In the past ten years, Europe has encountered several notably arid summers, such as in 2015, 2018, 2019, and 2022, which have already been extensively examined in prior research 2,8,22, 30–32. To gain a broader understanding of the drought events over the last decade and put them in a long-term context, we have calculated the magnitude and the extent of drought-affected regions at the European level (Fig. 2 and Fig. 3), as well as over the Mediterranean (MED, Figure S1, and Figure S2), Central Europe (CEU, Figure S4, and Figure S5), and Northeastern Europe (NEU, Figure S7, and Figure S8), separately. The drought covered area is computed over the grid points where the values of each drought indicator fall below the threshold defined by the 10th (Q90) percentile of the whole time series. The times series for the definition of Q90 have been computed for each separate region (i.e., MED, CEU, and NEU) as well as for the whole Europe (i.e., EURO). The time series used to compute Q90 are shown in Fig. 2 (EURO), Figure S1 (MED), Figure S4 (CEU), and Figure S7 (NEU) and the threshold for each indicator is shown in Table 1.
Table 1
Q90 a threshold for each indicator (i.e., PDSI, SPEI and SMI) and for each analyzed region (i.e., EURO, MED, CEU and NEU) and the analyzed period.
| EURO | MED | CEU | NEU | Period |
PDSI | -1.74 | -1.79 | -1.17 | -1.14 | 1401–2018 |
SPEI | -0.64 | -0.68 | -0.81 | -0.59 | 1601–2018 |
SMI | 0.33 | 0.32 | 0.31 | 0.36 | 1766–2018 |
For the European region as a whole, there are no significant changes either in the magnitude or the area covered by drought, for any of the analyzed indices (Fig. 2 and Fig. 3). The only exception is for the SMI index, where a significant increase (0.52%/decade, p > 0.05) in the drought covered area has been observed over the period 1766–2018 (Fig. 3c). For the European region, the driest year, both in terms of magnitude (M) and drought covered area (A) differs when looking at the different drought indicators, namely 1659 (PDSI, M = -2.07, A = 75.1%), 1783 (SPEI, M = -1.16, A = 90.1%) and 1868 (SMI, M = 0.21, A = 83.4%) (Table S1 and Fig. 4), respectively. For MED, only SMI indicates an increasing drying trend (~-0.025/decade, p > 0.05, Figure S1c), as well as a significant increase in the area covered by drought (~ 1.1%/decade, p > 0.001, Figure S2c), while for PDSI and SPEI no significant trends have been found over the analyzed period (i.e., 1401–2018 for PDSI and 1601–2018 for SPEI). The driest year (both in terms of magnitude and drought covered area), over MED, based on the three different indicators are: 1945 (PDSI, M = -3.60, A = 96.6%); 2012 (SPEI, M = -1.78, A = 98.3%) and 1945 (SMI, M = 0.07, A = 96.5%) (Figure S4 and Table S1). For Central Europe (CEU), no significant trends have been found for any of the analyzed indicators (Figure S5), while in the case of the drought covered area (Figure S6), only SMI indicates a significant increasing trend (~ 0.6%/decade, p < 0.001). In the case of CEU region, the driest year (both in terms of magnitude and drought covered area) was found to be 1921 for PDSI (M = -3.77, A = 94.6%), 2015 for SPEI (M = -1.52, A = 97.8%) and 1868 for SMI (M = 0.15, A = 92.7%) (Figure S7 and Table S1). The spatio-temporal coverage of droughts for NEU shows a relatively different picture compared to MED and CEU. PDSI and SMI indicate a wetting trend over NEU, although the magnitude is small and the trend is not significant (Figures S7 and S8). Opposite to PDSI and SMI, SPEI indicates an insignificant drying trend (Figure S7). In terms of the area covered by drought, no significant changes are observed for any of the three indicators (Figure S8). The driest year (both in terms of magnitude and drought covered area), over NEU, based on the three different indicators are: 1826 (PDSI, M = -3.19, A = 89.2%); 1868 (SPEI, M = -1.69, A = 75,4%) and 1826 (SMI, M = 0.06, A = 98.1%) (Figure S4 and Table S1). For MED and NEU, both PDSI and SMI indicate the same most extreme individual years (i.e., 1945 for MED and 1826 for NEU), while for CEU and EURO each indicators shows different individual extreme dry years. A detailed description of each extreme event identified for the different indicators is given in the supplementary file.
