a. Evolution and structure of B1

1) Thermodynamics

A sounding from the MGLASS-1 (see Figs. 4b and 6 for the location) was acquired at 2201 UTC within the trailing density current of B1, 105 min prior to collision. Figure 7 shows profiles within the cold air over the 740 m depth of the outflow, indicated by the shaded region (wind was not available). The coldest air was confined to the lowest 300 m. Using this sounding and the observed ground-relative propagation speed (V_gf = 11 m s⁻¹), the Froude number (F_r) was estimated using the relations

View Expanded

where u_amb is the ambient flow along the direction of gust-front propagation over the depth (h = 740 m) of the density current, b is an experimentally determined adjustment factor (0.62, included to account for the airflow in advance of the density current) determined from the laboratory experiments of Simpson and Britter (1980), and θ_υ−c (301.6 K) and θ_υ−w (305.1 K) are the respective virtual potential temperatures averaged over the depth h within the density current and in advance of the density current. Insertion of these measured parameters into Eq. (1) yields F_r = 1.3, close to the theoretical upper limit of 2 (Benjamin 1968).

The ASOS surface station at Mitchell, South Dakota, also recorded the gust-front passage 24 min earlier at 2137 UTC (Fig. 8). The virtual temperature dropped 3.5°C (32.5° to 29°C) and the mixing ratio increased from 11.4 to 12.8 g kg⁻¹. Wakimoto (1982) has shown that maximum surface pressure rise is within 10% of the value calculated hydrostatically. Therefore, the observed surface pressure increase can be used to estimate the Froude number of the density current according to the relation (Seitter 1986)

View Expanded

where Δp is the assumed hydrostatic pressure increase (∼0.5 hPa) measured at surface. Surface density prior to the boundary arrival is used for ρ_w, even though it is defined as an average value of environmental atmospheric density over the depth of cold outflow. The observed ground-relative propagation speed and these surface observations applied to Eq. (2) produce a Froude number of 1.4, in close agreement with the 1.3 value computed from Eq. (1).

b. Evolution and structure of B2

The passage of B2 over the Sioux Falls—Pavilion surface station (5.5 km south of KFSD radar; see Fig. 5 for location), produced a virtual temperature deficit of only 1.6°C (1.4°C less than that sampled in B1) and an increase in r_υ of 1.4 g kg⁻¹ around 2245 UTC (Fig. 12). The surface virtual temperature behind B2 was 30.8°C, 1.8°C warmer than that within the B1 outflow. Unfortunately the pressure sensor at this School Net site was not sensitive enough to resolve the relatively small surface pressure change. An average propagation speed of B2 during the 2200–2345 UTC time period was 4.2 m s⁻¹. The sounding G2 (Fig. 2) was used to determine the average ambient wind and virtual temperatures from surface to 500-m height. The average virtual temperature was assumed to be linearly decreasing within the lowest 500 m. The depth of B2 (∼500 m; Fig. 10b) was applied to Eq. (1), which yields F_r of 1.6, above the upper limit of the theoretical F_r for density currents.

The kinematic structure of boundary B2 is best illustrated (relative to the structure of B1) in Fig. 10b. In contrast to B1, B2 did not exhibit significant mass transport (a system relative feeder flow) behind its head (u′ ≈ 0 m s⁻¹). Although B2 exhibited higher reflectivity values (∼23 dBZ) the maximum updraft of 5 m s⁻¹ was 1–2 m s⁻¹ less than that within B1. The updraft width (3 km) appears to be greater than that diagnosed in B1. That the radar fineline in B2 is both wider and more intense than in B1 suggests that B2 was more effective in concentrating and lofting insects upward into the boundary layer. This is consistent with the greater number of CI events produced by B2. Above 1 km AGL, the relative flow overturns toward the east behind the gust frontal head, creating a return flow that descends and feeds the outflow of B2 in the form of a rotor circulation (Fig. 10b). This rotor was a persistent feature of B2. Mueller and Carbone (1987) found that significant vertical shear along the interface of two fluids is conducive to shear instability, the creation of inflection points, and a subsequent evolution to vortex rolls.

