Downstream modulation of North Pacific atmospheric river activity by East Asian cold surges



[1] An East Asian cold surge (EACS) is characterized by rapid advancement of a polar airmass toward the east coast of the Eurasian continent in boreal winter. Over the east coast of Asia, extratropical cyclogenesis and the amplitudes of atmospheric disturbances ranging from synoptic to subseasonal timescales are immediately enhanced as the cold air approaches. This study investigates for the first time the impact of these EACS-excited disturbances on the activity of atmospheric rivers (ARs) over the North Pacific. Applying a new AR detection algorithm to the NASA MERRA dataset, we show that the daily occurrence probability of ARs over the eastern North Pacific near the west coast of U.S. is effectively modulated by EACS. In particular, this downstream dynamical modulation goes through two distinct stages: during the period Day 0–3, where Day 0 corresponds to the time of the peak intensity of an EACS event, high-frequency (HF, <6-day) baroclinic disturbances developed over the western North Pacific and Gulf of Alaska lead to significant poleward moisture transport over these two regions, and during the period Day 4–6, intermediate-low frequency (IF-LF, >12-day) barotropic disturbances developed from the merging of high-frequency troughs increase the daily occurrence probability of ARs near the west coast of the U.S. by 50%, relative to the climatological value. The results reported here demonstrate the critical role of IF and LF disturbances in establishing the link between the predictability of EACS and that of the AR-related extreme precipitation events in the western U.S. in boreal winter.

1. Introduction

[2] An East Asian cold surge (EACS) is characterized by rapid advancement of cold and dry air from the interior lowlands of Siberia toward the east coast of Asia in boreal winter [Wang, 2005]. Accompanying EACS are abrupt temperature drops and severe snowstorms with major socioeconomic impacts in East and Southeast Asia [Chang, 2004]. Upon the arrival of the polar airmass at the east coast, intense atmospheric disturbances are generated over the East China Sea and the Sea of Japan [Kung and Chan, 1981]. Some of these disturbances, for example, high-frequency disturbances occurring on sub-weekly timescales and closely coupled to surface cyclogenesis, can induce heavy precipitation in southeastern China, Korea and Japan, and they tend to propagate northeastward into high latitudes of the north Pacific [Hanson and Long, 1985]. Lau and Lau [1984], as well as Lau and Li [1984] showed that in association with EACS events, high-frequency disturbances experience baroclinic growth and barotropic decay during the course of their poleward propagation, while low-frequency disturbances that occur on timescales greater than 5 days demonstrate equatorward propagation and more pronounced downstream development.

[3] Atmospheric rivers (ARs) are narrow moisture conveyor belts in the atmosphere that account for over 90% of the cross-latitude water vapor transport in the midlatitudes [Zhu and Newell, 1998]. In boreal winter, ARs are key sources of extreme precipitation and triggers of flooding along the west coast of the U.S., and they also significantly affect the snowpack across the Sierra-Nevada range [Guan et al., 2010; Ralph and Dettinger, 2011]. Climatologically, nearly twice the amount of precipitation falls when winter storms exhibit AR-like features [Neiman et al., 2008]. AR activity over the North Pacific is known to be modulated by atmospheric disturbances of various time and spatial scales. For example, there is a tendency for ARs to form in the warm sectors of mature extratropical cyclones [McGuirk et al., 1987; Ralph et al., 2004]. Bao et al. [2006] reported that ARs with significant entrainment of tropical moisture can also develop in the absence of extratropical cyclones. Ralph et al. [2011] showed that atmospheric phenomena over the North Pacific ranging from planetary-to mesoscale collectively contributed to the formation of an AR that produced heavy precipitation in the Pacific Northwest. A recent study by Guan et al. [2011] revealed significant modulation of AR activity along the California coast by the Madden-Julian Oscillation (MJO).

[4] Since an EACS is capable of exciting multi-scale disturbances across the North Pacific, it is scientifically interesting to identify its dynamical footprints in the downstream AR activity. This study aims at documenting and understanding this downstream modulation. It serves as a first step toward establishing the dynamical connection between extremes in a regional energy cycle (i.e., EACS) and extremes in the hydrological cycle of a remote area (i.e., eastern North Pacific AR).

