Friday 17 July 2015

Lakes are warming at a surprisingly fast rate

Map with lake temperature trends. As so often the trend is strongest in the mid-latitudes of the Northern Hemisphere. Two seasons are used to minimize cloud blockage: JAS (July, August & September) and JFM (January, February & March) for the dry season.

Many changes in the climate system go faster than expected, which fits to my hunch that the station temperature trend may have a cooling bias. The coming time I would like to blog about a few example changes. This first post is about lakes and rivers, their temperature changes and changes in the date they freeze and the date the ice breaks up. In this case it is hard to say whether this goes faster than expected, because there is not much research on this, but the temperature changes sure are surprisingly fast.

We will see more research on this in future. There is now a Global Lake Temperature Collaboration (GLTC), which is collecting and analyzing lake temperatures. It is easy to complain about the weather services and how they keep on making changes to the meteorological networks and fail to share many important observational datasets (and I will keep complaining), but at least they have systematically made such historical observations. For other observations, such a lake temperatures and especially ecological datasets it is much harder to obtain long and stable observations lacking institutional support. Now going into the century of climate change this institutional failure becomes even more problematic. I feel it should be part of the international climate change treaties to set up organizations that can provide long term, well-documented and climate quality stable measurements of a large range of environmental systems.

One of the reasons to found the Global Lake Temperature Collaboration was a scientific article by Schneider and Hook in 2010. It analyzed the temperature trends of lakes using thermal infra-red satellite images ([[AVHRR]] and [[ATSR]]). Between 1985 and 2009 the satellite lake temperatures increased by 1.13°C, which they report is more than the regional air temperature increases. For comparison this amounts to 4.5°C per century; see graph below. This is stronger than the land temperatures, although one would expect less warming of the lakes. For the same period the Northern Hemisphere temperature of Berkeley Earth increased by 3.9°C per century.

Trend in lake temperature anomalies. This is an average over all 113 water bodies that were large enough and had at least 15 years of data. The trend over the period 1985 to 2009 is 4.5±1.1°C per century.

Satellite data is great for their global overview (see graph at the top of this post), but are tricky when it comes to trends. Trend estimates from satellite data are difficult due to degradation of the instruments in a harsh space environment, changes in the orbit and height of the satellite, while at the same time the limited life span of satellites means that the instruments are regularly replaced and technological improvement often lead to new designs. All this happens while the small number of satellites means that there is minimal redundancy to study such data problems. Schneider and Hook (2010) did their best to study such data problems, used data from seasons with few clouds, so that the satellite can see the lakes more often. They compared their data with surface observations and found only small biases between the satellites and no indications of trend biases. And they used night-time observations to reduce the influence of orbital drift.

Still an astonishing outlier trend from satellite data calls for ground validation (in situ measurements). The GLTC now provides a dataset with both satellite and in-situ lake temperature measurements. The paper describing the dataset is out now (Sharma et al., 2015). The paper analyzing the trends has still to be published, but Philipp Schneider of the GLTC wrote to me that the in-situ trend is similar. Further papers explaining the trends are in preparation.

At the moment it is not clear yet what is the reason for the stronger increase. Many lakes are close to the Arctic, thus it could be ice albedo feedback for Northern lakes. The summer surface temperature of Lake Superior over the interval 1979 - 2006 has, for example, increased by 11±6°C per century, faster than regional atmospheric warming. This is thought to be due a reduction in the albedo of the lakes due to a reduction in ice cover (Austin and Colman, 2007). The air in the Arctic is also warming faster than elsewhere. Changes in the observations are naturally also possible — as always — and the land surface temperature trend might also be underestimated.

Stronger insolation may especially heat lakes, which typically reflect less solar radiation than the land surface. This may especially be important for the recent decades in the industrialised countries, where air pollution has been reduced considerably.

Surface temperatures may change due to less mixing with deep cold water: as the top temperatures warm more the warm surface water mixes less well with the deep cold and dense water and reductions in the wind speed can reduce mixing (Butcher et al., 2015). Also changes in the transparency of the water can influence where the solar warming ends up (Butcher et al., 2015).

In other words, no observation is ever completely straight forward and the air temperature is just one factor influencing lake water temperatures. Just like with station measurements, you always need to study which part of the changes in the raw observations is the part you are interested in.

On the other hand, because more of the additional greenhouse warming goes into evaporation rather than to warming, the warming of the air over land is expected to be stronger than the lake water temperatures. Butcher and colleagues (2015) estimate that the surface water temperature increases are only about 77% of increase in average air temperature change. Previously Schmidt and colleagues (2014) estimated this to be between 70 to 85%.