a. Radar analysis of the gust-front collision

Figure 9 illustrates a plan view of the collision process from 15 min before to 70 min after collision. Maximum updraft strengths within B1 and B2 increased from 5–6 m s⁻¹ to 7–8 m s⁻¹ between 2330 and 2345 UTC. During and after the collision, new cells were initiated at isolated points along the collision line 30–40 km south and north of the collision boundary axis. A boundary identified as AC_B2 (after-collision boundary of B2) in Figs. 9c,d appears as two parallel radar finelines (along the dashed line AB in Fig. 9d) exhibiting wavelike characteristics. Intense deep convection located 30–40 km southwest of the KFSD radar (denoted with a letter S) formed a new gust front that propagated toward the northwest to east. Figure 9d reveals that the east-moving segment of this gust front intersected boundary AC_B1 (after-collision boundary B1). Details of the structure of this gust front and the modified AC_B1 are examined with MIPS measurements in section 7.

The kinematic structures of B1 and B2 about 5 min prior to collision are presented in Fig. 13a. The main characteristics of B1 include a deep updraft, and an oscillatory return flow at 3–4 km AGL west of the updraft. In contrast, B2 exhibits two horizontal vortices east of its main updraft by 2345 UTC (Fig. 13b). Thus B2 resembles a gravity current and gravity wave combination as opposed to the pure gravity current structure of B1.³ As shown earlier by means of Froude number calculations [Eq. (1)], B2 was not a pure density current, but rather was a hybrid phenomenon of gravity current-gravity wave.

By 2345 UTC the updrafts of B1 and B2 had completely merged near x = −12.5 km (Fig. 13b) into a stronger and wider updraft and reflectivity core. At the time of collision, the convergent zone was wider and stronger (between −4 × 10⁻³ and −8 × 10⁻³ s⁻¹), and the depth of convergence increased to 1.2 km AGL. The updraft exhibits a maximum value of 12.5 m s⁻¹ at 1.2 km AGL and extends to 4 km AGL. (Clouds were observed along this convergence line.) Reflectivity values greater than 12 dBZ also reached maximum heights of 1.6 km AGL at this time. These observations confirm significant enhancement in updraft intensity and horizontal width (during the collision event) as observed by Kingsmill and Crook (2003) and simulated in the laboratory by Simpson (1997).

At 2350 UTC, 5 min after the collision, two separate updrafts emerged on each side of the Z maximum, one centered at x ≈ −10.5 km with a w_max of 7 m s⁻¹ and the other at x ≈ −13 km with a w_max of 4 m s⁻¹ (Fig. 13c). Even though the density was not identical within each gravity current, we note that two surface rooted CBZ, propagating in opposite directions, resulted from the collision event.

b. Radar analysis of the postcollision flows

Two separate updrafts with individual Z maxima are apparent 10 min after the collision at 2355 UTC in Fig. 14a. Like its predecessor B2, AC_B2 displayed a steep updraft with a w_max of 6 m s⁻¹ near x = −13 km. Although the updrafts associated with AC_B1 (x = −9 km) exhibit a tilt similar to that of the B1 updrafts, the subsequent evolution is ambiguous due to lack of complete radar coverage at close radar range. The two postcollision boundaries are separated with a broad downdraft peaking at −2 m s⁻¹. It is noteworthy that the updrafts of AC_B1 are surface rooted to low-level convergence shortly after the collision—at least in this vertical plane. The analysis of MIPS data presented in section 7 reveals a contrasting structure—an elevated wave confined to the 2–3 km AGL layer—about 1 h later and 40 km south of the analysis presented in Fig. 14a.

