With upcoming release of IPCC Fifth Assessment Reports beginning late in September, there will be a sharp focus on specific issues like projected sea-level rise but also on broader issues like climate sensitivity and the decade-and-a-half-long slow-down in the rate of overall warming. Let’s begin by examining that slow-down in depth, and just what is involved in taking Earth’s temperature …
Global surface temperatures have warmed more slowly over the past decade than previously expected. Some in the media have seized this warming pause in recent weeks, and the UK’s Met Office has just released a three-part series of white papers looking at the causes and implications.
While there is still no definitive cause identified, the Met Office scientists point to a combination of more heat going into the deep oceans and downturns in multi-decadal cycles in global temperature — natural variability — as the primary drivers of the pause. Others argue that a plethora of recent small volcanoes, changes in stratospheric water vapor, and a downturn in solar energy reaching the Earth may also be contributing to the slow-down. While few expect the pause to persist much longer, it has raised some questions about the growing divergence between observed temperatures and those predicted by climate scientists.
To understand the decline in the rate of warming requires understanding the different ways of measuring global temperature. These include measurements taken from land-based temperature stations (mostly using mercury thermometers), ocean buoys, ships, satellites, and weather balloons.
Land/Ocean Temperatures
The most common estimate of global temperatures comes from a combination of land temperature stations with sea surface temperature data from ships and buoys. There are three main global land/ocean surface temperature series, produced by NOAA’s National Climate Data Center (NCDC), NASA’s Goddard Institute for Space Studies (GISTemp), and the UK’s Hadley Center (HadCRUT).
As shown in the figure above, all three series agree quite well on global temperatures. The short dashed grey line at the upper right shows the trend in temperatures since 2001, while the long dashed black line shows the trend since 1970. Although the rise in global temperatures has slowed in recent years, it is not obviously divergent from the underlying long-term trend. On the other hand, there are no periods with similar temperature stagnation other than ones associated with a major volcano (e.g. Pinatubo in 1992 or El Chichón in 1982).
Land Temperatures Only
Land and ocean temperatures have diverged notably in recent years. Ocean temperatures generally rise more slowly than land temperatures as a result of the large thermal inertia of the oceans. Since 2001, land temperatures have continued to rise, though slightly more slowly than in prior years.
Here again the short dashed grey line at the top right shows the trend since 2001, while the dashed black line shows the trend over the whole period. The major land series used are CRUTEM4 (the land component of HadCRUT4), NCDC, GISTemp, and Berkeley Earth.
Ocean Surface Temperatures Only
Most of the decline in global surface temperatures in recent years has been concentrated in the oceans. The figure below shows two of the major sea surface temperature records: HadSST3 from the Hadley Center, and NCDC’s ERSST series.
Ocean temperatures have cooled slightly in recent years, after a large jump upwards in 2000/2001.
Lower Tropospheric Temperatures
Global temperatures can also be estimated based on data from satellites in orbit. These use instruments to measure radiance from Earth to determine temperature, and they tend to have quite good spatial coverage of the planet (excluding some high-latitude regions). While there is still some uncertainty regarding how to best correct for issues like orbital drift and transitions to different satellites, satellite-based records now fairly closely mirror surface-based records, though with slightly lower trends. Scientists are still trying to resolve the discrepancy between satellite and surface trends.
Satellite records show some stagnation of temperatures in recent years, somewhere between the land and ocean surface records.
Deep Ocean Temperatures
In recent years, a global network of automated buoys has offered a much-improved picture of what is going on below the surface in the ocean. These buoys automatically dive deep down into the ocean every day, taking temperature measurements as they slowly rise, and transmitting that data back to a central database via satellite. The figure below, via Argo, shows the location of buoys currently active in the world’s oceans.
While measurements of deep-ocean temperatures existed further back in the past, they were taken only in limited locations until 1999, when Argo buoys were widely deployed. However, scientists for far longer have been able to use more limited data to reconstruct temperatures down to depths of 2,000 meters, as shown in the figure below.
