Connecting atmospheric blocking to European temperature extremes in spring

Atmospheric blocking is an important contributor to European temperature variability. It can trigger cold and warm spells, which is of speciﬁc relevance in spring because vegetation is particularly vulnerable to extreme temperatures in the growing season. The spring season is investigated as transition period from predominant connections of blocking with cold spells in winter to predominant connections of blocking with warm spells in summer. Extreme temperatures are termed cold or warm spells if temperature stays outside the 10th to 90th percentile range for at least 6 consecutive days. Cold and warm spells in Europe over 1979 to 2014 are analyzed in observations from E-Obs data and the connection to blocking is examined in geopotential height ﬁelds from ERA-Interim. A highly signiﬁcant link between blocking and cold and warm spells is found which changes during spring. Blocking over the north-eastern Atlantic and Scandinavia is correlated with the occurrence of cold spells in Europe, particularly early in spring, while blocking over central Europe is associated with warmer conditions, particularly from March onwards. The location of the block also impacts the spatial distribution of temperature extremes. More than 80 % of cold spells in south-eastern Europe occur during blocking whereas warm spells are correlated to blocking mainly in northern Europe. Over the analysis period, substantial interannual variability is found but also a decrease in cold spells and an increase in warm spells. The long-term change to a warmer climate holds the potential for even higher vulnerability to spring cold extremes.

side the 10th to 90th percentile range for at least 6 consecutive days. Cold and warm spells in Europe over 1979 to 2014 are analyzed in observations from E-Obs data and the connection to blocking is examined in geopotential height fields from ERA-Interim. A highly significant link between blocking and cold and warm spells is found which changes during spring. Blocking over the north-eastern Atlantic and Scandinavia is correlated with the occurrence of cold spells in Europe, particularly early in spring, while blocking over central Europe is associated with warmer conditions, particularly from March onwards. The location of the block also impacts the spatial distribution of temperature extremes. More than 80 % of cold spells in south-eastern Europe occur during blocking whereas warm spells are correlated to blocking mainly in northern Europe. Over the analysis period, substantial interannual variability is found but also a decrease in cold spells and an increase in warm spells. The long-term change to a warmer climate holds the potential for even higher vulnerability to spring cold extremes. European weather and climate is strongly influenced by large-scale circulation patterns such 39 as the Atlantic storm tracks, the jet stream, and atmospheric blocking (e.g., Woollings 2010). 40 Atmospheric blocking describes a meteorological situation in which a persistent and stationary 41 high pressure system blocks the climatological westerly flow at mid-latitudes for several days to 42 weeks (Rex 1950;Tibaldi and Molteni 1990;Pelly and Hoskins 2003;Barriopedro et al. 2006;43 Croci- Maspoli et al. 2007). 44 Extremes on both ends of the temperature distribution are especially closely connected to atmo-  Surface temperatures can be impacted by atmospheric blocking via radiative forcing or advec-51 tion. Radiative effects are mainly constrained to the center of the block where clear-sky conditions 52 favor positive temperature anomalies. The anticyclonic circulation of the block affects tempera-53 tures especially on the eastern and southern flanks by advection of cold air from the north and east 54 (e.g., Trigo et al. 2004;Bieli et al. 2015). A range of studies has either focused on the predominant 55 cooling effect of blocking in winter (Trigo et al. 2004;Barriopedro et al. 2008;Cattiaux et al. 56 2010;Buehler et al. 2011;Sillmann et al. 2011;Whan et al. 2016) or on the warming effect in 57 summer (Xoplaki et al. 2003;Cassou et al. 2005;Pfahl and Wernli 2012;Stefanon et al. 2012 potential of earlier spring green-up to also impact European warm spells via feedback processes.

69
In this study we analyze the connection of blocking and extreme temperature occurrences, their 70 spatial distribution and change over the last decades. We focus on spring on a month-by-month 71 basis, but also show results for the seasonal mean of other seasons. We describe data and methods 72 in section 2. Results are presented in section 3 and a summary is given in section 4.

74
The detection of temperature extremes is based on E-Obs version 12.0 (Haylock et al. 2008 in the E-Obs temperatures to remove the long-term temperature trend. Daily 10th/90th percentiles 82 4 of T min /T max are computed over the 36 year period using a 21 day sliding window. A grid point 83 with T min below the 10th percentile or T max above the 90th percentile for at least 6 consecutive 84 days is identified as cold or warm extreme, respectively. This study focuses on large scale events 85 on a daily basis. Therefore we define a cold spell day (CSD) or warm spell day (WSD) if at least 86 400 grid points (i.e., 5 • × 5 • ) simultaneously are found to be exposed to a cold or warm extreme 87 criterion on a given day. Resulting cold/warm spells are found to be spatially highly coherent, so 88 no separate adjacence-criterion was applied.

