1. Introduction
Mesoscale convective complexes (MCCs) are characterized by both their long lifetimes (>6 h) and the large quasi-circular cloud shields they produce when observed at infrared wavelengths by satellite (−32°C contiguous area >104 km2; Maddox 1980). These complexes commonly occur over land in the lee of major mountain ranges and in association with low-level jets (LLJs; Stensrud 1996a), and make significant contributions to local and global hydrologic budgets (Laing and Fritsch 1997). Over South America, a high incidence of MCCs occurs in southeastern South America (SESA) during the warm season (Velasco and Fritsch 1987). Zipser et al. (2006) further suggest that SESA has the most intense thunderstorms in South America, making this area an important region for the study of deep convection.
It is well known that North American MCCs develop within conditionally unstable environments in association with strong low-level warm advection, a LLJ, a weak midlevel short-wave trough, and a quasi-stationary frontal boundary (Maddox 1983). Over SESA, the Andes Mountains play a large role in producing the mean low-level northerly flow that is observed throughout much of the year (Byerle and Paegle 2002; Campetella and Vera 2002). Often embedded in this northerly flow is a LLJ (Marengo et al. 2004). This South American low-level jet (SALLJ) is characterized by a poleward penetration of the northerly wind, promoting moisture and heat transport from the Amazon region into SESA (Saulo et al. 2004). Numerous studies show that the SALLJ is important to the hydrologic cycle in the La Plata basin of SESA (James and Anderson 1984; Rasmusson and Mo 1996; Marengo et al. 2004). Owing to this favorable mean low-level flow and advection pattern, it is not surprising that MCCs are a common occurrence in this region.
A more common type of organized convective region, containing a contiguous precipitation region of at least 100 km in horizontal extent, is the mesoscale convective system (MCS) of which the larger MCC is a subset. Over SESA, Nicolini et al. (2002) find a high correlation between the presence of MCSs and the SALLJ, with 81% of the 27 precipitating MCSs investigated occurring during SALLJ events. Salio et al. (2007) further find that 81% of the spring MCSs and 67% of the summer MCSs develop during SALLJ events. Using regional modeling experiments, Saulo et al. (2007) explore the linkages between LLJs and organized convection. Their results show that the LLJ generates moisture convergence and warm advection, thereby providing a favorable environment for the triggering of deep convection. The subsequent latent heat release from the convection reinforces the convergence, helping to extend the life of the MCS. Using satellite, radar, radiosonde, and surface observations in combination with selected regional modeling experiments over the southwest Amazon basin, Silva Dias et al. (2002) show that only a few deep and intense tropical convective cells are necessary to explain the overall convective line formation and that the production of multiple convective lines may be related to discrete cell propagation and their coupling with upper-atmosphere circulations.
Examining an important subset of all MCSs over SESA, Anabor et al. (2008) document 10 long-lived serial MCS events (lifetimes >18 h) over South America using satellite data and National Centers for Environmental Prediction–National Center for Atmospheric Research (NCEP–NCAR) reanalysis data. They show that a series of individual MCSs in the prefrontal region develop upstream relative to the flow at all levels of the troposphere and move away from the frontal boundary to produce these long-lived events. These serial MCSs form within an environment of strong low-level warm advection and strong moisture advection from the Amazon region southward, a situation very reminiscent of the environments of North American MCCs (Maddox 1983). A strong surface anticyclone off the Brazilian coast influences the low-level flow during the early stages of the event. The MCSs develop on the anticyclonic side of the entrance region to an upper-level jet as also seen in South American MCS environments by Salio et al. (2007). Unfortunately, the use of the low-resolution NCEP–NCAR reanalysis data to document the environments for the upstream-propagating MCS events in Anabor et al. (2008) does not allow for the investigation of the physical mechanisms that produce the upstream propagation of the convective region.
The upstream propagation of convective regions over the United States is shown to be a fairly regular occurrence by Porter et al. (1955). An examination of a single long-lived upstream-propagating convective event by Stensrud and Fritsch (1993, 1994) shows that these southward burst systems propagate upstream with respect to the flow at all levels of the atmosphere as also documented over South America by Anabor et al. (2008). This upstream propagation occurs via a combination of cold pool propagation and internal gravity waves that move ahead of the convective line and initiate new convective development. Thus, while the individual MCSs move downstream, the convective region as a whole propagates upstream (Stensrud and Fritsch 1993).
The importance of MCSs in SESA is emphasized by noting that the rainfall from MCSs contributes nearly 90% of the total rainfall in the La Plata basin (Nesbitt et al. 2006). This result strongly suggests that these systems are very important to both the local climate and human populations. Berbery and Barros (2002) indicate that ∼80% of the total precipitation in the La Plata river basin occurs in the austral warm season (October–April) when the main concentration of MCSs is observed over SESA (Velasco and Fritsch 1987; Laing and Fritsch 2000; Zipser et al. 2006). The SESA region also produces about 70% of the total combined gross national product of Brazil, Uruguay, Argentina, Paraguay, and Bolivia, and houses about half of their combined population. Thus, the SESA region has a large impact on energy production, water resources, agriculture, and livestock in South America (Vera et al. 2006), suggesting that the accurate prediction of MCS activity in this region would be very beneficial to a number of users of weather information.