Freezing and melting of lakes and rivers

The lake temperature observations are unfortunately not very long. To put their warming into perspective there are also observations of the freezing and breakup dates of lakes and rivers. These sometimes go back many centuries. Magnuson and colleagues (2000) have gathered 39 observational dataset on lakes and rivers with more than 150 years of data. They found that all but one of them showed later freezing dates and earlier breakup dates. The freeze dates are 5.8 days per century later and the breakup dates are 6.5 days per century earlier. This is comparable to a warming of the regional air temperature of about 1.2°C (2°F) per century, but with a large confidence interval. For comparison, the Berkeley Earth dataset shows a warming of the NH land temperatures for the same period between 1846 and 1995 of 0.67°C per century. For this comparison it should be reminded that these rivers and lakes are in high latitude regions that warm more.

Time series of freeze and breakup dates from selected Northern Hemisphere lakes and rivers (1846 to 1995). Data were smoothed with a 10-year moving average. Figure 1 from Magnuson (2002).

For one dataset they mention a small non-climatic influence (a power plant) and one dataset (a harbor) is excluded because of its much stronger trend. Magnuson and colleagues seem confident that warm waste water is not the reason for the trends in the other series, but do not explicitly write about that in their 4-page Science article.

Melting lakes showing a clear contrast between ice and water. This makes is relatively easy to use historical aircraft and satellite observations. Figure 5 from Duguay et al. (2003).
The advantage of such datasets is that they provide yearly information, that the observations are often long and that they are relatively precise. For the recent decades, the observations can be made for a large number of lakes from space. These freezing and breakup dates are also nicely determined by the temperature averaged over a longer period, which removes a lot of variability: Freezing is determined by the temperature in the last two months before the event.

What complicates matters is that snow isolates the ice and slows down freezing and thawing. Thus the ice breakup of lakes is also influenced by snow on top of the ice. Also the depth of the lake is important; deeper lakes thaw later (Duguay and colleagues, 2003). The breakup date of rivers is also determined by the timing and amount of spring river run-off.

Concluding. Lake temperature are rising fast over the last decades, likely faster than the air temperatures, while one would expect that lakes warm slower because more heat goes into evaporation. This could be in part due to an albedo feedback or due to more sunshine from reductions in air pollution during these decades. Furthermore, also the length of the period that lakes and rivers are frozen decreases rapidly since 1850. Converting this in a temperature signal introduces a large uncertainty, but also this temperature increase seems to be faster than the current estimates of land surface temperature increases.

Now if I were a political activist, I would call climate science a hoax or claim that all scientists are stupid. Just listen to lonely brilliant Galileo me, forget science, the oceans will boil soon.

But, well, I am sorry for being such a scientist, I would just say, that we found something very interesting. Apparent discrepancies help one to understand a problem better. Only once we understand the warming of the lakes better can we know whether climate science was wrong.

As will be the case for most of this blog series on faster changes, the topic of this post goes beyond my expertise. Thus if anyone knows of good studies on this topic, I would be very grateful if you could leave a comment or would write me. Especially if anyone knows of comparisons between air and water temperature trends or between models and observations for lake and river temperatures or their ice cover. There are a number of interesting papers coming up, thus I will probably have to write an update in a few months.

Related resources

Climate Central reports on a new study (December 2015) on global lake temperature changes.

John Lenters (coordinator of the GLTC) at Nature's Scientific Data blog: Author’s Corner: Are lakes warming?

Global Lake Temperature Collaboration

Why raw temperatures show too little global warming

A recent study shows that since 1970 the ocean heat content of the upper 700m has increased 15% more than climate models have predicted.

Chris Mooney has an interesting piece in the Washington Post on related snow cover observations: Northern Hemisphere snow cover is near record lows. Here’s why that should worry you.

More scientific articles on lake ice by the group of John Magnuson.

The river and like ice phenology database: Benson, B. and J. Magnuson. 2000, updated 2012. Global lake and river ice phenology database. Boulder, Colorado USA: National Snow and Ice Data Center. doi: 10.7265/N5W66HP8.


Austin, J. A. and S. M. Colman, 2007: Lake Superior summer water temperatures are increasing more rapidly than regional air temperatures: A positive ice-albedo feedback, Geophysical Research Letters, 34, art. no. L06604, doi: 10.1029/2006GL029021.

Butcher, Jonathan B., Daniel Nover, Thomas E. Johnson, and Christopher M. Clark, 2015: Sensitivity of lake thermal and mixing dynamics to climate change. Climatic Change, March 2015, 129, Issue 1-2, pp 295-305, doi: 10.1007/s10584-015-1326-1.

Duguay, Claude R., Greg M. Flato, Martin O. Jeffries, Patrick Ménard, Kim Morris, and Wayne R. Rouse, 2003: Ice-cover variability on shallow lakes at high latitudes: model simulations and observations. Hydrological Processes, 17, pp. 3465-3483, doi: 10.1002/hyp.1394.

