Global Temperature in 2010: Is it the Hottest Year on Record, and Does it Matter?

All five of the major temperature indices — NASA’s GISTemp, National Climate Data Center (NCDC), Hadley Centre/UAE (HadCRUT3v), University of Alabama Huntsville (UAH), and Remote Sensing Systems (RSS) — have published their estimates of 2010 global surface or close-to-surface temperatures.

NASA reports that 2010 was tied with 2005 as the hottest year on record. NCDC also reports that 2010 was tied with 2005 for the hottest year on record, and Hadley, UAH, RSS reported 2010 as the second hottest year on record.

In all cases, except perhaps RSS, the 2010 temperature was close enough to other years to be within the margin of measurement error, so the ranking of individual years as hottest is not necessary the most meaningful metric. Rather, looking at the trends in temperature, and how recent years compare to long-term changes, can give a much clearer picture of how the global climate is changing.

Examining the Temperature Record

View larger image

While 2010 temperatures makes for headlines, the study of the climate is necessarily focused on multi-decadal trends. A single year can be strongly affected by factors like El Niño events, volcanoes, or other sources of natural variability. Over the course of decades, these factors tend to even out and the underlying trend is evident. As shown in the figure above, Earth has been warming since reliable global temperature records were first available around 1880 (Hadley produces a temperature record back to 1850, but there are some concerns about its reliability in the first 30 years because of sparse measurements and instrument changes.).

View larger image

A detailed look at the modern warming period from 1970 to present shows clear agreement among different temperature series but also some notable differences. Both satellite records (UAH and RSS) respond much more strongly to warm El Niño events (1998 and 2010) and cold La Niña events (2008) than surface temperature records. One satellite series (UAH) shows about 12 percent less warming than other series, while the rest largely agree on a warming rate of around 0.16 degrees C per decade over the past 30 years.

NASA’s GISTemp shows a bit more warming over the past decade than other series, mostly because of how it treats Arctic areas with little or no temperature station coverage. Specifically, for areas of the Arctic without temperature records, GISTemp infills them with average distance-weighted data from surrounding stations up to 1,200 kilometers away. All other series simply ignore Arctic areas not having coverage, which implicitly assigns the global average temperature trend to those areas. As there is general agreement in the literature that the Arctic should be warming faster than the global average, most series likely slightly underestimate Arctic warming, though there is also some concern that GISTemp’s method overestimates warming.

View larger image

Analysts can break down the global temperature into land and ocean (sea surface) components to better understand what is driving global temperature change. The figure above shows NCDC’s land record, ocean record, and combined global record. We see that land temperatures are increasing much more rapidly than ocean temperatures, which is somewhat unsurprising given the high thermal inertia of water. Because it takes a lot of energy to heat the oceans, and heat is transferred between surface waters and deep waters, a considerable part of Earth’s energy imbalance resulting from greenhouse gas forcing goes to warming deep ocean waters rather than to Earth’s surface.

Because the ocean covers so much of the planet’s surface, the global temperature record tends to be much more similar to sea surface temperatures than land temperatures. This is an important point, as it suggests that the global temperature record is not necessarily the best measure of the effect of climate change on people and land-based ecosystems. Indeed, climate models suggest that global land areas will be expected to warm considerably faster than the global average in a world where greenhouse gas emissions continue unabated.

Testing for Potential Biases

While the official temperature record paints a bleak picture of rapid modern warming, some “skeptics” have criticized the data for not taking into account biases resulting from urbanization and other factors. Much of this criticism has taken the form of pictures or surveys of individual stations, with little aggregate analysis of the effect of station location on the temperature record. With the use of various GIS databases, it is possible to classify stations as urban or rural based on various factors like nightlight brightness from space (via satellite observations), impermeable surfaces mapping, and population density data.

View larger image

The chart above shows the global temperature record controlled for a number of factors. It uses the Global Historical Climatological Network (GHCN v3) land stations and the ERSST sea surface temperature record used by NCDC. It examines what happens if we use the adjusted land temperature data (corrected for station moves and other inhomogenieties), the raw land data, the raw land data from stations that are located in areas that have no night lights visible from space, and finally the raw land data from dark stations that are not located at airports (in all cases, the same sea surface temperature series is used). As shown, even looking at the most conservative set of land stations, the difference in global temperature records is negligible. This result strongly suggests that recent increases in global surface temperatures are not in any way an artifact of data adjustments or station siting issues.

