Table 1.
Investigated systems and their abbreviations.
Table 2.
EWSs for systems with absolute noise.
Table 3.
EWSs for systems with relative noise.
Fig 1.
System QG with absolute noise.
Both standard deviation (center row, left) and autocorrelation (center row, right) show a significant positive trend.
Fig 2.
Both standard deviation and autocorrelation show a significant positive trend.
Fig 3.
System LND with relative noise.
Standard deviation shows a sharp increase, while autocorrelation increases more slowly.
Fig 4.
System TD with relative noise.
Both standard deviation and autocorrelation show a significant positive trend.
Fig 5.
System LNG with absolute noise.
For this system no EWS were detected. Standard deviation as well as autocorrelation decrease over time, while skewness and kurtosis show random fluctuations, but no clear trend.
Fig 6.
System LOD with relative noise after global detrending.
Detrending successfully removes the false positive: The standard deviation now decreases, while the autocorrelation only shows random fluctuations.
Fig 7.
Global surface temperature change.
Both standard deviation and autocorrelation increase, which can be interpreted as an EWS.