The persistent and quasi-stationary nature of atmospheric blocking is associated with long-lasting extreme weather conditions that influence much of the Northern Hemisphere during boreal winter. The Madden–Julian oscillation (MJO) has been previously shown to influence important factors for blocking, including Rossby wave breaking and the North Atlantic Oscillation (NAO). However, the extent to which the MJO influences blocking across the Northern Hemisphere is not yet fully understood.
Utilizing a two-dimensional blocking index, composites of North Pacific, North Atlantic, and European blocking are generated relative to MJO phase. In the west and central Pacific, all MJO phases demonstrate significant changes in blocking, particularly at high latitudes. A significant decrease in east Pacific and Atlantic blocking occurs following phase 3 of the MJO, characterized by enhanced convection over the tropical East Indian Ocean and suppressed convection in the west Pacific. The opposite-signed MJO heating during phase 7 is followed by a significant increase in east Pacific and Atlantic blocking. A significant decrease in European blocking follows MJO phase 4, with an increase after phase 6. The phase 6 European blocking is hypothesized to result from two preexisting conditions: 1) an anomalous anticyclone over the Atlantic and 2) a preceding negative Pacific–North American (PNA) pattern initialized and influenced by MJO heating.
Blocking anticyclones have been studied for many decades due to their ability to significantly disrupt the mean westerly flow in the midlatitudes (Berggren et al. 1949; Masato et al. 2012). The upper-tropospheric jet in the midlatitudes can become “blocked” for several weeks in the presence of a blocking anticyclone. During boreal winter, these persistent anticyclones redirect polar air toward the midlatitudes, leading to extreme cold weather outbreaks (e.g., Hoskins and Sardeshmukh 1987; Buehler et al. 2011). During boreal summer, blocking anticyclones can result in heat spells and episodes of drought (e.g., Dole et al. 2011).
Given the influence of atmospheric blocks on extreme events, understanding what regulates blocking is crucial to improving the prediction and understanding of these extremes. Many studies have demonstrated that anomalous heating in the tropics can generate stationary Rossby waves (e.g., Hoskins and Karoly 1981; Hoskins and Ambrizzi 1993; Jin and Hoskins 1995; Trenberth et al. 1998) that influence blocking. One example is the heating associated with the El Niño–Southern Oscillation (ENSO); using a general circulation model, Hinton et al. (2009) demonstrated that a stationary Rossby wave response to ENSO forcing leads to a dipole pattern in geopotential height anomalies over the northeast Pacific resulting in favorable conditions for blocking during the warm phase and unfavorable conditions during the cold phase. In the Southern Hemisphere, Australian blocking and rainfall was found to be influenced by ENSO heating and by intraseasonal convection anomalies near Indonesia (Pook et al. 2013; Marshall et al. 2014).
On intraseasonal time scales, tropical heating variability associated with the Madden–Julian oscillation (MJO) has been demonstrated to significantly perturb the midlatitude flow, with consequences for blocking including changes in Pacific wave breaking (Moore et al. 2010) and the North Atlantic Oscillation (NAO; Cassou 2008; Lin et al. 2009). The MJO is the dominant form of intraseasonal variability in the tropics with a period of approximately 30–90 days (Madden and Julian 1971, 1972). During an MJO event, coupled circulation and convection anomalies propagate eastward across the equatorial Indian and west Pacific Oceans, with flow anomalies becoming decoupled to convection at the date line. Anticyclonic Rossby gyres form west of the enhanced convection while cyclonic gyres develop behind areas of suppressed convection (Salby and Hendon 1994; Zhang 2005). Numerous studies have demonstrated that MJO-induced Rossby waves significantly alter the global extratropical circulation in both hemispheres (e.g., Kiladis and Weickmann 1992; Higgins and Mo 1997; Matthews et al. 2004). However, the impact of these MJO teleconnections on Northern Hemisphere blocking has received much less attention.
Some previous studies have presented links between the MJO and Northern Hemisphere blocking. In the Pacific, Moore et al. (2010) demonstrated that the subtropical jet maximum has a tendency to shift in conjunction with the eastward-propagating MJO Rossby gyres, leading to variability of the location and amplitude of the jet that in turn influences wave breaking. These changes in wave breaking coincide with changes in blocking. However, the blocking definition employed in that study allows for the propagation of anticyclones, which may produce different results if the analysis is focused on quasi-stationary blocks. Hoskins and Sardeshmukh (1987) investigated the long-lasting blocking event that was associated with the European sudden winter cold snap of 1985. The authors noticed a coincident strong intraseasonal oscillation over the equatorial Indian and Pacific Oceans and hypothesized that the phenomenon could have altered the Euro-Atlantic circulation leading to the European blocking event. In more recent work, Cassou (2008) found a significant relationship between the positive phase of the NAO, European blocking, and the MJO. This study hypothesized that the MJO acts as a trigger for the positive NAO that then transitions to a European blocking pattern, although it is still unclear whether the role of MJO heating in European blocking goes beyond triggering the NAO pattern. Additionally, Hamill and Kiladis (2014, hereafter HK14) found that certain phases of the MJO (i.e., the longitudinal location of MJO heating) coincide with active blocking periods in the Northern Hemisphere, and other phases coincide with suppressed periods. The mechanisms that bring about these relationships were not discussed, however. Furthermore, HK14 used a one-dimensional (1D) instantaneous blocking index (i.e., no blocking persistence criteria was used) and diagnosed blocking at the same latitude band for all longitudes, which may be a particularly poor assumption for regions such as the North Pacific (e.g., Pelly and Hoskins 2003).
