Skip to content

Schneider et al 2012 in a poster presentation to the two-day, “Taking the temperature of the Earth Conference,” that ends today, has the clever idea of looking at the temperatures of lakes and reservoirs around the world. They point out that in situ observations of lake surface temperatures are very rare on a global scale, but infrared imagery from space can be used to infer water surface temperatures of lakes and reservoirs.

They provide data for 169 of the largest inland water bodies world- wide using three satellite-borne instruments. Together they provide daily to near-daily data from 1981 through to the present, allowing them to calculate 25-year trends of nighttime summertime/dry-season surface temperature.

They find that the surface temperatures of the water bodies have been “rapidly warming” with an average rate of 0.350 ± 0.11 deg C per decade for the period 1985–201.

Two years ago Schneider et al published what was then described as the first global survey of lake temperatures. Then the researchers found a decadal trend of 0.45 deg C.

The researchers say the results provide a critical new independent data source on climate change that indicates lake warming in certain regions is greater than expected based on air temperature data.


Their graph of temperature anomaly (click on the image to enlarge) looks very familiar to anyone who is knows the global temperature datasets over the past thirty years. However, I don’t think their regression line is a good description of the data. My preliminary calculations suggest that there is no statistically significant trend post-1997. Hence an alternate description of their findings is that the world’s large bodies of water show the well known standstill of the past decade or so seen in global temperatures.

Note: While it is possible to draw a linear regression line between 1997 and 2011 (you can draw a trendline through almost anything) that yeilds 0.1 deg per decade it is statistically meaningless given the large variance of the data. The error on the trendline is several times its magnitude, and it is highly sensitive to moving the start and endpoint by a year or two. Conclusion: No statistically significant trend post-1997. Since 1997 the data is best represented by a straight line of mean 0.21 deg with a large standard deviation of 0.95 deg. Below is the post-1997 portion of the researcher’s graph. It is easy to see that the trendline calculated from the 1985-2011 data does not fit this section of the data in which there is no trend. Click on image to enlarge.


Plotted without any extra information, here is the post-1997 data. Click on image to enlarge.