The implications of the recently recognized mid-20th century shift in the Earth system

There is an increasing realization that human activity has led to a profound, long-term shift in the operation of the Earth system that exceeds natural variability (Ripple et al., 2019). These changes appear to occur several decades after the start of the ‘Great Acceleration’ in global economic activity, consumption and human population after the Second World War (Steffen et al., 2015), an acceleration that was paralleled by increases in Earth System indicators such as atmospheric carbon dioxide (CO₂), methane and nitrous oxide, along with declining stratospheric ozone concentration (Steffen et al., 2020). The onset of global, geological-scale impacts arising from human activity since the mid-20th century are widely considered to mark the start of the Anthropocene epoch, although the precise onset is an area of ongoing research (Lewis and Maslin, 2015; Turney et al., 2018; Waters et al., 2016, 2018).

Whilst satellite observations offer highly-resolved measurements of recent, and often dramatic, environmental changes, continuous observations are only available from 1979 (Caesar et al., 2018; Jones et al., 2016), limiting our ability to understand the drivers(s) of feedbacks, tipping elements and teleconnections across the Earth System (Chung et al., 2019; Steffen et al., 2018, 2020; Thomas, 2016). Surface records through the Anthropocene offer to extend some of these time series but are geographically sparse before the 1950s (Mauelshagen, 2014). Here we focus on the relatively short duration of scientific (instrumental) observations of the Earth System and describe the insights offered by alternative records of environmental change that provide considerable potential for reducing projection uncertainties in an accelerating and more extreme, warmer Anthropocene world (Caesar et al., 2018; Steffen et al., 2018, 2020).

There is a particular dearth of scientific observations in the mid to high latitudes (notably the Antarctic), where some of the greatest rates of change are currently seen (Jones et al., 2016; Kennicutt et al., 2019; Medley and Thomas, 2019; Miles et al., 2018; Trusel et al., 2018) and which have the potential to cause irreversible global impacts (IPCC, 2018). Ironically, the large rates of change appear to have taken place in many regions since continuous instrumental stations were first established (for instance, in the Antarctic during the International Geophysical Year 1956/57). This has driven efforts to exploit the relatively untapped instrumental records archived around the world (for instance, using ship logs as highlighted by the Atmospheric Circulation Reconstructions over the Earth programme; ACRE) to improve climate model reanalysis and reconstruct past weather back to the 19th century. This work is being supplemented by ‘natural archives’ of sub-annual to multi-decadal variability derived from climate-sensitive proxies preserved in high-resolution archives such as ice cores, tree-rings and peat sequences, many from regions where limited instrumental observations exist. These natural archives can extend the instrumental record to gain new insights into the Anthropocene (Emile-Geay et al., 2017; Mann et al., 2008; PAGES 2k Consortium, 2013; Palmer et al., 2015.) Whilst not offering the spatial coverage of satellite observations available since the late 20th century, integration of natural archives with long instrumental observations provides an important baseline against which to detect long-term change and the emergence of a new ‘anthropogenic’ state as the Earth system shifts beyond what might be considered ‘natural’.

Although numerous studies have identified a long-term global temperature increase from the late 19th century (the ‘pre-industrial period’), the relative roles of natural variability and anthropogenic warming remain uncertain globally and regionally (Abram et al., 2016; Turney et al., 2019). The potential offered by natural archives to help resolve this issue has recently been highlighted by a number of studies combining high-quality instrumental and natural archive observations that have identified a mid-20th century change across the Earth system that appears to exceed natural variability and pre-date continuous satellite observations (Caesar et al., 2018; Marvel et al., 2019; Roxy et al., 2014; Smith et al., 2017; Steffen et al., 2018; Trusel et al., 2018; Turney et al., 2017) (Figure 1). This recognition of a global step-change in the Earth system has far-reaching implications for our understanding of ice-ocean-atmospheric interactions and give further urgency to calls for substantial reductions in carbon emissions to limit the impacts of future warming.

OPEN IN VIEWER

Figure 1.

