Abstract

Trajectory analysis using a weather prediction model is performed for five cases to interpret the formation of enhanced bands of vertically integrated water vapor (IWV) in the central and eastern Pacific that are frequently seen in satellite images from the Special Sensor Microwave Imager. The connection of these enhanced bands with poleward water vapor transport from the Tropics is also examined. It is shown that the leading end of the enhanced IWV bands (defined as the most eastward and poleward end) is the manifestation of moisture convergence in the warm conveyor belt associated with extratropical cyclones, while the bands away from the leading end result mainly from moisture convergence along the trailing cold fronts. There is evidence that some enhanced IWV bands may be associated with a direct poleward transport of tropical moisture along the IWV bands from the Tropics all the way to the extratropics. The trajectory analysis, together with the seasonal mean sea level pressure analysis, indicates that a favorable condition for the occurrence of a direct, along-IWV band transport of tropical (defined as south of 23.5°N) moisture to the U.S. West Coast in the eastern Pacific is a weakened subtropical ridge in the central Pacific with an enhanced southwesterly low-level flow. The authors hypothesize that the direct poleward transport of tropical moisture within an enhanced IWV band in the eastern Pacific is most possible in the neutral El Niño–Southern Oscillation (ENSO) phase and is least possible in the El Niño phase.

1. Introduction

During the wintertime, satellite integrated water vapor (IWV) images from the Special Sensor Microwave Imager (SSM/I) over the central and eastern Pacific frequently show narrow enhanced bands of IWV (where the magnitude of the IWV is greater than 2.0 cm). These IWV bands, as the local maximum of vertically integrated moisture, are typically on the order of 200 km in width and several thousands of kilometers in length (e.g., Ralph et al. 2004). Extensive upper- and midtropospheric cloud cover coexists with the leading end of the IWV bands (defined as the most eastward and poleward end of the bands) such that the latter can be easily identified even from infrared satellite imagery (e.g., McGuirk et al. 1987). A close examination of the enhanced IWV bands (Ralph et al. 2004) indicates that they evolve and move with the development and transience of extratropical cyclones, suggesting that the IWV bands are dynamically related to the water vapor transport by extratropical cyclones.

A fundamental question that emerged from our attempt to understand the role of the IWV bands in water vapor transport is whether these bands originate solely from local convergence of preexistent moisture, or from a long-distance transport of moisture, or from both. In fact, in some cases of landfalling extratropical cyclones that produced heavy precipitation on the U.S. West Coast, the IWV bands extend far enough south that it appears possible that tropical (defined as south of 23.5°N) moisture is directly transported poleward along the IWV bands and contributes to the precipitation on the coast (e.g., Ralph et al. 2004). However, based only on the satellite images, it is not clear how the IWV bands are related to water vapor transport between the Tropics and extratropics, and whether or not the landfalling cyclones ever directly transport the tropical moisture along the IWV bands to produce heavy precipitation on the U.S. West Coast. Therefore, it remains a question as to where the moisture sources are for the precipitation produced by landfalling extratropical cyclones on the U.S. West Coast during wintertime.

Diagnosis of operational analyses of global meteorological observations indicates that there exist transient filaments of large vertically integrated horizontal water vapor fluxes corresponding to the IWV bands, which are termed “atmospheric rivers” by Newell et al. (1992). Despite the name, the atmospheric rivers in general do not represent trajectories associated with water vapor transport because they correspond to vertically integrated, transient streamlines (Wernli 1997). Therefore, snapshots of the so-called atmospheric rivers cannot distinguish whether or not the filaments of large integrated horizontal water vapor fluxes originate from local moisture convergence or from a long-distance, river-like moisture transport. On the other hand, the integrated horizontal water vapor fluxes corresponding to the mean atmospheric flow patterns, such as seasonally averaged ones, are a good representation of time-averaged water vapor transport because they change slowly with time. The seasonally averaged mean and transient vertically integrated horizontal fluxes of water vapor by the atmospheric flow over the Pacific Ocean during the wintertime are characterized by 1) a westward transport throughout the Tropics and an eastward transport in the midlatitudes associated with the dominant zonal winds, and 2) a poleward transport from the Tropics by extratropical cyclones along the storm track (e.g., James 1994, chapter 7, section 5). However, this seasonally averaged picture does not provide details about the connection between the observed, individual IWV bands and the poleward water vapor transport from the Tropics.

Wintertime extreme precipitation events on the U.S. West Coast are caused by landfalling extratropical cyclones. Various observational analyses (e.g., Higgins et al. 2000 and references therein) indicate that tropical climate variation strongly affects the intensity and location of the cyclonic activity in the extratropics, leading to the interannual variability in the poleward transport of tropical moisture. Accordingly, there is interannual variation of extreme precipitation events on the U.S. West Coast. The major, if not the largest, tropical climate variation is associated with the El Niño–Southern Oscillation (ENSO) cycle. The tropical influence on the extratropics related to the ENSO cycle can be understood in terms of teleconnections epitomized by Rossby wave trains propagating away from the anomalous convective heating due to anomalous sea surface temperature changes in the Tropics (e.g., James 1994, chapter 8, section 2). In this way, the ENSO events create circulation anomalies over the midlatitude Pacific, among other regions of the globe. In particular, when compared with its climatological position, the Pacific storm track often extends eastward during warm episodes and retracts westward during cold episodes. One major consequence of the circulation anomalies is that the storm track in the central and eastern Pacific undergoes a north–south displacement from one winter to another, leading to a great variation in annual precipitation and flood frequency along the U.S. West Coast corresponding to the ENSO cycle (e.g., Cayan et al. 1999; Andrews et al. 2004). Because of the apparent strong linkage between the IWV bands and poleward moisture transport from the Tropics, it is natural to wonder whether the relationship between the IWV bands and poleward transport of tropical moisture varies with the ENSO cycle, and how such a variation is linked with the possibility of extreme precipitation events on the U.S. West Coast.