The aim of this study is to investigate the relationship between the SALLJ and the upstream-propagating convective region as documented by Anabor et al. (2008) and to analyze the physical mechanisms for upstream propagation. Section 2 contains a description of the numerical model, while the methods used to create the initial and boundary conditions are discussed in section 3. Results are presented in section 4 followed by an examination of the mechanisms of upstream propagation. A final discussion is found in section 6.
2. Model description
The nonhydrostatic Advanced Research version of the Weather Research and Forecasting Model (ARW-WRF) is the model selected for use in this study (Skamarock et al. 2005). The ARW-WRF uses a terrain-following vertical coordinate and has a variety of physical process parameterization schemes available. The physics options used in this study are the Lin et al. (1983) microphysics scheme, the Kain and Fritsch (1993) convective scheme, the Mlawer et al. (1997) rapid radiative transfer model for longwave and the Dudhia (1989) model for shortwave radiation, the Dudhia (1996) five-layer soil model, and the Yonsei University (YSU) planetary boundary layer (PBL) scheme (Noh et al. 2003) combined with a Monin–Obukov-based similarity theory surface layer approach.
The three-dimensional model grid used in this study contains 500 × 500 × 32 grid points using 10-km horizontal grid spacing and encloses an area from 10°–60°S to 85°–35°W (Fig. 1). The model top is at 100 hPa, no damping layer is used, and the vertical grid of 32 layers is stretched with more layers near the ground surface. Some sensitivity tests to ascertain the importance of these model choices were performed, with the results indicating that some scheme combinations do not produce convective activity ahead of the frontal boundary as observed. The lack of prefrontal convective activity (the focus of this study) is considered a significant flaw in any simulation and so these options were not used. Thus, the results reported below depend on the physics options selected. The physics options selected provide the most reasonable evolution of the convective activity in comparison with the available observations.
3. Composite initial and boundary conditions
Inspired by Coniglio and Stensrud (2001), realistic horizontally nonhomogeneous initial and boundary conditions are created using a simple composite approach on the 10 serial upstream-propagating MCS events documented by Anabor et al. (2008). This approach is selected to capture the features that are common to the environments of these serial MCS events, while removing features that are unique to individual cases. Thus, this approach focuses attention on the key, repeatable features that can most easily be used in developing forecast techniques. The composite approach also allows us to maintain a more idealized modeling perspective where the focus is on the physical processes producing upstream propagation instead of the ability of the model to replicate all observed aspects of a particular event. Since Nicolini et al. (2002) indicate that using NCEP–NCAR reanalysis lacks the horizontal resolution needed for the realistic simulation of SALLJ convective events, we choose to use the 1.0° × 1.0° NCEP Global Final (FNL) analyses (see online at http://dss.ucar.edu/datasets/ds083.2/) available every 6 h, as the dataset for constructing the composite.
Anabor et al. (2008) use the Geostationary Operational Environmental Satellite-12 (GOES-12) satellite images to track the geometric center of 10 serial upstream-propagating MCS events throughout their lifetimes. Since we are most concerned with the physical mechanisms of upstream propagation, the location of the first storms position is used to center the spatial location of the composite (Table 1). The FNL analyses are interpolated to a 60° × 60° grid with 1° resolution centered on the first storm position for each of the 10 serial MCS events. To produce the composite initial condition, the resulting 10 three-dimensional grids are simply averaged. As shown in Table 1, half of the serial MCS events start in the morning hours and half start in the evening hours. Owing to the importance of the Andes Mountains to the development of the low-level flow patterns, the data are composited with respect to the diurnal cycle instead of the time relative to storm initiation or MCS life cycle stage. While this is not a perfect solution, results indicate that the simulation correctly reproduces the evolution of the large-scale environment over the 3-day period, suggesting that the composite technique is successful in allowing for an accurate evolution of the large-scale features important to these serial MCS events in the model.
Since convection in SESA typically develops in the morning hours (Velasco and Fritsch 1987), the initial condition is chosen in the local midafternoon at 1800 UTC. Thus, the FNL analyses from the 1800 UTC time prior to the time of first storms for each of the 10 cases are interpolated to the 60° × 60° grid and used for the composite initial condition (see Table 1). Boundary conditions are created in a similar manner using FNL analyses from each subsequent 6-h time period from each serial MCS event. Composite boundary conditions are provided out to 72 h, allowing for a full 3 days to be simulated. While the composite initial condition at 1800 UTC may have some imbalances because of the compositing procedure, the model adjusts to its own balance within a few hours. The resulting evolution of the large-scale pattern is constrained by the composite boundary conditions. By convention, the simulation starts at 1800 UTC on day 0 and extends through 1800 UTC on day 3. The calendar day used to define the amount of solar radiation is chosen to be 3 December, in the middle of the dates of the observed serial MCS events (Table 1).
The resulting simulations, shown in detail in the next section, reproduce the large-scale features seen in Anabor et al. (2008) in association with the serial upstream-propagating MCS events. Thus, it appears that the composite analysis method produces a realistic representation of the large-scale environment for these cases.