Total ocean heat content has increased by around 170 Zettajoules since 1970, and about 255 Zettajoules since 1955. This increased temperature has caused the oceans (0-2,000 meters) to warm about 0.09 C over this period. As the UK’s Met Office points out, if the same amount of energy had gone into the lower atmosphere it would of caused about 36 C (nearly 65 degrees F) warming! The oceans are by far the largest heat sink for the Earth, absorbing the vast majority of extra heat trapped in the system by increasing concentrations of greenhouse gases.
It’s important to point out that overall deep-ocean heating (0-2,000 meters) shows no sign of a slow down in recent years, though shallower layers (0-300 meters and 0-700 meters) do. That the slowdown in surface warming has been concentrated in the ocean-surface (and shallow-ocean) temperatures has led a number of scientists (including the Met Office) to posit that the pause in ocean surface warming may be driven in part by increased heat uptake in the deep ocean.
Multi-Decadal Cycles
There are a number of inter-decadal and multi-decadal cyclical patterns observable in the climate system, particularly in ocean surface temperatures. These include the Atlantic Multidecadal Oscillation (AMO) and Pacific Decadal Oscillation (PDO), both operating over a period of a few decades, and the El Niño Southern Oscillation (ENSO), which has a period of three to seven years. While ENSO tends to average-out over periods of more than a decade, the AMO and PDO can both potentially impact the climate over longer periods of time.
The AMO is generally calculated by subtracting out the linear trend from 1880 to present in North Atlantic sea surface temperatures. This approach isn’t perfect, and it may overestimate the magnitude of natural variability in recent years, but there is a clear cycle present in the North Atlantic that can contribute to variations in global temperatures. The Met Office, for example, cites a paper arguing that variations in the AMO can change global temperatures by around 0.1 C. While the AMO has not changed much in the past 10 years, the strong increase in North Atlantic temperatures between 1970 and 2000 may have contributed to the rapid rise in global temperatures over that period, and the leveling-out of the AMO may help make the observed pause in warming more likely.
The Pacific Decadal Oscillation (PDO) is calculated rather differently from the AMO. The PDO is calculated by examining the difference in temperatures of the northern Pacific from global ocean temperatures as a whole in order to isolate changes specific to that region. This approach more effectively removes any anthropogenic signals affecting the whole world than the simple linear detrending used in calculating the AMO.
The PDO transitioned to a cold phase around the year 2000. While there still is quite a bit of uncertainty surrounding the effects of the PDO on Earth’s climate, the U.K. Met Office says that “decadal variability in the Pacific Ocean may have played a substantial role in the recent pause in global surface temperature rise.” It argues that Global Climate Models (GCMs) that show decadal-scale pauses in surface temperature warming tend to exhibit sea surface temperature patterns similar to those of the PDO in a cold phase.
Incoming Solar Radiation
The sun has a well-known roughly 11-year cycle in solar output that can have an influence on global temperatures, though most solar scientists consider the impact to be relatively minor. Estimates of the difference in temperatures between the peak (high point) and trough (low point) of the solar cycle range between about 0.05 C to 0.1 C, holding everything else equal. If solar cycles hold steady, this won’t really impact trends over periods of 11 years, as the peaks and troughs will cancel-out.
However, the most recent solar cycle has been notable for the extended trough and low peak (at least so far). The figure below shows the PMOD compilation of solar irradiance measurements from a number of different satellites.
The most recent trough in solar activity likely plays a role in depressing short-term trends, and the overall decline in total solar irradiance (TSI) in recent years relative to past solar cycles may be a small contributing factor in the current slow-down in the rate of warming. As the Met Office explains, “There is no doubt that the declining phase of the 11-year cycle of total solar irradiance has contributed to a reduction in incoming energy over the first decade of the 21st century, but still not enough to explain the pause in global surface temperature rise.” On the other hand, climate models using up-to-date solar forcings don’t show noticeably lower temperatures in the past decade, and that data runs counter to the idea that longer-term changes in the solar cycle are playing a major role in the pause.
Stratospheric Water Vapor
Water vapor in the upper atmosphere plays an important role in Earth’s climate. The figure below, from a paper in Science by Susan Soloman and her colleagues, shows a notable decline in stratospheric (high-atmosphere) water vapor after the year 2000. They argue that this “very likely made substantial contributions to the flattening of the global warming trend since about 2000” and that temperatures between 2000-2009 would have warmed about 25 percent had stratospheric water vapor remained constant. The figure below shows various different estimates of stratospheric water vapor content, with the pre- and post-2001 periods highlighted.
