Global Temperature Data
Measuring the temperature of our planet has many problems and there is currently no accepted method for measuring the Earth's temperature. The first issue is that widespread temperature data spread over the entire globe dates at best to 1880. There are a few longer-term records, but they are scarce. Thus, we must use other indirect methods referred to as a proxy if we want to examine temperatures of say the past two thousand years.
Proxy methods vary from things like tree-rings from very old trees to sediment and ice core data. Perhaps the best reconstruction that I have seen is that of Loehle (Energy & Environment, Vol. 18, pp. 1049 – 1058) seen below.
A couple of things are apparent in this reconstruction. First, global temperatures have fluctuated significantly over the last 2,000 years. The period from around 850 to 1050 AD was quite warm – much warmer than today. This time is referred to as the Medieval Warm Period. Anecdotal evidence from the time period supports this as grapes were grown in England and the Vikings inhabited Greenland. During this time, prosperity was widespread as an abundance of food was grown. On the other hand, the period from around 1600 to 1750 AD was quite cold and is called the Little Ice Age. Life for humans was very difficult and survival became the top priority.
Even in recent times, temperature records have issues of quality, coverage, and continuity. To produce an accurate record of temperature each and every day, the thermometer or temperature probe must be sited properly. A project by Anthony Watts of surfacestations.org is attempting to document all of the 1221 US Historical Climate Network stations. He has found numerous location problems with the temperature probes including some located in or near parking lots, on rooftops, and near air-conditioners. I encourage you to visit his website to see his pictures and statistics. Station coverage has fluctuated quite a bit over the years. While North American coverage is quite good, other parts of the world are woefully inadequate. Brazil, for example, has a land mass almost identical to our lower forty-eight states. Yet, in one data set used to calculate global temperatures, only six stations are used. In the former Soviet republic, hundreds of stations were shuttered when the republic spilt apart in the early 1990's. Pictured below is a map of the Global Historical Climate Network.
While land data has issues already pointed out, sea surface temperatures also are problematic. Much of the historical sea surface temperatures were recorded by ships as the traveled across the water. During World War's I and II, commercial ship traffic was greatly reduced, which produced a lack of data in those years. More recently, data has been collected by stationary buoys. These buoys, though, take temperatures right at the surface whereas ships pulled water from beneath the surface.
There are currently four major efforts at monitoring global temperatures. A brief discussion and explanation of each is included.
GISS - Produced by NASA's Goddard Institute for Space Studies. It is based on the work of Dr. James Hansen. Outside of North America, raw data is used that may include a bias of the Urban Heat Island effect. North American data is adjusted (or manipulated, as some might suggest) for this bias. This version also includes polar data not found in the other measurements. In August of 2007, a Canadian blogger by the name of Steve McIntyre posted at his website climateaudit.org that he had discovered a Y2K programming error in the GISS program. The effect of this discovery was to reduce all of the yearly records after 1999. While Hanson claims that it is no big deal, others would disagree. In my opinion, because of this and other known issues, I think that this temperature record should be ignored.
HadCRUT3 - Produced by the Climate Research Unit of the University of East Anglia in the United Kingdom. They use over 3,000 land-based monthly temperature averages along with sea surface temperatures.
UAH Satellite - Produced by the University of Alabama-Huntsville. Uses satellite data that began in late 1978 to measure the temperature of the entire atmosphere and is based on the work of John Christy and Roy Spencer. "Ours is the only dataset that has been compared to non-satellite data," said Christy. "This gives us confidence in its results." In my opinion, this is the best of all of the records, but it only goes back about thirty years.
RSS Satellite - Produced by the Remote Sensing Systems - a business in Santa Rosa, California. Uses satellite data as well to determine both land and sea surface temperatures.