4. Results
a. Synoptic setting and convective evolution
The dominant low-level feature during day 1 is the South Atlantic subtropical high as it strongly influences the synoptic-scale circulation. This high helps to induce the northeasterly flow, seen both at the surface and at 850 hPa (Figs. 2a,b and 3a,b) in the northern portion of SESA, that advects the warm and moist air from the Amazon region to the south (Paegle 2000; Nicolini and Saulo 2000; Salio et al. 2007; Anabor et al. 2008). A weak LLJ (∼10 m s−1 maximum wind speed) is observed along the eastern slopes of the Andes near 22°S (Fig. 3b). Warm advection at 850 hPa occurs downstream and to the south of the LLJ region with magnitudes near 2.8°C (12 h)−1 (not shown). In contrast, the vertically integrated moisture flux is a maximum in the LLJ region, reaching values in excess of 200 kg m−1 s−1 (Fig. 4b). At upper levels the flow is westerly with a trough suggested to the west of the Andes (Figs. 5a,b).
A thermal low forms in the lee of the Andes Mountains between 24° and 32°S at 0000 UTC on day 2 (Fig. 2c) that closely resembles the northern Argentine low (NAL; Lichtenstein 1980; Seluchi et al. 2003). The NAL has a distinct diurnal pressure cycle with lower pressures near 2100 UTC and higher pressures near 1200 UTC (Seluchi et al. 2003). Simulated sea level pressures in the NAL region decrease over 3 hPa from near 1009 hPa at 1200 UTC on day 1 to 1006 hPa at 0000 UTC on day 2 (Figs. 2b,c), following the diurnal cycle of observed NALs. The development of this low helps to expand the region of the LLJ southward (Fig. 3c), yielding vertically integrated moisture fluxes exceeding 400 kg m−1 s−1 along the slopes of the Andes (Fig. 4c). The strongest region of warm advection at 850 hPa is again located to the south of the LLJ region with magnitudes near 2.2°C (12 h)−1 (not shown). At upper levels, the winds at 250 hPa have accelerated south of 32°S to speeds above 30 m s−1 (Fig. 5c).
By 1200 UTC on day 2, the NAL is slightly weaker, likely owing to its thermal nature, and a cold front is now present over southern SESA (Fig. 2d) in association with a cyclone that is crossing the south Andes and the arrival of an upper-level shortwave with wind speeds in excess of 45 m s−1 (Fig. 5d). This synoptic pattern is a common feature during MCS development over SESA (Salio et al. 2007). The LLJ as represented at 850 hPa has expanded in size during the past 12 h and now has wind speeds of over 20 m s−1 in the lee of the Andes Mountains near 20°S (Fig. 3d). Wind speeds above 10 m s−1 are seen as far south as 36°S across an east–west band nearly 300 km wide. As expected, the wind speeds in this simulated LLJ are stronger than in the composite analyses of Anabor et al. (2008). Silva Dias et al. (2001) show that the structure of simulated LLJs is dependent on the horizontal grid spacing, with results indicating that the LLJ becomes more confined to the slopes of the Andes as grid spacing is decreased. The strongest 850-hPa warm advection continues to occur downstream of the LLJ region with magnitudes near 2.2°C (12 h)−1 (not shown). Vertically integrated moisture fluxes are now above 500 kg m−1 s−1 along the slopes of the Andes in the LLJ core with values above 100 kg m−1 s−1 located as far as 36°S (Fig. 4d).
Convection in the model simulation starts just after 1200 UTC on day 1 across the 27°–33°S latitude band and stretching from the Atlantic coast to the eastern slopes of the Andes Mountains (not shown). While this timing of convective activity agrees with the mean time of first storms as observed from satellite data (Table 1), the simulated convection in this broad area remains unorganized and largely dissipates by 0000 UTC on day 2. The first organized region of convection develops 3 h later along the cold front near 36°S, 66°W at 0300 UTC on day 2, while the first organized region of convection in the warm sector ahead of the cold front initiates ∼6 h later near 30°S, 58°W (Fig. 2d) over SESA between 0600 and 0900 UTC on day 2 (Figs. 6a,b). This position is still in reasonable agreement with the mean location of the first storms at 33°S, 60°W as observed via satellite for the 10 observed events (Table 1). However, since only satellite observations are available, it is impossible to ascertain when the observed storms begin to organize and determine if this model timing of convective evolution is correct. Yet the general evolution of the convective region agrees with the available satellite observations, yielding some confidence in the results.
The simulated convective region occurs within the zone of strong warm advection and also is in the exit region of the LLJ (Fig. 2d). The large-scale composite at the time of first storms from Anabor et al. (2008) shows a jet streak to the south of the convective region, suggesting that this feature may play a role in developing an environment favorable for convective initiation. Similarly, the 250-hPa jet streak in the model simulation arrives to the south of SESA between 0000 and 1200 UTC on day 2 (Fig. 5d) near the time of the first storms. Thus, the large-scale environment in the convective initiation region has both low-level and upper-level forcing for upward motion.
Over the next 12 h, the convection becomes organized within the warm sector as the cold front pushes northward (Fig. 2e). As discussed in the next section, this organized convective activity propagates upstream relative to the flow at all levels of the troposphere. The cyclone continues to cross the Andes and induces a low-level, ageostrophic southerly flow that produces the equatorward advection of cold air as described by Garreaud and Wallace (1998) and enhances convergence ahead of it. Wind speeds at 850 hPa are northerly and above 10 m s−1 across the warm sector, with the strongest winds confined to the higher terrain (Fig. 3e). The vertically integrated moisture fluxes resemble those reported by Salio et al. (2007) in typical SESA MCSs, with the LLJ channeling moisture toward the MCS region with values above 500 kg m−1 s−1 across a broad area from 20° to 30°S (Fig. 4e). The convective region (Fig. 6f) remains within the exit region of the LLJ (Fig. 3e) and is slowly encroached upon by the cold front to the south.