2. Data and Methods

[5] The primary fields analyzed in this study include daily surface air temperature (SAT), precipitation rate, column integrated total water vapor (TWV), isobaric-surface winds, geopotential height and specific humidity, all of which were obtained from the NASA Modern Era Retrospective-Analysis for Research and Applications (MERRA) dataset. MERRA provides high quality representation of the hydrological cycle and covers the period 1979 to present [Rienecker et al., 2011]. The native MERRA grid has a spatial resolution of equation image° latitude by equation image° longitude with 72 hybrid model levels in the vertical. The high resolution of MERRA makes it possible to detect filamentary structures characteristic of ARs in the TWV field.

[6] A daily SAT index in October-March for the period 1979/80 to 2008/09 was first constructed by averaging daily SAT over East Asia (100°E–130°E, 20°N–40°N). Applying a 30-day highpass Lanczos filter [Duchon, 1979] with 203 weights to the SAT index, we identify extreme EACS events based on the criteria that local minima (relative to the previous day and the day that follows) in the daily anomaly of the index must be less than-3 standard deviations and successive minima (events) must be at least 5 days apart. Based on these criteria, a total of 42 extreme EACS events were identified, which is consistent with results based upon different EACS definitions [e.g., Park et al., 2011]. The dates on which local minima occurred correspond to the peaks of individual EACS events and are denoted as “Day 0” in the subsequent composite analysis.

[7] The algorithm of AR detection used in this study is based on TWV that has been demonstrated by Neiman et al. [2008] as an appropriate field for capturing AR activity. Specifically, a grid point is declared as an AR grid when the observed TWV value at the grid point (Qr) satisfies:

equation image

[8] In the above expressions, Qzmean denotes the zonal mean TWV along the latitude of the grid point under consideration and Qzmax is the maximum value of TWV found along that latitude. Similarly, Qmmean denotes the meridional average of TWV between 0° and 90°N along the longitude of the grid point and Qmmax is the maximum value of TWV found along that longitude between 0° and 90°N. A and B are two adjustable parameters and their values (A = 0.3, B = 1.0) are chosen to best capture filamentary structures of prominent ARs in the observational record. This detection algorithm differs from that of Zhu and Newell [1998] in terms of the formulation (zonal and meridional threshold versus zonal only threshold) and the variable of interest (TWV vs. moisture transport vector (MTV)). Three ARs that have been discussed by Ralph et al. [2011] were detected by this algorithm and plotted here in Figures 1 (top left), 1 (top right), and 1 (bottom left) in terms of TWV values at the AR grids. The basic features of these three ARs, including origin and horizontal extent, are consistent with those identified via a direct inspection of satellite images. Applying this algorithm to the daily TWV field of MERRA, we count the number of days in a single winter when AR is present at a grid point. Dividing this number by the number of days in a winter we get an estimate of the daily occurrence probability of ARs at that location. The 30-winter (DJF, 1979/80-2008/09) climatological distribution of this occurrence probability is given in Figure 1 (bottom right), where a band of enhanced probability extends from the tropical-subtropical western North Pacific into the midlatitudes of the eastern North Pacific. The mean probability of AR presence on any single day in winter is approximately 12% along the west coast of the U.S.

Figure 1.

Total Water Vapor (TWV) of three prominent ARs detected through our algorithm (unit: mm) that occurred on (top left) 09 January 2005, (top right) 07 November 2006, and (bottom left) 13 October 2009. (bottom right) the climatological distribution of the daily occurrence probability of ARs in boreal winter (in percentage), based upon data in 30 winters (DJF, 1979/80-2008/09).

[9] Daily composite anomalies and regression analysis are used throughout the study to illustrate the evolution of the large-scale circulation and hydrological fields over the North Pacific following the peak of extreme EACS events (i.e., Day 0s defined before). Lanczos filters with 203 weights are employed to isolate atmospheric disturbances occurring on different timescales. Statistical significances of most composite fields are derived using a Monte Carlo method based upon random sampling.

3. Results

3.1. Circulation and Moisture Transport Anomalies Over the North Pacific Following EACS

[10] The typical evolution of large-scale atmospheric disturbances over the North Pacific following an extreme EACS is illustrated in Figure 2 in terms of the composite maps of 500 mb geopotential height anomalies (color shading). On Day 0 (when the cold surge peaks), an intense trough (Trough I) is found over Japan (Figure 2a) together with a ridge downstream of it. This negative height anomaly is meridionally elongated indicating its high-frequency (sub-weekly timescales) nature. As Trough I propagates northeastward, it starts taking a more zonally-elongated form characteristic of low-frequency disturbances (Figures 2b2e). The development of Trough I is directly related to the enhancement of surface baroclinicity along the east coast of Asia due to the arrival of cold air [Boyle, 1986; Neiman et al., 2008]. Starting from Day 1, a second trough (Trough II) appears in Gulf of Alaska ahead of the ridge. After continued intensification on day 2 and 3 (Figures 2c and 2d), Trough II becomes zonally-elongated on Day 4 (Figure 2e) and merges with Trough I to form a negative height anomaly occupying the region north of 50°N extending from Northeast Asia to western North America. This planetary-scale negative height anomaly persists for 12 days after the peak of the EACS event and starts expanding into lower latitudes of the eastern North Pacific on day 4.

