Climate Models Simulate Global Warming Pause

Some climate models simulate the global warming pause if they take account of cooling Pacific waters, according to new research.

The apparent historic inability of climate models to simulate the pause in global warming – which on some measures has been going on since 1995 – has been problematic for climate scientists and has been repeatedly highlighted by climate change sceptics.

But now a team of researchers from the US National Center for Atmospheric Research (NCAR) and the Centre for Australian Weather and Climate Research (CAWCR) has shown that, in certain conditions, these climate models can simulate the pause. They present their results in a paper published in Nature Climate Change.

The solution to the problem lies in the way that climate models are used, according to the researchers. Computer climate forecasts using the Coupled Model Intercomparison Project Phase 5 (CMIP5) consist of ensembles of multiple model runs that are averaged together to iron out the effects of naturally occurring internal climate variations and leave the warming impact of greenhouse gases (GHGs).

This means that if the pause is due to natural climate variability then the ensemble approach will remove it in the averaging process, the researchers point out. The “average of all those simulations for the early 21st century would, and indeed does, lie above the actual plateau of warming that occurred in the observations,” they write.

However, when the individual ensemble members are examined it is apparent that some do simulate the pause, say the researchers led by NCAR’s Gerald Meehl. The analysis shows that 12 enemble members from a set of 262 recreated a pause between 2000 and 2012, ten between 2012 and 2013, nine continue through 2000 to 2014, six from 2000 to 2015, and six from 2000 to 2016, one of which from 2000 to 2017 continues to 2018 (a hiatus of 19 years), the scientists report.

The common factor linking the ensemble members that forecast the pause is that they feature a cooling of the Pacific Ocean sea surface temperatures; a negative phase of the Interdecadal Pacific Oscillation (IPO) – a long term 15 to 30 year cycle in sea surface temperatures affecting both the north and south Pacific. This contrasts with the overall average of the larger total set of all ensemble members that shows mostly warming in the tropical Pacific. In simple terms, if climate models are set up with an assumption that the IPO is in a negative phase then they can replicate the pause.

The authors write: “This is a compelling application of the result derived from other analyses, in that tropical Pacific surface temperatures in the negative phase of the naturally-occurring IPO can temporarily counteract the warming from in-creasing GHGs to produce a hiatus of warming in globally averaged surface air temperatures that can last for a decade or more, even as the climate system is still trapping excess heat of about 0.51.0Wm-2”.

On a technical note the authors say they analysed all available uninitialized CMIP5 climate model simulations, with all possible ensemble members for all four greenhouse warming scenarios adopted by the Intergovernmental Panel on Climate Change – the so called representative concentration pathways (RCPs). This amounted to 262 possible realizations from 45 models, with up to 10 ensemble members for the period 2000 -2020. These model simulations all start from some pre-industrial state in the nineteenth century, and use observations for natural (volcanoes and solar) and anthropogenic (GHGs, ozone, aerosols, land use) forcings through 2005 with the four RCP scenario forcings after 2005. They point out that for the period of the early 2000s, there is little difference among the RCP scenarios for this short-term time frame, so all were used.

Today’s tools would have foreseen warming slowdown

September 8, 2014 | If today’s tools for multiyear climate forecasting had been available in the 1990s, they would have revealed that a slowdown in global warming was likely on the way, according to new research.

The analysis, led by NCAR’s Gerald Meehl, appears in the journal Nature Climate Change. It highlights the progress being made in decadal climate prediction, in which global models use the observed state of the world’s oceans and their influence on the atmosphere to predict how global climate will evolve over the next few years.

Such decadal forecasts, while still subject to large uncertainties, have emerged as a new area of climate science. This has been facilitated by the rapid growth in computing power available to climate scientists, along with the increased sophistication of global models and the availability of higher-quality observations of the climate system, particularly the ocean.

Although global temperatures remain close to record highs, they have shown little warming trend over the last 15 years, a phenomenon sometimes referred to as the “early-2000s hiatus”. Almost all of the heat trapped by additional greenhouse gases during this period has been shown to be going into the deeper layers of the world’s oceans.

The hiatus was not predicted by the average conditions simulated by earlier climate models because they were not configured to predict decade-by-decade variations.