Before I show some of this data, I will share a smaller study that I did some time ago to satisfy my own sanity. The US Historical Climate Network or USHCN is a collection of 1,221 data stations. I picked twenty-five of these stations from across the country and downloaded the yearly averages for the past 120 years or so. The stations I picked were all rural to avoid as much of the Urban Heat Island effect. Also, I picked all northern or middle of the country sites as these should be most affected by the increase in carbon dioxide. Those locations include:
|
Minonk, IL
|
Gardiner, ME
|
Miles City, MT
|
Alma, MI
|
McCook, NE
|
|
Eureka Springs, AR
|
Willows, CA
|
Tarboro, NC
|
Gannvalley, SD
|
Clarinda, IA
|
|
Glenville, WV
|
Greensburg, KY
|
Montevideo, MN
|
Fillmore, UT
|
Lowville, NY
|
|
Selinsgrove, PA
|
Lancaster, WI
|
Port Townsend, WA
|
Sweet Springs, MO
|
Ellsworth, KS
|
|
Princeton, IN
|
McConnelsville, OH
|
Payette, ID
|
Waynesboro, TN
|
Cheyenne Wells, CO
|
The 120 year graph of the aggregated data is seen below:
The fuchsia-colored line represents the trend line for this data. Notice that it is nearly horizontal, which means that temperatures are not that much different today than they were 120 years ago.
The increase in carbon dioxide levels has been occurring over the last sixty years or so. Thus, let's examine only that portion of the data as seen below.
While hard to discern, there is a slight up tick in the trend line. The equation for this line (ignoring the intercept) is y = 0.0033x. This works out to a wimpy 0.2 degrees Celsius increase for the last sixty years - hardly a catastrophic increase!
Now, let us look at some global data. I have chosen the HadCRUT data, which I believe is better than NASA's GISS data (see earlier discussion). One important note - data can and is manipulated in its presentation to deceive the reader. As I did above, I chose to start from 1945 because this correlates to when the observed carbon dioxide levels began to increase. Many of the pro-AGW crowd, though, will choose a much later date like 1979 to present for their graphs. This is important because a closer look at the data shows global temperatures actually fell from the 1940's to the 1970's. Starting from the late 1970's magnifies the warming since then and makes it seem to correlate strongly with the increasing carbon dioxide. Another trick to is to "squash" one of the axes - in this case - the Year or x-axis. Doing this distorts the data and makes the trend line appear with a much higher slope. Put these two together, here is what a graph the might look like:
When the average person sees this, they would probably believe that the sky is falling! Now let's see that same graph displayed without squashing the x-axis:
Not as dramatic is it! This still shows temperatures have increased from 1979 to present. However, notice the last few data points have dropped significantly. Is this the start of a cooling trend? Only time will tell.
One thing is certain, from the data presented thus far, you may have noticed a spike in temperatures in 1998. That year ranks as the second hottest globally in the last 120 years or so (1934 beats it out by a small margin). It should be noted, though, that this year had an unusually strong magnitude El Nino in the eastern Pacific ocean. (An El Nino is a warming of the water's off of South America and extending into the middle of the Pacific. El Nino's are natural phenomena not associated with man.) Since 1998, temperatures dropped in 1999 and have remained fairly constant since as seen below.
For even more recent, since 2002 the data shows a cooling trend may be starting.
One final way to look at temperature data is to average out some of the individual peaks and valleys. In this final look, I have used a three month running average. To do this, I add up the current month with the month before and after. The sum is then divided by three and then repeated for every month in the data series. The reason I did this is twofold. First, the trend line tends to be very "noisy" using each month as a data point and this produces a poor correlation coefficient for the trend line. Second, weather patterns can often run in about a one month cycle and this can help to smooth out some of those extremes. The result of this for the last six or so years is:
Once again, a small, but noticeable decrease in the trend line is seen.
Okay, so this is one of the four Global Temperature surveys. Do they all produce the same results? Well, three of the four do - with the outlier being GISS. Let's take a look at the UAH data for the past six and a half years.
I should make one note here - not every temperature metric uses the same thirty year period to establish a temperature anomaly of zero. Other than that, this graph is remarkably similar to the HADCrut data with an almost identical down trend. The three-month running average also looks the same.
One final intriguing aspect of the recent data is the sharp downturn since January of 2007. While similar dips have occurred over a relatively short period before, this one appears to be the start of a longer trend. Only time will tell if this is true.