Figure 2.

Composite 500 mb geopotential height anomalies (color shading, unit: m) and moisture transport vector (MTV) anomalies following the peak of the EACS (Day 0). (a–i) Day 0 to Day 8, respectively. The anomalous MTVs are rescaled to unit magnitude to emphasize the direction of transport. Only MTV anomalies significant at the 90% level are plotted.

[11] Also shown in Figure 2 are the composite maps of anomalous moisture transport vectors (MTVs) defined as the vertical integral (1000 mb to 300 mb) of isobaric-surface moisture fluxes. Immediately following the peak of the EACS on Day 0, a significant amount of moisture is transported from the tropical-subtropical western Pacific into Northeast Asia ahead of Trough I in the warm sector of the corresponding surface cyclone (Figures 2a and 2b). This northward transport of moisture is responsible for most of the severe weather and flooding that take place in Korea and Japan after EACS [e.g., Chen et al., 2002]. During the period Day 2 to Day 4, enhanced poleward moisture transport is found over Gulf of Alaska on the eastern edge of the intensifying Trough II (Figures 2c2e). On Day 4, a narrow band of significant moisture transport emerges between 40°N and 50°N. This band brings moisture of tropical-subtropical origin into the west coast of the United States. The formation of this MTV anomaly is directly associated with the anomalous southwesterly ahead of the persistent planetary-scale height anomaly that eventually expands into midlatitude eastern Pacific after Day 4. This MTV anomaly is visible until Day 12 (not shown).

3.2. Modulation of the Eastern North Pacific AR Activity and Coastal Precipitation

[12] The moisture transport anomalies shown in Figure 2 suggest potential modulation of downstream AR activity by EACS. In fact, the 500 mb height anomalies during the period Day 4 to Day 6 closely resemble the composite trough-ridge pattern characteristic of the occurrence of an AR near the U.S. west coast [Neiman et al., 2008]. To quantify the effect of EACS on the eastern Pacific AR activity, we plot in Figure 3a the composite anomalies of the daily occurrence probability of ARs along 124°W. Around the time the EACS peaks (Day 0 and 1), the AR activity along 124°W is suppressed. Following this relatively quiet phase is a significant increase in the chance of AR occurrence south of 50°N between Day 2 and Day 8. The transition from a suppressed AR phase to an active one is also clearly visible on a latitude-longitude map (Figures 3c and 3d) where a statistically significant decrease of AR occurrence probability at 40°N near northern California during Day 0 to Day 2 (Figure 3c) changes into an increase during Day 4 to Day 6 (Figure 3d). The increase of the probability here is clearly due to the tendency of AR formation in association with the organized band of moisture transport shown in Figures 2e2i. The magnitude of the change of the probability is on the order of 6%. It seems small; however, given the climatological value of approximately 12% near the west coast of U.S., 6% amounts to a 50% change. The phase transition is equivalent to a 100% jump in the chance of AR occurrence with respect to climatology. Corresponding to the phase transition of AR activity is the increase of precipitation along 124°W starting from Day 2 (Figure 3b). It is worth pointing out that the 3 mm per day increase in rainfall would translate into a much higher increase over land given the orographic enhancement of precipitation by the coastal range. Another important feature in Figures 3c and 3d is the broad area over western North Pacific that has exhibited an increase of AR occurrence probability on the order of 20%.

Figure 3.

(a) Composite anomalies of the daily occurrence probability of ARs along 124°W (in percentage); (b) composite anomalies of the precipitation rate along 124°W (unit: mm/day); (c) composite anomalies of the daily occurrence probability of ARs averaged over Day 0 to Day 2 (in percentage); (d) same as Figure 3c, but for average over Day 4 to Day 6. Areas enclosed by solid contours are significant at the 90% level.

