3.1. Time Series Analysis
The temporal evolution of the EAWR index, for the period 1950–2012, is shown in
Figure 1a. The time series of JFM EAWR index presents pronounced interannual variability and higher amplitude for the negative phase. Persistent negative anomalies were observed during the mid-winters of 1970, 1977 and 1979, periods which were also characterized by enhanced precipitation over the western part of Europe (
Figure 1a). The wavelet analysis reveals a significant periodicity in the 2–6 years band (
Figure 1b) and a weaker periodicity at around 8 years. The Morlet wavelet spectrum (
Figure 1b) of the JFM EAWR index shows a nonstationary behavior, especially in the 2–6 years band. This nonstationary behavior emphasizes the fact that the strength of the interannual variability of the JFM EAWR index has varied with time.
Figure 1.
(a) Annual January to March (JFM) value of the East Atlantic/Western Russia (EAWR) Index and the 5 years running mean (black line); (b) The continuous wavelet power spectrum of the time series of JFM EAWR Index. The thick black line contour is the 5% significance level against red noise.
Figure 1.
(a) Annual January to March (JFM) value of the East Atlantic/Western Russia (EAWR) Index and the 5 years running mean (black line); (b) The continuous wavelet power spectrum of the time series of JFM EAWR Index. The thick black line contour is the 5% significance level against red noise.
3.2. Relationship of the EAWR with the Atmospheric Circulation
The relationship between the JFM EAWR index and geopotential height at 500 hPa is investigated using lagged composite maps. The use of lagged composite makes it possible to detect the prior development and subsequent decay of Z500 anomalies associated with the JFM EAWR structure. Here we use seasonal anomalies for Z500 defined as: October–November–December (OND), December–January–February (DJF) and February–March–April (FMA)). The seasonal anomalies were calculated by subtracting mean annual cycle. After that the long-term mean of every calendar month was subtracted from each individual month and the seasonal anomalies were calculated as 3-month averages. To obtain the normalized anomalies all the data sets employed were normalized by their standard deviation.
There is little EAWR signature in the 500-hPa geopotential height anomalies during the late autumn and early-winter (OND) (
Figure 2a). By mid-winter (DJF), the classic EAWR pattern starts to develop, particularly over Europe and eastern Russia (
Figure 2b). There are significant 500 hPa height anomalies over Europe and the Scandinavian Peninsula, negative anomalies over eastern Russia and positive anomalies over northern part of China and Mongolia. By late winter, the geopotential height anomalies pattern becomes more pronounced and all the centers of action characteristic to the EAWR pattern become much stronger (
Figure 2c). During late spring (February–April), the overall geopotential height anomalies related to EAWR persist, but their intensity decreases (
Figure 2d).
Figure 2.
Geopotential Height at 500 hPa composite map (difference between the averaged maps for which the EAWR JFM index was higher (lower) than 1 (–1) standard deviation) for (a) October to December (OND); (b) December to February (DJF) (c) January to March (JFM) and (d) February to April (FMA). Hatching displays areas at or over 95% sign significance level using a two tailed t-test.
Figure 2.
Geopotential Height at 500 hPa composite map (difference between the averaged maps for which the EAWR JFM index was higher (lower) than 1 (–1) standard deviation) for (a) October to December (OND); (b) December to February (DJF) (c) January to March (JFM) and (d) February to April (FMA). Hatching displays areas at or over 95% sign significance level using a two tailed t-test.
3.3. TheInfluence of EAWM on the Hydroclimatology over the European Region
The relationship between JFM EAWR and PP, TT and SPEI3 over Europe, is investigated using lagged correlation maps, with JFM EAWR index leading the aforementioned fields. The seasonal averages of PP, TT and SPEI3 were computed in a similar way as for Z500. For PP, TT and SPEI3 we used seasonal averages for the following seasons: January–February–March (JFM), February–March–April (FMA) and March–April May (MAM).