In this study we utilize a two-dimensional (2D) blocking index to provide statistical relationships between the MJO and large-scale blocking events over the Pacific, Atlantic, and European regions. We will also discuss possible physical mechanisms for the relationships observed. Additionally, we propose a new mechanism for the link between the MJO and European blocking, which suggests that the importance of the MJO goes beyond the NAO mechanism hypothesized by Cassou (2008). Section 2 describes the development of the blocking index used as well as the compositing techniques. Sections 3–6 describe the MJO phase composite blocking frequencies and discusses possible physical mechanisms for the relationships observed for the west-central Pacific, east Pacific, Atlantic, and European regions, respectively. Last, section 7 summarizes the main findings and discusses future work.
Boreal winter [December–February (DJF)] 6-hourly global data are obtained from the ERA-Interim reanalysis (Dee et al. 2011) and daily averages are calculated from December 1979 to February 2010, for a total of 31 winter seasons comprising 90 days each. We use 500-hPa geopotential height and 200-hPa zonal and meridional winds with a 1.5° × 1.5° horizontal resolution. Streamfunction is calculated from the 200-hPa wind fields. Satellite-based daily outgoing longwave radiation (OLR) with a 1° × 1° horizontal resolution was acquired from the National Oceanic and Atmospheric Administration/National Climatic Data Center (NOAA/NCDC; Lee et al. 2011).
To generate MJO phase composites, the eastward propagation of the MJO is divided into eight phases using the Real-Time Multivariate MJO (RMM) indices of Wheeler and Hendon (2004; http://www.bom.gov.au/climate/mjo/), where each phase gives an approximate representation of the longitudinal location of MJO convection anomalies. The two RMM indices, known as RMM1 and RMM2, are the principal components of the first two combined empirical orthogonal functions of equatorially averaged 200- and 850-hPa zonal wind and OLR anomalies. An MJO phase is determined by , where phase 1 is associated with initial MJO convection in the west Indian Ocean, and phase 8 represents MJO enhanced convection reaching the central Pacific. Only the days with an RMM amplitude greater than 1, defined by , are used to assess blocking frequency in our analysis.
b. Two-dimensional blocking index
MJO anomalies are known to shift the jet from its climatological position (e.g., Moore et al. 2010). Therefore, predefining an average latitude at which to observe blocking may not adequately detect blocks that develop at other latitudes. This is largely resolved by utilizing a 2D blocking index. The Northern Hemisphere blocking index used here is based on Masato et al. (2013b), which detects blocking using the average difference of 500-hPa geopotential height across a latitude, , given by
A longitude is defined as instantaneously blocked when the difference between the integrated 500-hPa geopotential height () above is greater than the integrated height below so that , where is varied between 40° and 70°N. A value of 30° latitude is used for , consistent with Masato et al. (2013b) and Pelly and Hoskins (2003). Equation (1) is applied separately to four sectors: the west-central Pacific (140°E–160°W), east Pacific (160°–95°W), Atlantic (90°–20°W), and Europe (20°W–45°E), using the additional criteria discussed below. Local daily positive maxima of are found within each sector to represent the approximate center of each block. A similar approach to Masato et al. (2013b) is then followed to track each blocking event within each sector. The first instance of a positive maximum is considered the first day of a blocking event. The maximum of the next day must be within 13.5° latitude and 18° longitude of the previous day’s maximum. If more than one maximum is detected, the maximum with the smallest great-circle distance to the previous day’s maximum is used. All maxima must be within the longitudinal bounds of each region as defined above, but the block, defined as the region of positive values around each maximum, may extend beyond these bounds. A large-scale block is then defined if at least 15° of longitude are consecutively blocked. Only the times meeting these criterion for at least 5 days are retained.
The mean DJF blocking frequency for the Pacific, Atlantic, and European regions is shown in Fig. 1. For brevity, the west-central Pacific and east Pacific sectors are combined in this and all subsequent figures. Although the mean west-central Pacific high-latitude blocking overwhelms the mean east Pacific blocking in Fig. 1, results with the separate Pacific sectors were found to provide the same information as results combining the two sectors.
The Pacific comprises 2023 total blocked days (419 of these over the east Pacific sector), the Atlantic 922 blocked days, and Europe 973 blocked days (Fig. 1). These blocking frequency climatologies are consistent with those of Masato et al. (2013a,b). Figure 1 demonstrates that at high latitudes over the North Pacific a geopotential height reversal is found approximately 45% of the time during DJF, a similar total as that obtained by Woollings et al. (2008). This can be partially explained by the blocking criterion, which is more easily met in the high latitudes where the meridional gradient of geopotential height is relatively weak (Masato et al. 2013a). Woollings et al. (2008) found this high-latitude blocking to be dynamically significant, although unlike classical midlatitude blocking, high-latitude blocking does not block the jet but rather diverts its flow.
c. Composite analysis
Blocking events are composited relative to MJO phase for four different lags, where an n-day lag represents blocking occurrence n days after a given MJO phase. The full record of each blocking event is analyzed, so it is possible that a continuation of the same blocking event exists across multiple lags and/or MJO phases. For each MJO phase blocking frequencies are calculated by taking the total number of blocked days during (or after, if lagged) a given MJO phase and dividing it by the total number of days within that MJO phase. This study examines blocking in DJF; hence, composites that examine blocking after an MJO phase will also utilize the MJO index of late November. (Blocking composites are shown in Figs. 2, 7, and 8 for the Pacific, Atlantic, and European regions, respectively, and will be discussed in detail in the subsequent sections.)
