Trajectories toward the 1.5°C Paris target: Modulation by the Interdecadal Pacific Oscillation
Abstract
Global temperature is rapidly approaching the 1.5°C Paris target. In the absence of external cooling influences, such as volcanic eruptions, temperature projections are centered on a breaching of the 1.5°C target, relative to 1850–1900, before 2029. The phase of the Interdecadal Pacific Oscillation (IPO) will regulate the rate at which mean temperature approaches the 1.5°C level. A transition to the positive phase of the IPO would lead to a projected exceedance of the target centered around 2026. If the Pacific Ocean remains in its negative decadal phase, the target will be reached around 5 years later, in 2031. Given the temporary slowdown in global warming between 2000 and 2014, and recent initialized decadal predictions suggestive of a turnaround in the IPO, a sustained period of rapid temperature rise might be underway. In that case, the world will reach the 1.5°C level of warming several years sooner than if the negative IPO phase persists.
Plain Language Summary
Global temperature is rapidly approaching the 1.5°C Paris target. In this study, we find that in the absence of external cooling influences, such as volcanic eruptions, the midpoint of the spread of temperature projections exceeds the 1.5°C target before 2029, based on temperatures relative to 1850–1900. We find that the phase of the Interdecadal Pacific Oscillation (IPO), a slow-moving natural oscillation in the climate system, will regulate the rate at which global temperature approaches the 1.5°C level. A transition to the positive phase of the IPO would lead to a projected exceedance of the target centered around 2026. If the Pacific Ocean remains in its negative phase, however, the projections are centered on reaching the target around 5 years later, in 2031. Given the temporary slowdown in global warming between 2000 and 2014, and recent climate model predictions suggestive of a turnaround in the IPO, a sustained period of rapid temperature rise might be underway. In that case, the world will reach the 1.5°C level of warming several years sooner than if the negative IPO phase persists.
1 Introduction
The Paris Agreement of the United Nations Framework Convention on Climate Change (UNFCCC) in December 2015 agreed to pursue efforts to limit the temperature increase to 1.5°C above pre-industrial levels, recognising that this would significantly reduce the risks and impacts of climate change [UNFCCC Conference of the Parties, 2015]. In response, the Intergovernmental Panel on Climate Change (IPCC) announced a special report on the 1.5°C target, due to be completed by September 2018. Currently, there is a relative lack of scientific knowledge around the implications of 1.5°C warming [Mitchell et al., 2016; Hulme, 2016; Schleussner et al., 2015; Schleussner et al., 2016; King et al., 2017], resulting in uncertainties for policymaking. One aspect of this uncertainty is how soon the world will reach the 1.5°C warming mark and how the timing of the approach and overshoot is influenced by internal modes of climate variability.
Over the last century, we have observed decadal to multidecadal variability in global mean surface temperature (GMST), along with a persistent long-term rise [Hartmann et al., 2013] (Figure 1a). Periods of around 1–2 decades of acceleration and slowdown in the rate of global warming are associated with, and partially caused by, the Interdecadal Pacific Oscillation (IPO) (Figure 1b) [Meehl et al., 2013; England et al., 2014; Dai et al., 2015; Kosaka and Xie, 2016; Fyfe et al., 2016]. The positive (negative) phase of the IPO is associated with anomalously warm (cool) sea surface temperatures (SSTs) in the tropical Pacific Ocean and anomalously cool (warm) SSTs in the subtropical North and South Pacific Oceans. Although closely associated with the El Niño–Southern Oscillation (ENSO), the IPO has a stronger extratropical signature than ENSO [Henley et al., 2015]. The closely related Pacific Decadal Oscillation is thought to be due to the combined effects of ENSO and several extratropical atmospheric and oceanic processes [Newman et al., 2016]. The IPO has memory on decadal-to-multidecadal timescales, as opposed to the 2–7 year oscillation timeframe of ENSO [Power et al., 1999; Henley et al., 2015]. Several studies have attributed at least part of the temporary slowdown in temperature increase since around 1998 to the IPO [Meehl et al., 2013; England et al., 2014; Dai et al., 2015; Kosaka and Xie, 2016; Fyfe et al., 2016]. Other analyses have found potential influences from Atlantic [Dai et al., 2015; McGregor et al., 2014] and Indian [Nieves et al., 2015] Ocean heat uptake, anthropogenic aerosols [Schmidt et al., 2014], a minor influence of the prolonged solar minimum [Meehl, 2015], and observational uncertainty [Karl et al., 2015] and instrumental data biases [Hausfather et al., 2017] on the slowdown. However, agreement is emerging that the dominant influence on at least the recent slowdown was the Pacific decadal variability related to the IPO [Dai et al., 2015; Fyfe et al., 2016; Meehl et al., 2016a].
