83the strength of AMOC are considered. The ZT13 estimate of AAWR without
consideration of AMOC is in close agreement with the value published by FR2011,
and disagrees with our value for the reasons described above.
The importance of the ZT13 study is that if their value of AAWR found upon con-
sideration of AMOC (0.07 °C/decade) is correct, one would conclude that the CMIP5
GCMs warm a factor of three more quickly than the actual climate system has
responded to human influence. We are also able to reproduce the results of ZT13, but
we argue their estimate of AAWR is biased low because they used a single linear func-
tion to describe ΔTHUMAN over the entire 1860–2010 time period. As illustrated on the
second rung of the Figs. 2.4 and 2.5 ladder plots, ΔTHUMAN varied in a non-linear man-
ner from 1860 to present. The time variation of ΔTHUMAN bears a striking resemblance
to the rise in population over this period of time. For the determination of AAWR, not
only should a model for ΔTHUMAN be used, but this model must correspond to the
actual shape of the time variation of radiative forcing of climate caused by humans.
2.4 Global Warming Hiatus
The evolution of ΔT over the time period 1998–2012 has received enormous atten-
tion in the popular press, blogs, and scientific literature because some estimates of
ΔT over this period of time indicate little change (Trenberth and Fasullo 2013 ).
Various suggestions had been put forth to explain this apparent leveling off of ΔT,
including climate influence of minor volcanoes (Schmidt et al. 2014 ; Santer et al.
2014 ; Solomon et al. 2011 ), changes in ocean heat uptake (Balmaseda et al. 2013 ;
Meehl et al. 2011 ), and strengthening of trade winds in the Pacific (England et al.
2014 ). The major ENSO event of 1998, which led to a brief, rapid rise in ΔT due to
suppression of the upwelling of cold water in the eastern Pacific, must be factored
into any analysis of the hiatus.^30
Karl et al. ( 2015 ) have questioned the existence of a hiatus. They showed an
update to the NCEI record of GMST, used to define ΔT, which exhibits a steady rise
from 1998 to 2012, despite the ENSO event in 1998. The main improvement was
extension to present time of a method to account for biases in SST, introduced by
varying techniques to record water temperature from ship-borne instruments.
Figure 2.14 compares measured ΔT over 1998–2012 to simulations of ΔT from
the EM-GC. The EM-GC simulations were conducted for the entire 1860–2015
time period: the figure zooms in on the time period of interest. Figure 2.14a–c shows
results using the latest version of ΔT from CRU, GISS, and NCEI (footnotes 1 to 3
provide URLs, data versions, etc.). Each panel also includes the slopes of a linear fit
to the data (black) and to modeled ΔT (red), over 1998–2012.
For the first time in our extensive analysis, the choice of a data center for ΔT
actually matters. The observed time series of ΔT from CRU in Fig. 2.14 exhibits a
(^30) The effect of ENSO on ΔT in 1998 is readily apparent on the fourth rung of Figs. 2.4 and 2.5
ladder plots.
2.4 Global Warming Hiatus