The moisture transport reaches its maximum at 0000 UTC on day 3 (Fig. 4e) in agreement with the results of Salio et al. (2007). At upper levels, the convection is on the anticyclonic side of the entrance region to an upper-level jet (Fig. 5e) in agreement with the results of Salio et al. (2007) and Anabor et al. (2008). The evolution of the upper-level features, including the deepening of the trough during the past 12 h and the increased winds extending northward toward the convective region, may be due in part to the upscale feedbacks of the convective region. Maddox et al. (1981), Wolf and Johnson (1995), and Stensrud (1996b) all demonstrate that a long-lived MCS can significantly alter the upper-troposphere environmental conditions, producing a jet streak poleward and upstream of the MCS region.
The organized convective region reaches its mature stage by 1200 UTC on day 3, with a cold surface mesohigh clearly seen underneath it (Fig. 2f). The cold front is still located over 200 km to the south of the warm sector convection with an intensifying cyclone now present over the Atlantic Ocean. The LLJ at 850 hPa remains strong, with the strongest winds starting to move off the elevated terrain (Fig. 3f). However, warm advection into the convective region is greatly reduced from previous times and the vertically integrated moisture fluxes also have decreased (Fig. 4f). The upper-level trough is advancing across the Andes, with the convective region remaining on the anticyclonic side of the entrance region to the jet (Fig. 5f). Over and downstream of the convective area at 250 hPa, a divergent flow pattern is clearly evident, again highlighting the potential feedbacks of the convection on the large-scale environment. Over the next few hours, the surface mesohigh expands in size and appears to control the evolution of the simulated convection. However, the convective region still moves upstream relative to the flow at all levels of the troposphere (Fig. 6m). Total rainfall from this simulation is above 80 mm in some locations (Fig. 7), close to half a typical mean monthly rainfall for many locations in this region (Berbery and Barros 2002). The precipitation amount and spatial distribution is consistent with the MCSs rainfall patterns during SALLJ events (Nicolini et al. 2002; Saulo et al. 2007; Salio et al. 2007). The mechanisms of upstream propagation that helped produce these large rainfall totals are now explored.
5. Mechanisms of upstream propagation
During the first 36 h of the model simulation, only unorganized convection associated with the diurnal cycle is produced. However, close inspection of the 3-hourly accumulated precipitation fields on days 2 and 3 after convection starts to become organized shows several interesting features (Fig. 6). As mentioned in the previous section, convection develops along the cold front by 0300 UTC on day 1, with a few small areas of convection in the warm sector to the north. However, 6 h later an arc-shaped region of convection is seen in the warm sector between 30° and 32°S along 58°W (Fig. 6b). This arc-shaped convective line moves west-northwestward and expands outward over the next 6 h until it stretches over 600 km from 28° to 34°S along 60°W at 1800 UTC on day 2 (Fig. 6d). The extent of this region of rainfall is large enough to be considered an MCS. With the tropospheric winds from the northwest at all levels (Figs. 2 –5), this west-northwestward movement of the convective region is in an upstream direction.
Over the next 9 h the western edge of the convective region continues to move slowly west-northwestward while the convection transitions from north–south-oriented convective lines at 1800 UTC on day 2 (Fig. 6d) to an northeast–southwest-oriented convective line at 0300 UTC on day 3 (Fig. 6g). The largest rainfall amounts increase during and immediately after this transition in convective line orientation (Figs. 6f–h). After the transition occurs, the convection is organized as a convective line that moves in a more northerly direction through the end of the simulation (Figs. 6i–m). Since the winds throughout the troposphere are still from the north and west, with the possible exception of the winds in the upper troposphere by 1200 UTC on day 3 that likely are influenced by the long-lived convective activity (Fig. 5f), the convective region is still considered to be propagating upstream through the end of the simulation.
The initial arc-shaped convective region simulated between 1200 and 2100 UTC on day 2 (Figs. 6b–e) is very suggestive of a wavelike feature. Indeed, an examination of the 850-hPa vertical motion and sea level pressure fields at 0800 UTC on day 2 indicates two arc-shaped regions of upward motion preceded by a surface pressure trough and followed by an arc-shaped region of downward motion preceded by a pressure ridge (Fig. 8a). Tracking these regions back in time suggest that they initiate from a small region of convection that developed to the east in a region of low-level upslope flow over the previous hour. These upward–downward motion couplets move west-northwestward at ∼9 m s−1 and expand outward in size over the next 5 h. The values of upward motion reach 0.4 m s−1 in some locations along the arc-shaped line. A second packet of two upward–downward motion couplets appears to the east of the first packet at 1200 UTC on day 2 (Fig. 8d) and is also associated with the development of a small region of convection in an area of low-level upslope flow. This couplet also expands outward with time and moves west-northwestward at the same speed as the earlier wavelike features.