In this study, we examine model projections of global temperature toward 1.5°C. We then isolate and explore the influence of the IPO on future global temperature trajectories. Specifically, we investigate how the next phase of the IPO will affect the rate and timing of global warming reaching 1.5°C.
2 Data and Methodology
2.1 Data
We use observed global land and ocean mean annual (calendar year) surface temperature data (1880–2016) from NOAA [National Oceanic and Atmospheric Administration, 2017], the HadCRUT4 data (1850–2016) from the UK Met Office [Morice et al., 2012], and GISTEMP (1880–2016) from NASA [Hansen et al., 2010]. The tripole index (TPI) [Henley et al., 2015] is used to represent the IPO and is calculated using the ERSSTv4 data set [Huang et al., 2015]. GMST and TPI are calculated from monthly gridded surface air temperature data in 70 century-long simulations from 32 climate models under RCP8.5 (high emissions scenario) in the fifth phase of the Climate Model Intercomparison Project Phase 5 (CMIP5) [Taylor et al., 2012] archive (supporting information Table S1) stored at the Australian National Computing Infrastructure (NCI) repository. Although underestimating the duration of decadal phases, the models adequately capture broad characteristics of the IPO [Henley et al., 2017] and our results are not overstated due to this likely bias. The models adequately capture global temperature variability [King et al., 2016].
2.2 Baseline
We use the 1850–1900 period as our quasi-preindustrial baseline, as it is the earliest possible 51-year baseline using instrumental data and is deemed sufficiently long to dampen the influence of decadal variability. This baseline was used by the IPCC to compare global mean temperature under RCP scenarios [Collins et al., 2013, Table 12.3, IPCC 2013]. We first use the period 1961–1990 to align our three observed data sets. Since the HadCRUT4 data set extends back to 1850, we use this dataset to compute the difference between 1850–1900 and 1961–1990. We then apply this baseline shift to all three data sets to compute the anomaly from 1850 to 1900. We note that there is no ideal preindustrial baseline [Hawkins et al., 2017] and that our results should be interpreted in the context of the selected baseline.
2.3 Analysis
We compute global temperature sequences from the CMIP5 ensemble and plot their future trajectories. For each sequence, we first express the global temperature as an anomaly from the commencing year. We estimate that the 2015–2016 El Niño-induced short-term anomalous warmth of approximately 0.1°C (see supporting information). We then add each sequence to the observed temperature series, commencing our sequences 0.1°C cooler than the 2016 value. We then compute the tripole index (TPI) [Henley et al., 2015] of the IPO and the annual (January–December) GMST in each model simulation for the period 2006–2100 (Table S1). We apply a correction to the GMST sequences to account for the higher warming rates in the latter part of the century (see supporting information). We then isolate the annual sequence of global mean surface temperatures within IPO positive and negative phases, commencing the sequences 5 years prior to the start of each IPO phase and ending them 5 years after the end of the IPO phase (Figure S1 in the supporting information shows this definition schematically). IPO phases are defined as the TPI years above (for IPO positive) or below (for IPO negative) a threshold of ±1 standard deviations away from the long-term TPI mean, after detrending using a power fit (y = axb + c). We conduct sensitivity tests on these analysis settings in the supporting information. Our method produces ensembles of future temperature trajectories in IPO positive and negative phases.
3 IPO Modulation of Global Temperature Trajectories
Here we examine our future global temperature simulations from CMIP5. The mean of our sequence of global temperatures reaches the 1.5°C warming level in around 2029, with an interquartile range (IQR) of 2026–2032 (Figure 2a). For the NASA and Hadley Centre observed data sets we see small differences in the timing of up to 3 years (supporting information).
We also compare composites of global mean surface temperature sequences in each of the two phases of the IPO. We find a statistically significant difference (p < 0.05) between the ensemble mean of the GMST sequences in IPO positive and negative phases (Figure 2b). In IPO positive phases, global temperatures rise substantially faster than for IPO negative. For 14 consecutive years, from 2019, we find that in the IPO positive phase, global temperatures are statistically significantly higher than in IPO negative phases. This means that the rate that global temperatures approach the 1.5°C level is likely to be significantly quicker, or slower, depending on the IPO.
The projected timing of global warming reaching 1.5°C above the preindustrial level can be expressed in a number of ways. These include but are not limited to (a) the year in which an ensemble mean of temperature simulations reaches 1.5°C, (b) the year in which the global mean first reaches 1.5°C, (c) the year in which a longer-term (e.g., 5-year) mean first reaches 1.5°C, or (d) the year in which the global mean reaches and does not return below 1.5°C, referred to as an expulsion from history [Power, 2014].