To assess statistical significance, a moving-blocks bootstrap is employed independently at each lag and MJO phase. The technique is similar to a traditional bootstrap in that both sample from the original dataset with replacement in order to approximate the characteristics and distribution of the data in the bootstrap samples (e.g., Marchand et al. 2006; Wilks 2011). However, unlike a traditional bootstrap, which samples individual elements from the dataset, the moving-blocks bootstrap samples blocks of consecutive elements in order to preserve the data autocorrelation in the bootstrap samples. The moving-blocks bootstrap is implemented by taking the full blocking record and dividing it into overlapping blocks of size l beginning on each day. A bootstrap sample of size N is generated by joining N/l randomly sampled blocks, where here N represents the total number of days in each MJO phase across all 31 seasons. This process is repeated 5000 times in order to obtain sufficient bootstrap samples.
To implement the moving-blocks bootstrap, block length l must be chosen. A block length that is too large may result in a bootstrap sample that is not sufficiently random, and a block length that is too small may not properly capture the data’s persistence. Block length should also increase with increasing N (e.g., Wilks 2011). To calculate block length, the number of days in each phase is calculated for all MJO events, where days in the same phase within one MJO event may not be consecutive. The block length is then the average number of days per phase rounded to the nearest integer. This is a similar procedure as that of Marchand et al. (2006), who randomly chose a point in time and looked for the adjacent elements following the same regime to determine block length. The block lengths calculated in this way are for MJO phases 1–8, respectively. A block length that is too small can result in an unwarranted rejection of the null hypothesis (Marchand et al. 2006). Therefore, the largest calculated block length of was tested for all MJO phases and similar results were obtained, suggesting that the results are robust to small variations in this parameter. Values of N are provided in the blocking composites above each phase panel (Figs. 2, 7, and 8) as well as a value NB denoting the total number of days within or after (if lagged) each MJO phase found to contain a block. A ratio R = NB/N is also provided, representing the percentage of MJO phase days at a given lag that experienced blocking within the region.
d. Wave vector analysis
Possible explanations for the relationships found in the blocking composites are supported by analyzing the composite atmospheric patterns associated with MJO events. Whereas blocking composites are calculated and shown in days, upper-level anomalous streamfunction and 500-hPa geopotential height anomalies associated with MJO phases are shown as pentad means (i.e., 5-day averages) in Figs. 3–6. Pentad 0 indicates the field average of days 0–4 following an MJO phase, pentad 1 corresponds to days 5–9, and so forth. Assuming a normal distribution, geopotential height anomalies found to be significantly different from zero are dotted based on a two-tailed Student’s t test with N/5 independent samples, where 5 is approximately the average number of days per phase in an MJO event (e.g., Alaka and Maloney 2012).
To examine the Rossby wave pathway for each MJO phase, we calculate a horizontal wave activity flux, or vectors, derived from the conservation of wave-activity psudomomentum by Takaya and Nakamura (2001) defined as
where is the 200-hPa lagged perturbation pentad streamfunction; and are the mean DJF 200-hPa zonal and meridional winds, respectively; and the subscripts indicate partial derivatives of in the corresponding x and/or y directions. The term is approximately parallel to the group velocity of a stationary Rossby wave, thereby giving a snapshot of the direction in which wave packets travel at a given moment in time. Here vector divergence indicates a Rossby wave is emitted (a wave source) and vector convergence implies Rossby wave absorption (a wave sink).
3. West-central Pacific (140°E–160°W)
Blocking in the central and west Pacific is predominant at high latitudes (Fig. 1), where a geopotential height reversal corresponds to diverted rather than blocked flow (e.g., Woollings et al. 2008). Midlatitude blocking is far less common, with a mean blocking frequency of 2%–3% between 40° and 45°N. The anomalous blocking frequency for each MJO phase at lag 0 is shown in Fig. 2, where dotted regions indicate a blocking frequency that is outside of the 95% confidence bounds of the moving blocks bootstrap distribution. Figure 2 suggests that significant changes in west-central Pacific blocking frequency are observed during all MJO phases, with a blocking frequency increase during phases 6–8 and a decreased frequency during phases 1–5.
Phase 1 of the MJO corresponds to suppressed tropical convection over the Maritime Continent and is associated with the initialization of MJO convection in the west Indian Ocean. The enhanced convection propagates east and reaches the Maritime Continent in phase 5. The propagating convection and associated Rossby gyres appear as an eastward propagation of the subtropical jet (e.g., Moore et al. 2010). During phase 1, positive MJO-induced geopotential height anomalies are observed between 30° and 55°N over the western North Pacific with anomalously negative geopotential height to the north over eastern Russia (Fig. 3, top row), indicative of a strengthened climatological gradient. The dipole pattern persists, moving east with the MJO Rossby gyres through phase 5, as demonstrated in Fig. 3.
The eastward movement of the MJO-induced dipole and associated strengthened climatological gradient coincides with reduced blocking frequency (Fig. 2) for MJO phases 1–5. Blocking frequency decreases of 19.2% occur in northeast Russia in phase 1, where the frequency change is defined (here and henceforth) as the maximum difference in blocking frequency from the DJF mean in the region discussed. Phases 3–4 demonstrate significant decreases in blocking frequency over the central Pacific, with the greatest reduction in high-latitude blocking days occurring during phase 4, producing anomalies of almost −20%. During phase 5, the strengthened geopotential height gradient coincides with reduced blocking (−17%; Fig. 2) in southern Alaska and the central Pacific.