89
The detection of blocking is based on daily geopotential height (GPH) fields from the European   In addition, we analyse selected subdomains and investigate the importance of the location of 104 cold/warm spells and blocking for their connection. For selection for CSDs/WSDs in subdomains 105 we adjust the spatial criterion to consider CSDs/WSDs with more than half of their grid points in 106 5 the selected subdomain. For selection of blocking in subdomains we consider blocks with at least 107 one blocked grid point in the selected subdomain.

108
In order to test any co-occurrence of CSDs/WSDs and blocked days for significance we perform 109 a Monte-Carlo test. Given N CSDs/WSDs in a period (i.e., month or season), we draw 1000 ran-  the trend in the underlying temperature time series is not removed (Fig. 1c, bottom) we find more spells also occurs after top), pointing at the role of internal variability.

131
Over the spring season, the number of WSDs and with it the number of blocked WSDs increases 132 towards summer (Fig. 1b, right). Over the analysis period, the seasonal mean time series also show 133 considerable interannual variability for WSDs (Fig. 1d, top). If the trend is not removed from the 134 underlying temperature time series (Fig. 1d, bottom)  (anti-correlation) as well as in winter (correlation) and in summer (anti-correlation; cf. Table 1).

147
Regarding WSDs in spring, a statistically significant fraction of 54 % is blocked and about 21 % 148 blocked spring days coincide with a WSD. Also most individual months of the extended spring 149 show a significant correlation with blocking (as do summer months), except February on the transi-150 tion from winter to spring exhibits a significant anti-correlation (as do winter months; cf. Table 1).

151
Analyzing blocking on a grid point basis, the climatological blocking frequency in the Euro-

152
Atlantic region is generally between 2 % and 6 % of spring days. The blocking frequency coin-153 the British Isles (i) and over northern Scandinavia (ii) the blocking frequency is up to three times 155 higher for CSDs than for climatological conditions and differs statistically significantly from the 156 random sample. This is consistent with cold advection during such blocks into central and west-157 ern Europe. Over central and eastern Europe (iii) there is significantly less blocking during CSDs 158 (<2 %) than in the climatology since blocking occurring there tends to lead to warmer, fair weather 159 conditions.  The blocking frequency coinciding with WSDs in spring is found to be up to three times higher 174 than during climatological conditions (Fig. 3a) and statistically significantly different from the 175 random sample in most of Europe. Blocks linked to warm spells are distributed across Europe, 176 while there are less than average blocking days associated with WSDs west of the British Isles. For a closer investigation of the dipole feature we divide Europe into two subdomains for 204 CSDs/WSDs: northern (> 50 • N) and southern (< 50 • N) Europe (cf. Fig. 4c, d). Selecting only 205 CSDs/WSDs in these subdomains we show the corresponding blocking frequency in Fig. 5. For 206 the 163 CSDs in northern Europe hardly any blocking is found in the entire Euro-Atlantic block-207 ing region (Fig. 5a) indicating that blocking tends to counteract CSDs here. CSDs (136 days) in 208 southern Europe (Fig. 5c)   The location of the block is found also essential for its impact on European extreme tempera-250 tures. Blocking west of the British Isles and over northern Scandinavia is clearly connected with 251 cold spells in southern Europe while blocking over central Europe and southern Scandinavia is 252 associated with warm spells in northern Europe. This is consistent with the role of cold advec-253 tion at the edges of blocks leading to cold spells outside blocked regions and with increased solar 254 radiation leading to warm spells in blocked regions.

255
The spatial distribution of cold and warm spells during blocking reveals a distinct dipole pattern.

256
Cold spells in south-eastern Europe are found highly correlated with blocking, and more than 80 % 257 of cold spell days co-occur with a blocking. In contrast, cold spells in northern Scandinavia and 258 blocking are anti-correlated with regionally less than 30 % co-occurrence. Warm spells show 259 the opposite relationship with locally more than 80 % of warm spell days in northern Europe co-260 occurring with blocking, but anti-correlation in southern Europe. An increased occurrence of both, 261 warm and cold spells during blocked conditions is found around 50 • N indicating that blocking 262 increases the probability for both high and low temperature extremes here.

263
The occurrence of atmospheric blocking in the European region is found to be crucial for the 264 development of both, extended cold and warm spells, in spring. We provide insight into the chang- The climatological blocking frequency is indicated by black contour lines.  warm spell days (WSDs) that occur over northern (top) and southern (bottom) Europe. The split into north/south is made at 50 • N as indicated by the gray boxes. Values that are statistically significantly larger than the number of blocks from random days (above 95th percentile) are marked with a plus sign and values that are statistically significantly lower (below 5th percentile) are marked with a times sign, respectively. The climatological blocking frequency is indicated by black contour lines.