During the period between 2355 and 0015 UTC, AC_B1 propagated faster than AC_B2 (Fig. 14). The reflectivity fields associated with each boundary decreased in height and increased in width. The maximum w values within AC_B2 and AC_B1 were about 5 and 3.5 m s⁻¹, respectively. The outflow of B1 created a stable layer with uniform flow indicated west x = −20 km in Fig. 14c. As AC_B2 propagated into the residual outflow of B1, two successive updraft–downdraft couplets formed behind AC_B2. Radial velocity observations (not shown) suggest that an outbound flow associated with AC_B2 was first detected at 2355 UTC between 0.4 and 0.8 km AGL. The flow within AC_B2 at 0000 UTC resembles a hydraulic jump that characterizes a bore (e.g., Knupp 2006); that is, air parcels rise in the updraft and remain elevated, exhibiting wave oscillations within the westerly flow between 1 and 2 km AGL. However, a more prominent up–down flow 15 min later at 0015 UTC (Fig. 14c) presents a pattern more consistent with that of a solitary wave, with a primary crest centered near x = −15 km and a secondary crest at x = −8 km. Both the updrafts and downdrafts within the westernmost wave exhibit a magnitude of about 5 m s⁻¹. The depression of the Z contours within reduced Z near x = −13 is physically consistent with this interpretation. This relative minimum clearly separates maxima in Z associated with the crest of each wave.

The secondary updraft associated with the ill-defined vortex later intensified and developed a more prominent circulation, completely separated from the circulation of AC_B2. AC_B2 propagated westward at a faster rate, 5–6 m s⁻¹, than its earlier form (B2). It moved 60 km west-northwest of KFSD before disappearing from radar around 0200 UTC. Unfortunately, there were no surface stations along the path of the AC_B2 to characterize its thermodynamic features.

Figure 15 shows that two lines of positive radial velocity were apparent by 0055 UTC. Their separation distance and average westward speed are estimated at about 8.2 km (similar to the spacing at 2355 UCT) and 5.5 m s⁻¹, respectively. The north-pointing arrows in Fig. 15 depict locations of updrafts associated with convergent boundaries for each wave. A vertical section through these features (Fig. 16) reveals maximum low-level westerly flow (7 m s⁻¹) west of AC_B2, and easterly flow (−4 m s⁻¹) that trails this feature (east of x = −20). Therefore, AC_B2 resembles a dual-crested solitary wave consisting of a downdraft (2–3 m s⁻¹) separating two updrafts of about 5 m s⁻¹ intensity.

This scenario reveals a greater complexity than that indicated in idealized laboratory simulations (Simpson 1997) and observations (Kingsmill and Crook 2003). The height of the solitary wave (h₁) and the wavelength (λ) seen in Fig. 16 were observed to be about 1.4 and 7.2 km, respectively. The type of the bore structure depends on the Froude number, F_r, and the ratio d₀/h₀, where d₀ is the depth of the cold outflow and h₀ is the prefrontal height of the stable layer (Rottman and Simpson 1989). As seen earlier, the depth of the most stably stratified layer, prior to the B1 passage, was 0.75 km (h₀). Figure 10b suggests that the depth of B2 is 0.5 km. One of the difficulties specific to this case was to determine the height of the collided boundary seen at Fig. 13b. Upward displacements of enhanced reflectivity values suggest that the mixed boundary at the time of the collision was lifted to the height of 1.5 km AGL at 2345 UTC. There was a considerable turbulent mixing between the two boundaries at that time. Therefore, the depth of the boundary AC_B2 during the collision period was roughly between 1–1.2 km AGL. The ratio, d₀/h₀, then becomes 1.3–1.6. To compute F_r as B2 passed over Sioux Fall surface station (Fig. 12), reduced gravity, g′, was computed by using averaged virtual potential temperatures within the lowest 0.5 km layer based upon the sounding MGLASS-G2 at 2235 UTC for the ambient air and surface observations and adiabatic lapse rate assumption behind the boundary B2. Considering boundary normal westerly wind of 2.2 m s⁻¹, Eq. (1) results in F_r number of 0.9. Simpson (1997) describes the bore character with respect to the ratio between the bore’s mean depth and depth of the stable layer. After these values were plotted on the parameter space between F_r and d₀/h₀, the bore strength h₁/h₀ are found to be 1.9–2.0, which according to laboratory bore simulations (Simpson 1997) indicates type A, smooth undular, bore (see also Figs. 3a,b of Rottman and Simpson 1989). Then, the bore depth can be deduced from the following relation: d_b/h₀ = the bore strength ∼1.9–2.0. The theoretical values of the bore height yield values of 1.4–1.5 km, which is in close agreement with the observed bore height seen in Fig. 16. The radar reflectivity field confirmed that this dual radar finelines disappeared within the two hours from their formation. A close agreement between observed mean bore depth and the bore depth from hydraulic theory has been documented by previous observational and numerical studies (Koch et al. 1991; Ralph et al. 1993; Koch and Clark 1999; Knupp 2006; Koch et al. 2008).