The relationship between the sea level pressure perturbations and upward motion, with the zone of upward motion occurring between the pressure trough and pressure ridge (Fig. 8), is consistent with an internal gravity wave (Eom 1975; Gossard and Hooke 1975). Closer inspection reveals that the convection and the waves propagate in phase one another (Fig. 9). Upward motion associated with the gravity waves occurs prior to the development of rainfall in the model as the waves move toward the northwest (Fig. 9), as indicated by the upward motion leading (i.e., located to the west of) the hourly rainfall totals. Thus, the simulated upstream-propagating gravity waves are acting to trigger the model convective parameterization. The convective scheme is deactivated in the region of descending motion associated with the gravity wave, although the timing of the scheme deactivation may be a function of the convective scheme adjustment time scale. A west–east cross section through the waves shows that the potential temperature perturbations tilt downstream in the vertical (Fig. 10), indicating that these waves are untrapped (Gossard and Hooke 1975). The simple two-layer gravity wave model of Eom (1975) is used to further examine the wave characteristics. This analytic model divides the atmosphere into two vertical layers with uniform properties in order to determine reasonable gravity wave characteristics. Selecting the 800-hPa level as the level that separates the two vertical model layers, the gravity wave model yields an upstream propagation speed of 12 m s−1 in reasonable agreement with the simulated wave propagation speed of ∼9 m s−1 [differences of a few m s−1 between calculated and simulated wave speeds are common (see Eom 1975; Stensrud and Fritsch 1993, 1994)]. Using a typical surface pressure perturbation from the model simulation, the Eom (1975) model also diagnoses upward motion between 1.16 and 2.3 m s−1, roughly twice the values seen in the simulation (Figs. 8, 12, and 15). A model sounding upstream of the gravity wave activity (Fig. 11) shows that there is no stable layer to trap the wave energy (Lindzen and Tung 1976). The waves propagate almost parallel to the surface–700-hPa winds in a weak-to-moderate wind shear environment (900–500-hPa wind shear >2 s−1) that is convectively unstable with values of convective available potential energy near 500 J kg−1 (Fig. 11). Untrapped internal gravity waves also are seen to produce upstream propagation in Stensrud and Fritsch (1994).
Between 2100 UTC on day 2 and 0300 UTC on day 3, the southern portions of the convective lines merge into a more cohesive convective cluster (Jirac et al. 2003; Figs. 6e,f). The transition from north–south-oriented convective lines to a single east–west-oriented convective line appears to be linked to the creation of a cold pool at the surface (Fig. 12). The cold pool acts to organize the simulated convection along its leading edge, or outflow boundary, and is produced by the development of explicit convective processes in the model simulation (prior to this time the majority of the convection in the model is produced by the convective parameterization scheme). Rain is consistently produced along and to the south of the outflow boundary as the cold pool expands and intensifies (Fig. 13), leading to the creation of the northwest–southeast orientation of the main convective line. While internal gravity waves are still present in the warm sector to the north of the outflow boundary and influence convective activity, the cold pool begins to dominate the convective region evolution as it strengthens and grows upscale (Fig. 13). Rainfall is even produced in regions of descending motion associated with the gravity waves owing to the dominating influence of upward motion along the cold pool outflow boundary. With strong northerly environmental winds to the north of the cold pool helping to advect warm and moist air toward the cold pool, the rising motion along the outflow boundary is as large or larger than that associated with the gravity waves. Along the eastern edge of the cold pool, in the region where the cold pool and gravity waves meet, the values of rising motion are maximized (Fig. 12). Another group of internal gravity waves are seen to the southeast of the cold pool and are moving toward it. Thus, the evolution of convection in the warm sector is dominated initially by internal gravity waves and later by the interactions between internal gravity waves and a growing and strengthening cold pool (Figs. 9 and 13).
The cold pool continues to grow and strengthen over the coming hours, stretching from near the Atlantic Ocean to the Andes Mountains by 1200 UTC on day 3 (Fig. 14). The cold pool boundary is located nearly 400 km ahead of the cold front, and the cold pool mesohigh is over 2 hPa in magnitude. Upward motion along the outflow boundary exceeds 0.5 m s−1 in many locations and continues to be a focus for initiating convective activity as indicated by a vertical cross section (Fig. 15). Indeed, the cold pool appears to be the dominant feature in controlling the convective development and this explains the shift in the propagation of the convective region to a more northeasterly direction.
The important interactions between internal gravity waves and cold pools that together control the evolution of convection is also discussed by Stensrud and Fritsch (1993, 1994). They show that while internal gravity waves can lead to convection jumping ahead of an active convective line, thereby forming a new convective line, the cold pool plays the dominant role once moist downdrafts develop and organize into a cold pool density current. The present study outlines a slightly different picture in which internal gravity waves lead to the initial convective development and upstream propagation until downdrafts from several convective elements merge and organize into a cold pool that grows and slowly dominates the further evolution of convection. Thus, similar to the results of Stensrud and Fritsch (1993, 1994) for North American serial MCS events, the combined effects of internal gravity waves and cold pools control the upstream propagation of serial MCS events over SESA.
A conceptual model of satellite-observed serial MCS events over SESA by Anabor (2004) and shown in Anabor et al. (2008) captures average MCS behavior, and shows successive storms developing at roughly 5-h intervals 370 km to the northwest of the previous MCS, while the whole MCS trajectory is toward the northeast. Low-level cloud bands oriented from southwest to northeast are seen in some cases (Fig. 16a), suggesting that upstream-propagating internal gravity waves may be present. These gravity waves need to propagate upstream at 19 m s−1 to explain the average distance in time and space between successive storms found by Anabor (2004). Since this magnitude of upstream wave propagation speed is possible in the atmosphere (Eom 1975), this conceptual model is very suggestive of the importance of internal gravity waves to the development of successive MCSs in the upstream direction. However, field campaigns, radar observations, and a high-resolution observational network are needed to evaluate the hypotheses raised in this study.