Analyzing Temperatures over Recent Years

Another common argument is that global temperatures have leveled-off over the past decade, or that there has been no global warming since 2000. It is true that a linear trend fit over just the past 10 years of data shows little to no warming, but this argument is somewhat misleading. As mentioned earlier, climate change is a multi-decadal phenomenon, and short-term temperatures are strongly affected by natural variability. To best assess if the warming over the past 40 years has continued into the most recent decade, analysts can do a simple test:  Calculate the trend in temperatures for the period from 1970 to 2000, and use it to predict what temperatures over the last decade would be expected to have been prior to actually knowing them. Reviewing actual temperatures from 2001 to 2010 indicates how the trend changes. If the 1970-2010 trend is higher than the 1970 to 2000 trend, then the last decade was warmer than expected.

View larger image

The figure above shows global land temperatures from 1970-2000 (in black) and 1970-2010 (in red), with the trends for both. The results show clearly that global land temperatures over the past decade were considerably warmer than would have been expected given the prior temperature record.

View larger image

Ocean temperatures over the past decade have been pretty much consistent with the prior temperature trend, with similar variability as seen in the past.

View larger image

Combining both land and ocean temperatures shows that global temperatures over the past decade have been warming slightly faster than would otherwise have been expected given the prior temperature trend. This analysis should help put to rest spurious arguments that global warming somehow “stopped” over the past decade. The more interesting questions are how variability over the last decade compares to past variability, and how consistent recent temperatures have been with climate model projections.

Both of those questions are current areas of focus for the climate science community.

Zeke Hausfather

Zeke Hausfather, a data scientist with extensive experience with clean technology interests in Silicon Valley, is currently a Senior Researcher with Berkeley Earth. He is a regular contributor to Yale Climate Connections (E-mail:, Twitter: @hausfath).
Bookmark the permalink.

6 Responses to Global Temperature in 2010: Is it the Hottest Year on Record, and Does it Matter?

  1. Wayne J says:

    Your temperature charts are worthless. We must know what the lower temperature readings were in order to see whether more or less heating has occurred. In the US, there has occurred over two dozen double record days. That is, they were both the hottest and coldest on record. So, were those days hotter or colder. Don’t know. It could have been very hot for most of the day, then cooled a lot for the last minute. So, mostly a hotter day. On the other hand, it could have been cold for most of the day, then got hot in the last hour. So, that made it mostly a colder day. You see, one point temp readings do not tell you whether the day was actually hotter or colder than normal.

  2. Wayne J says:

    Post Script: What is needed to get a better handle on global temps is to use FDD, Freezing Degree Days, or HDD, Heating Degree Days (the opposite of FDD). This quantifies the amount of hot or cold for winter heating or summer cooling requirements.

  3. logicophilosophicus says:

    The analysis only “works” because of the linear assumption. TAR projected accelerating warming this century. Describing temperatures which fall below all the IPCC model projections for 2001-2011 as “higher than expected” is frankly odd.

  4. Zeke says:


    Indeed, except for the words immediately following “higher than expected” being “given the prior temperature record”. This post was arguing against the idea that warming had somehow ceased by pointing out that temperatures over the last decade were consistent with if not higher than we would naively expect given the trend over the prior three decades, rather than comparing models and observations.

    I haven’t done a recent comparison of modeled and observed temperatures (its somewhere on my long to-do list), but a version I created in 2010 looked like this:

  5. Zeke says:

    Actually, here is a slightly more recent version that goes through the end of 2010:

  6. logicophilosophicus says:


    The graph everyone refers to for decelerated warming is here

    and what worries me is that (carefully ignoring the 1998 El Niño) the, say, 1976-1997 trend has faltered in 2001-2011. This is clear in the graph, but can easily be hidden by adding in the early 1970′s and then “going linear”. That way, however, there would be no uptick – nothing resembling the hockey-stick.