a. Surface and general fineline characteristics

MIPS surface observations (Fig. 18) indicate that pressure steadily increased in the presence of weak southerly flow, while virtual potential temperature and mixing ratio remained nearly constant at 305 K and 10.5 g kg⁻¹, respectively, during the 0035–0042 UTC period. A 90° wind shift to 270°, and wind gusts to 14 m s⁻¹, (∼3 m s⁻¹ greater than boundary propagation speed) accompanied the GF passage at 0046 UTC. Based on surface observations, the CBZ passage occurred between 0041 and 0049 UTC. This time interval is also characterized by the updraft and downdraft branches of the GF measured by 915 MHz profiler. Pressure and virtual potential temperature across the CBZ varied by +0.64 hPa and −1.4 K, respectively. The initial pressure increase of 0.36 hPa is attributed to dynamic effects,⁴ while the subsequent increase of 0.28 hPa following the GF passage represents the hydrostatic component (Charba 1974; Wakimoto 1982; Droegemeier and Wilhelmson 1987).

b. Vertical structures of the boundaries

MIPS 915-MHz wind profiler observations between 0037 and 0133 UTC are shown in Fig. 19. The profiler sampled a low-level updraft–downdraft circulation associated with the GF, followed by an intriguing w oscillation near 2.5 km AGL, apparently associated with AC_B1. After the passage of the GF, a prolonged enhancement in 915-MHz signal-to-noise ratio (SNR) (Fig. 19a) and in ceilometer backscatter (Fig. 19d) was measured below 500 m AGL, and perturbations in SNR are associated with the w field below 1.5 km. Horizontal flow derived the 915-MHz profiler (Fig. 20) shows westerly flow within the GF head, followed by a much shallower layer of westerly flow below 0.5 km AGL that coincides with the enhanced 915-MHz SNR. Relatively strong southwest (SW) flow was present above the (distinct) top of the density current. Thus, this pattern is consistent with the shallow density current indicated by the 915 and ceilometer backscatter (Figs. 19a,d).

The w pattern within the gust front (Fig. 19b) is unusual in two respects: (i) The updraft is not contiguous and first appears aloft within the 0.5–1.5 km layer, and then below 0.5 km after 0045 UTC, just before the GF passage at 0046 UTC. (ii) The trailing downdraft centered near 1 km AGL has a magnitude of 6 m s⁻¹ that is considerably stronger than that of the updraft (2.6 m s⁻¹), and is immediately adjacent to the updraft. This latter feature is in contrast to a more typical pattern (based on a comprehensive analysis of gust fronts summarized by Karan 2007) in which the trailing gust front downdraft follows a narrow, vertically continuous updraft by 2–10 min. While it is tempting to interpret the oscillation in w at the 1 km level as an internal gravity wave (see Fig. 19f for greater details in the w time series at 1.0 km AGL), a sounding launched 7 min later shows a nearly neutral stratification at this level. However, the lowest several hundred meters was rendered stable by the earlier passage of B2 and cloud shading [e.g., the surface temperature prior to the gravity wave (GW) passage was about 27°C, 4°C less than the maximum temperature; see Fig. 21] so this stable layer would have been lifted by the gust front. Moreover, the time series of surface pressure (Fig. 19c) is in phase with that expected from the w pattern at 1 km (e.g., relative high and low pressure beneath respective ascending and descending portions of the wave).