The purpose of this study is to address the following three questions: 1) How are the enhanced satellite IWV bands formed? 2) Is it possible for there to be a direct poleward transport of tropical moisture to the U.S. West Coast in the eastern Pacific along an IWV band stretching from the coast to the Tropics? 3) Is it possible that the connection between the IWV bands and the direct transport of tropical moisture in the eastern Pacific is modulated by the ENSO cycle? These questions are not about overall poleward transport of moist air originating in the Tropics. Rather, they are concerned with whether tropical moist air can be transported poleward within the IWV band from the Tropics all the way to the extratropics. By answering these questions, insight is gained into the hydrological link between tropical climate variation and moisture sources of wintertime precipitation on the U.S. West Coast and the connection between the occurrence of extreme precipitation events on the U.S. West Coast and the direct transport of tropical moisture from the eastern Pacific along IWV bands from the Tropics all the way to the extratropics.

Numerical simulations of five cases using the fifth-generation Pennsylvania State University–National Center for Atmospheric Research (NCAR) Mesoscale Model (MM5; Grell et al. 1994) are carried out in this study, along with observational analysis, to interpret the connection between the evolution of the enhanced satellite IWV bands and the moisture transport associated with extratropical cyclones, and to reveal the possible modulation of the connection by the ENSO cycle. Specifically, trajectory analysis is performed using the model output to determine whether or not air parcels at low levels move along the enhanced IWV bands as in a river-like transport. Then, the National Centers for Environmental Prediction (NCEP)–NCAR reanalysis (Kalnay et al. 1996) is used to interpret how water vapor transport as revealed by the trajectory analysis is controlled by the large-scale background flow, and thus how it may be affected by the ENSO cycle.

The remainder of the paper is arranged as follows: the case selection and the configuration of the numerical model are summarized in section 2; the results of trajectory analysis based on model simulations are presented in section 3; a hypothesis on the possible modulation of the trajectory analysis results by the ENSO cycle is provided in section 4; and the summary and discussion of the conclusions appear in section 5.

2. Selected cases and the weather prediction model

Five cases of landfalling cyclones have been selected for this study because they all resulted in heavy precipitation on the U.S. West Coast and had great societal impacts. The first two cases (15–16 February 2004 and 15–16 October 2003) shed light on how the IWV bands evolve with the water vapor transport on various scales. The third case (1–5 January 1997) exemplifies the possibility of heavy precipitation in California being fed by a direct, along-IWV band transport of tropical moisture in the eastern Pacific. The last two cases (2–3 February 1998 and 3–5 March 2001), along with the third case, will be used together with the NCEP–NCAR reanalysis to show that there may be interannual variation associated with the ENSO cycle in the connection between the IWV bands and the direct transport of tropical moisture in the eastern Pacific.

Trajectory analysis is carried out to physically interpret the connection between the poleward water vapor transport and the evolution of the IWV bands. The four-dimensional datasets required for the trajectory analysis are provided by numerical simulations using MM5. The initial and boundary conditions for the model are prescribed using both NCEP and European Centre for Medium-Range Weather Forecasts analyses. The simulations using the initial and boundary conditions specified by the two analyses are qualitatively similar in the evolution of the synoptic flow. The results shown here are from the simulations initialized with the NCEP analyses. The model domain is made of 200 × 200 horizontal grid points with 36-km grid spacing and 50 vertical levels and covers the central and northeast Pacific, with the southern boundary extending to the Tropics (Fig. 1). The configuration of MM5 physics used in this study includes the Reisner-1 mixed-phase cloud microphysics, the Grell convective scheme, the Burk–Thompson 1.5-order planetary boundary layer scheme, and the MM5 simple shortwave and longwave radiation parameterization schemes.

Fig. 1.

Geographical coverage of the 36-km 200 × 200 grid.

Fig. 1.

Geographical coverage of the 36-km 200 × 200 grid.

3. Interpretation of trajectory analysis

This section presents the interpretation of the evolution of observed satellite IWV bands using trajectory analysis based on model simulations of the five cases mentioned in section 2. In particular, the trajectory analysis is applied to illustrate the connection between the evolution of the observed satellite IWV bands and poleward transport of tropical moisture along the bands. An apparent dependence of such a connection on the ENSO cycle is also revealed by the case studies.

a. Trajectory analysis procedure

Trajectory analysis has proven to be an effective tool to depict the Lagrangian transport by airflows (e.g., Reap 1972; Wernli 1997; Eckhardt et al. 2004). In this study, hourly MM5 output is used to perform backward trajectory analysis in conjunction with the application of the well-established conceptual model of three-dimensional airflows associated with extratropical cyclones (e.g., Browning 1986; Wernli and Davies 1997; Schultz 2001). The question to be addressed by the trajectory analysis is whether it is possible for any of the trajectories released along the U.S. West Coast within the IWV band to have moved along the IWV bands from the Tropics all the way to the coast. All trajectories are released at 1 km above the surface along the coast and within the simulated IWV bands when the frontal rainband makes landfall, and the number of the trajectories moving along the IWV bands from the Tropics all the way to the U.S. West Coast is examined.

The reason that the air parcels are released at 1 km above the surface is that our model simulations indicate that before reaching the coast, the leading end of the simulated IWV bands corresponds to the low-level jet in the convex poleward boundary of the warm conveyor belt, which is defined as the warm and moist slantwise- ascending flow ahead of the cold front (Browning 1986). On average, before the warm conveyor belt ascends over the warm front, the moisture content is highest in the lowest kilometer above the surface, and the transport of this low-level moisture is closely associated with the low-level jet. The level of 1 km above the surface is near the altitude of the greatest horizontal moisture flux associated with the low-level jet that is within the warm conveyor belt (and thus the corresponding IWV band) before its slantwise ascent (Ralph et al. 2004). It is also the altitude critical to the moisture convergence associated with orographic precipitation enhancement in the coastal mountains of the western United States (Neiman et al. 2002). Significant moisture is still present within the IWV band between the low-level jet and the middle troposphere, and it plays a significant role in the production of frontally forced precipitation through a deep layer (e.g., Browning 1986). The moisture between the low-level jet and the middle troposphere is mostly the result of the rearward ascent of moist air in the warm conveyor belt that is forced by the cold front. During the rearward ascent, the warm, moist air in the warm conveyor belt ahead of the surface cold front is lifted rapidly to about 2–3 km above the surface and then undergoes a much slower slantwise ascent (Browning 1986). It takes a short time (on the order of 1–2 h) for the moisture within the warm conveyor belt at the lowest kilometer above the surface to be transported to the middle troposphere, which is fast in comparison with the time scale of horizontal transport along the warm conveyor belt (on the order of 24 h).

