Abstract

Ninety-four outflow boundary (OB) collisions were documented in north-central Alabama over the summers of 2005–07 using the Advanced Radar for Meteorological and Operational Research (ARMOR) dual-polarimetric radar located at the Huntsville, Alabama, airport. These data were used to extend and verify previous research and to look for new correlations among the various factors that lead to convective initiation (CI) from OB collisions more frequently. For this study, CI is defined as the first occurrence of a ≥35-dBZ radar echo at an elevation angle of 0.8° and within 10 km of the point of collision, from a convective cloud. The radar reflectivity and angle of collision between both OBs along with time of day at which CI occurs most often were analyzed. Also, the presence of cumulus clouds along either/both OBs, or within the area of collision, was examined using Geostationary Operational Environmental Satellite-12 (GOES-12) visible imagery. A more detailed analysis of 23 of the 94 OBs that passed over the Mobile Integrated Profiling System instruments examines the relation among radar reflectivity, updraft magnitude, and water vapor enhancements. This analysis indicates that OB updraft magnitude is positively correlated with OB reflectivity factor. The main findings are that when OBs collide in a more head-on manner, when both colliding OBs have radar reflectivity values of 15 dBZ or greater, or when cumulus clouds preexist along at least one OB, CI is produced at a greater rate. These results, using a much larger dataset than had previously been used for colliding OBs, are subsequently compared with two existing studies.

1. Introduction

Collisions between two separate outflow boundaries (OB) have been shown to induce convective initiation (CI), which sometimes results in severe thunderstorms and tornadoes (e.g., Droegemeier and Wilhelmson 1985; Wilson and Schreiber 1986; Koch and Ray 1997; Frank and Kucera 2003; Karan and Knupp 2009). Other studies reveal how the flow behaves toward producing CI for colliding gust fronts and bores (Clarke 1984; Kingsmill 1995). For this study, “CI” will be defined as any collision that generates a new cell with a reflectivity factor exceeding 35 dBZ within 10 km of the point of collision at a radar elevation angle of 0.8°, as in Wilson and Schreiber (1986), Mahoney (1988), and Weckwerth and Parsons (2006). Predicting when these collisions will result in the initiation of convection is difficult, and significant progress has not been made in this area of research over the years (Wilson and Mueller 1993).

Intrieri et al. (1990) used Doppler lidar to help to determine what occurs during a collision of OBs. It was determined that vertical air motion associated with a single gust front is enhanced when two separate outflows collide. Each of the collisions that were studied resulted in the slightly warmer boundary overriding the colder, denser boundary. Intrieri et al. (1990) also noted that dewpoints in OBs are usually higher than in the boundary layer air in which they are propagating. As a result, the lifted condensation level may be reached by an air parcel at lower heights than boundary layer air when both are forced vertically into the atmosphere. The results from that study implied that collisions involving outflows that extended to a greater depth produced updrafts that penetrated higher into the atmosphere. They also modeled the flow that would take place if two OBs of the same strength and density collided (Fig. 2.2 in Intrieri et al. 1990), and discovered that the cold pools would both deflect upward, causing a strong updraft of air. Additionally, the presence of cumulus clouds in the area of collision between two OBs, or occurring along one or both of the boundaries, appears to be important.

Similar to the Intrieri et al. work, Wilson and Mueller (1993) completed a study on nowcasts of thunderstorm initiation using OB collisions, in which they found that cumulus clouds are a good indicator of instability, because these clouds imply that there is deep boundary layer moisture present; Mueller et al. (1993) also showed this deep boundary layer moisture to be important for CI. In model sensitivity studies of CI processes that are associated with convergence zones, it was found that low-level vertical gradients of temperature and moisture were important in predicting whether or not CI occurs with an outflow collision. A change in the vertical temperature gradient of only 1°C made the difference in whether CI occurred from the resulting collision or not. The strength of the resulting simulated storm varied significantly when the magnitude of the low-level moisture gradient changed by only 1 g kg−1 (Weckwerth and Parsons 2006).

The radar reflectivity of the OBs just before impact has also been shown to influence storm formation from these collisions, as Frank and Kucera (2003) found, with higher-reflectivity boundaries leading to CI more often. Wilson and Schreiber (1986) and Frank and Kucera (2003) examined the angle of collision between two OBs just before impact and found that those that collided in a head-on fashion resulted in CI more often (as will be discussed later).

