HadCRUT4 combined land and sea surface temperature anomaly data is aligned at the beginning of each year the major stratospheric eruptions that are estimated to have made a significant effect on optical properties of the atmosphere.
1883 1902 1912 1963 1982 1991
The resulting overlay of these major eruptions provides an average response where random variations will be reduced. What it shows is a surprising pattern both before and after the events. A similar pattern is shown in HadSST3 :
Some means to eliminate this repetitive variation is needed in order to correctly assess the net impact of the eruptions on surface temperature.
To assess any net cooling effect, the cumulative integral (CDF) of mean temperatures is calculated for both tropical and extra-tropical regions.
Since the data are from separate periods with both positive and negative anomalies relative to the arbitrary reference period for the dataset, the averaged temperature anomaly does not have any climatological significance. It is simply a convenient reference point.
Since a constant average anomaly will result in linear slope in the CDF this average has been estimated for the 4 years preceding the “average” eruption and removed to give the horizontal reference line.
This does not introduce a rate of change of temperature, it simply zeroes out arbitrary average “anomaly” reference point.
The horizontal lines represent the case where temperatures remained constant after the eruption.
To preserve effects on the seasonal variation dates are aligned to start of year not the exact date of the eruption.
The extra tropical region is easiest to read.
The fairly linear downward slope represents a constant lower temperature. Before 2 years has elapsed there is a curvature, levelling out the CDF. This curvature represents re-warming. The flat period from 2.5 to 5 years after eruption signifies temperatures are constant and equal to the pre-eruption average.
The further cooling and warming that follows is after the residency of volcanic aerosols and is probably variations in climate unrelated to volcanism.
The delay after zero reflects the fact that the data were aligned 1st of January of eruption year. So the onset of cooling is the average time of year of eruption plus any delay in ejections spreading enough to start having measurable effect.
Similar interpretation of the tropical SH plot shows the initial climate response is a clear warming of the tropical SST.
There is a return to the pre-eruption mean with a similar timing to that in the extra-tropical region.
Again, the later strong warming is likely to be independent of volcanic effects.
The clear difference between the two regions is noteworthy. The tropical regions not only recovers to the pre-eruption temperature but has also maintained (perhaps slightly gained) the level of degree.days. The concept of degree.days is used in farming since it relates to plant growth days.
The regional difference could be said to show tropical plants would come out with no deficit in growth where as extra-tropical plants would loose some growth days and experience a net negative effect.
In terms of understanding system behaviour, this constancy of the time integral of temperature shows the tropics to be totally self regulating , suffering no net impact from the volcanic disruption of the incoming radiative energy.
This shows a level of recovery and self-regulation that is much stronger than simply returning to previous average temperature as would happen with a passive, linear negative feedback.
For both regions, it is clear that idea of a net negative temperature response to volcanic forcing is unfounded. This has deep reaching implications for our understanding of climate and the current climate modelling paradigm.
A similar processing of the northern hemisphere data is shown here:
Similar patterns are seen which larger variations in extra-tropical regions. This is probably attributable to the greater volatility of land temperatures, with land area being significantly larger proportion of the total area in the northern hemisphere.