The above concept of moisture transport within the warm conveyor belt supports the notion that the moisture between the low-level jet and the middle troposphere comes from the lowest kilometer above the surface, and that this moisture very likely originates in the same geographical area as that in the IWV band at 1 km above the surface. To further verify this notion, the trajectories released at 1 km above the surface at the time when the leading end of the IWV band makes landfall are compared with trajectories released at levels above and below 1 km above the surface and at various times before and after the landfall. Results from this comparison confirm that those trajectories released at 1 km above the surface at the time when the leading end of the IWV band makes landfall are representative of the origin of the moisture feeding the precipitation along the warm conveyor belt. Thus, in the following discussion of the trajectory analysis results we only show representative samples of the trajectories that are released within the IWV band (where the magnitude of the IWV is greater than 2.0 cm) at 1 km above the surface at the time when the leading end of the IWV band makes landfall.

b. Evolution of IWV bands and associated water vapor transport

Figure 2 shows two consecutive daily SSM/I images for the 15–16 February 2004 case that produced flooding rain in California’s coastal mountains (up to ∼250 mm in 2 days). The IWV band in this case extends from the California coast to the central Pacific and is linked to the IWV filament that extends northeastward from the tropical IWV reservoir in the western Pacific. An animated loop of the satellite images for this case indicates that the evolution of the IWV band is consistent with the average wintertime water vapor transport in the Pacific depicted by James (1994, chapter 7, section 5). According to James, the transport of water vapor by the zonal winds is dominant in the Pacific during wintertime, with a strong westward transport throughout the Tropics and an eastward transport in the midlatitudes. The maxima of the eastward and poleward transport of water vapor in the midlatitudes are associated with the storm track; the total horizontal moisture flux is roughly parallel to the storm-track axis. The loop of satellite images also shows that the IWV bands in the Northern Hemisphere propagate eastward and poleward such that the tropical moisture appears to flow from the western Pacific to the eastern Pacific along the IWV bands.

Fig. 2.

SSM/I composite image of IWV (cm) on (a) 15 Feb 2004 and (b) 16 Feb 2004.

Fig. 2.

SSM/I composite image of IWV (cm) on (a) 15 Feb 2004 and (b) 16 Feb 2004.

Figures 3a and 3b are two snapshots of the model-simulated IWV (color-shaded background), sea level pressure (contours), and the backward trajectory locations at 12 (Fig. 3a) and 36 h (Fig. 3b) into the simulation. It is obvious that in this case the air parcels that are within the IWV band when reaching the coast have not always been within the IWV band. Instead, they originate from both the cold and warm sides of the cold front and move into the warm conveyor belt, and thus the IWV band, because of convergence. The trajectories of all the air parcels, except for parcel 5, start from locations in the warm sector with lower IWV and slightly descend (as indicated by the decrease in the size of the arrowheads) before they enter into the IWV band in the warm conveyor belt at low levels as the cyclone progresses. Air parcel 5 originates from the cold sector close to the cold front and descends into the IWV band. The divergence field at the lowest model level (Fig. 3c) indicates that the enhanced IWV band coincides with a low-level convergence band that is associated with the cold front. This low-level convergence band results in significant moisture convergence. Keeping in mind the well-established conceptual model of three-dimensional airflows associated with extratropical cyclones over the ocean (see the discussion in section 3a), all of the above reveal that the IWV band is the result of moisture convergence along the cold front, and the leading end of the IWV band is closely associated with the warm conveyor belt rising over the warm front.

Fig. 3.

The positions of backward trajectories valid at (a) 12 h (1200 UTC 15 Feb 2004) and (b) 36 h (1200 UTC 16 Feb 2004) into the simulation and (c) the convergence areas at the surface (i.e., the lowest model level) valid at 36 h into the simulation (1200 UTC 16 Feb 2004) at 1 km above the surface. The color-shaded background in all of the panels is the simulated IWV. The solid black contours in (a) and (b) are sea level pressure with a contour interval of 4 mb (1 mb − 1 hPa), while in (c) they are 0 and −1 × 10−5 s−1 isopleths of the divergence field. The parcels for the backward trajectory analysis are released at 36 h into the model simulation (1200 UTC 16 Feb 2004) at 1 km above the surface. The arrowheads indicate the trajectory locations at each hour, and their size (as indicated by the legend) represents the elevation of air parcels, with bigger arrowheads corresponding to higher elevations.

Fig. 3.

The positions of backward trajectories valid at (a) 12 h (1200 UTC 15 Feb 2004) and (b) 36 h (1200 UTC 16 Feb 2004) into the simulation and (c) the convergence areas at the surface (i.e., the lowest model level) valid at 36 h into the simulation (1200 UTC 16 Feb 2004) at 1 km above the surface. The color-shaded background in all of the panels is the simulated IWV. The solid black contours in (a) and (b) are sea level pressure with a contour interval of 4 mb (1 mb − 1 hPa), while in (c) they are 0 and −1 × 10−5 s−1 isopleths of the divergence field. The parcels for the backward trajectory analysis are released at 36 h into the model simulation (1200 UTC 16 Feb 2004) at 1 km above the surface. The arrowheads indicate the trajectory locations at each hour, and their size (as indicated by the legend) represents the elevation of air parcels, with bigger arrowheads corresponding to higher elevations.

Figure 4 shows two sequential daily SSM/I images for another case, which occurred on 15–16 October 2003 and produced extreme flooding in Washington State and British Columbia, Canada. As in the previous case, the IWV band extends from the tropical western Pacific to the U.S. West Coast along the Northern Hemispherical storm track, such that the tropical moisture appears to be flowing from the western Pacific to the eastern Pacific along the IWV band.

Fig. 4.

SSM/I composite image of IWV (cm) on (a) 15 Oct 2003 and (b) 16 Oct 2003.

Fig. 4.

SSM/I composite image of IWV (cm) on (a) 15 Oct 2003 and (b) 16 Oct 2003.