6. Discussion
The mechanisms of upstream propagation during serial MCS events over SESA are explored using a numerical simulation of the composite environmental conditions from 10 observed events. Results indicate that the simulation from the composite yields a reasonable evolution of the large-scale environment throughout the 3-day simulation and produces a large region of organized convection in the warm sector over an 18-h period. Most importantly, the model simulation shows that the initial convective development and upstream propagation of the early convection is tied to the development and evolution of untrapped internal gravity waves. As convective downdrafts develop and begin to merge to form a surface cold pool in the simulation, the cold pool and its interaction with the environmental low-level flow starts to play a role in convective evolution. The internal gravity waves and cold pool interact over several hours to influence the convective development, but the cold pool eventually dominates and controls the propagation of the convective region by the end of the simulation. This upstream propagation of the South American convective region resembles the southward burst convective events described by Porter et al. (1955) and studied by Stensrud and Fritsch (1993, 1994) over the United States.
Serial MCS events over SESA are important contributors to the local hydrologic cycle as they can provide roughly half of the total monthly summer precipitation. Thus, the forecasting of these events deserves serious attention in the region. Unfortunately, our conclusions that interactions between internal gravity waves and cold pools largely control convective development and evolution suggests that correctly forecasting these events may be challenging. The model results presented also may be sensitive to the parameterization schemes selected.
While the simulation outlines one mechanism of upstream propagation and explains how these serial MCS events could be produced, a number of questions remain. What is so special about the environmental conditions in the warm sector that allows for development of upstream-propagating gravity waves of sufficient amplitude to initiate convection? Is the answer to this question linked to the presence of the South Atlantic anticyclone and the SALLJ that together advect warm and moist air into the MCS region? Are there other mechanisms that could lead to upstream propagation? Now that one possible mechanism behind these serial MCS events has been identified, attention needs to turn to developing a better understanding of the combination of environmental conditions that allows such interesting convective events to develop. Better observations within SESA are sorely needed to answer many of these questions. The ability of real-time modeling systems to predict these events also deserves exploration.
Acknowledgments
The first author was supported by the Conselho Nacional de Desenvolvimento CientÃfico e Tecnológico (CNPQ; National Council of Scientific and Technological Development), Brazil. We also gratefully acknowledge the support given by the ITS group at NSSL. The helpful and constructive comments of three anonymous reviewers are greatly appreciated and led to an improved discussion of these results.
REFERENCES
Anabor, V., 2004: Descriptive analyses of meso-α convective systems by GOES-8 satellite images. M.S. thesis, Departament of Remote Sensing, Universidade Federal do Rio Grande do Sul, Brazil, 78 pp.
Anabor, V., D. J. Stensrud, and O. L. L. de Moraes, 2008: Serial upstream-propagating mesoscale convective system events over southeastern South America. Mon. Wea. Rev., 136 , 3087–3105.
Berbery, E. H., and V. R. Barros, 2002: The hydrologic cycle of the La Plata Basin in South America. J. Hydrometeor., 3 , 630–645.
Byerle, L., and J. Paegle, 2002: Description of the seasonal cycle of low-level flows flanking the Andes and their interannual variability. Meteorologica, 27 , 71–88.
Campetella, C. M., and C. S. Vera, 2002: The influence of the Andes Mountains on the South American low-level flow. Geophys. Res. Lett., 29 , 1826. doi:10.1029/2002GL015451.
Coniglio, M. C., and D. J. Stensrud, 2001: Simulation of a progressive derecho using composite initial conditions. Mon. Wea. Rev., 129 , 1593–1616.
Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment using a mesoscale two-dimensional model. J. Atmos. Sci., 46 , 3077–3107.
Dudhia, J., 1996: A multi-layer soil temperature model for MM5. Preprints, Sixth PSU/NCAR Mesoscale Model Users’ Workshop, Boulder, CO, PSU–NCAR, 49–50. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307-3000].
Eom, J. K., 1975: Analysis of the internal gravity wave occurrence of 19 April 1970 in the Midwest. Mon. Wea. Rev., 103 , 217–226.
Garreaud, R. D., and J. M. Wallace, 1998: Summertime incursions of midlatitude air into subtropical and tropical South America. Mon. Wea. Rev., 126 , 2713–2733.
Gossard, E. E., and H. W. Hooke, 1975: Waves in the Atmosphere. Elsevier, 472 pp.
James, I. N., and D. L. T. Anderson, 1984: The seasonal mean flow and distribution of large-scale weather systems in the Southern Hemisphere: The effects of moisture transports. Quart. J. Roy. Meteor. Soc., 110 , 943–966.
Jirak, I. L., W. R. Cotton, and R. L. McAnelly, 2003: Satellite and radar survey of mesoscale convective system development. Mon. Wea. Rev., 131 , 2428–2449.
Kain, J. S., and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The Kain-Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor. Monogr., No. 46, Amer. Meteor. Soc., 165–170.
Laing, A. G., and J. M. Fritsch, 1997: The global population of mesoscale convective complexes. Quart. J. Roy. Meteor. Soc., 123 , 389–405.