The characteristics of AC_B1, as indicated in Fig. 19, differ substantially from that of AC_B2 and the earlier form of AC_B1, both of which assumed a surface-based bore or solitary wave structure (see Fig. 14). Figure 19a indicates that significant variations in 915 SNR and w above 1.6 km (the top of the residual layer) were associated with the passage of AC_B1 over the MIPS around 0050 UTC. A relative maximum in SNR within the 2–3 km AGL layer between 0050 and 0055 UTC is collocated with a wavelike oscillation around AC_B1. Although the initial w oscillation is most significant (updraft of 2.5 m s⁻¹, downdraft of −5 m s⁻¹), subsequent oscillations in both SNR and w are evident through 0125 UTC within this layer following the passage of AC_B1 (see Fig. 19f for more details of the w time series at 2.6 km AGL). The arrival of AC_B1 was associated with a sudden 0.5 km increase in cloud-base height. The significant variations in cloud-base height (indicated in Fig. 19d and also drawn as black dots in Figs. 19a,b) suggest that the w oscillations within this layer were quite turbulent in character.

A balloon sounding launched from the MIPS site at 0053 UTC, 7 min after the GF passage (Fig. 21), sampled the trailing part of this wave region at 0100 UTC (assuming a sounding ascent rate of 5 m s⁻¹). This sounding reveals the shallow depth of the cold outflow trailing the gust front (consistent with the SNR patterns in Fig. 19a), a residual layer between 0.4 and 1.6 km, a stable layer centered near 1.8 km, and a second mixed layer between 2 and 3 km. Unfortunately, this sounding terminated prematurely near 670 hPa. It does suggest stable layers near 800 and 680 hPa. A stable layer is also implied by the layer of enhanced 915-MHz SNR centered near 3.0 km between 0035 and 0055 UTC, and then extending up to 3.5 km by 0100 UTC (Fig. 19a). The main perturbation of AC_B1 is located within this upper mixed layer. Small-scale variability in the sounding profiles of θ and r_υ corroborate the large turbulent eddies inferred within this layer from the 915 MHz SNR and w, and ceilometer cloud-base measurements. Additional validation of relatively intense turbulence within AC_B1 is indicated by high values of spectrum width (2.5–5.0 m s⁻¹, not shown) measured by the 915-MHz profiler, and by clear air return (between −10 and 0 dBZ) measured near this level (3.5° elevation angle) by the KFSD radar (not shown).

These observations indicate a radical change in the structure of AC_B1 from that produced by the collision of B1 and B2. Figure 14a indicates that AC_B1 exhibited low-level convergence immediately after the collision at 2355 UTC, about 1 h earlier and 40 km north of the MIPS observations. It seems plausible that the vigorous gust front lifted AC_B1 above the surface-based stable layer produced by cool air behind B2 and subsequent cloud shading. Based on the 915 measurements, the gust front head was about 1.3 km deep. If we assume that (i) the gust front head represents the maximum vertical displacement, (ii) the potential temperature of the mixed layer two hours earlier is 307 K (as indicated in the sounding shown in Fig. 2), and (iii) the surface temperature prior to the gust front was 28°C (as measured by the MIPS), then adiabatic lifting of the hypothetical sounding (shown in Fig. 21) by a vertical distance of 1.3 km would nearly duplicate the measured sounding. Since the gust front moved faster than AC_B1, it would have intersected and then lifted the bore circulation about 1.3 km A continued upward propagation to the next stable layer (e.g., the top of the mixed layer near 700 hPa) would have occurred because of the near-neutral stability of the intervening layer. Although the MGLASS sounding in Fig. 21 represents the atmosphere modified by the gust front and passage of the elevated AC_B1, the measured profiles of T and T_d (θ and r_υ) are consistent with the observed preexisting conditions and structure of the gust front.

This secondary collision event (gust front collision with a bore, which was previously produced by a collision between two gust fronts) represents the first detailed observation of a secondary collision. Although the detailed MIPS observations provide a comprehensive picture of the effects of this secondary collision, a numerical simulation would be required to clarify the physics of this interaction and the evolution AC_B1. In this case, the active “weather” associated with AC_B1 assumed the form of moderate to intense turbulence at scales ranging from about 10³ m (as directly measured by the 915-MHz velocity) to about 10 m (as indicated by large values of spectrum width).