The model-simulated IWV-band evolution and trajectory locations within approximately the same 24-h period as shown in Fig. 4 are presented in Figs. 5a and 5b. It is clear that in this case the air parcels that are within the IWV band when they reach the coast are not exclusively transported along the IWV band. They move along the band only after they descend and converge into the IWV band in the warm conveyor belt. The trajectories of air parcels 2, 5, 8, and 9 start from the area under the influence of a subtropical high. These air parcels initially move westward and later turn and move northeastward as they are caught by the pre-cold-frontal southwesterly flow within the warm conveyor belt and converge into the IWV band. Air parcels 6, 7, and 11 originate at lower levels and stay within the IWV band all the way to the coast. The trajectories of the rest of the air parcels (1, 3, 4, and 12) are from the upper levels and follow the descending flow behind the cold front. An examination of the model-simulated low-level divergence field (Fig. 5c) indicates that, as in the previous case, the IWV band results from moisture convergence associated with the warm conveyor belt and cold front, and the band moves and extends eastward because the warm conveyor belt moves with the cyclone. The part of the IWV band that is away from the leading end is the result of moisture convergence along the trailing cold front.

Fig. 5.

Same as in Fig. 3, but for the 16 Oct 2003 case valid at (a) 36 h (0000 UTC 17 Oct 2003), and (b) and (c) 60 h (0000 UTC 18 Oct 2003) into the simulation. The parcels for the backward trajectory analysis are released at 60 h into the simulation (0000 UTC 18 Oct 2003) at 1 km above the surface.

Fig. 5.

Same as in Fig. 3, but for the 16 Oct 2003 case valid at (a) 36 h (0000 UTC 17 Oct 2003), and (b) and (c) 60 h (0000 UTC 18 Oct 2003) into the simulation. The parcels for the backward trajectory analysis are released at 60 h into the simulation (0000 UTC 18 Oct 2003) at 1 km above the surface.

In summary, the model simulations of the 15–16 February 2004 and 15–16 October 2003 cases indicate that the evolution of the satellite-enhanced IWV bands is directly linked with water vapor transport by airflows associated with extratropical cyclones; this transport appears to be superimposed on the planetary-scale water vapor transport by the mean latitudinal winds along the storm track. According to the trajectory analysis based on the model simulations, although the IWV bands appear to move and extend eastward when the satellite images are looped, their leading ends are actually the manifestation of moisture convergence associated with the warm conveyor belts that move with extratropical cyclones. Ralph et al. (2004) show that the mesoscale characteristics of the IWV band are indeed those that are typical for the low-level jet within the warm conveyor belt (see also Browning and Pardoe 1973; Browning 1986; Wernli and Davies 1997; Schultz 2001): a few hundreds of kilometers wide and 1–2 km deep with warm, moist air that is subject to rearward slantwise ascent where the warm air moves up over the cold front, and forward-sloping ascent where the warm air remains within the conveyor belt, eventually rising over the warm front. The moisture convergence along the trailing cold front leads to the formation of the rest of the IWV band away from the leading end, which is often linked westward with the leading end of another IWV band.

c. Direct poleward transport of tropical moisture in the eastern Pacific

The above study of the 15–16 February 2004 and 15–16 October 2003 cases reveals that the enhanced satellite IWV bands over the central and eastern Pacific are the direct result of moisture convergence associated with extratropical cyclones, and thus are linked with water vapor transport by extratropical cyclones across the Pacific. Such transport, as a part of the atmospheric general circulation, occurs as successive frontal systems associated with extratropical cyclones pass through the moisture reservoir in the tropical western Pacific and transport the tropical moisture toward the northeastern Pacific as the cyclones advance. It is observed, however, that sometimes a single IWV band extends from the U.S. West Coast southward into the tropical eastern Pacific. This gives rise to the question of whether it is possible for the tropical moisture in the eastern Pacific to be directly transported poleward along the IWV band toward the U.S. West Coast by a single cyclone. We have found that such a possibility exists. Here we show an example by using the MM5 simulation of the 1–5 January 1997 case, which occurred during the near-neutral phase of the ENSO cycle with a multivariate ENSO index (MEI) value of −0.45 and produced heavy rainfall along the U.S. West Coast, particularly along the California coast and the Sierra Nevada (see the analysis presented in Ralph et al. 1998). Figure 6 shows the enhanced satellite IWV band associated with the case in which the IWV band associated with a landfalling extratropical cyclone extends from the southern California coast to the tropical eastern Pacific.

Fig. 6.

SSM/I composite satellite image of IWV (cm) on 2 Jan 1997.

Fig. 6.

SSM/I composite satellite image of IWV (cm) on 2 Jan 1997.

Figures 7a and 7b present the model-simulated IWV (color-shaded background), sea level pressure (contours), and trajectory locations at 72 and 96 h into the simulation, respectively, for this case. It is seen that the trajectories of air parcels 6, 7, 10, 15, 16, and 19 originated in the Tropics and end near the U.S. West Coast, where the heavy rainfall took place, indicating a direct transport of moisture from the high IWV region in the Tropics all the way to the U.S. West Coast along the IWV band. Figure 7c shows the divergence field at the lowest model level along with the IWV field valid at the same time as Fig. 7b. The fact that the IWV band coincides with the low-level convergence indicates that the simulated IWV band is formed mainly by the convergence associated with the warm conveyor belt and the cold front, even though there is a direct transport of tropical moisture. Thus, in this case the IWV band is associated with a direct transport of tropical moisture as well as the convergence of the subtropical and extratropical moisture in the eastern Pacific.

Fig. 7.

Same as in Figs. 3a–c, but for the 1–5 Jan 1997 case valid at (a) 72 h (0000 UTC 31 Dec 1996), and (b) and (c) 96 h (0000 UTC 1 Jan 1997) into the simulation. The parcels for the backward trajectory analysis are released at 96 h into the model simulation (0000 UTC 1 Jan 1997) at 1 km above the surface.

Fig. 7.

Same as in Figs. 3a–c, but for the 1–5 Jan 1997 case valid at (a) 72 h (0000 UTC 31 Dec 1996), and (b) and (c) 96 h (0000 UTC 1 Jan 1997) into the simulation. The parcels for the backward trajectory analysis are released at 96 h into the model simulation (0000 UTC 1 Jan 1997) at 1 km above the surface.