This present study also examines closely the different factors that cause deep convection in these collisions, and why some of these colliding cold pools do not result in storm development or enhancement. Primarily, this study extends the past research done on OB collisions, confirms previous analyses, and shows possible new correlations between different factors that lead to CI more often from these collisions. However, what makes this analysis unique is that a larger dataset (significantly larger than in previous studies) of 94 radar-identified OB collision events is assembled, while surface-based remote sensing and cloud analyses are included, as a means of better understanding the background thermodynamic conditions that can explain why CI occurs in some OBs and not in others.

In this study, the following questions about OB collisions are addressed:

  1. Do colliding OBs with higher reflectivity (∼20 dBZ) exhibit a higher probability of CI, when compared with OBs with lower reflectivities?

  2. In relation to the previous question, do OBs of higher reflectivities imply a more intense boundary (i.e., a boundary with a stronger updraft)?

  3. Is the probability of CI dependent on the angle of collision between two OBs?

  4. Do small cumulus clouds, either along an OB or in the area of collision, indicate a more likely chance of CI than if there were no small cumulus?

  5. When combining the answers to the first four questions, are there any significant conclusions that can be drawn about OB collisions that will improve prediction of CI?

Question 2 has not yet been addressed by previous literature, while the other questions have been examined before but have continued uncertainty. Previous studies have been unable to decisively answer these questions for various reasons. First, with a large database of 94 collisions, patterns and correlations will likely appear. Second, comparisons with the different examinations involving colliding OBs (such as an examination of reflectivity factor of the OBs, an examination of the angle of collision between the two OBs, and an examination of the cumulus clouds within the collision and OBs) have not been performed for the most part. Third, the relationship between reflectivity factor Z and updraft magnitude within OBs has not been previously studied. Herein, we find that such a relationship exists, which can be used to improve prediction of CI. Fourth, although different angles of collision for colliding OBs have been considered in previous studies (e.g., Wilson and Schreiber 1986; Frank and Kucera 2003), none has offered an explanation of the trends that emerge from these data. Also, the study area for this examination is located in a humid subtropical environment, a region with limited previous research concerning colliding OBs (e.g., Frank and Kucera 2003).

2. Method

In this study, 94 OB collisions were documented in the north-central Alabama region during the summers of 2005–07 between 1700 and 0000 UTC (1100–1800 LST). The sections below detail our analysis procedures.

a. Outflow boundary collision analysis

For the radar reflectivity analysis, the Advanced Radar for Meteorological and Operational Research (ARMOR) was utilized. This radar is located at the Huntsville International Airport near Huntsville, Alabama (34°38′13.9″N, 86°46′30.2″W). Most of the time, an elevation scan of 0.8° and a maximum range of 50 km was used, because it was the one scan that was consistently available for which the OBs and their collisions could be seen clearly and with great detail. An OB’s reflectivity factor magnitude is measured in reflectivity decibels, and typical values range from 0 (low reflectivity) to 20 (high reflectivity) dBZ on radar imagery, with moderate-reflectivity OBs having 10-dBZ reflectivity. The reflectivity value was obtained by finding the maximum value in the OB just prior to collision. In addition, to be counted as a case of CI, the new cell will have to have occurred within 20 min and 10 km of the collision.

The next examination was an analysis of the angle of collision (AOC) between two OBs upon impact. A nomenclature of AOC types was developed by dividing AOC into five categories shown in Fig. 1: S-, V-, T-, W-, and B-type collisions. The letters V and T were chosen because of the shapes that are created from the collision of OBs for both of these types (OB collision forms a V shape for the V type, and collision forms a T shape for the T type). The letter B was chosen because these collisions involve one OB merging with another boundary from behind. Letters S and W were chosen arbitrarily because no other letters or symbols described the shape of these collisions well. For each collision, the category was noted along with the fraction of cases in which CI resulted from the collision. For this study, “CI%” will be defined as the percentage of time that CI occurred from an OB collision.

Fig. 1.

Diagram of all five AOC types. Cold-front symbols represent the OBs, and the arrows indicate the direction of propagation of the OBs.

Fig. 1.

Diagram of all five AOC types. Cold-front symbols represent the OBs, and the arrows indicate the direction of propagation of the OBs.