Positive EAWR phase is typically associated with wetter than normal conditions over the western coast of the Scandinavian Peninsula and drier than normal conditions over central Europe and the Balkans (
Figure 3a). In early spring (
Figure 3b) the EAWR influence remains significant just over the central and the eastern part of Europe. In late spring the influence of EAWR persists over the eastern part of Europe and over the northern part of the Scandinavian Peninsula (
Figure 3c). As it can be inferred from
Figure 3a–c there is a propagation of the influence of EAWR on the European precipitation from the west towards east, from mid-winter towards late spring.
Figure 3.
Correlation between the JFM EAWR and CRU T.S. 3.2 Precipitation (PP) for (a) January to March (JFM); (b) February to April (FMA) and (c) March to May (MAM). Hatching displays areas at or over 95% significance level using a two tailed t-test.
Figure 3.
Correlation between the JFM EAWR and CRU T.S. 3.2 Precipitation (PP) for (a) January to March (JFM); (b) February to April (FMA) and (c) March to May (MAM). Hatching displays areas at or over 95% significance level using a two tailed t-test.
Mid-winter temperatures during the positive phase of EAWR are characterized by warmer than normal conditions over the Scandinavian Peninsula and colder than average conditions over the Balkan region. Outside these regions the correlation is low and insignificant (
Figure 4a). For the early spring the positive anomalies persist and they are as significant and extended as in the case of mid-winter (
Figure 4b). Furthermore, the influence of EAWR starts to be seen also on the temperature over the central Europe. By late spring the warmer than normal conditions are prevailing mostly over the central part of Europe and the Scandinavian Peninsula (
Figure 4c), remaining significant at the 95% level. The influence over the Balkan regions starts to disappear. As in the case of precipitation, there is also a north-west to south-east propagation of the influence of EAWR on the European temperature, from mid-winter to late spring.
Figure 4.
Correlation between the JFM EAWR and CRU T.S. 3.2 Temperature (TT) for (a) January to March (JFM); (b) February to April (FMA) and (c) March to May (MAM). Hatching displays areas at or over 95% significance level using a two tailed t-test.
Figure 4.
Correlation between the JFM EAWR and CRU T.S. 3.2 Temperature (TT) for (a) January to March (JFM); (b) February to April (FMA) and (c) March to May (MAM). Hatching displays areas at or over 95% significance level using a two tailed t-test.
The influence of EAWR pattern on the occurrence of mid-winter droughts follows the same path as precipitation (
Figure 5a), but for some regions the correlations are higher (e.g., the eastern part of Europe). This can be due to the fact that SPEI integrates information from PP as well as TT. During the positive phase of EAWR most of the European regions are exposed to meteorological drought, while the northern part of the Scandinavian Peninsula is characterized by wet conditions. In early spring the correlation between the JFM EAWR index and FMA SPEI3 becomes higher over the European region (
Figure 5b), meaning that the influence of mid-winter EAWR has a stronger impact over the central Europe in the upcoming early spring. For the late spring (
Figure 5c), the influence of EAWR persists over the same regions, but the correlation decreases over the central part of Europe, but remains still significant at 95% level.
Figure 5.
Correlation between the JFM EAWR and SPEI3 for (a) January to March (JFM); (b) February to April (FMA) and (c) March to May (MAM). Hatching displays areas at or over 95% significance level using a two tailed t-test.
Figure 5.
Correlation between the JFM EAWR and SPEI3 for (a) January to March (JFM); (b) February to April (FMA) and (c) March to May (MAM). Hatching displays areas at or over 95% significance level using a two tailed t-test.
3.4. Physical Mechanism
To identify the physical mechanism by which EAWR influences the hydroclimatology variability over Europe, we constructed the composite maps between the normalized and standardized time series of EAWR JFM and the seasonal average (JFM) of the zonal component of mean wind at 200 hPa (U200) and the vertically integrated water vapor transport (WVT), for the years of high (>1 standard deviation) and low (<−1 standard deviation) values of the normalized EAWR index. As in the case of Z500, this threshold was chosen as a compromise between the strength of climate anomalies associated with EAWR and the number of maps that satisfy this criterion.