Figure 8 depicts time series of vertical displacement, hourly precipitation, and the water vapor mixing ratio of air parcels originating in the Tropics (i.e., air parcels 6, 7, 10, 15, 16, and 19). It indicates that in this case tropical moisture is indeed transported to the coast along the IWV band without much depletion via precipitation. The noticeable descent (beginning at 30 h into the simulation) along the trajectories of the air parcels originating in the Tropics (see Fig. 8a) indicates that these air parcels do not come from the near-surface tropical atmosphere where there is abundant moisture. Additionally, some of the air parcels (9, and 12–14) reaching the U.S. West Coast originated in the extratropics. Nevertheless, all the air parcels from both sides of the cold front merge into the IWV band. It is seen in these figures that although there is precipitation along the trajectories of the air parcels as they move within the IWV band, the overall amount of the precipitation along these trajectories is light when the air parcels are away from the coast. More importantly, the water vapor mixing ratio of air parcels 6, 7, 10, and 15 at the ending time of the trajectories is not significantly different from that at the beginning time (see Fig. 8c), while the moisture in air parcels 16 and 19 at the ending time is significantly greater than that at the beginning time, indicating that the initial moisture cannot be depleted entirely and at least some of the moisture acquired by the air parcels in the Tropics is carried all the way to the coast along the IWV band.

Fig. 8.

Vertical displacement in terms of (a) pressure (mb), (b) hourly precipitation (mm), and (c) water vapor mixing ratio (g kg−1) along the trajectories of the air parcels originating in the Tropics.

Fig. 8.

Vertical displacement in terms of (a) pressure (mb), (b) hourly precipitation (mm), and (c) water vapor mixing ratio (g kg−1) along the trajectories of the air parcels originating in the Tropics.

The fact that the moisture in air parcels 16 and 19 at the ending time is significantly greater than that at the beginning time (Fig. 8c) also indicates that these parcels gain a significant amount of moisture mainly due to their turbulent mixing with ambient air when the parcels move into the high IWV band, since the water vapor mixing ratio in a parcel is conserved in the absence of phase change and turbulent mixing. It is interesting to note that on average precipitating storms on the U.S. West Coast feed on the moisture already in the subtropical and extratropical atmosphere that is transported on the scale of 1000 km by the warm conveyor belt ahead of cold fronts (e.g., Trenberth 1998). However, in this case, the tropical moisture is transported up to 3000–4000 km along the IWV band and contributes to the precipitation on the coast.

d. Variability of direct transport of tropical moisture in the eastern Pacific

The previous case study shows that direct transport of moisture from the Tropics into extratropical cyclones can occur along an IWV band during the near-neutral phase of the ENSO cycle. This finding raises an interesting question: Does the possibility of a landfalling extratropical cyclone directly transporting the moisture in the high IWV region in the eastern Pacific along an IWV band vary with the ENSO cycle? To begin answering this question, we examine two more cases: 2–3 February 1998, which occurred during a strong El Niño phase of the ENSO cycle with an MEI value of 2.57, and 3–5 March 2001, which occurred during a weak La Niña phase of the ENSO cycle with an MEI value of −0.58. Heavy rainfall along the California coast was produced during both cases. In particular, in the El Niño case, flood damage resulted from the heavy rainfall (see Neiman et al. 2004). Figures 9 and 10 are the SSM/I images for the two cases, which show that the IWV bands morphologically connect the Tropics and the coastal area. The possibility of the extratropical cyclones in these two cases directly transporting the moisture in the tropical high IWV region along the IWV bands is investigated by examining whether or not air parcels along the backward trajectories released within the IWV bands in the extratropics originate in the tropical high IWV region and move along the IWV bands.

Fig. 9.

SSM/I composite satellite images of IWV (cm) on 2 Feb 1998.

Fig. 9.

SSM/I composite satellite images of IWV (cm) on 2 Feb 1998.

Fig. 10.

SSM/I Composite satellite images of IWV (cm) on 4 Mar 2001.

Fig. 10.

SSM/I Composite satellite images of IWV (cm) on 4 Mar 2001.

The model-simulated IWV-band evolution and trajectory positions within a 24-h period for the El Niño case are shown in Fig. 11. The air parcels are released for the backward trajectory analysis at 1200 UTC 3 February 1998 (60 h into the simulation). The trajectories of air parcels released from the coast do not always stay within the IWV band. Instead, they originate from both the cold and warm sides of the cold front and converge into the warm conveyor belt. As in the previous cases, although the leading end of the IWV band is associated with the warm conveyor belt rising over the warm front, the rest of the IWV band is the result of moisture convergence along the cold front (not shown). There is no long-distance water vapor transport along the IWV band. It should be pointed out that the position of the IWV band is transient and the background of the trajectory figures are snapshots of the IWV band at a particular time. Only the last position of each trajectory in the trajectory figures corresponds to the time at which the IWV field is valid. Thus, the trajectories of air parcels 9 and 10 only appear to be within the IWV band in Fig. 11a because of the fact that the paths of the two trajectories coincide with the position of the IWV band at the time the IWV field is valid, but these parcels do not always move within the IWV band. The trajectories of air parcels 4–6, 8, and 20 descend within the subtropical high (i.e., in the area to the south of the cold front with lower values of IWV) during the earlier time of the 24-h period and then converge into the warm conveyor belt as the cold front advances southeastward. The trajectories of air parcels 9, 10, 18, and 22 have their origins at lower levels than 4–6, 8, and 20, and they enter into the warm conveyor belt earlier than 4–6, 8, and 20. The air parcels coming from north of the cold front also descend as they move toward the coast, following the dry descending flow behind the cold front.

Fig. 11.

Positions of trajectories for the 2–3 Feb 1998 case valid at (a) 36 h (1200 UTC 2 Feb 1998) and (b) 60 h (1200 UTC 3 Feb 1998) into the simulation. The color-shaded background is the simulated IWV, and the solid black contours are sea level pressure with a contour interval of 4 mb. The parcels for the backward trajectory analysis are released at 60 h into the model simulation (1200 UTC 3 Feb 1998) at 1 km above the surface.

Fig. 11.

Positions of trajectories for the 2–3 Feb 1998 case valid at (a) 36 h (1200 UTC 2 Feb 1998) and (b) 60 h (1200 UTC 3 Feb 1998) into the simulation. The color-shaded background is the simulated IWV, and the solid black contours are sea level pressure with a contour interval of 4 mb. The parcels for the backward trajectory analysis are released at 60 h into the model simulation (1200 UTC 3 Feb 1998) at 1 km above the surface.