During MJO phase 5, suppressed convection exists in the west Indian Ocean and propagates eastward behind the enhanced convection that exists to the east, with the enhanced convection reaching the date line in phase 8. Anomalous negative geopotential height begins to develop between 30° and 50°N in the west Pacific during phase 5 (Fig. 3) and strengthens as it moves east in phase 6. By phases 7–8, the negative geopotential height anomaly expands over much of the North Pacific with anomalously high geopotential height to the northeast, and phase 8 anomalies form a positive PNA-like pattern. These MJO-induced anomalies imply a weakened climatological gradient north of the anomalous cyclonic center.
Enhanced blocking frequency is observed where the climatological gradient is weakened (Fig. 2; phases 6–8). During phase 6, a significant increase in blocking frequency occurs in the west Pacific between 45° and 50°N, where the mean blocking frequency is relatively low (Fig. 1). Moore et al. (2010) demonstrated that phase 6 coincides with an increase in cyclonic wave breaking. Cyclonic wave breaking in the west Pacific leads to blocking north of the jet, resulting in a northward flux of warm and moist subtropical air and thereby increased precipitation over the western North Pacific, with anomalously cold and dry conditions over China, Korea, and Japan (Masato et al. 2013a). During phase 8, the highest increase in high-latitude blocking is observed, with an anomalous blocking frequency peak near +17%. Phase 8 shows blocking anomalies of about −9% from 45° to 50°N, coinciding with a strengthened geopotential height gradient (Fig. 3).
One issue is determining cause and effect. It is possible that the geopotential height anomalies presented in Fig. 3 are the result of the blocking anomalies rather than the cause. Such a scenario was previously discussed in Woollings et al. (2008) in the context of Atlantic blocking. We attempt to address this issue by resampling (with replacement) the phase 8 dates to match the climatological ratio of blocking to no-blocking dates for the north-central Pacific (34% blocking days, 66% no-blocking days) and averaging to create new geopotential height anomaly composites as in Fig. 3. Resampling the phase 8 blocking and no-blocking dates to match the north-central Pacific climatology will remove any excess north-central Pacific blocks that make up the anomalous blocking increase shown in Fig. 2. If the phase 8 increase in blocking is responsible for the Pacific geopotential height anomalies in Fig. 3, the new geopotential height anomaly composite with the climatological ratio would not be able to reproduce the Fig. 3 anomalies over the north-central Pacific. This assumes that north-central Pacific blocks that form during phase 8 have the same structure as blocks that occur at all other times. However, the new phase 8 geopotential height composite is almost identical in structure and amplitude to that of Fig. 3, and so it is not show here. This analysis suggests that the increased blocking over the north-central Pacific during phase 8 is not responsible for the geopotential height anomalies in Fig. 3. The same analysis was repeated for MJO phase 4, which has the greatest decrease in north-central Pacific blocking, and again the new geopotential height anomalies are almost identical in structure and amplitude to that of Fig. 3, so it is not shown here. We repeated this analysis for both MJO phases but with the maximum climatological ratio of the entire North Pacific (44% blocking days, 56% no-blocking days) and the same results were obtained. Our contention is also supported by previous studies using simple models that do not resolve blocking, which demonstrate similar eastward-propagating geopotential height anomalies as a response to MJO forcing (e.g., Bao and Hartmann 2014; Seo and Son 2012; Matthews et al. 2004). However, we do note that the blocking frequency anomalies in Fig. 2 may still amplify the geopotential height anomalies relative to the mean.
In contrast to our results, Moore et al. (2010) found increased midlatitude central Pacific blocking when MJO convection is active over and east of the Maritime Continent, or approximately RMM phases 2–5. Their study utilizes the blocking index of Croci-Maspoli et al. (2007) that allows for blocking propagation, hence the eastward-propagating anticyclone discussed for phases 1–5 likely corresponds to their increase in midlatitude blocking. In contrast, our blocking index allows only minimal propagation so that the propagating anticyclone shown in Fig. 3 is not considered a block. Moore et al. (2010) did, however, find decreased blocking between 40° and 50°N in the central Pacific when MJO convection is active just east of the date line, which approximately corresponds to RMM phases 7–8, in agreement with our results.
Although the relationship between blocking and the MJO are not discussed in detail in HK14, their study calculates the change in instantaneous blocking frequency relative to MJO phase (lag 0) based on the 1D index of Tibaldi and Molteni (1990), where blocking is calculated along a constant blocking latitude near 50°N. Our results agree in some aspects with HK14, who found suppressed instantaneous blocking during MJO phases 1–5. The unmodified Tibaldi and Molteni (1990) index used in HK14 looks for blocking as high as 54°N; based on our results it is likely that their phase-1–5 decrease in blocking frequency coincides with instances of high-latitude blocking. In contrast to our results, HK14 demonstrates a blocking increase west of the date line only during phase 5, whereas we show that midlatitude blocking increases further during phase 6. These differences may arise from the lack of a persistence requirement in their blocking index.
4. East Pacific (160°–95°W)
We now analyze the relationship between MJO phase and east Pacific blocking frequency. Figure 2 suggests that significant decreases in east Pacific blocking frequency are strongest and most widespread from phases 3–5, and enhancement of blocking is most prominent during phases 7 and 8. Persistent significant changes in blocking frequency are observed for MJO phases 3 and 7, where a significant difference from the mean is observed out to a 15-day lag (e.g., see Fig. 12 below). We discuss MJO phases 3 and 7 and subsequent time lags here to help explain physical mechanisms for the enhancement and suppression of blocking at different points in the MJO life cycle.
a. MJO phase 3
During phase 3 of an MJO event enhanced convection is observed in the east Indian Ocean, and suppressed convection is located in the west Pacific (Wheeler and Hendon 2004). This heating pattern is associated with a large-scale stationary Rossby wave train that propagates north over the Pacific Ocean in a great-circle route to the North Pacific Ocean, North America, and to the Atlantic (Fig. 4), in agreement with the wave path presented by Lin et al. (2009). At pentad 0, the wave activity flux associated with the MJO teleconnection is strongest over the Pacific and North America, where the teleconnection geopotential height anomalies are significantly different from zero and remain significant through pentad 1.