Laing, A. G., and J. M. Fritsch, 2000: The large-scale environments of the global populations of mesoscale convective complexes. Mon. Wea. Rev., 128 , 2756–2776.
Lichtenstein, E. R., 1980: La depresion del noroeste Argentino (The northwestern Argentina low). Ph.D. dissertation, University of Buenos Aires, Buenos Aires, Argentina, 223 pp.
Lin, Y-L., R. D. Farley, and H. D. Orville, 1983: Bulk parameterization of the snow field in a cloud model. J. Climate Appl. Meteor., 22 , 1065–1092.
Lindzen, R. S., and K. K. Tung, 1976: Banded convective activity and ducted gravity axes. Mon. Wea. Rev., 104 , 1602–1617.
Maddox, R. A., 1980: Mesoscale convective complexes. Bull. Amer. Meteor. Soc., 61 , 1374–1387.
Maddox, R. A., 1983: Large-scale meteorological conditions associated with midlatitude, mesoscale convective complexes. Mon. Wea. Rev., 111 , 126–140.
Maddox, R. A., R. D. Perkey, and J. Fritsch, 1981: Evolution of upper tropospheric features during the development of a mesoscale convective complex. J. Atmos. Sci., 38 , 1664–1674.
Marengo, J. A., W. R. Soares, C. Saulo, and M. Nicolini, 2004: Climatology of the low-level jet east of the Andes as a derived from the NCEP–NCAR reanalyses: Characteristics and temporal variability. J. Climate, 17 , 2261–2280.
Mlawer, E. J., S. J. Taubman, P. D. Brown, M. J. Iacono, and S. A. Clough, 1997: Radiative transfer for inhomogeneous atmosphere: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102 , (D14). 16663–16682.
Nesbitt, S. W., R. Cifelli, and S. A. Rutledge, 2006: Storm morphology and rainfall characteristics of TRMM precipitation features. Mon. Wea. Rev., 134 , 2702–2721.
Nicolini, M., and A. C. Saulo, 2000: ETA characterization of the 1997–98 warm season Chaco jet cases. Preprints, Sixth Int. Conf. on Southern Hemisphere Meteorology and Oceanography, Santiago, Chile, Amer. Meteor. Soc., 330–331.
Nicolini, M., A. C. Saulo, J. C. Torres, and P. Salio, 2002: Enhanced precipitation over Southeastern South America related to strong low-level jet events during austral warm season. Meteorológica, 27 , 89–98.
Noh, Y., W. G. Cheon, S-Y. Hong, and S. Raasch, 2003: Improvement of the K-profile model for the planetary boundary layer based on large eddy simulation data. Bound.-Layer Meteor., 107 , 401–427.
Paegle, J., 2000: American low level jets in observations and theory: The ALLS project. Preprints, Sixth Int. Conf. on Southern Hemisphere Meteorology and Oceanography, Santiago, Chile, Amer. Meteor. Soc., 161–162.
Porter, J. M., L. L. Means, J. E. Hovde, and W. B. Chappell, 1955: A synoptic study on the formation of squall lines in the north central United States. Bull. Amer. Meteor. Soc., 36 , 390–396.
Rasmusson, E. M., and K. Mo, 1996: Large-scale atmospheric moisture cycling as evaluated from NMC global analysis and forecast products. J. Climate, 9 , 3276–3297.
Salio, P., M. Nicolini, and E. J. Zipser, 2007: Mesoscale convective systems over southeastern South America and their relationship with the South American low-level jet. Mon. Wea. Rev., 135 , 1290–1309.
Saulo, A. C., M. E. Seluchi, and M. Nicolini, 2004: A case study of a chaco low-level jet event. Mon. Wea. Rev., 132 , 2669–2683.
Saulo, A. C., J. Ruiz, and Y. G. Skabar, 2007: Synergism between the low-level jet and organized convection at its exit region. Mon. Wea. Rev., 135 , 1310–1326.
Seluchi, M. E., A. C. Saulo, M. Nicolini, and P. Satyamurty, 2003: The northwestern Argentinean low: A study of two typical events. Mon. Wea. Rev., 131 , 2361–2378.
Silva Dias, M. A. F., and Coauthors, 2002: A case study of convective organization into precipitating lines in the Southwest Amazon during the WETAMC and TRMM-LBA. J. Geophys. Res., 107 , 8078. doi:10.1029/2001JD000375.
Silva Dias, P. L., D. Moreira, and M. F. Silva Dias, 2001: Downscaling resolution and the moisture budget of the Plata Basin. Proc. IX Congreso Latinoamericano e Iberico de MeteorologÃa y VIII Congreso Argentino de MeteorologÃa, La MeteorologÃa y el Medio Ambiente en el Siglo XXI, Buenos Aires, Argentina, Centro Argentino de Meteorológos, 8.C.27-341.
Skamarock, W. C., J. B. Klemp, J. Dudhia, D. O. Gill, D. M. Barker, W. Wang, and J. G. Powers, 2005: A description of the Advanced Research WRF version 2. NCAR Tech. Note TN-468STR, 88 pp. [Available from NCAR, P.O. Box 3000, Boulder, CO 80307.].
Stensrud, D. J., 1996a: Importance of low-level jets to climate: A review. J. Climate, 9 , 1698–1711.