Geostationary Operational Environmental Satellite-12 (GOES-12) visible images (spatial resolution about 1 km) were examined in all of the collision cases to determine whether there were cumulus clouds along either/both of the OBs, or within the area of collision. Three different aspects of the cumulus cloud coverage were noted. First, whether or not there were small cumulus clouds in the area where two OBs would eventually collide was documented. Second and third, it was noted whether each of the OBs separately was marked by cumulus cloud lines.

b. Ground-based analysis

In addition, 23 OBs of varying reflectivities were examined as they passed over the Mobile Integrated Profiling System (MIPS; Karan and Knupp 2006), located 14 km northeast of the ARMOR (see Fig. 2). MIPS instruments used in this study include a 915-MHz five-beam profiling radar (915), the 12-channel Microwave Profiling Radiometer (MPR), and surface sensors. The 915 vertical motion data were used primarily to determine the relationship between updraft magnitude within the OB and ARMOR reflectivity.

Fig. 2.

Plan position indicator (PPI) image of the reflectivity factor at 0.8° elevation angle from the ARMOR radar at 1910 UTC. An OB is moving over the MIPS site at this time.

Fig. 2.

Plan position indicator (PPI) image of the reflectivity factor at 0.8° elevation angle from the ARMOR radar at 1910 UTC. An OB is moving over the MIPS site at this time.

For each of the OBs used for this part of the study, the time that it was over the MIPS, the reflectivity, and the date of the observation were cataloged. For the MIPS surface instrumentation data (as well as for the 915-MHz profiler and the MPR measurements) with regard to boundary passage time, “pre” is defined as occurring just before the OB passage (from 3 to 10 min), and “post” is defined as occurring after the OB passage (from 3 to 10 min). These values were not peak to peak or averaged, but were simply the values taken at the specifically stated times that the change in surface parameters from pre-OB passage to post-OB passage was calculated. Intense OBs are defined as those with strong thermodynamic gradients. For example, an intense boundary is classified if one of the following threshold differences is measured: temperature T decrease of >4°C, dewpoint temperature Td increase of >2°C, wind direction change >100°, wind speed increase of ≥3 m s−1, or pressure p increase of >0.1 hPa.

For the 915 measurements, an intense OB is defined if one of the following is analyzed: 1) vertical velocity w of >2 m s−1, 2) spectrum width of >2 m s−1, or 3) signal-to-noise (SNR) increase of 10 dB. For the MPR measurements an intense boundary is classified if the water vapor density increase at 1 and 2 km AGL exceeds 1.5 and 1.0 g m−3, respectively. The water vapor density increase was determined by finding the water vapor density value at the specific heights of 1 and 2 km just before the collision and just after collision, and finding the change in those values at those two heights (vertical resolution of 2 km). However, note that the vertical resolution decreases with increasing height (Guldner and Spänkuch 2001).

c. Sources of error

As with any research study, there are some aspects of the work in which improvements could be made. We note these now, ahead of the analysis to follow as a means of highlighting the limitations of this study while helping to guide the discussion to follow and recognizing the comparisons we can make with previous related research.

Although 94 collision cases over three summers is a sizable dataset, there were a few instances in which it would have been nice to have more documented collisions of certain types. With the T-, W-, and B-type collisions, the size of the dataset was considerably smaller than for the S- and V-type collisions.

One very important analysis that was not done was a dual-Doppler analysis with the ARMOR and National Weather Service Weather Surveillance Radar-1988 Doppler (WSR-88D) radar in Hytop, Alabama (KHTX). This kind of analysis would prove to be very useful in a study such as this, as it would allow a researcher to do a cross-sectional examination of what exactly is occurring inside these OB collisions. Doing this was not feasible because of the long baseline of ∼65 km between the KHTX and ARMOR radars.

Another problem was that the GOES-12 images were only available every 15 min, so there were times when the nearest satellite image to a given OB collision was 7.5 min away. This could cause problems if the sky at the time of the image changed significantly by the time the collision actually occurred. Also, some of the collisions did not have images that came within 15 min of the collision, so they were not used.

Last, the depth and speed of the OBs were not taken into account. It is likely that these two factors could play an important role in whether or not CI results from the OB collision.

3. Results

a. Radar analysis

Of all of the OB collisions that were documented over the summers of 2005–07 in north-central Alabama, 34 (or 36%) produced CI. Figure 3 indicates an increasing CI% as the Z value within the OB increases. When the Z of each OB exceeds 20 dBZ, the CI% is 75 (9 of 12 cases). The probability of CI is similar (71%, 10 of 14 cases) when the Z value exceeds 15 dBZ in one OB, and 20 dBZ in the other OB. When the Z of each OB exceeded 15 dBZ (prior to collision), the CI% decreases to 59 (13 out of 22 cases). This decreasing trend continues as the Z value further decreases. For example, when the Z value within each OB is less than 15 and 10 dBZ, the CI% is 26 (15 out of 57 cases).