So far, we focused our analysis on the extremeness of individual years, but the impact of a drought depends on how long it lasts, and this aspect must be taken into account when assessing the spatio-temporal evolution of drought events. While single year events might have also a high impacts on forestry, agriculture, water management, and ecosystems, among others, multi-year droughts are the ones with the most extensive impacts 7,28,33. For example, while ecosystems like forests can endure single-year droughts, the repeated stress of multi-year droughts may have significant consequences on their functioning 6. In this respect, in the following, we will analyze the extremeness and changes of multi-year droughts in Europe by averaging the gridded fields (i.e., PDSI, SPEI, and SMI) over 10 consecutive years. The choice for 10 years average was motivated by previous studies which have shown that multi-years droughts most of the time occur in clusters of 4 up to 10 consecutive years 26,30,33,34 and it also allows us to make a proper comparison between the different drought indicators at decadal time scale. Only the events which show extremeness, both in terms of magnitude and spatial extent, are discussed in the current study. For the sake of simplicity, we categorize only the first five extreme drought events based on their magnitude and spatial coverage (see Table S2).
The spatio-temporal analysis of the largest droughts events in Europe, at decadal time scales are summarized in Fig. 5 and Fig. 6, respectively. At decadal time-scales, the strongest drought events in terms of magnitude, as captured by PDSI, are 1941–1950 (M = -0.82), followed by 1551–1560 (M = -0.80). In terms of spatial extent, the roles are reversed, with the drought event of 1551–1560 covering the largest area (A = 44.5%), followed by the 1941–1950 decade (A = 42.1%) (Fig. 5a, 6a and Table S2). The drought event over the period 1551–1960 was focalized mainly over the central and eastern parts of Europe (Fig. 7a), while the 1941–1950 drought event was focalized over the central and southern parts of Europe (Fig. 6b). When looking at each analyzed region, separately, there are clear divergences in terms of the driest and most extensive drought events at the decadal time scales. In the case of MED, PDSI indicates the same multi-year event as in the case of the whole Europe, namely 1941–1950 (Figure S10a and Table S2). This is not surprising since the spatial extent of the 1941–1950 events indicates that this decade was the driest one over the central and southern parts of Europe, including the Mediterranean region (Fig. 7b). The same decade, namely 1941–1950 is the second driest one both in terms of magnitude (M = -0.95) and spatial extent (A = 34.2%) over CEU (Figure S11a and Table S2). The driest decade, over CEU was 1551–1560 (M = -1.49 and A = 44.7%, Figure S11a and Table S2). For NEU, the driest decade, both in terms of magnitude and coverage, was 1531–1540 (M = -083 and A = 44.3%), followed by 1651–1660 (M = -0.66 and A = 41.2%, Figure S12a and Table S2)
In the case of SPEI, the strongest drought events, at decadal time scales, in terms of magnitude and spatial extent are 2011–2018 (M = -0.36 and A = 46.6%), followed by 1681–1690 (M = -0.23 and A = 38.1%), respectively (Fig. 5b and 6b). Similar results have been found by 26 who have shown that the period 2015–2018 was unusually dry over the last 400 years over large parts of central and western Europe. The 1681–1690 drought event was mainly focused over the eastern part of Europe (Fig. 7c), while the 2011–2018 was covering the central and southern parts of Europe (Fig. 7d). In the case of SPEI similar results have been found for the MED and CEU regions, with the period 2011–2018 being the driest (MED: M = -0.65, CEU: M = -0.44) and having the largest spatial extent for both regions (MED: A = 49.4%, CEU: A = 44.6%) (Table S2 and Figure S10 and S11). For NEU, the driest decade was 1851–1860 (M = -0.41, A = 31.6%). The second driest decade, in terms of magnitude, was 1861–1870 (M = -0.34, A = 22.1%), while in terms of coverage, it was 1891–1900 (M = -0.26, A = 30.5%) (Figure S12 and Table S2). As in the case of the interannual time scales, there are discrepancies also at decadal time scales regarding the extremeness of different decades, based on the drought indicator used. This is another indication that one needs to treat with care the use of different indicators in assessing the long-term evolution of drought either at larger spatial scales (e.g., the whole Europe) or at more regional scales (e.g., the Mediterranean region or Central Europe). SPEI, being of the indicators in which temperature and evapotranspiration, plays a crucial role 1,23,35, is more prone to identify the recent decades, especially 2011–2018 as the driest and most spatially extended decade, while PDSI is more prone to identify as extreme drought prone periods, time scans when the rainfall deficit was more important than the temperature.
The same analysis was performed for the overlaying time period of the three datasets, namely 1766–2018. At interannual time-scales the results vary slightly (i.e., some of the extreme years identified for PDSI have changed to due the reduction in the time period analyzed), but at decadal time-scale the results are more do not change for the driest decades (i.e., 1941–1950 for PDSI, 2011–2018 for SPEI and 1781–1790 for SMI), mainly because all these extremely dry decades happened after the beginning of the common period (Table S5 and S6). The same holds trued for the regional analysis, with the exception of PDSI for CEU and NEU (Table S5 and S6). In the case of CEU, the driest decade over the common period is 1941–1950, while for NEU the driest decade is 1971–1980.