Figure 12 shows the simulated evolution of the IWV band and trajectory movement within a 24-h period for the La Niña case (3–5 March 2001). The air parcels are released for the backward trajectory analysis at 72 h into the simulation (0000 UTC 4 March 2001). Again in this case, none of the trajectories originated in the Tropics, indicating that the tropical moisture is not directly transported into the cyclone along the IWV band. It is interesting to note that because of the cyclone’s weak intensity and slow movement, the parcels whose backward trajectories originate in the warm sector do not travel a great distance. The area of high IWV near the southern California coast is an indication of the occlusion, while the narrow IWV band to the west of the cyclone is the result of moisture convergence associated with the trailing cold front. This clearly indicates that in this case the observed IWV band is formed by moisture convergence. To confirm that the narrow IWV band is indeed due to local moisture convergence and not to a river-like transport of tropical moisture, two additional trajectories (15 and 16) are released from the cold front near the point of 30°N, 130°W at 1 km above the surface, and as seen in Fig. 12, they do not always stay within the IWV band.

Fig. 12.

Same as in Fig. 11, except for the 3–5 Mar 2001 case valid at (a) 48 h (0000 UTC 3 Mar 2001) and (b) 72 h (0000 UTC 4 Mar 2001) into the simulation. The parcels for the backward trajectory analysis are released at 72 h into the model simulation (0000 UTC 4 Mar 2001) at 1 km above the surface.

Fig. 12.

Same as in Fig. 11, except for the 3–5 Mar 2001 case valid at (a) 48 h (0000 UTC 3 Mar 2001) and (b) 72 h (0000 UTC 4 Mar 2001) into the simulation. The parcels for the backward trajectory analysis are released at 72 h into the model simulation (0000 UTC 4 Mar 2001) at 1 km above the surface.

One important result from the previous trajectory analyses is that a direct poleward transport of moisture from the Tropics along the IWV band only occurred in the case of the neutral ENSO phase. Because of the natural dynamic linkage between the Tropics and extratropics, this result is unlikely a coincidence and suggests that the possibility of a direct poleward transport of tropical moisture along an IWV band is ENSO-cycle dependent.

4. Hypothesis of ENSO-cycle dependence

We hypothesize that the possibility of a direct poleward transport of tropical moisture occurring along an IWV band is strongly dependent on the ENSO cycle. In particular, it is the highest in the neutral ENSO phase and the lowest in the El Niño phase. This hypothesis is based on the fact that the large-scale displacement of the convection zones in the Tropics associated with the ENSO cycle not only changes the latitudinal Walker circulation, but also has an impact on the meridional Hadley circulation (James 1994, chapter 8). Consequently, the interaction between an individual extratropical cyclone approaching the U.S. West Coast and the Tropics depends on the strength of the Hadley circulation in the eastern Pacific. Whether or not the cyclone directly transports the tropical moisture over the eastern Pacific along the IWV band all the way to the coast is determined by whether or not the intensity of the descending branch of the Hadley circulation in the eastern subtropical Pacific is strong enough to block the southward extension of the warm conveyor belt into the Tropics. If the intensity of the descending branch of the Hadley circulation is weakened, it may allow the warm conveyor belt of a strong extratropical cyclone to connect with the IWV reservoir in the Tropics, resulting in the warm conveyor belt directly transporting tropical moisture along the IWV band all the way to the extratropics. Consequently, low-level tropical moisture over the eastern tropical Pacific is transported northeastward and contributes to the formation of the precipitation along the U.S. West Coast. On the other hand, if the descending branch of the Hadley circulation is enhanced in such a way that the cold front of an extratropical cyclone cannot penetrate it, then air parcels over the eastern tropical Pacific are prevented from being transported poleward by the extratropical cyclone. Therefore, through the influence on the intensity of the Hadley circulation, the ENSO cycle may have a direct impact on the poleward transport of water vapor from the eastern tropical Pacific to the U.S. West Coast.

The influence of the interannual variability in the Tropics associated with the ENSO oscillation (as well as the Madden–Julian oscillation) on the extratropical transient meridional transport has long been confirmed in previous analysis studies (see Matthew and Kiladis 1999a, b; Higgins et al. 2000; and references therein), all of which are in favor of our hypothesis. To further support our hypothesis, analysis of 3-month-averaged sea level pressure patterns associated with the last three cases (Figs. 13) is performed using the NCEP–NCAR reanalysis (Kalnay et al. 1996).

Fig. 13.

Seasonal mean sea level pressure (mb) for (a) December 1996–February 1997 (neutral ENSO), (b) December 1997–February 1998 (El Niño), and (c) January–March 2001 (La Niña). The NCEP reanalysis data used in this figure are provided by the National Oceanic and Atmospheric Administration–Cooperative Institute for Research in Environmental Sciences (NOAA–CIRES) Climate Diagnostics Center, Boulder, CO, from their Web site at http://www.cdc.noaa.gov/.

Fig. 13.

Seasonal mean sea level pressure (mb) for (a) December 1996–February 1997 (neutral ENSO), (b) December 1997–February 1998 (El Niño), and (c) January–March 2001 (La Niña). The NCEP reanalysis data used in this figure are provided by the National Oceanic and Atmospheric Administration–Cooperative Institute for Research in Environmental Sciences (NOAA–CIRES) Climate Diagnostics Center, Boulder, CO, from their Web site at http://www.cdc.noaa.gov/.

In the neutral ENSO case (Fig. 13a), the ridge associated with the descending branch of the Hadley circulation in the central Pacific is weakened, and the western fringe of the subtropical high does not extend westward beyond Hawaii. When an intense extratropical cyclone approaches the U.S. West Coast and slowly moves poleward, it is possible for the southwesterly low-level flow of the warm conveyor belt at the western fringe of the subtropical high to directly transport moisture from the Tropics to the extratropics. In the El Niño case (Fig. 13b), the ridge associated with the descending branch of the Hadley circulation is enhanced in the central Pacific in such a way that it makes the direct transport of tropical moisture by an extratropical cyclone in the eastern Pacific least possible. In the La Niña case (Fig. 13c), the ridge associated with the descending branch of the Hadley circulation is stronger than in the neutral ENSO phase, but weaker than in the El Niño phase, rendering an intermediate possibility of direct transport of tropical moisture to the U.S. West Coast by an extratropical cyclone in the eastern Pacific.