Phase 3 geopotential height anomalies over the east Pacific sector demonstrate a persistent north–south dipole structure suggestive of a strengthened climatological geopotential height gradient with increased westerlies between 45° and 65°N, a pattern especially apparent by pentad 1. The strengthened climatological gradient coincides with the decrease in blocking frequency observed for all lags following phase 3 (e.g., see Fig. 12). During phase 3, east Pacific blocking frequency is decreased by up to 4.2% (Fig. 2). By lag 5, blocking frequency decreases by as much as 6.1% (not shown), consistent with the amplification of negative blocking anomalies shown during phase 4 (Fig. 2), since MJO phases average about 5 days apart. MJO phases 4 and 5 geopotential height anomaly composites (Fig. 3) also show a similar north–south dipole structure in the east Pacific associated with reduced east Pacific blocking. Using a general circulation model, Hinton et al. (2009) demonstrated that an anomalous dipole structure (similar to Fig. 4) coincides with faster cyclone speed along the region of increased westerlies. The increase in cyclone speed reduces the tendency of the cyclones to locally advect subtropical low potential vorticity (PV) air poleward that supports blocking onset and maintenance. This mechanism aids in the suppression of blocking, in agreement with Fig. 2.
b. MJO phase 7
Phase 7 is characterized by opposite sign heating as phase 3, with suppressed convection in the east Indian Ocean and enhanced convection in the west Pacific. Likewise, the teleconnection pattern in the Pacific associated with phase 7 (Fig. 6) is similar to that of phase 3 but with weakened anomalies of the opposite sign. Over the east Pacific, a significant positive geopotential height anomaly is observed to the north of 45°N for all pentads shown. The anticyclone is flanked by negative anomalies to the south and west, suggesting that the climatological geopotential height gradient is weakened with anomalously weak westerly flow between the geopotential height anomalies, which is near the climatological location of the storm track (e.g., Pelly and Hoskins 2003). The weakened gradient coincides with a significant increase in blocking frequency observed for all subsequent lags of MJO phase 7 (e.g., see Fig. 12). At lag 0, an east Pacific blocking frequency increase of +7% is observed for phase 7 (Fig. 2). The anomalous blocking frequency is highest for lags 10 and 15, reaching +8.6% and +9.3%, respectively (not shown). With a DJF climatological blocking frequency peaking near 7%, Fig. 2 suggests that blocking frequency is approximately doubled during and following phase 7. Out of the 419 total east Pacific blocking days of the 31 winter seasons, approximately 15.3% (lag 0) to 17.2% (lag 15) occur during and following MJO phase 7.
In addition to a weakened geopotential height gradient, Hinton et al. (2009) demonstrated that a dipole pattern such as that observed in the east Pacific during phase 7 coincides with reduced cyclone speed along the weaker westerlies near the climatological location of the storm track, thereby increasing the tendency for weather systems to deposit low PV air from the subtropics into higher latitudes. The poleward flux of low PV air from storms in the east Pacific would aid in the formation and persistence of blocks during and following phase 7.
Our results agree in some respects with HK14. While differences are expected due to the different methods used, our east Pacific results at lag 0 agree fairly well with HK14 for MJO phases 3–5. However, the increase in blocking associated with phase 7 is largely absent in their study. Their blocking frequency calculation suggests that the increase in blocking during phase 7 occurs only west of 150°W. In addition to using a constant blocking latitude, the blocking index of HK14 also imposes no persistence or large-scale requirements, which may result in a different mean blocking frequency from ours.
5. Atlantic (90°–20°W)
In the Atlantic sector, a significant blocking frequency increase occurs during and/or after MJO phases 6–8, while a significant decrease is shown during and/or after phases 1–5 (e.g., see Fig. 12 below). To most clearly demonstrate the MJO’s dynamical influence on Atlantic blocking, we have chosen to focus on MJO phases 3 and 7, each of which captures the sign of the blocking response of the other MJO phases, respectively [similar to what was done by Lin et al. (2009) for the NAO pattern]. The blocking frequency anomalies for lags 5, 10, and 15 are shown in Fig. 7 for phase 3 (top row) and phase 7 (bottom row).
a. MJO phase 3
During and following MJO phase 3 (Fig. 4, pentads 0 and 1), significant high geopotential height anomalies exist over eastern North America with negative geopotential height to the northeast, indicating a strengthened climatological geopotential height gradient. The dipole coincides with the suppressed high-latitude blocking west of Greenland observed soon after phase 3 (Fig. 7, first panel), which peaks in magnitude at a value of −9.4%. The Rossby wave train (Fig. 3) strengthens over the Atlantic after one pentad and becomes more robust after two pentads, with a pattern that develops reminiscent of the positive phase of the NAO (e.g., Lin et al. 2009 ). The NAO pattern indicates that the climatological pressure gradient is strengthened, with increased westerlies between 45° and 55°N and a more zonal flow. The positive NAO coincides with significantly decreased blocking frequency in the midlatitudes and the high latitudes (Fig. 7). With a DJF climatological mean blocking frequency peaking near 17%, Fig. 7 implies that following phase 3 the probability of blocking over the North Atlantic is more than halved, dropping to a 5%–7% total blocking frequency 10–15 days after phase 3, with some regions reaching blocking probabilities near zero.
