.0% over 2010 to 2024. Costa Rica swung most (12.1%) and the United States least (3.0%). Growth in the fleet is removed from the trend, so a swing is a year that came in above or below its neighbours.
Do many countries have a weak hydro, wind or solar year at the same time?
Yes, in some years. In 2019, 14 of 55 scored hydro countries were at least 10% below trend. In 2022, 14 of 55 scored hydro countries were at least 10% below trend. In 2012, 5 of 17 scored wind countries were at least 10% below trend. In 2016, 7 of 24 scored wind countries were at least 10% below trend. A shared shortfall points to weather, but annual data cannot prove it.
Does a low year mean a drought or a calm year?
Not necessarily. Annual output also moves with plant outages, new capacity coming online part way through a year, curtailment and changes in how Ember or national statistics report a source. This page measures how far a year sits from its trend, not why.
Is this a forecast?
No. It describes how output varied in the past against its own trend. It does not predict next year, and a country with a small past swing can still have a bad year.
Source: Ember (ember-energy.org), licensed under CC BY 4.0. Data as of 2025. Annual statistics from Ember; each country scored over its own years.