As it can be inferred from
Figure 6 a pronounced difference is found in the structure of the near-tropopause Atlantic subtropical jet, represented here by the averaged seasonally magnitude of the 200 hPa wind vector. During the positive phase of EAWR the degree of separation between the Atlantic and the North African segment of the subtropical jet is considerably more pronounced (
Figure 6a), comparative to the negative phase of EAWR (
Figure 6b). As a consequence, during the positive phase of EAWR, the central and southern part of Europe is characterized by dry conditions (see
Figure 3a), while the northern part of the Scandinavian Peninsula is characterized by wet conditions due to the fact that the jet exit points towards Scandinavia (
Figure 6a). According to [
23] the poleward side of the jet exit regions is preferred for cyclonic growth, which in turn can induce a large amount of precipitation. During the negative phase of EAWR the axis of the Atlantic Jet is shifted southward and the two jets are almost coupled (
Figure 6b). In this case, the western part of Europe receives more precipitation than normal, being situated to the west side of the jet exit.
Figure 6.
The composite maps between the normalized time series of JFM EAWR and JFM zonal component of mean wind at 200 hPa for (a) High (JFM EAWR > 1 std. dev.) and (c) Low (JFM EAWR < −1 std. dev.) The composite maps between the normalized time series of JFM AO and JFM vertically integrated water vapor transport for (b) High (JFM EAWR > 1 std. dev.) and (d) Low (JFM EAWR < −1 std. dev.) For (a,c) Climatological means are shown as contour lines. Units are m/s; For (b,d): The shaded areas indicate the distribution of the horizontal divergence of the total water vapor transport (units 10−6 kg/m2s).
Figure 6.
The composite maps between the normalized time series of JFM EAWR and JFM zonal component of mean wind at 200 hPa for (a) High (JFM EAWR > 1 std. dev.) and (c) Low (JFM EAWR < −1 std. dev.) The composite maps between the normalized time series of JFM AO and JFM vertically integrated water vapor transport for (b) High (JFM EAWR > 1 std. dev.) and (d) Low (JFM EAWR < −1 std. dev.) For (a,c) Climatological means are shown as contour lines. Units are m/s; For (b,d): The shaded areas indicate the distribution of the horizontal divergence of the total water vapor transport (units 10−6 kg/m2s).
To further prove that the jet axis plays a crucial role in the variability of the climate over Europe we have applied the same analysis (composite maps) to the field of WVT. The WVT is strongly related with the Jet variability. In the exit regions of the Atlantic Jet a meridional circulation carries negative vorticity southward in the upper troposphere and the return branch of this circulation enhances the baroclinicity on the lower troposphere through the northern thermal advection. During the positive phase of EAWR (
Figure 6c), the axis of the maximum transport is directed towards Scandinavia and there is a significant reduction of the water vapor transport from the Atlantic towards Europe. In the same time a convergent zone develops over the northern part of the Scandinavian Peninsula, which in turn enhances precipitation. The path of the axis of WVT follows the same direction as the Atlantic Jet axis. During the negative phase of EAWR the axis of the WVT is directed towards Europe (
Figure 6d), enhancing precipitation over the western and southern part of Europe (in agreement with
Figure 3b) due to the convergent zones which are developing.
To support this idea we have plotted the composite maps between the EAWR JFM index and the 200 hPa and 850 hPa velocity potential and the divergent wind (
Figure 7). At upper levels, during the positive phase of EAWR, there is a well organized zonal quadruple structure over the Atlantic Ocean, central and southern part of Europe, Caspian Sea and India (
Figure 7a). At 200 hPa level there is a region of positive velocity potential anomalies over the central and southern part of Europe, which means convergent winds at upper levels and divergent winds at lower levels (
Figure 7b) over these areas, flanked (at 200 hPa level) by two regions of convergence over the Atlantic Ocean and one over the Caspian Sea. Based on
Figure 7, during the positive phase of EAWR, the central Europe is characterized by surface divergence and upper-levels convergence, meaning subsiding air over central Europe and reduced precipitation. This is in agreement with the results from
Figure 3 and
Figure 4, respectively, based on which, during the positive phase of EAWR the central and southern part of Europe are characterized by higher than normal temperatures and reduced precipitation.
Figure 7.