To further examine our hypothesis on the ENSO-dependent connection of the IWV band to tropical moisture, the wintertime mean sea level pressure composites of the El Niño, neutral ENSO, and La Niña phases from 1950 to 2000 (Fig. 14) are made according to the yearly ENSO classifications made by Compo et al. (2001), which are based on a sea surface temperature index time series and are consistent with those based on the MEI time series. The multiyear sea level pressure composites of the El Niño and the La Niña phases (Figs. 14b,c) are very similar to those for the El Niño phase of 1997/98 and the weak La Niña phase of early 2001 (Figs. 13b,c). That is, the ridge associated with the descending branch of the Hadley circulation is enhanced in the central Pacific in the El Niño phase, while the ridge in the La Niña phase is significantly weaker than in the El Niño phase. Although the multiyear sea level pressure composite of the neutral ENSO phase (Fig. 14a) shows that the ridge associated with the descending branch of the Hadley circulation appears to be a little stronger than that in the composite for the neutral ENSO of 1996/97 (Fig. 13a), the northern Pacific low in the multiyear composite is still stronger and the subtropical high is still weaker than those in the multiyear composite of the La Niña phase. All these further support the hypothesis derived from the individual case studies that a direct poleward transport of tropical moisture within an enhanced IWV band in the eastern Pacific is more possible in the neutral ENSO phase than in the La Niña phase.

Fig. 14.

Multiyear composite of seasonal mean sea level pressure (mb) from 1950 to 2001 for (a) neutral ENSO, (b) El Niño, and (c) La Niña. The years included in the multiyear composites are indicated above the color bar. The NCEP reanalysis data used in this figure are provided by the NOAA–CIRES Climate Diagnostics Center (http://www.cdc.noaa.gov/).

Fig. 14.

Multiyear composite of seasonal mean sea level pressure (mb) from 1950 to 2001 for (a) neutral ENSO, (b) El Niño, and (c) La Niña. The years included in the multiyear composites are indicated above the color bar. The NCEP reanalysis data used in this figure are provided by the NOAA–CIRES Climate Diagnostics Center (http://www.cdc.noaa.gov/).

Although this study has presented evidence to support our hypothesis, to ultimately test it requires running trajectory analysis similar to that carried out by Eckhardt et al. (2004) for all the winter seasons using the same NCEP–NCAR reanalysis from which Fig. 14 is made. Since this study is intended to address the interpretation of the IWV formation and its role in the moisture interaction between the Tropics and extratropics, and to reveal the possible interannual variability in the occurrence of tropical moisture transport along IWV bands, we leave further confirmation of our hypothesis for another ongoing study.

5. Summary and discussion

Trajectory analysis using a weather prediction model is performed for five cases to interpret the formation of the enhanced IWV bands in the central and eastern Pacific that are frequently seen in satellite SSM/I images and to address whether tropical moist air can be transported poleward within the IWV band from the Tropics all the way to the extratropics. It is shown that the model-aided trajectory analysis is an effective tool to provide a detailed picture of how low-level moisture over the eastern tropical Pacific can directly contribute to the heavy precipitation over the U.S. West Coast. It is important to note that such a detailed Lagrangian water vapor transport cannot be easily discerned from either seasonally averaged horizontal water vapor fluxes (as shown in James 1994) or the satellite SSM/I images. Two important conclusions have been obtained from the study of the five cases:

  1. Local moisture convergence is primarily responsible for the formation of the enhanced IWV bands, although there is evidence that some enhanced IWV bands are associated with direct poleward transport of tropical moisture.

  2. A direct poleward transport of tropical moisture in the eastern Pacific to the U.S. West Coast may occur along an IWV band, and a weakened subtropical ridge in the central Pacific with enhanced southwesterly low-level flow is a favorable condition for such a direct transport to occur.

The results from the study of the first two cases indicate that the IWV bands are formed within the convergence zone associated with the warm conveyor belt and the cold front of extratropical cyclones. The leading end of the IWV bands is the manifestation of moisture convergence associated with the warm conveyor belt rising over the warm front, while the bands away from the leading end are the result of moisture convergence associated with cold fronts. It has been shown that only in some cases (such as the 1–5 January 1997 case) does moisture flow within the IWV bands as in a river-like transport. This result is consistent with the trajectory analysis by Eckhardt et al. (2004, their Fig. 3) that showed only in a few cases over the 15-yr period did the airflow in the warm conveyor belt originate in the Tropics.

Results from the study of the remaining three cases suggest that the possibility of a direct, river-like transport along IWV bands appears to be dependent on the ENSO cycle. We hypothesize that the direct poleward transport of tropical moisture within an enhanced IWV band in the eastern Pacific is most possible in the neutral ENSO phase and is least possible in the El Niño phase. Our analysis of the ENSO-related interannual variability in the intensity of the descending branch of the Hadley circulation in the central and eastern Pacific, along with previous studies of the connection between the tropical interannual variability and the extratropical high-frequency transients (e.g., Matthew and Kiladis 1999a, b; Higgins et al. 2000), favors this hypothesis.

Our hypothesis has an implication for understanding the climate–weather interaction in the U.S. west coastal region. Other studies have indicated that El Niño winters typically bring the most rain (e.g., Cayan et al. 1999) and the greatest possibility of flooding to the U.S. West Coast (Andrews et al. 2004). Yet, according to our hypothesis, these same winters are least likely to have extratropical cyclones that directly transport the tropical moisture in the eastern tropical Pacific all the way to the U.S. West Coast along IWV bands. Instead, it is most likely for extratropical cyclones to transport the tropical moisture during the winters in the neutral phase of the ENSO cycle, which are typically drier than the El Niño winters. The fact that tropical moisture from the eastern Pacific is directly transported poleward along the IWV band all the way to the coast and contribute to the precipitation in the 1–5 January 1997 case suggests that the direct transport of tropical moisture can cause a single, very extreme precipitation event, while a season of a series of heavy precipitation events is not likely to be associated with direct transport of tropical moisture from the eastern Pacific.