A significant decrease in high-latitude blocking frequency occurs as early as lag 5 (Fig. 7, first panel), which is prior to the development of the positive NAO anomalies (Fig. 4), suggesting it may not be the NAO pattern that creates unfavorable conditions for high-latitude blocking. Previous studies have suggested that the pattern associated with a positive NAO is actually a result of infrequent high-latitude blocking (Woollings et al. 2008; Davini et al. 2012), indicating that the accumulated suppression of high-latitude blocking after phase 3 (Fig. 7) may cause the positive NAO pattern in Fig. 4. The NAO-blocking cause and effect relationship is difficult to diagnose and is beyond the scope of this study.
The MJO has been suggested to influence the NAO pattern by directly forcing the tropospheric Rossby wave teleconnection to the Atlantic as shown in Fig. 4 (e.g., Lin et al. 2009; Cassou 2008), as well as by a more indirect influence through the stratosphere. Garfinkel et al. (2012) proposed that MJO-induced poleward and vertically propagating Rossby waves modify the polar vortex, which in turn impact the NAO; a mechanism previously suggested for the ENSO–NAO link (Bell et al. 2009; Ineson and Scaife 2009). The downward coupling of the stratospheric polar vortex and the tropospheric NAO pattern is a longer pathway of MJO influence with time scales up to one month, with MJO phase 3 preceding a cooling of the polar lower stratosphere and consequently a more positive NAO by 10–25 days after phase 3 (Garfinkel et al. 2014). Given the longer time scale of the stratospheric pathway, it is possible that North Atlantic blocking can be influenced at time scales longer than what are discussed here.
b. MJO phase 7
The teleconnection pattern associated with phase 7 (Fig. 6) results in opposite-signed anomalies compared to phase 3, with an Atlantic pattern indicative of a negative NAO. During a negative NAO, a weakened climatological pressure gradient occurs in the Atlantic with higher-amplitude wavelike flow (e.g., Shabbar et al. 2001; Woollings et al. 2008). Following phase 7, the negative NAO coincides with an increase in Atlantic blocking frequency (Fig. 7, bottom row), reaching +16.5% at lag 15, an almost doubling of the mean blocking frequency. Of the 922 total Atlantic blocked days in the 31 winter seasons, approximately 14%–15% occur within 5–15 days after MJO phase 7.
MJO phase 7 has been shown to influence the negative NAO pattern via a tropospheric pathway as shown in Fig. 6 (e.g., Lin et al. 2009), as well as through the stratosphere. Garfinkel et al. (2014) found that the MJO-induced low pressure over the North Pacific (Fig. 6, pentad 0) corresponds to greater meridional tropospheric and stratospheric heat flux and a weakening of the polar vortex, and consequently a more negative NAO 10–25 days following MJO phase 7.
As a comparison to our results, the instantaneous 1D blocking index of HK14 shows a decrease in Atlantic blocked days during MJO phases 4–6, whereas we find a significant decrease only during phase 5, lag 0, north of 50°N (not shown), which may also be partially reflected in the lags following phase 3 (Fig. 7). HK14 also shows an increase in blocked days during MJO phase 8, in agreement with our results (see Fig. 12 below).
6. Europe (20°W–45°E)
The greatest reduction in European blocking is shown to occur during and/or following MJO phases 3–5, while a significant blocking increase exists for MJO phases 6–8. Here we have chosen to focus on phases 4 and 6 to represent these two groups, as these phases demonstrate the greatest and most persistent changes in blocking frequency (Fig. 8). It is worth noting that MJO phase 8 demonstrates a similar increase in blocking frequency at lag 0 (not shown) as the phase 6 composites during lags 10 and 15 (Fig. 8, bottom row). Phase 8 coincides with the time frame of the phase-6 lag-10–15-day composites and therefore a similar physical mechanism can be argued. Phase 7 also shows a significant increase in blocking frequency at lag 0, although it is confined to west of the Prime Meridian (not shown).
a. MJO phase 4
During and following MJO phase 4 (Fig. 3, only pentad 0 shown), the Euro-Atlantic region consists of the positive NAO pattern that is initiated following phase 3. Figure 3 demonstrates a significant positive geopotential height anomaly over Europe with a negative anomaly to the northwest during phase 4. The dipole straddles the climatological location of the storm track (e.g., Pelly and Hoskins 2003) and indicates a strengthened climatological gradient. The dipole is associated with a significant decrease in blocking frequency, with blocking frequency anomalies of approximately −9.5% five days after phase 4 (Fig. 8, top row). This corresponds to a frequency reduction of more than two-thirds of the climatology. The decrease continues through lag 10 east of the Prime Meridian, with an anomalous blocking frequency reaching −8.5%.
In agreement, HK14 and Cassou (2008) show reduced blocking during comparable MJO phases. HK14 found a higher-amplitude decrease, peaking at 15%–20% below the DJF mean during phases 4–5. The higher frequency in HK14 is expected as they included all instantaneously blocked days with no large-scale requirements. A greater decrease is found by Cassou (2008), who found up to a 30% decrease in Scandinavian blocking occurrence relative to climatology following phase 4. Cassou (2008) defined each day in their study as belonging to one of the four Atlantic regimes (one of which is Scandanavian blocking) even if the flow is weakly correlated to the assigned regime, resulting in higher blocking frequencies.
b. MJO phase 6
An increase in blocking is observed following phase 6 as early as lag 5, with a significant blocking frequency increase of 8%–9% for lags 10–15 (Fig. 8, bottom row). Approximately 8.3% (lag 5) to 11.4% (lag 15) of all 973 European blocked days follow phase 6. These results are in agreement with those of HK14 who found an increase in instantaneous blocking during MJO phase 7, which can be interpreted as corresponding to a lagged phase 6. Our results also show an increase during phase 7, although weaker than phase 6 (not shown). A significant relationship between the MJO and European blocking at such early lags following phase 6 is also presented by Cassou (2008).