(a) 200 hPa velocity potential (shaded) and divergent wind (wind bars) composite map (difference between the averaged maps for which the normalized EAWR JFM index was higher (lower) than 1 (−1) standard deviation); (b) 850 hPa velocity potential (shaded) and divergent wind (wind bars) composite map (difference between the averaged maps for which the normalized EAWR JFM index was higher (lower) than 1 (−1) standard deviation). Units are: 106 m2/s for the velocity potential and m/s for the divergent wind.
Figure 7.
(a) 200 hPa velocity potential (shaded) and divergent wind (wind bars) composite map (difference between the averaged maps for which the normalized EAWR JFM index was higher (lower) than 1 (−1) standard deviation); (b) 850 hPa velocity potential (shaded) and divergent wind (wind bars) composite map (difference between the averaged maps for which the normalized EAWR JFM index was higher (lower) than 1 (−1) standard deviation). Units are: 106 m2/s for the velocity potential and m/s for the divergent wind.
3.5. Non-Stationarity Issues
In order to test the strength of the relationship between the EAWR index and the precipitation, temperature and SPEI3 fields over Europe, we have calculated the correlation coefficients between the JFM EAWR index and the PP, TT and SPEI3 fields for a 21-year moving window. Prior to the correlation analysis, all the time series were linearly detrended and normalized by their standard deviation. The non-stationarities in the climate system can lead to high correlation and anti-correlation, for different periods, e.g., before and after the 1970s climate shift [
24]. Another example of a non-stationary relationship is the correlation between NAO and Northern Hemisphere SLP. For the last fifteen years the correlation is weak in the Pacific sector [
25], however, significant correlations between NAO and Pacific climate variability during the 1930s to 1960s have been identified [
26]. In order to overcome this problem, we have applied the stability maps criterion [
20,
22] to the relationship between EAWR and the PP, TT and SPEI3 fields.
Figure 8 shows the stability maps between JFM EAWR and PP, TT and SPEI3, which are defined as regions where the running correlation in a 21-year window does not change the sign. The green (blue) and yellow (red) regions coincide with the regions where the correlation between JFM EAWR and the PP field is statistically significant at 80% (90%) level. The stability map between JFM EAWR and JFM PP (
Figure 8a) shows small stable regions over the western part of Europe and Balkans. During early spring (
Figure 8d) these regions extend all over the southern part of Europe (negative correlations) and the north-western part of the Scandinavian Peninsula (positive correlations), while in late spring (
Figure 8g) there is no EAWR signal in the precipitation variability, except some small regions (e.g., northern part of Romania).
Figure 8.
Stability maps of the correlation between JFM EAWR and Precipitation (first column), Temperature (second column) and SPEI3 (third column) for (a–c) January to March; (d–f) February to April and (g–i) March to May. Regions where the correlation is stable, positive, and significant at the 90% (80%) level for at least 80% of the windows are shaded red (yellow). The corresponding regions where the correlation is stable but negative are shaded blue (green).
Figure 8.
Stability maps of the correlation between JFM EAWR and Precipitation (first column), Temperature (second column) and SPEI3 (third column) for (a–c) January to March; (d–f) February to April and (g–i) March to May. Regions where the correlation is stable, positive, and significant at the 90% (80%) level for at least 80% of the windows are shaded red (yellow). The corresponding regions where the correlation is stable but negative are shaded blue (green).
For mid-winter the stability map between JFM EAWR and JFM TT (
Figure 8b) shows stable regions just over Bulgaria (negative stable correlation). By mid-spring the negative stable regions remain almost the same (
Figure 8e), but significant positive correlations appear over the Scandinavian Peninsula. EAWR influence persists by late-spring especially over the central part of Europe (positive stable correlations), while over the Balkans the influence of EAWR almost disappears (
Figure 8h).
In the case of SPEI3 the stability maps show persistent stable regions, from late winter to late spring (
Figure 8c,f,i) over the central and the eastern part of Europe. The stable regions identified for PP, TT and SPEI3 can give a potential predictive skill based on the JFM EAWR index for the upcoming hydroclimate in late spring over Europe.