Global Warming Melts Away

Wondering where things stand on global warming?  Let's go to the science of it.  And by "science" I mean physical observations.  Nothing complicated.  What do thermometers say?  What's happening with polar ice caps?  And carbon dioxide?

In the following analysis I use publicly available data as provided by government scientists at places like NASA, the U.K. Met office, and NOAA.  This is the data we're told to use and trust.  I'm using it and trusting it.

Measured Temperatures.  The two charts below show recent global temperatures (NASA/GISS data) in units of hundredths of a degree above the 1951-1980 average.  The lines are linear regression trends for the extent shown.  March 2012 was the most recent month of data at the time of writing.

Based on NASA/GISS data going back to 1880:

  • The trend of annual mean temperatures since 1880 is warming of 0.59 degrees Celsius per century.
  • The trend over the last nine years (since 2002) is one of cooling.
  • There has been no statistically significant (at 95% confidence level) warming in 14 years (since 1997).
  • The trend in March temperatures for the last 12 years (since 2000) is one of cooling.
  • There has been no statistically significant warming in March for 17 years (since 1995).
  • March 2012 was the coolest March in 13 years (since 1999).

Similar observations can be drawn from Hadley Center data from the U.K., which goes back to 1850.

  • The trend of annual mean temperatures since 1850 is warming of 0.45 deg. C per century.
  • The trend over the last 13 years (since 1998) is one of cooling.
  • There has been no statistically significant warming in 16 years (since 1995).

In short, both temperature data sets (NASA and Hadley Center) show

  1. Minimal global warming over the last 130 to 160 years: about half a deg. C per century.
  2. No statistically significant global warming in the last 14 to 17 years.
  3. Global cooling in the last 9 to 13 years.

Sea Ice.  The following two charts show Northern Hemisphere and Southern Hemisphere winter sea ice extent, in units of millions of square miles.

While Northern Hemisphere winter sea ice extent was shrinking over much of the time that satellite measurements have been available (since 1979), the trend over the last eight years has been growth.  There has been no statistically significant shrinkage of winter sea ice in the Northern Hemisphere for 13 years (since 1999).

Southern Hemisphere winter sea ice extent has grown over all the 32 years that satellites have been measuring it.  And that growth is statistically significant.

The chart below shows total global sea ice, north and south, over the whole time that satellites have been measuring it.  There is not now, nor has there ever been, any apparent alarming trend in either direction in the amount of sea ice on our planet in that time.  The latest measurements put the amount of sea ice a bit above the 1979-2008 average.

Correlating Temperature and CO2.  The chart below shows global temperatures vs. measured atmospheric CO2.  The data cover the years from 1959 through 2011.  The correlation between CO2 and temperature is clear to the eye.  The calculated correlation coefficient is 0.91.

In blaming man's activities for global warming, this is as close to the "smoking gun" as it gets.  A correlation coefficient of 0.91 appears high.  But the correlation coefficient between temperatures and year is also a high 0.89.  Simply put, both CO2 and temperatures went up over time.  Inferring that CO2 is the cause of the temperature increase is a bit of a stretch.  Anything that went up in that timeframe would have similar correlation.

For example, here is another chart: global temperature anomaly vs. federal debt (in millions of dollars).

The correlation coefficient between global temperature and U.S. federal debt is also 0.91.  If you use correlation to convict CO2 of climate change, then federal debt is just as guilty.

Even if you believe in the correlation-is-causation theory, that correlation has broken down recently.  Look at the chart below.  It shows that the correlation between CO2 and temperature has entirely disappeared in the last decade.  In fact, the correlation is negative since 2002.

What the data show.  Whether you think the globe is getting warmer or not largely depends on your timeframe.  If your timeframe is the entire time that modern temperature records have been kept, meaning since the 19th century, then the trend is indeed warming: about half a degree Celsius (about one degree Fahrenheit) per century.  That is less than the difference in average July temperatures between Cape Hatteras and Charlotte, NC, for example.

If your time frame is more recent, like the last 15 years, then no warming can be detected.  In fact, the trend has been cooling for about the last dozen years.  This is apparent in NASA's data, the U.K.'s Met Office data, and measurements of both Arctic and Antarctic sea ice.  Indeed, Antarctic sea ice has been growing for the 32 years since satellite measurements have been available.

The main link between man's activities and global temperatures, other than pure (all-else-equal) theory and assumption-filled computer programs, has been the correlation between measured atmospheric CO2 and measured temperatures.  But that correlation suffers from at least two drawbacks: (1) correlation is not causation, and the correlation between federal debt and temperatures is just as strong and that between CO2 and temperatures.  And (2) that correlation has broken down in the last decade, coinciding with the cooling trends in temperatures noted above.

In short, the data show nothing alarming at all: very mild warming over the long term, and actual cooling over the short term.

Please look at just one more chart, below.  That chart is the global temperature anomaly as provided by the Hadley Center.  Usually you might see it plotted on a scale that spans about one degree Celsius, making the trend look alarming.  On this chart I chose the total vertical scale to match the difference in average June temperatures between Flagstaff and Yuma, Arizona (about 16 deg C).

Does it strike you as just a bit odd that the data of this chart causes so much fuss?

Randall Hoven can be followed on Twitter.

Sources & Methods:

(1) NASA/GISS data is available at  I used the global mean data of the "Combined Land-Surface Air and Sea-Surface Water Temperature  Anomalies (Land-Ocean Temperature Index, LOTI)" product.

(2) Hadley Center of the UK Met Office data is available at  I used the HadCRUT3 global (GL) product.

(3) Sea ice data from the National Oceanic and Atmospheric Administration (NOAA) is available via the University of Colorado here:  I used the peak winter month for both poles: March for north (N_03_area.txt) and September for south (S_09_area.txt).

(4) The chart of total global sea ice is available from the University of Illinois here:  The website of the Polar Research Group in the Department of Atmospheric Sciences at the University of Illinois is available here:

(5) CO2 measurements taken at Mauna Loa, Hawaii are available from NOAA here:

(6) Daily mean temperatures for each month for several U.S. cities is available from NOAA here:

(7) Linear regression lines, regression analysis, and correlation coefficients were done with Microsoft Excel's built-in functions.  All references to "statistical significance" are at the 95% confidence level.  All references to "trend" are to linear regression.  All temperatures are given in degrees Celsius.

If you experience technical problems, please write to