However, to challenge the assumption that no climate model could have foreseen the hiatus, Meehl posed this question: “If we could be transported back to the 1990s with this new decadal prediction capability, a set of current models, and a modern-day supercomputer, could we simulate the hiatus?”

Looking at yesterday’s future with today’s tools

To answer this question, Meehl and colleagues applied contemporary models in a “hindcast” experiment using the new methods for decadal climate prediction. The models were started, or “initialized,” with particular past observed conditions in the climate system. The models then simulated the climate over previous time periods where the outcome is known.

The researchers drew on 16 models from research centers around the world that were assessed in the most recent report by the Intergovernmental Panel on Climate Change (IPCC). For each year from 1960 through 2005, these models simulated the state of the climate system over the subsequent 3-to-7-year period, including whether the global temperature would be warmer or cooler than it was in the preceding 15-year period.

Starting in the late 1990s, the 3-to-7-year forecasts (averaged across each year’s set of models) consistently simulated the leveling of global temperature that was observed after the year 2000. (See image at bottom.) The models also produced the observed pattern of stronger trade winds and cooler-than-normal sea surface temperatures over the tropical Pacific. A previous study by Meehl and colleagues related the observed hiatus of globally averaged surface air temperature to this pattern, which is associated with enhanced heat storage in the subsurface Pacific and other parts of the deeper global oceans.

Letting natural variability play out

Although scientists are continuing to analyze all the factors that might be driving the hiatus, the new study suggests that natural decade-to-decade climate variability is largely responsible.

As part of the same study, Meehl and colleagues analyzed a total of 262 model simulations, each starting in the 1800s and continuing to 2100, that were also assessed in the recent IPCC report. Unlike the short-term predictions that were regularly initialized with observations, these long-term “free-running” simulations did not begin with any particular observed climate conditions.

Such free-running simulations are typically averaged together to remove the influence of internal variability that occurs randomly in the models and in the observations. What remains is the climate system’s response to changing conditions such as increasing carbon dioxide.

However, the naturally occurring variability in 10 of those simulations happened, by chance, to line up with the internal variability that actually occurred in the observations. These 10 simulations each showed a hiatus much like what was observed from 2000 to 2013, even down to the details of the unusual state of the Pacific Ocean.

Meehl pointed out that there is no short-term predictive value in these simulations, since one could not have anticipated beforehand which of the simulations’ internal variability would match the observations.

“If we don’t incorporate current conditions, the models can’t tell us how natural variability will evolve over the next few years. However, when we do take into account the observed state of the ocean and atmosphere at the start of a model run, we can get a better idea of what to expect. This is why the new decadal climate predictions show promise,” said Meehl.

Decadal climate prediction could thus be applied to estimate when the hiatus in atmospheric warming may end. For example, the UK Met Office now issues a global forecast at the start of each year that extends out for a decade.

“There are indications from some of the most recent model simulations that the hiatus could end in the next few years,” Meehl added, “though we need to better quantify the reliability of the forecasts produced with this new technique.”

Meehl, Gerald A., Haiyan Teng, and Julie M. Arblaster, “Climate model simulations of the observed early-2000s hiatus of global warming,” Nature Climate Change (2014), doi:10.1038/nclimate2357

Writer/contact
Bob Henson, NCAR/UCAR Communications

Collaborating institutions
Center for Australian Weather and Climate Research
National Center for Atmospheric Research

Funders
Australian Bureau of Meteorology
National Science Foundation
U.S. Department of Energy Regional and Global Climate Modeling Program

End of press release

The slowdown in the rate of global warming in the early 2000s is not evident in the multi-model ensemble average of traditional climate change projection simulations. However, a number of individual ensemble members from that set of models successfully simulate the early-2000s hiatus when naturally-occurring climate variability involving the Interdecadal Pacific Oscillation (IPO) coincided, by chance, with the observed negative phase of the IPO that contributed to

the early-2000s hiatus. If the recent methodology of initialized decadal climate prediction could have been applied in the mid-1990s using the Coupled Model Intercomparison Project Phase 5 multi-models, both the negative phase of the IPO in the early 2000s as well as the hiatus could have been simulated, with the multi-model average performing better than most of the individual models. The loss of predictive skill for six initial years before the mid-1990s points to the need for consistent hindcast skill to establish reliability of an operational decadal climate prediction system.

Leave a Reply