The residual dynamics left after adjusting global surface temperature anomalies (1950-2014) for short-term variability from El Niño Southern Oscillation (ENSO) and volcanic eruptions have a staircase pattern. Linear trends for three quasi-stable periods 1950-1987, 1988-1997 and 1998-2014 are near zero with nearly all warming occurring during two step-like shifts in the years 1987/1988 and 1997/1998.
Figure 2. Staircase consisting of regimes and shifts in various climate parameters: (a) adjusted for ENSO and volcanoes: yearly global surface temperature anomalies (HadCRUT4), 1950-2014 years; (b) adjusted for ENSO monthly SST anomalies (ICOADS), 1950-2016 years and aerosol optical depth (reflecting the influence of major volcanic eruptions); (c) The same as „b‟ for satellite based SST measurements (NCEP OI v2), 1983-2016 years; (d) The same as ‘b‟ for databases of satellite measurements of lower troposphere temperature (UAH MSU v6.0 and RSS MSU v3.3), 1980-2016 years; (e) Comparison of global meridional wind at 300mb height of NCEP/NCAR reanalysis (green line) and adjusted surface temperatures (red line), 1950-2014 years; (f) Comparison of satellite measurements of outgoing longwave radiation (UMD/NCEI) at 30N-70N (green line), at 30S-70S (blue line) and adjusted surface temperatures (red line), 1950-2014 years.
The residual dynamics left after adjusting global surface temperature anomalies (1950-2014) for short-term variability from El Niño Southern Oscillation (ENSO) and volcanic eruptions have a staircase pattern. Linear trends for three quasi-stable periods 1950-1987, 1988-1997 and 1998-2014 are near zero with nearly all warming occurring during two step-like shifts in the years 1987/1988 and 1997/1998. We analysed several global datasets: HadCRUT v4.5 – land and sea surface temperature (SST) anomalies; ICOADS v2.5 – SST anomalies measured from ships; NCEP OI v2 – SST measured by satellite instruments; UAH MSU v5.6 and RSS MSU v3.3 – two satellite datasets measuring temperature of the lower troposphere (TLT). The ENSO signal was removed by EOF analysis, and gave comparable results for all datasets. A similar staircase behavior was found in global NCEP/NCAR reanalyses of 300mb meridional wind and outgoing longwave radiation (OLR) in northern and southern midlatitudes. These many different sources confirm the reality of the regime-shift staircase structure of recent warming, which is masked by short-term ENSO variability and the effects of volcanic eruptions.
Regime-shift like structures in decadal climate change have been detected in many studies of temperature and related climatic variables (Yasunaka and Hanawa 2002; Chavez et al. 2003; Lo and Hsu 2010; Reid and Beaugrand, 2012; Jones, 2012; Reid et al., 2016; Jones and Ricketts, 2017). Much attention has been given to a pause in warming (hiatus) during 1998-2014 (Tollefson 2014). We recently suggested and applied a simple method to adjust HadCRUT4 surface temperature anomalies for ENSO effects (Belolipetsky et al. 2015). After this adjustment for major ENSO and volcanic effects, the hiatus is even more pronounced. Moreover, we observed similar quasi-stable periods during 1950-1987 and 1988-1997 and almost all the warming occurred during the ~1987 and ~1997 shifts. It should be mentioned that a similar shift has likely occurred in 2015-2016, but this issue is outside the scope of this short paper. Here we want to demonstrate the reality of the staircase pattern using different measurements and climate parameters. It is well known that most short term global temperature variability is due to the well-defined ENSO natural oscillation (see: Wang and Fiedler, 2006). During strong El Niño events global average temperature rises by a few tenths Kelvin and reverts back subsequently. If spatial patterns are considered however, it is seen that global average temperature is most influenced by changes in the eastern tropical Pacific (Wang and Fiedler, 2006). Our 2015 study (Belolipetsky et al. 2015) takes into account variability in the spatial pattern of ENSO. Here we have applied the same method that uses a simple linear regression model to a broader set of parameters to reveal a clear staircase pattern in warming global temperature in the middle of the 20th century. [….]
Discussion and conclusion
Yasunaka and Hanawa (2002) define a regime shift as an “abrupt transition from one quasi-steady climatic state to another, and this transition period is much shorter than the length of the individual epochs of each climatic state”. Subtracting the ENSO signals from global temperature time series is not the only technique that has been used to detect 1987/1988 and 1997/1998 shifts. Shifts with the same timing have been detected by other methods and for many other parameters (Reid et al. 2016; Lo and Hsu 2010; Li et al 2010; Hare and Mantua 2000). Reid et al. (2016) for example analysed 72 different biological and physical time series from different regions over the period 1946 to 2012. They detected 165 statistically significant step changes, 40% of which were close to 1987 and 25% close to 1997. The identification of the shifts in so many different parameters provides further support for the reality of the 1987 and 1997 global temperature shifts revealed here. Thus a very simple model almost totally reproduces global surface temperature dynamics from 1950 to 2014 – a linear sum of ENSO and the staircase signal with two shifts.