Fig. 3.

Value of CI% as a function of the value of Z within each colliding OB. Number of cases for each situation is given just below each point.

Fig. 3.

Value of CI% as a function of the value of Z within each colliding OB. Number of cases for each situation is given just below each point.

The CI% for all V- and S-type collisions are 48 (13 out of 27 cases) and 40 (15 out of 37 cases), respectively (Fig. 4). When both categories are combined, the CI% is 44. The more oblique T- and W-type collisions exhibit CI% of 29 (5 out of 17 cases) and 17 (1 out of 6 cases), respectively. Interestingly, CI was not observed in any of the seven B-type collisions.

Fig. 4.

Value of CI% as a function of the value of Z within colliding OBs, stratified according to the AOC type defined on the right.

Fig. 4.

Value of CI% as a function of the value of Z within colliding OBs, stratified according to the AOC type defined on the right.

For the GOES-12 analysis, several examinations were performed. There were a total of 28 cases in which cumulus clouds (hereinafter abbreviated “Cu”) were found in the area of collision (11 resulting in CI), 36 cases in which Cu were found along either/both of the OBs just before collision (17 resulting in CI), 22 cases in which Cu were found in both OBs (8 resulting in CI), and 24 cases in which Cu were found in the area of collision and along either/both of the OBs (9 resulting in CI). Meanwhile, there were 50 cases in which there was no Cu along either/both OBs or in the area of collision (15 resulting in CI). The presence of cumulus clouds around the OB collision area increases the CI% for all OB categories to 39. For clear conditions, the CI% is lower at 30. More specifically, when cumulus clouds exist along either OB or both of the OBs, the CI% is 47. If the previous two conditions are combined (i.e., Cu in the area or along either boundary) the CI% is 38. These results suggest that Cu in the area of collisions and/or along either/both of the OBs increases the chances of CI occurring.

b. MIPS observations

Surface measurements indicate that high-reflectivity OBs (Z ≈ 15–20 dBZ) exhibit an average temperature decrease of 4.4 K and a Td increase of 2.0 K. Lower-reflectivity OBs having Z values of ∼10 and 0–5 dBZ show temperature decreases of 2.4 and 3.0 K, respectively, while Td increases for each were +1.5 and +0.9 K. The mean wind direction change, 115°, was also greater for intense (15–20 dBZ) OBs, versus 90° and 76° for the respective moderate- and low-reflectivity OBs. The mean wind speed change for the intense- and moderate-reflectivity (10 dBZ) OB was 3.1 m s−1. OBs in the 0–5-dBZ reflectivity range produced an average increase in wind speed of 2.7 m s−1. Finally, the average pressure change was an increase of 0.3 hPa for high-reflectivity OBs, and 0.03 and 0.08 hPa for moderate- and low-reflectivity OBs, respectively. See Table 1 for a summary of the MIPS surface data.

Table 1.

Average MIPS surface change vs reflectivity. Higher-reflectivity OBs show stronger gradients in the various parameters (ΔW Dir and ΔW Spd are changes in wind direction and speed, respectively).

Average MIPS surface change vs reflectivity. Higher-reflectivity OBs show stronger gradients in the various parameters (ΔW Dir and ΔW Spd are changes in wind direction and speed, respectively).
Average MIPS surface change vs reflectivity. Higher-reflectivity OBs show stronger gradients in the various parameters (ΔW Dir and ΔW Spd are changes in wind direction and speed, respectively).

Figure 5a shows the kinematic characteristics of OBs. The mean updraft strength increases with Z (from 0.7 m s−1 for low-reflectivity OBs to 2.6 m s−1 for high-reflectivity OBs). The relation with spectral width (an indicator of subgrid-scale vertical motion w) shows a minimum (1.5 m s−1) for moderate-reflectivity Z and a maximum (2.7 m s−1) for 15–20 dBZ (Fig. 5b). Thus, the updrafts are stronger and more turbulent for the higher-Z OBs. The backscattered power in Fig. 5c suggests that SNR is larger for the lower-reflectivity cases. This is an interesting result and implies that Bragg scatter contributes more importantly to the lower-reflectivity OB cases. Note that the SNR was not range normalized in this research. Table 2 summarizes the 915-MHz profiler database.