While phase composites of upper-level streamfunction and geopotential height demonstrate a clear wave train during other phases (e.g., phase 3), phase 6 composites (Fig. 5) portray a confusing evolution of anomalies leading to a strong dipole over the Euro-Atlantic region resembling a negative NAO pattern. The wave activity flux vectors do not portray a clear Rossby wave pathway, suggesting two possibilities: 1) The Euro-Atlantic anomalies develop independent of the MJO and the timing is coincidental, or 2) the MJO link in the composites is more complex than in other regions and phases.
An initial examination of the phase 6 composites (Fig. 5) suggests the first possibility, particularly since the increase in European blocking frequency is observed as early as 5 days after phase 6 (Fig. 8). This provides only a short time for a Rossby wave induced by phase 6 MJO heating in the west Pacific to reach the Euro-Atlantic region. Cassou (2008) proposed that the increase in blocking following phase 6 demonstrates the tendency for the North Atlantic to follow a preferred set of regimes as discussed by Vautard (1990). The argument states that the increase in blocking is an indirect consequence of the positive NAO pattern previously set up by MJO phase 3. We test this by taking all phase 3 days and calculating the European blocking frequency forward in time, with the idea that if the only role of the MJO is to force the positive NAO, the high frequency of blocking observed following phase 6 should be reproducible with phase 3 data at a large lag. We generated European blocking composites every five days for phase 3 lags 10–40 (not shown). However, the blocking increase of Fig. 8 could not be reproduced. This suggests that in order for increased blocking to coincide with MJO phase 6, the MJO may play a role beyond that suggested by Cassou (2008).
The connection between MJO and European blocking is further investigated by taking the 227 days that lag phase 6 by 5 days and separating them into two categories: European blocking (NB = 81 days) and no European blocking (146 days). Similar results were obtained with lag 10, and hence only lag 5 is presented. Daily composites are shown in Figs. 9 and 10 for the blocking and no-blocking cases, respectively, with the purpose of uncovering key differences that may suggest an MJO-European blocking link. Since blocking appears soon after phase 6, we examine the days preceding MJO phase 6, where phase 6 is indicated by lag 0.
Four days prior to phase 6 in the blocking composites (Fig. 9; lag −4) there is a Rossby wave train propagating north over the North Pacific and east over North America. The wave activity flux vectors follow the direction of the Rossby wave packets, which appear to display energy dispersion toward the Atlantic high latitudes as well as southeast toward the anticyclonic center over the east coast of the United States. The latter resembles a negative PNA pattern, which is clearly absent in the no-blocking composites (Fig. 10). By lag 0 (Fig. 9, bottom-left panel), Rossby wave energy propagates from the large minimum over North America toward the European blocking, with vectors depicting the eastward Rossby pathway over Greenland. The negative PNA pattern provides a link from the Pacific to the Euro-Atlantic region prior to phase 6 and serves as a possible explanation for the increase in European blocking that follows phase 6. Previous studies have demonstrated that the MJO can act as a trigger for the PNA pattern (e.g., Mori and Watanabe 2008; Franzke et al. 2011). Given that both the blocking and no-blocking composites are associated with strong MJO events, it is worth investigating the days preceding the negative PNA pattern to examine why the pattern is absent in the no-blocking composites.
Figure 9 demonstrates that MJO enhanced convection is initially concentrated over the Bay of Bengal (lags −24 to −20 days) and is associated with a wave train that develops within the subtropical jet (plotted in red). An anticyclonic anomaly forms poleward of the jet exit region, which acts as a source of Rossby wave energy, as indicated by the divergence of W vectors, for the wave train that develops across the Pacific in lag −12. The anticyclone shifts east and grows north of the jet by lag −4, becoming part of the negative PNA pattern. This sequence of events is in agreement with the PNA life cycle suggested by Mori and Watanabe (2008), who argued that PNA growth is initiated over the subtropical jet exit region triggered by MJO-induced divergence north of the Bay of Bengal. In contrast, Fig. 10 does not follow this life cycle. Lags −24 to −20 demonstrate weaker OLR anomalies in the tropics relative to Fig. 9, suggesting that on average the no-blocking case is preceded by weaker early MJO phase heating. Although the composite difference in OLR between the blocking and no-blocking cases was not found to be statistically significant based on a difference of means test, MJO heating strength and structure in the early phases of an MJO event may still be important for the PNA life cycle and in turn, European blocking. Furthermore, a zonal wind average for the 30 days preceding phase 6 for the blocking and no-blocking cases (not shown) demonstrate that the subtropical jet in the blocking case is shifted equatorward over the central Pacific relative to the no-blocking case. This difference suggests a possible change in the basic state that may influence the pathway of the MJO-induced anomalies and necessitates further investigation.
Figures 9 and 10 also demonstrate a second key difference between blocking and no-blocking cases: a preexisting anticyclone over the Euro-Atlantic region exists at all lags preceding European blocking. In fact, the composite structure of Fig. 9 suggests that the anticyclone shifts northeast toward Europe, eventually becoming the blocking anticyclone as shown by an asterisk in each panel. The asterisks represent the maxima in Atlantic daily streamfunction (where the maximum closest to the previous day’s maximum is marked), and tracing this feature demonstrates the evolution of the anomalous anticyclone. The maxima’s daily trajectory is shown in Fig. 11 from 12 days prior to 5 days after phase 6, demonstrating the transition followed by stagnation of the anticyclone. The anticyclone strength is found to be significantly different from the no-blocking composites at the 90% level based on a two-tailed difference of means test prior to and after transition, although at certain lags the anticyclone is weakened and does not pass the difference of means test. This transition may be indicative of the positive NAO to blocking transition suggested by Cassou (2008). It is possible that the negative PNA pattern and the preexisting Euro-Atlantic anticyclone are necessary precursors for the MJO-blocking link suggested by Fig. 9.