Magnuson, John J., Dale M. Robertson, Barbara J. Benson, Randolf H. Wynne, David M. Livingstone, Tadashi Arai, Raymond A. Assel, Roger B. Barry, Virginia Card, Esko Kuusisto, Nick G. Granin, Terry D. Prowse, Kenton M. Stewart, and Valery S. Vuglinski, 2000: Historical trends in lake and river ice cover in the Northern Hemisphere. Science, 289, pp. 1743-1746, doi: 10.1126/science.289.5485.1743.

Schmid, Martin, Stefan Hunziker, and Alfred Wüest , 2014: Lake surface temperatures in a changing climate: a global sensitivity analysis. Climatic Change, 124, pp. 301–315, doi: 10.1007/s10584-014-1087-2.

Schneider, Philipp, and Simon J. Hook, 2010: Space observations of inland water bodies show rapid surface warming since 1985. Geophysical Research Letters, 37, art. no. L22405, doi: 10.1029/2010GL045059.

Schneider, Philipp and Simon J. Hook, 2012: Global Trends of Lake Surface Temperatures Observed From Space. Geophysical Research Abstracts, 14, EGU2012-2858, EGU General Assembly 2012.

Schneider, Philipp, Simon J. Hook, Derek K. Gray, Jordan S. Read, Stephanie E. Hampton, Catherine M. O’Reilly, Sapna Sharma, and John D. Lenters, 2013: Global lake warming trends derived from satellite and in situ observations. Geophysical Research Abstracts, 15, EGU2013-2235, EGU General Assembly 2013.


  1. Just a question, did they account for waste heat input, for example from industries?

  2. Arjan, good question. That was also one of my first thoughts. You often hear that industry is heating streams and canals too much and that that causes ecological problems. Similarly also waste salt used to be a problem for the Rhine and would influence freezing.

    The ice data Science article explicitly mentions that one series may have a small contamination and that one series has been removed. And the dataset (now linked above) contains information on the size of nearby urban areas and power plants. Thus I guess they studied it and did not find it to be important enough to mention, except for those two series.

    The lake temperature paper and some others on lake temperatures do not mention waste heat at all. The lakes discussed are quite large, otherwise you cannot see them from space. I guess that it is hard to warm such large lakes with waste heat. Quite complicated factors influencing lake temperatures were mentioned in the papers I read, but waste heat was not even mentioned as something that might be worth of future study. Thus it sounds as if lake hydrologists are quite confident that it is not important, otherwise they would have at least discussed it.

    I am happy to be corrected by more knowledgeable people. This is new to me.

  3. Turbidity changes (via runoff from ag or urban development) seems like a better candidate for affecting temperatures than direct industrial heating, although you mentioned they checked for that (transparency).

    This is a really great article, Victor, and the series concept is fantastic!

    Possibly try to solicit comments from the subject matter experts (allowing anonymity if desired)?

  4. I would say they thought about transparency changes, checked may be a too big word at this stage of the research.

    Asking experts is not much easier than writing a post yourself. Then you have to assess who is credible, which is also hard when you do not know the community working on that topic. I guess I would prefer to write a post myself with many disclaimers as to accidentally post a confident post of someone who presents an outlier opinion. That is especially difficult when you are talking about changes that go faster than expected. You do not want to get a post by a Wadhams by accident.

    And I have tried many times to ask for guest posts from people I trust. Unfortunately many do not see writing a blog post as sufficiently important.

    That being said, if someone reading this has a proposal for a guest post that fits to this blog, do contact me.

  5. TBC, I was suggesting not asking for independent posts but simply for comments on yours (*after* posting). That way it wouldn't be a strain on you, and if they decided to write a comment it could be minimal.

  6. Steve Bloom, ah, I have send emails to five researchers whose work I discuss above. Thus I guess that if I was WUWT-wrong, they would comment and explain what is wrong.

    It is a pity that many scientists seem to be rather shy about discussing their work on blogs, but then WUWT & Co. have put a lot of work in making sure that blog are not seen as a productive and serious medium.

  7. This also means that lakes are becoming a less friendly place for fish and other things that live there as pO2 decreases in the warming water. Recently Deutsch, et al pointed this out for the oceans

  8. Some might even say, Eli surely not, that not only temperature and dissolved gasses change, but that these initial changes will change the entire ecosystem and also affect species that in principle would be fine with the new temperature, CO2 and oxygen.

    Such changes in the competition between species will normally take time, especially when new species establish themselves (punctuated equilibrium), and may be influenced to a large part by changes in the probability of extreme weather (freezing, hot water, strong run-off, lack of water and low flow), not just the long term mean.


Comments are welcome, but comments without arguments may be deleted. Please try to remain on topic. (See also moderation page.)

I read every comment before publishing it. Spam comments are useless.

This comment box can be stretched for more space.