Browsing a few of my favorite Internet news sources earlier this week (some left leaning, some right leaning, some not leaning, and one 'fair and balanced') I came across Drudge's link to this DailyTech blogpost on evidence for global cooling. That's right, there is scientific evidence for global cooling. Or so DailyTech claims:
Over the past year, anecdotal evidence for a cooling planet has exploded
...and the reason for cooling? The sun.
Scientists quoted in a past DailyTech article link the cooling to reduced solar activity which they claim is a much larger driver of climate change than man-made greenhouse gases. The dramatic cooling seen in just 12 months time seems to bear that out. While the data doesn't itself disprove that carbon dioxide is acting to warm the planet, it does demonstrate clearly that more powerful factors are now cooling it.
Intrigued I decided to dig further. Is it possible that sun activity swamps any effects of CO2? Let's take a look.
DailyTech's compiled evidence focuses on this graph (click for larger):
As you can see (and as highlighted by the big blue line on the graph), there seems to be a dramatic decrease in temperatures from January 07 to January 08. Pretty convincing when viewed in isolation. But I'm not one to view things in isolation (unless it's the Sports Illustrated Swimsuit Edition...oops, TMI). So I decided to go the source data and see for myself.
DailyTech's evidence focuses on a compilation of readings from four sources* of global surface temperature data. Not knowing much about three of the four sources I will focus my attention on the data from NASA--a fairly reputable source in my opinion.
On their web-site, NASA provides the monthly Global Land-Ocean Temperature Index from 1880 to Present. To convince myself I know what I'm doing Using that data, I duplicate below the equivalent of DailyTech's graph using NASA's data (global surface temperatures 1988-January 2008):
So far so good, it looks like the NASA data shows the same thing and I can draw a graph in excel and export it as a jpg for your viewing pleasure.
Yep, there's a dip in 2007-2008. One point for DailyTech. But wait, there's more. If you expand the graph back to 1880, a different pattern emerges. I've taken the liberty to add a trend line to make my point.
Even though there appears to be a noticeable dip in temperatures over the past year, there is a long-term trend at work, and that trend is noticeably upward.
Don't get me wrong, I'm not claiming that we're not in a cooling cycle. Just that one year out of 220 does not a trend make.
But, that's not the bigger point. The DailyTech post cites articles that claim that sun activity is responsible for the downturn and seems to imply that in the long-run, sun activity has much more influence on temperature than CO2. Is that true? Keep reading, because I don't think the evidence backs it up.
To check the relationship between global surface temperatures and atmospheric CO2 concentrations, I will use the National Oceanic and Atmospheric Administration's monthly mean CO2 concentrations** measured at the Mauna Loa (yummy macadamia nuts!) Observatory in Hawaii. Observations are only available from January, 1958 to December, 2007 but that's at least 30 years longer than the timeframe the DailyTech sources show.
The graph to the left maps the trend in CO2 concentrations (right axis) and the trend in temperatures (left axis)on the
same graph. The trends are clear. CO2 and temperatures are both increasing. Some might be tempted to conclude that this is evidence that CO2 concentrations cause increasing temperatures. I am not one of them. First, the two vertical axes can't be compared, other than to note they increase from bottom to top. Different scales render direct comparisons difficult. Second, even if scaled comparisons were possible, the graph simply implies correlation, not causation. In econspeak, correlation is a necessary condition for causality, but not sufficient. More on this in a minute.
A scatter plot of CO2 concentrations and temperatures provides further evidence of this correlation (left). Clearly there is positive correlation between temperatures and CO2 concentrations, but there are three possibilities for this correlation: 1) CO2 causes higher temperatures, 2) Higher temperatures cause higher CO2 concentrations, 3) Something else causes higher temperatures and higher CO2 concentrations (spurious correlation). Without much more information we can't figure out which of these three are true and this is beyond the scope of what I want to talk about here. Instead, I will leave it as at least an unproven possibility that higher CO2 concentrations cause higher temperatures. That will be enough to prove my point later.
So what about sun activity? To investigate the relationship between sun activity and temperatures I will use more data from NASA. This time from the Solar Physics Program at the Marshall Space Flight Center. The next graph shows the monthly average sunspot count and the global temperature data on the same graph. To make a relevant comparison I only use 1958 to present although sunspot data is available much longer.
The cyclical nature of sunspots is clear. Since 1958 there have been four peaks and four valleys and consistent with the DailyTech post, it looks like we are currently in a low activity cycle.
Less clear, but at least at first glance, it also looks like there is little trend in long-term sunspot activity. So I am tempted to claim that any effect of sun activity on temperatures would have to be short-lived because of the non-trending cycle. But not yet. First, take a look at the scatter plot of sunspots versus temperature (left).
While there is no obvious correlation (no detectable upward or downward trend in the data), a very interesting pattern emerges
. In periods of low sunspot activity, the variability in temperatures is much greater than periods of high sunspot activity***. Sunspot activity may add to temperature variability, but there appears to be no contribution to the trend. In other words, sunspot activity may cause temperatures to fluctuate more or less around a long-term trend, but something else is causing the trend.
Finally, to rule out sunspots as the cause of spurious correlation between CO2 and temperatures, the last graph shows the relationship between monthly changes in CO2 concentrations and monthly changes in sunspot activity. No correlation (as shown by the flat trend line). 
So what's my conclusion? First a disclaimer:
Disclaimer: I have proven nothing here. Just providing anectotal evidence for discussion. If I were trying to prove my result I would run a regression and show that everything I have said here holds up statistically--in the env-econ blogging sense of statistics. But that would just be bragging. Feel free to draw your own conclusions.
Follow the bouncing ball of logic. C02 concentrations and temperatures are correlated. Sunspot activity and temperatures are not and neither are CO2 concentrations and sunspot activity. Therefore, sunspots activities are not a causal factor in determining long-term temperature trends.
The current downturn in temperatures may be caused by a valley in the sunspot cycle. But that doesn't mean that global cooling is taking place. It just means that temperatures are likely to be more variable until sunspot activity increases again.
In summary:
- Is there a long-term upward trend in temperatures? Yes.
- Is there a one year downturn in temperatures? Yes.
- Is the downturn caused by decreased sunspot activity? Possibly.
- Is it just as likely that the same decrease in sunspot activity will cause a temperature increase next year? I'm nodding. But it might cause a lower temperature too. That's what variance does.
- Do sunspots have an effect on the long-term temperature trend? Nope.
- Does CO2 cause temperature changes? I have no idea, but I know they're correlated.
*UK's Hadley Climate Research Unit, NASA's Goddard Institute for Space Studies, University of Alabama Huntsville and Remote Sensing Systems of Santa Rosa
**According to NOAA, "Data are reported as a dry mole fraction defined as the number of molecules of carbon dioxide divided by the number of molecules of dry air, multiplied by one million (ppm)."
***Those versed in econometrics will notice the classic heteroskedasticity cone shape.