Key climate and environmental datasets showing the mid-20th century shift in the Earth system. Panel a: Tasmania’s Cape Grim observations of atmospheric carbon dioxide (CO₂) back to the 1970s (https://www.csiro.au/en/Research/OandA/Areas/Assessing-our-climate/Latest-greenhouse-gas-data) and trapped air measurements reconstructed from the Antarctic Law Dome ice core (Etheridge et al., 1996), with onset of the Great Acceleration (Steffen et al., 2015) and the duration of continuous satellite observations (since 1979). Panel b: Observed changes in Arctic sea-ice extent (Trusel et al., 2018; Walsh et al., 2017). Panel c: Observed and reanalysis-derived measures of Atlantic Meridional Overturning Circulation (AMOC) (Caesar et al., 2018). Panel d: Mean annual tropical Atlantic (20°S to 20°N) sea-surface temperatures (SSTs) derived from HadISST and extracted using KNMI Climate Explorer (Li et al., 2014b; Rayner et al., 2003; Trouet and Van Oldenborgh, 2013). Panel e: Mean annual West Tropical Indian Ocean (WTIO; 10°S to 10°N, 50 V to 70 VE) SSTs derived from HadISST and extracted using KNMI Climate Explorer (Rayner et al., 2003; Roxy et al., 2014; Trouet and Van Oldenborgh, 2013). Panel f: Reconstructed subantarctic Campbell Island austral growing-season temperature (October-March) (Turney et al., 2017). Panel g: High southern latitude changes represented by increased south Atlantic (Stanley, Falkland Islands) austral spring/summer precipitation (Turney et al., 2016), reconstructed onset of retreat of Pine Island Glacier in West Antarctica (Smith et al., 2017) and Cook Glacier retreat in East Antarctic (Miles et al., 2018). Panel g: Reconstructed normalised snowfall on Antarctic Peninsula derived using a network of snow and ice cores (Thomas et al., 2017).

Alignment of natural archive climate reconstructions with long observational records from key parts of the Earth system have detected a mid-20th century shift across both hemispheres, in some regions prior to the onset of the Great Acceleration in 1950 (Figure 1). For instance, temperature-sensitive tree-ring records from the Southern Ocean reveal a sustained increase in climate variability from the 1930s, and appear to be associated with increasing projection of tropical Pacific warming on the mid-latitude westerly wind belt (Turney et al., 2017). Similar trends and associations have been observed in the Indian Ocean since the mid-20th century (McIntosh and Hendon, 2018; Rodrigues et al., 2019; Roxy et al., 2014). These teleconnections have extended to higher latitudes via an atmospheric Rossby wave train (a so-called ‘atmospheric bridge’), driving temperature and precipitation trends over the subantarctic Atlantic and Antarctic Peninsula (Medley and Thomas, 2019; Thomas et al., 2018) that are unprecedented over recent millennia and have led to a greening of ice-free areas (Kennicutt et al., 2019; Turney et al., 2016). Recent work has highlighted that feedbacks associated with southern high-latitude warming can also extend the other way, reaching into the tropics, with global implications (England et al., 2020). Of particular concern is these changes appear to have resulted in increasing sub-surface ocean temperatures near the grounding lines of numerous Antarctic glaciers, including the western Thwaites Glacier (Smith et al., 2017) and eastern Cook Glacier (Miles et al., 2018), the latter a part of the Wilkes Subglacial Basin that holds an estimated 3–4 m of equivalent global sea level rise (Mengel and Levermann, 2014).