Stensrud, D. J., 1996b: Effects of a persistent, midlatitude mesoscale region of convection on the large-scale environment during the warm season. J. Atmos. Sci., 53 , 3503–3527.
Stensrud, D. J., and J. M. Fritsch, 1993: Mesoscale convective systems in weakly forced large-scale environments. Part I: Observations. Mon. Wea. Rev., 121 , 3326–3344.
Stensrud, D. J., and J. M. Fritsch, 1994: Mesoscale convective systems in weakly forced large-scale environments. Part III: Numerical simulations and implications for operational forecasting. Mon. Wea. Rev., 122 , 2084–2104.
Velasco, I. Y., and J. M. Fritsch, 1987: Mesoscale convective complexes in the Americas. J. Geophys. Res., 92 , 9591–9613.
Vera, C. S., and Coauthors, 2006: The South American low-level jet experiment. Bull. Amer. Meteor. Soc., 87 , 63–77.
Wolf, B. J., and D. R. Johnson, 1995: The mesoscale forcing of a midlatitude upper-tropospheric jet streak by a simulated convective system. Part I: Mass circulation and ageostrophic processes. Mon. Wea. Rev., 123 , 1059–1087.
Zipser, E. J., D. J. Cecil, C. Liu, S. W. Nesbitt, and D. P. Yorty, 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87 , 1057–1071.
Model domain used in this study. Topography is shaded as indicated from sea level to 1000 m above sea level. The Andean Mountains (Andes) and La Plata Basin (LPB) areas are indicated.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Sea level pressure (hPa, solid lines) and 900-hPa wind vectors at (a) 0000 UTC on day 1, (b) 1200 UTC on day 1, (c) 0000 UTC on day 2, (d) 1200 UTC on day 2, (e) 0000 UTC on day 3, and (f) 1200 UTC on day 3. The filled circle indicates the position of the convective activity centroid starting at 1200 UTC on day 2.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
As in Fig. 2, but for 850-hPa geopotential height (solid lines) and including wind barbs. Isolines of geopotential height every 30 m. Shading denotes wind speed (see key at bottom).
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Vertical integrated moisture flux (kg m−1 s−1, vectors and shading) and 850-hPa wind magnitude (solid black lines). Filled circle as in Fig. 2. The key at the bottom indicates integrated moisture flux values.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
As in Fig. 2, but for 250-hPa geopential heights (m), winds, and wind speed shaded (see key at bottom). Filled circle as in Fig. 2. Isolines of geopotential height every 50 m.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Accumulated precipitation (mm) over 3-h periods beginning 0900 UTC on day 2–1800 UTC on day 3. Amounts are indicated by key at bottom.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Total accumulated precipitation (mm) over the 3-day model simulation. Amounts in excess of 80 mm are seen in some locations.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Sea level pressure (hPa, solid lines) and 850-hPa vertical velocity (m s−1, shaded, see key at bottom) every hour from 0800 to 1300 UTC on day 2. Isolines of sea level pressure are every 0.15 hPa. (f) The location of the cross section depicted in Fig. 9 is shown.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Sea level pressure (hPa, solid lines) and 850-hPa vertical velocity (m s−1, shaded, see key at bottom) every hour from 1500 to 1800 UTC on day 2. Isolines of sea level pressure every 1 hPa. The hatched semitransparent region outlines accumulated >5 mm precipitation over 1-h periods.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Vertical cross section of potential temperature (shaded every 5 K and black lines) and vertical velocity (m s−1) isolines every 0.2 m s−1 along the line depicted in Fig. 8f. Negative values of vertical velocity are dashed.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Model grid point sounding from point A in Fig. 8.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Sea level pressure (hPa) isolines every 1- and 850-hPa vertical motion (m s−1, see key at bottom) at 0300 UTC on day 3. Locations of mesoscale troughs (T) and ridges (R) are indicated by the dashed and solid lines, respectively. The location of the outflow boundary is also indicated by thick solid line. The hatched region at the center of the grid encloses an area of upward motion >1.0 m s−1. The arrow indicates the direction of gravity wave motion.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Sea level pressure (hPa, solid lines) and 850-hPa vertical velocity (m s−1, shaded, see key at bottom) every hour from 0300 to 0600 UTC on day 2. Isolines of sea level pressure every 1 hPa. The hatched semitransparent region outlines accumulated >5 mm precipitation over 1-h periods.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
As in Fig. 11, but at 1200 UTC on day 3. Note the 1011-hPa mesohigh located in the center of the plot with a coherent region of upward motion to the north along the edge of the cold pool.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Cross section of potential temperature (K) with isolines every 5 K and vertical motion (m s−1, see key at bottom) across the mesoscale cold pool. Location of the cross section shown in Fig. 12. Note the strong region of upward motion along the northern edge of the cold pool.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
Infrared satellite images suggesting gravity waves on 3 Dec 2005 in the (a) pre-MCS environment. Also shown is the infrared imagery during (f) MCS initiation and at the time of the (g) MCS maximum as based on the criteria of Maddox (1983). The arrow in (a) indicates a region of apparent wave activity extending upstream from a region of early convection. Temperature thresholds are shown in the key.
Citation: Monthly Weather Review 137, 7; 10.1175/2008MWR2617.1
The 10 upstream-propagating serial MCS events that make up the model composite are listed by the date, time (UTC), and location (latitude–longitude) of the first storms along with the nearest FNL analysis time. The mean values for all storms also are indicated.