Once blocking is established, it is also possible that MJO anomalies could project onto the negative PNA pattern, thereby enhancing or maintaining the large minimum over North America that provides a Rossby wave energy pathway for the European blocking. It would be interesting to examine the impacts of the PNA pattern on the persistence of phase 6 European blocking. Further analysis is required to better understand these mechanism and relationships.
7. Summary and conclusions
Utilizing a 2D blocking index, we examined the statistical relationship between the MJO and boreal winter blocking over the Pacific, Atlantic, and European regions. Although the MJO-blocking relationship is not discussed in detail in HK14, the results presented here expand on the blocking composite shown in their study, which utilized a 1D index along a constant blocking latitude to demonstrate instantaneous blocking relative to MJO phase. Blocking composites were generated for eight phases of the MJO as defined by Wheeler and Hendon (2004), and statistically significant relationships were found for all regions. The blocking frequency associated with each MJO phase is summarized in Fig. 12, where the 2D blocking frequency calculated above is averaged between 40° and 60°N to highlight the mean midlatitude changes in blocking frequency. The DJF climatological mean midlatitude blocking frequency is shown in each MJO phase panel in black, and the lag 0–15-day average blocking frequency associated with each MJO phase is portrayed from cooler to warmer colors, respectively, with statistically significant regions marked with an asterisk. The main findings of this study are summarized below.
MJO heating is associated with significant circulation anomalies throughout the North Pacific. These anomalies were found to influence blocking in the west-central Pacific region during all phases, particularly in the high latitudes. Our results contrast those of Moore et al. (2010), whose study demonstrated an increase in Pacific blocking frequency coinciding with the eastward-propagating anticyclone during MJO phases 2–5. In the east Pacific, a doubling in blocking frequency follows MJO phase 7. The teleconnection results in a dipole pattern that weakens the climatological gradient and the westerly flow near the climatological location of the storm track, where cyclones slow down as they move poleward depositing low PV air. The favorable blocking conditions can be observed in the geopotential height composites of phase 7, which demonstrate a persistent anticyclone out to a 15-day lag. Approximately 15.3% of all east Pacific blocks coincide with phase 7. A stronger, opposite-signed wave train occurs following phase 3, resulting in a strengthened climatological gradient over the east Pacific and a decrease in blocking frequency.
The teleconnection patterns associated with phases 3 and 7 are also important for Atlantic blocking. In agreement with Lin et al. (2009) and Cassou (2008), 10–15 days after MJO phase 3 a positive NAO pattern develops over the Atlantic. We find that Atlantic blocking frequency is more than halved in association with the positive NAO pattern. In contrast, MJO phase 7 is followed by a negative NAO pattern, which coincides with a high-amplitude wavelike flow and an increase in blocking frequency. Atlantic blocking frequency is almost doubled following phase 7, reaching +16.5% relative to climatology. Approximately 14%–15% of all DJF Atlantic blocked days follow phase 7.
In Europe, a significant blocking increase follows phase 6, with up to 11.4% of all DJF European blocked days occurring within 15 days of MJO phase 6. We identify two possible precursors to the phase 6 blocking: 1) a preexisting anticyclone over the Atlantic and 2) a negative PNA pattern. The first precursor was discussed by Cassou (2008), who noted the positive NAO pattern from MJO phase 3 could provide the anticyclone that later would transition into the European block. While Cassou (2008) hypothesized the phase 6 European blocking is primarily a result of phase 3-driven NAO variability, we propose a second precursor, a negative PNA pattern, as a link between the MJO and European blocking. In agreement with the daily PNA life cycle discussed by Mori and Watanabe (2008), we argue that through forcing of the negative PNA by the MJO, European blocking coincides with phase 6. The negative PNA pattern then acts as a Rossby wave energy pathway influencing the European blocking, as indicated by the W vectors.
Future work should focus on further understanding the mechanisms between the MJO and blocking, in addition to the PNA pattern. More work is needed to understand how changes in MJO heating and the basic state influence the relationships observed between the MJO and blocking. This is particularly true for Europe, in which MJO heating differences and the basic state may play a role in the development of the negative PNA pattern preceding phase 6 blocking. The implications for general circulation model (GCM) simulations of blocking should also be examined, since GCMs tend to poorly simulate both the MJO (e.g., Zhang et al. 2006) and blocking (e.g., Masato et al. 2013b). Additionally, a two-way interaction between blocking and the MJO necessitates investigation. Blocking alters the flow impacting the extratropical waveguide, and in turn, extratropical waves may influence MJO convection and therefore future MJO Rossby wave energy dispersion (e.g., Roundy 2011).
We are grateful to three anonymous reviewers, whose insight greatly improved the quality of this manuscript. This work was supported by the Climate and Large-Scale Dynamics Program of the National Science Foundation under Grants AGS-1441916 and AGS-1025584, and the Science and Technology Center for Multi-Scale Modeling of Atmospheric Processes, managed by Colorado State University under Cooperative Agreement ATM-0425247. EAB was supported, in part, by the Climate and Large-scale Dynamics Program of the National Science Foundation under Grant 1419818. The statements, findings, conclusions, and recommendations do not necessarily reflect the views of NSF.