Mirroring the above, another step change that pre-dates satellite observations is seen in seasonal atmospheric circulation changes in the Northern Hemisphere (Figure 1). Here, the high northern latitudes experienced similarly large changes that pre-date satellite observations, substantially greater than the global mean (‘polar amplification’) (Bekryaev et al., 2010; Kim et al., 2019). For instance, a sustained change in seasonal atmospheric circulation has been recognized as a major cause of Arctic warming since the 1940s, resulting in dramatic sea ice and Greenland ice mass loss, exceptional in the context of at least the last 350 years (Kinnard et al., 2011; Trusel et al., 2018) (Figure 1). Observations made by shipping across the North Atlantic since the 19th century, in combination with climate model simulations, suggest a parallel 15% weakening of the Atlantic Meridional Overturning Circulation (Caesar et al., 2018), resulting in pronounced surface cooling from the reduced northward transportation of warm waters (Figure 1). Such changes in ocean circulation have potentially global impacts: here, the warming equatorial waters have been linked to increased penetration of Rossby wave trains into the higher latitudes (Kennicutt et al., 2019; Li et al., 2014a), similar to that observed in the Pacific, possibly accounting for the spatial contrasting trends in Antarctic sea ice since 1979 (Kennicutt et al., 2019). Improvements in the parameterization of Earth system models capture this newly observed sensitivity to greenhouse gas forcing (Caesar et al., 2018; Marvel et al., 2019).

Despite this recent work, there remain remarkably large knowledge gaps in some regions of the planet that based on our current understanding of the Earth system, are considered to be globally significant. There is an urgent need to prioritise the development of datasets that capture a more ‘natural’ state of variability during the first half of the 20th century, most notably in Antarctica and the Southern Ocean (Jones et al., 2016). Furthermore, these observed changes in ice-ocean-atmospheric teleconnections highlight the risk of producing multi-centennial reconstructions of climate modes of variability ‘calibrated’ by contemporary observations that represent a shifting climate state (Liu et al., 2017). As increasing modelling efforts are made to forecast future changes under different emission scenarios, it is imperative that a global interdisciplinary approach is undertaken to integrate these with instrumental and natural archive records to truly capture natural variability, teleconnections and trends. A focus on developing datasets in the first half of the 20th century – before the full emergence of an anthropogenically-changed climate state – will likely improve with the calibration of proxy records for palaeoclimate reconstructions. Such coordinated international efforts will greatly help determine the trajectories and identify thresholds across all parts of the Earth system, supporting societal decision making into the future (Ripple et al., 2019).

It therefore seems that until recently, our understanding of the human impact on the Earth system was skewed by continuous satellite observations which only began in 1979. If the relatively sparse instrumental records which extend back through the 20th century and beyond are combined with the growing number of ‘long’ records reconstructed from natural archives, a new perspective emerges: that the Earth system experienced a substantial shift across the mid-20th century (Figure 2), one that appears to have been underway before the Great Acceleration of human activities from the 1950s. These studies underline the importance of baseline observations that extend back in time to capture ‘natural’ conditions before the emergence of an anthropogenic influence (Caesar et al., 2018; Marvel et al., 2019).

OPEN IN VIEWER

Figure 2.

Global distribution of mid-20th changes in the Earth system. The globe shows mean surface temperature differences between the two periods 1901–1950 and 1951–2018, derived from Goddard Institute for Space Studies (GISTEMP v3). Hatching represents areas where the signal is larger than one standard deviation of natural variability, determined using KNMI Climate Change Atlas (https://climexp.knmi.nl). Red-coloured symbols denote long-term increases in one or more components of the Earth system described in the text; blue symbols denote decreases in the component of the Earth system.

A major question that this recent work immediately raises from a policy point of view is whether the scientific and political communities have been blindsided over recent observations? Whilst satellite observations provide exquisite details into how the Earth system has operated since 1979, natural and instrumental archives are providing a richer story with considerable implications:

Whether natural feedback systems will accelerate us into an irreversible ‘Hothouse Earth’ remains to be seen (Ripple et al., 2019; Steffen et al., 2018) but given the above changes occurred with global warming of <1°C, it seems likely that the 2°C target identified in the Paris Climate Agreement is an optimistic threshold for many, if not all, parts of the Earth system (Schellnhuber et al., 2016), raising serious concerns over society’s ability to successfully plan robust adaptation or mitigation strategies. Now the United Nations Climate Change Conference in Glasgow (COP26) has been delayed until November 2021, there is an urgent need to use the available time to build a global consensus for considerably more ambitious targets to cut carbon emissions. With the magnitude of environmental changes observed across the 20th century, humanity cannot delay any longer.