Results from all the case studies suggest that one should be cautious when using the term “atmospheric river” to refer to the enhanced IWV bands associated with extratropical cyclones. While the term “river” conventionally refers to a transport along a fixed two-dimensional path, moisture transport associated with the IWV bands in all the case studies involves slantwise ascent within the warm conveyor belt. Such a three-dimensional feature of moisture transport is in agreement with all the previous results of observational and modeling studies of extratropical cyclones (Browning 1986; Wernli and Davies 1997; Schultz 2001; and references therein). Perhaps, “moisture conveyor belt” is a better term to refer to the moisture transport associated with the IWV bands because it appears to be more consistent with the well-established conveyor belt model of extratropical cyclones than the term atmospheric river.

Acknowledgments

The authors gratefully thank Gary Wick at the NOAA/Environmental Technology Laboratory for making the composite images of IWV, Jeff Whitaker at the NOAA/Climate Diagnostics Center (CDC) for providing information on how to graph NCEP–NCAR reanalysis data using the CDC’s online analysis tool, and three anonymous reviewers for their constructive comments.

REFERENCES

REFERENCES
Andrews
,
E. D.
,
R. C.
Antweiler
,
P. J.
Neiman
, and
F. M.
Ralph
,
2004
:
Influence of ENSO on flood frequency along the California coast.
J. Climate
,
17
,
337
348
.
Browning
,
K. A.
,
1986
:
Conceptual models of precipitation systems.
Wea. Forecasting
,
1
,
23
41
.
Browning
,
K. A.
, and
C. W.
Pardoe
,
1973
:
Structure of low-level jet streams ahead of mid-latitude cold fronts.
Quart. J. Roy. Meteor. Soc.
,
99
,
619
638
.
Cayan
,
D. R.
,
K. T.
Redmond
, and
L. G.
Riddle
,
1999
:
ENSO and hydrologic extremes in the western United States.
J. Climate
,
12
,
2881
2893
.
Compo
,
G. P.
,
P. D.
Sardeshmukh
, and
C.
Penland
,
2001
:
Changes of subseasonal variability associated with El Nino.
J. Climate
,
14
,
3356
3374
.
Eckhardt
,
S.
,
A.
Stohl
,
H.
Wernli
,
P.
James
,
C.
Forster
, and
N.
Spichtinger
,
2004
:
A 15-year climatology of warm conveyor belts.
J. Climate
,
17
,
218
237
.
Grell
,
G. A.
,
J.
Dudhia
, and
D. R.
Stauffer
,
1994
:
A description of the fifth-generation Penn State/NCAR Mesoscale Model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 122 pp
.
Higgins
,
R. W.
,
J-K. E.
Schemm
,
W.
Shi
, and
A.
Leetmaa
,
2000
:
Extreme precipitation events in the western United States related to tropical forcing.
J. Climate
,
13
,
793
820
.
James
,
I. N.
,
1994
:
Introduction to Circulating Atmospheres.
Cambridge University Press, 422 pp
.
Kalnay
,
E.
, and
Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
.
Matthew
,
A. J.
, and
G. N.
Kiladis
,
1999a
:
Interactions between ENSO, transient circulation, and tropical convection over the Pacific.
J. Climate
,
12
,
3062
3086
.
Matthew
,
A. J.
, and
G. N.
Kiladis
,
1999b
:
The tropical–extratropical interaction between high-frequency transients and the Madden–Julian oscillation.
Mon. Wea. Rev.
,
127
,
661
677
.
McGuirk
,
J. P.
,
A. H.
Thompson
, and
N. R.
Smith
,
1987
:
Moisture burst over the tropical Pacific Ocean.
Mon. Wea. Rev.
,
115
,
787
798
.
Neiman
,
P. J.
,
F. M.
Ralph
,
A. B.
White
,
D. E.
Kingsmill
, and
P. O. G.
Persson
,
2002
:
The statistical relationship between upslope flow and rainfall in California’s coastal mountains: Observations during CALJET.
Mon. Wea. Rev.
,
130
,
1468
1492
.
Neiman
,
P. J.
,
P. O. G.
Persson
,
F. M.
Ralph
,
D. P.
Jorgensen
,
A. B.
White
, and
D. E.
Kingsmill
,
2004
:
Modification of fronts and precipitation by coastal blocking during an intense landfalling winter storm in southern California: Observations during CALJET.
Mon. Wea. Rev.
,
132
,
242
273
.
Newell
,
R. E.
,
N. E.
Newell
,
Y.
Zhu
, and
C.
Scott
,
1992
:
Tropospheric rivers? A pilot study.
Geophys. Res. Lett.
,
19
,
2401
2404
.
Ralph
,
F. M.
,
P. J.
Neiman
,
P. O. G.
Persson
, and
J-W.
Bao
,
1998
:
Observations of California’s New Years Day storm of 1997. Preprints, Second Conf. on Coastal Atmospheric and Oceanic Prediction and Processes, Phoenix, AZ, Amer. Meteor. Soc., 219–224
.
Ralph
,
F. M.
,
P. J.
Neiman
, and
G. A.
Wick
,
2004
:
Satellite and CALJET aircraft observations of atmospheric rivers over the eastern North Pacific Ocean during the El Niño winter of 1997/98.
Mon. Wea. Rev.
,
132
,
1721
1745
.
Reap
,
R. M.
,
1972
:
An operational three-dimensional trajectory model.
J. Appl. Meteor.
,
11
,
1193
1202
.
Schultz
,
D. M.
,
2001
:
Reexamining the cold conveyor belt.
Mon. Wea. Rev.
,
129
,
2205
2225
.
Trenberth
,
K. E.
,
1998
:
Atmospheric moisture residence times and cycling: Implication for rainfall rates and climate change.
Climatic Change
,
39
,
667
694
.
Wernli
,
H.
,
1997
:
A Lagrangian-based analysis of extratropical cyclones. II. A detailed case study.
Quart. J. Roy. Meteor. Soc.
,
123
,
1677
1706
.
Wernli
,
H.
, and
H. C.
Davies
,
1997
:
A Lagrangian-based analysis of extratropical cyclones. I: The method and some applications.
Quart. J. Roy. Meteor. Soc.
,
123
,
467
489
.

Footnotes

Corresponding author address: Jian-Wen Bao, NOAA/Environmental Technology Laboratory, 325 Broadway, Boulder, CO 80305-3328. Email: jian-wen.bao@noaa.gov