Soloman and her co-authors argue that El Niño has been one of the drivers of changes in stratospheric water vapor, noting that “The drop in stratospheric water vapor observed after 2001 has been correlated to sea surface temperature (SST) increases in the vicinity of the tropical ‘warm pool’ which are related to the El Niño Southern Oscillation (ENSO).”
That said, the models that Soloman and her co-authors use still show significant warming over the past decade even when stratospheric water vapor is declining (they give a rise of 0.10 C instead of 0.14 C, a 0.04 degree C difference). Based on these results, declining stratospheric water vapor would account for only about one-fourth of the slow-down in warming. They also point out that an increase in stratospheric water vapor during the 1990s may have led to about 30 percent more warming during that decade than otherwise would have occurred.
Small Volcanic Eruptions
Stratospheric aerosols — small air-borne particles in the upper atmosphere — play an important role in Earth’s climate. By scattering incoming solar radiation, they can significantly cool the planet. Historically, much of the study of stratospheric aerosols has focused on large volcanic eruptions, which inject large amounts of sulfur dioxide into the stratosphere.
A recent paper by Ryan Neely and coauthors argues that multiple small volcanoes also can have a notable impact on stratospheric aerosols. They point out that “recent studies using ground-based lidar and satellite instruments document an increase in stratospheric aerosol of 4–10 percent per year from 2000 to 2010.” The Neely research argues that “as much as 25 percent of the radiative forcing driving global climate change from 2000 to 2010 may have been counterbalanced by the increases in stratospheric aerosol loading over this period.” The authors examine various potential causes of aerosol increases and identify a number of small volcanoes over the last decade as the most plausible source.
The Met Office downplays these results in its report, arguing that the effect would only be around -0.02 C to -0.03 C during the 2008-2012 period and “will not be detectable above climate variability.” This conclusion relies on an analysis that has been submitted for publication but not yet published, so that analysis may still be subject to revision in the course of the peer review process.
Climate Models and Observations
While it is difficult to distinguish between the recent slow-down in global surface temperatures and the underlying long-term trend, the slow-down stands out much more vividly when compared to projections from the latest set of GCMs. These models predict warming of around 0.2 C per decade from 2000 to present on average.
The figure above compares the three major global surface temperature records to 105 unique runs involving 42 different GCMs used in the upcoming IPCC report. It shows the 5th and 95th percentile of model runs in light grey, and the 25th to 75th percentile in dark grey, with a black line representing the average of all models. While surface temperatures have generally remained fairly close to the multi-model mean in the past, the recent pause threatens to cause surface temperatures to fall below the 5th percentile of models in the next year or two if temperatures do not rise.
The current slow-down also stands out sharply if one looks at the full range of model projections, from 1880 to 2100. However, it’s important to remember that all models are not created equal. Some inevitably will have more realistic parameters, better physical models, higher resolutions, etc. Simply averaging all the models together may not provide an accurate picture of variations in individual model performance.
The figure above shows all 105 model runs, and reveals significant differentiation among the models. Generally speaking, models that are more consistent with recent temperatures tend to have slightly lower climate sensitivity than those that predict higher temperatures over the past few decades. A 2013 paper in Environmental Resource Letters used recent observations to argue that some of the highest sensitivity models may be inconsistent with the observational record.
There have been a number of new papers that use recent atmospheric, ocean, and surface temperature observations to argue that climate sensitivity may be lower than previously estimated (e.g. closer to 2 C than 4 C). These studies tend to be rather sensitive to the time period chosen, and a future warm decade could considerably change the picture. As with many things in science, there is still significant uncertainty surrounding climate sensitivity, and different approaches can obtain fairly different results. However, the longer the current slow-down continues, the more questions will arise about whether GCMs are getting either multi-decadal variability or climate sensitivity wrong.
What is clear is that there is still much we don’t understand about the many different factors impacting Earth’s climate system, especially over periods as short as a decade.