Fig. 5.

The 915-MHz profiler (a) maximum vertical velocity w, (b) maximum spectral width, and (c) maximum SNR vs ARMOR reflectivity factor.

Fig. 5.

The 915-MHz profiler (a) maximum vertical velocity w, (b) maximum spectral width, and (c) maximum SNR vs ARMOR reflectivity factor.

Table 2.

The 915-MHz profiler database, showing the average maximum vertical motion w, spectral widths, and SNRs. Higher values for w and spectral width correspond to higher Z (from ARMOR) within the OB, whereas the relation between SNR and Z exhibits low correlation.

The 915-MHz profiler database, showing the average maximum vertical motion w, spectral widths, and SNRs. Higher values for w and spectral width correspond to higher Z (from ARMOR) within the OB, whereas the relation between SNR and Z exhibits low correlation.
The 915-MHz profiler database, showing the average maximum vertical motion w, spectral widths, and SNRs. Higher values for w and spectral width correspond to higher Z (from ARMOR) within the OB, whereas the relation between SNR and Z exhibits low correlation.

The MPR data confirm that larger temperature drops and water vapor density ρυ increases occur with the higher-reflectivity OBs. Of particular interest are enhancements in water vapor produced by updrafts within and above the atmospheric boundary layer (ABL). Figure 6 shows that the increase in ρυ at both 1 and 2 km AGL increases as the OB reflectivity increases, and that the increases at 1 km exceed those at 2 km AGL for all categories. For high-reflectivity OBs (15–20 dBZ), the average increase in ρυ was 2.1 and 1.2 g m−3 at 1 and 2 km, respectively. In contrast, low-reflectivity OBs (0–5 dBZ) exhibit average ρυ increases of 0.9 and 0.7 g m−3 at 1 and 2 km, respectively. Figures 7a and 7b compare changes in ρυ during the passage of a low-reflectivity OB (5 dBZ) and a high-reflectivity OB (15 dBZ). In this comparison the difference is substantial.

Fig. 6.

Average increase in water vapor density ρυ during passage of OBs at 1 km (top line; diamonds) and 2 km (bottom line; squares) AGL vs ARMOR reflectivity factor.

Fig. 6.

Average increase in water vapor density ρυ during passage of OBs at 1 km (top line; diamonds) and 2 km (bottom line; squares) AGL vs ARMOR reflectivity factor.

Fig. 7.

Time–height sections of water vapor density ρυ (g m−3) obtained from the MPR for a (a) low-reflectivity OB and (b) high-reflectivity OB. The black rectangle in each panel depicts the time of OB passage.

Fig. 7.

Time–height sections of water vapor density ρυ (g m−3) obtained from the MPR for a (a) low-reflectivity OB and (b) high-reflectivity OB. The black rectangle in each panel depicts the time of OB passage.

c. Temporal variation of CI%

With regard to the temporal variation of CI for all OBs sampled, CI occurred in 42% (21 of 50 cases) of the OBs during the 1900–2200 UTC time period. In comparison, the CI% for the 1700–1900 UTC and 2200–0000 UTC time periods combined was 29 (13 of 45 cases). The increase in CI during the period from 1900 to 2200 UTC is caused by the fact that surface heating caused by solar radiation is at its greatest during this time. Thus, the instability is greater, providing more lift from the convergence of the OBs. Details of the temporal data are provided in Tables 3 –5.

Table 3.

Time of collision vs the S, V, and T AOC types and the resulting CI%.

Time of collision vs the S, V, and T AOC types and the resulting CI%.
Time of collision vs the S, V, and T AOC types and the resulting CI%.
Table 5.

Time of collision vs ARMOR reflectivity factor and Cu formation along OBs for Cu along either OB, Cu in neither OB (column labeled “No Cu”), and all Cu/non-Cu cases (column labeled “all”). Columns labeled “tot” include both CI collisions and non-CI collisions in the respective category. Note: two collision cases were not included because of a lack of satellite data.

Time of collision vs ARMOR reflectivity factor and Cu formation along OBs for Cu along either OB, Cu in neither OB (column labeled “No Cu”), and all Cu/non-Cu cases (column labeled “all”). Columns labeled “tot” include both CI collisions and non-CI collisions in the respective category. Note: two collision cases were not included because of a lack of satellite data.
Time of collision vs ARMOR reflectivity factor and Cu formation along OBs for Cu along either OB, Cu in neither OB (column labeled “No Cu”), and all Cu/non-Cu cases (column labeled “all”). Columns labeled “tot” include both CI collisions and non-CI collisions in the respective category. Note: two collision cases were not included because of a lack of satellite data.