3.3. Nature of the Atmospheric Disturbances Responsible for the AR Modulation

[13] EACS excites atmospheric disturbances of various time and spatial scales over the North Pacific. To understand the nature of the disturbances that are primarily responsible for the formation of ARs, we decompose the composite 500 mb height anomalies discussed before into three frequency bands: high-frequency (HF, 2–6 days), intermediate-frequency (IF, 12–25 days), and low-frequency (LF, 30–90 days). The time gaps between different bands are added here to minimize signal contamination due to the imperfection of the digital filter [Matthews and Kiladis, 1999]. To quantify the relative contribution of disturbances in a specific frequency band to the total disturbance field (i.e., the composite 500 mb height anomaly), we calculate the coefficients of projection [Michel and Rivière, 2011] onto the daily total disturbance field by the daily disturbance fields in the three frequency bands defined above. Specifically, the coefficient of projection Pn can be expressed as,

equation image

where n identifies different frequency bands; h′ is the composite height anomalies shown in Figure 2; hn represents the height anomalies in an individual frequency band; λ and ϕ are longitude and latitude, respectively. This projection calculation is done for two regions representing the western North Pacific (20°N–70°N, 120°E–160°E) and the eastern North Pacific (20°N–70°N, 160°W–120°W).

[14] In Figure 4, the projection coefficients are shown as a function of time (days since the peak of EACS). Over the western North Pacific (Figure 4a), the initial disturbance field is dominated on day 0 by HF disturbances corresponding to Trough I discussed in 3.1. The HF contribution declines after Day 0 and the projection becomes negligible (<0.1) on Day 3. IF (LF) disturbances become more important than HF disturbances starting from Day 1 (Day 3). IF and LF disturbances remain the main contributors to the total disturbance field until Day 10. This result is consistent with those reported by Compo et al. [1999], where they found that the IF band represents the largest portion of variance in the SLP anomalies over the western North Pacific and it is largely attributable to EACS events.

Figure 4.

Coefficients of projection onto the total disturbance field by disturbances in HF, IF and LF bands (detailed definition of the frequency bands is provided in 3.3). The evolution of the projection coefficients following the peak of the EACS (Day 0) is plotted in (a) for the western North Pacific and in (b) for the eastern North Pacific (detailed definition of the regions is provided in 3.3).

[15] Over the eastern North Pacific (Figure 4b), IF disturbances are the primary contributor to the total disturbance field throughout the entire period following the peak of the EACS. Contribution from the LF band catches up after Day 9 and HF disturbances play a minimal role in forming the circulation anomaly in this region. This result confirms the significance of the southward expanding, slowly-evolving negative height anomaly discussed in 3.1 in creating a large-scale background that favors AR formation near the U.S. west coast. Inspection of the three dimensional structure of this height disturbance reveals no vertical tilts indicating the barotropic nature of the disturbance. This is in contrast with Trough I over the western North Pacific and Trough II over Gulf of Alaska, both of which are well-defined high-frequency baroclinic disturbances.

4. Conclusions

[16] The modulation of the North Pacific AR activity by extreme EACS events is investigated in this study with a focus on uncovering the nature of the atmospheric disturbances responsible for the modulation. Winter climatology of the daily occurrence probability of ARs is first derived utilizing a new algorithm of AR detection in the TWV field. Within four days of the peak of the EACS, high-frequency baroclinic disturbances represented by two troughs generate intense poleward moisture transport over the western North Pacific and Gulf of Alaska. However, the development (the two roughs become zonally elongated and start merging) and southward expansion of an IF-LF barotropic disturbance is the primary factor making the eastern Pacific near the west coast of the U.S. a favorable place for AR formation up to twelve days past the peak of the EACS. Following the peak of the cold surge, the west coast of U.S. (40°N–50°N) experiences a transition from 50% less likely to have an AR (relative to the climatological condition) to 50% more likely to have an AR within two to three days. This transition from a suppressed AR phase to an active one is also clearly reflected in the near coast precipitation field. The results reported here demonstrate a close link between the predictability of the EACS and that of the AR-related extreme precipitation events in the western U.S. in boreal winter. A reliable projection of future AR activity along the U.S. west coast by a climate model depends on an accurate representation of IF and LF disturbances in the model- both the ones over the Eurasian continent that trigger an EACS and those that develop following an EACS over the North Pacific.


[17] The authors would like to thank Brad Hegyi and Eric Parker for their helpful suggestions during the drafting process. The MERRA data used in this study were provided through the NASA GSFC. This research was supported by the NASA Energy and Water Cycle Study (NEWS) under grant NNX09AJ36G.

[18] The Editor thanks the two anonymous reviewers for their assistance in evaluating this paper.