Considering (a), that is, the ensemble mean of our GMST sequences in each IPO phase (Figure 2b), if the world experiences a transition to an IPO positive phase, we expect global temperatures to reach the 1.5°C level by around 2027 (IQR: 2024–2029). If, however, the Pacific Ocean remains in an IPO negative phase, there would likely be a delay in reaching 1.5°C for 4–5 years, until around 2031 (IQR: 2026–2033). When the IPO positive GMST ensemble mean reaches the 1.5°C level, the mean temperature for the IPO negative scenario is around 1.3°C. The mean trajectories have a peak difference in global mean temperature of around 0.2°C (Figure 2b).
If we consider (b), the year in which the global mean first reaches 1.5°C, the distributions are shifted a little earlier, with a mean of 2025 for IPO positive and 2029 for IPO negative (Figure 3a). For the 5-year mean projections, method (c), the mean is around 2025–2026 for IPO positive and 2030–2031 for IPO negative. For method (d), the expulsion from history, we found too few long IPO sequences in our temperature sequences meeting this criterion to form a statistically powerful ensemble. As the expulsion above 1.5°C is further into the future, it is not yet possible to assess the IPO influence on the timing of that expulsion.
Figure 3c shows the distribution of trend rates of warming in each IPO phase in our ensemble. The global warming slowdown period (2000–2014), with a trend of 0.23°C/decade, is located near the mode of the distribution of IPO negative phase temperature trends. The previous IPO positive phase (1976–1998) is associated with a higher rate of warming of 0.34°C/decade. This is consistent with the IPO positive trend distribution in our model ensemble. We note that the ensemble uses future temperature sequences, so we would expect the modelled distributions to have slightly higher warming trends than past IPO phases.
The models vary in their characterization of decadal climate variability and tend to underestimate the duration and magnitude of the IPO variability [Henley et al., 2017; Power et al., 2016]. However, this means that our results are most likely to be conservative estimates of the duration of the IPO's impact on global temperatures. Meehl et al. [2016a] quantified the contribution of the IPO to multidecadal GMST trends. Importantly, here we find that the observed rate of GMST rise in both IPO negative (slowdown) and positive (acceleration) periods in the twentieth century is within, and consistent with, the distribution of the CMIP5 modelled trends (Figure 3c).
We note also that our multiyear multimodel mean trajectories, by design, dampen interannual and multiannual variability, and the actual trajectory of any one sequence in our ensemble has higher year-to-year variability. Our projections do not include the potential influence of unpredictable external cooling influences such as volcanic eruptions. However, the difference in global temperature trajectories between IPO phases is consistent across the ensembles of projections and robust to methodological choices (see supporting information).
4 Discussion and Conclusions
In the first decade of the 21st century the global climate was under the influence of a negative IPO phase. This may have provided a temporary buffer for the radiative forcing effect of continually rising atmospheric greenhouse gas concentrations on global temperatures. It is therefore possible that a negative phase of the IPO since the turn of the century has cushioned the impacts of global warming on extreme events, such as heat waves. A global temperature record was set in 2015, and a strengthening El Niño followed in 2016, with associated widespread coral bleaching [Hughes et al., 2017] and high temperatures leading to another global temperature record. Initialized decadal predictions are suggestive of a turnaround in the IPO to its positive phase [Meehl et al., 2016b], triggered by upper ocean heat content on decadal timescales in the off-equatorial western tropical Pacific.
A turnaround of the IPO to its positive phase could initiate a period of accelerated warming over the next one to two decades. This would likely lead to the Paris target of 1.5°C being surpassed within the next decade. Our analysis provides an illustration that decadal climate variability is likely to be a significant determinant on global temperature trajectories over the next 10 years. As a consequence, decadal variability will also influence when the global warming target of 1.5°C over a preindustrial climate is breached. Disregarding the influence of the IPO, it is likely that global mean temperatures will pass the 1.5°C warming mark within the next 10–15 years; with our ensemble mean projecting this will occur prior to 2030. Equilibrating the Earth's climate at 1.5°C above the preindustrial level will involve overshooting the target and then reducing atmospheric greenhouse gas concentrations and global temperatures on a net negative carbon emissions pathway.
Acknowledgments
B.H. receives funding from the Australian Research Council (ARC) Linkage Project LP150100062. A.K. receives funding through the Australian Research Council Centre of Excellence for Climate System Science (CE110001028). B.H. is an Associate Investigator of the ARC Centre of Excellence for Climate System Science. We acknowledge the support of the NCI facility in Australia. We acknowledge the World Climate Research Programme's Working Group on Coupled Modelling, which is responsible for CMIP, and we thank the climate modelling groups for producing and making available their model output. For CMIP the U.S. Department of Energy's Program for Climate Model Diagnosis and Intercomparison provides coordinating support and led development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. The data used in this study are publicly available from the CMIP5 model repositories, NOAA, NASA, and the UK Met Office Hadley Centre.
