by Frank Bosse and Nic Lewis

Not really.

A recent paper (M. B. Freund et al 2023, MBF23 thereafter) in “Nature communication earth and environment” investigates the variability of the summer drought events since 1600. It uses the method of “stable isotope analyses C13/O18” to extend the “Standardised Precipitation-Evapotranspiration Index (SPEI) from 1950 to now back to 1600.

The paper describes and uses a multi proxy network over large parts of Europe (see Fig. 1 of MBF23) to reconstruct the history of summer droughts for a longer historic period. It finds interesting results about the dependency of those events on volcanos and solar forcing. It’s a worthwhile read and we were interested in whether the headline title is justified and likewise this claim in the Abstract:

“We show that the recent European summer drought (2015–2018) is highly unusual in a multi-century context…”

Thanks to the authors the used SPEI reconstruction annual data are available, so we were able to perform calculations to check these assertions.

An apparent first “confirmation” of the headline title of the paper appears in Figure 3a in MBH23:

Fig.1: A reproduction of Fig. 3a of MBH23. Annual European mean SPEI-data in blue/red, the low pass filter output is shown in black.

The black line in this figure shows the effect of applying a 13-year low-pass smooth, so it relates to the recent past. Indeed, after 2010 the used 13-year Chebyshev filter shows a “dramatic” downward dip to a far lower precipitation index than at any other time during the 1600-2018 reconstruction period. However, when eyeballing one finds also dry periods, before 1950, the onset of the classical SPEI dataset marked with “SPEI”, or before 1880 marked with dark grey in Fig.1, and the low pass filter didn’t react in the way it did after 2010.

The reason for this behaviour is quite simple: All smoothing filters struggle with the beginning and the end of a filtered dataset. They estimate the output because there are no precursors/ successors in the raw data. To test the impact of this properties we used the same data with a similar filter (Loess) and made a comparison with Fig. 1 but stopped the filtering in 1949:

Fig. 2: Fig.1, but with the smoothed SPEI-Index ending in 1949.

If the paper was written in 1950 it would find “unusual recent hydroclimate”, in 2023 it finds the same for the recent conditions due to a filter issue. The beginning after 1600 is also very unusually wet in the filter output for the same reason.

The dip in the end in Fig.3a of MBH23 is not real, it’s an artefact of the used filter.

A simple running mean filter which while it has no output in the early years, is unartefacted, gives a fairer smoothing of fluctuations over 1600-2018:

Fig. 3: Summer SPEI-Data (black) filtered with a trailing running mean (red). The historical minimum of this filter is shown as a broken red line. Clear to see minima in the 1870s and 1680s in addition to at the end of the 1600-2018 period.

Fig. 3 gives the contrary result of the headline title of MBF23: to 2018 (the last datapoint in the set in MBF23) it indicates that the recent European summer hydroclimate was NOT unusual, the SPEI index was in the ballpark of natural variability.

To show that also the claim in the Abstract (“2015-2018 highly unusual”) is not true we had a deeper look in the data and calculated those 4 years averages over the whole timespan.

It turned out that during many periods the average of 4 years in the SPEI data was more negative than during 2015-2018, for which this average is -0.273:

Since 1900 there have been four such periods, all in the years leading up to 1950: 1947-1950; 1946-1949; 1945-1948; 1944-1947. The period before 1950 (not strongly influenced from anthropogenic forcing) was indeed marked by very dry summers, not mentioned in one word in MBF23.

Before 1900 there are also some periods:

1892-1895; 1760-1763; 1759-1762; 1738-1741; 1688-1691.

The “European summer drought 2015-2018” was NOT highly unusual in a multi-century context”, as falsely claimed in the abstract.

To further bolster this point we looked also if longer period averages were “highly unusual”.

It turned out that a trailing average of 5 years produces 10 periods during 1600-1950, a time span predominantly affected by natural variability, with more negative SPEI-values than the most recent period to 2018; a 10-years average gives 9 such pre-1951 periods.  And a 3-year trailing average produces no less than 57 pre-1951 periods with more negative SPEI values than the most recent period.

Furthermore we had a look at the variability of the annual data after 1950 (the time span of the “native SPEI”) and before this year, the time span of the reconstruction of the “European hydroclimate based on a network of tree-ring stable isotopes of oxygen and carbon ratios” in MBF23. We calculated running 21 years standard deviations (sigma) of the annual data (Fig.4):

Fig. 4: The variability of the annual SPEI data. The averages before 1950 and after this year are marked with a dotted line. Note the jump.

The lower time variability of the reconstruction cast some doubts, as to whether the reconstruction of the SPEI 1600…1950 is useful to compare 1:1 the newer native SPEI data with the historical reconstruction data pre- 1950. It looks as if the reconstruction, even if otherwise valid, significantly understates natural variability. This is a common problem with proxy-based reconstructions. It results in the extent of fluctuations during the post-1950 instrumental SPEI era being an exaggerated relative to natural variability, so that normal fluctuations can appear to be unusual.

Conclusion

MBF23 is a very valuable paper when it comes to the description of the variability of European summer droughts since 1600. However, neither its title “European tree-ring isotopes indicate unusual recent hydroclimate” nor the claim in its Abstract that “recent European summer drought (2015–2018) is highly unusual in a multi-century context” are justified by the data used in the paper.

The lower resolution in time and spatially of the reconstruction before 1950 in relation to the determined SPEI after 1950 casts some doubts if the comparison of some years after 1950 to the historical reconstructed values is appropriate.

MBF23 should be corrected and retitled because some key conclusions, including the headline claim in its title, are not supported by proper statistical analysis of the SPEI values that their reconstruction method produces. The recent European drought to 2018 remained within the range of natural variability.