4. Discussion and conclusions

The important points and conclusions, relative to the questions posed in section 1, are now discussed to summarize the findings.

There is a positive correlation between OB intensity and Z. High-reflectivity OBs are defined as those in which Z > 15 dBZ. High-reflectivity OBs exhibit stronger updrafts, greater enhancements in water vapor within and above the ABL, and larger temperature reductions at the 2-m level. In contrast, low-reflectivity boundaries (Z < 5 dBZ) are marked by weaker updrafts, less substantial water vapor enhancements at 1–2 km, and lower temperature reductions at the 2-m level. This behavior is consistent with the hypothesis that high-reflectivity OBs have a greater density contrast and, hence, a greater propagation speed, which increases updraft strength and vertical water vapor advection within and above the ABL. The increase in Z (derived from the 5.5-cm ARMOR radar) within the OB is due primarily to Rayleigh scatter from insects within the ABL. The 915-MHz profiler data indicate that the low-reflectivity OBs exhibit higher backscatter (SNR), an indication that Bragg scatter dominates for this wavelength (33 cm). Bragg scatter is sensitive to small-scale variations in refractive index and provides an important source of scattering from longer-wavelength wind-profiling radars. Since 915-MHz profilers are sensitive to both Rayleigh scatter from insects as well as Bragg scatter from refractive index gradients, this is an interesting result and implies that Bragg scatter contributes more importantly to the lower-reflectivity OB cases.

Next, it was observed that OB collisions involving high-reflectivity OBs appear to be much more likely to result in CI than is the case with lower-reflectivity OBs. Given that OB reflectivity is thought to arise from insect populations, higher reflectivity probably implies larger or more insects. Larger insects, or a greater concentration thereof, in turn imply a stronger updraft accompanied by greater horizontal convergence. Thus, OB reflectivity is generally a good indicator of updraft strength and hence CI likelihood. If two OBs with stronger updrafts collide, the resulting convergence will be even stronger, resulting in air parcels that have a high likelihood of reaching the level of free convection and convection forming.

It was then found that CI% is higher for S and V (head on or nearly head on) collisions than for the other AOC types (40 and 48, respectively). Meanwhile, there were no storms that formed from the B-type collisions, T-type collisions occasionally resulted in CI (29 CI%), and rarely did CI result from W-type collisions (17 CI%). In numerical simulations by Droegemeier and Wilhelmson (1985) in which the collision dynamics between two OBs were analyzed, they found that the modeled head-on collisions produce high convergence between the two OBs, which only gets stronger as the OBs continue to approach each other and eventually collide. This causes strong vertical lifting, moistening, and warming due to latent heat of condensation, which increases the chances of convection (Droegemeier and Wilhelmson 1985). Meanwhile, with the B- and W-type collisions, the OBs are moving in roughly the same direction, which results in the two OBs merging, decreasing the strength of the convergence. Thus, the updrafts during B- and W-type collisions were not observed to be as strong as those from the S- and V-type collisions.

Finally, it was found that CI probability is maximized when 1) the collision is of the S or V type, 2) the OBs possess high Z, and 3) Cu are present along either/both of the OBs prior to the collision. For high-Z V-type collisions and high-Z S-type collisions occurring when there are Cu found along either/both of the OBs, the probabilities of CI reach 60% and 67%, respectively. It is also found that combining just two of the aforementioned criteria also increases the chances of convection as opposed to when just one of the criteria is satisfied. The reason why having Cu along either/both of the OBs is an indication as to possible CI from an OB collision is that Cu are indirect evidence of regions of deep boundary layer moisture, which Mueller et al. (1993) showed to be important for CI (Wilson and Mueller 1993).

There are a couple of previous studies that have performed analyses similar to this research study. Frank and Kucera (2003) examined colliding OBs in July 2002 in Florida, during the Cirrus Regional Study of Tropical Anvils and Cirrus Layers Florida-Area Cirrus Experiment (CRYSTAL-FACE). They used the National Aeronautics and Space Administration polarimetric Doppler 10-cm radar to find the OBs and their imminent collisions with each other in southern Florida. They documented 20 OB collisions. Also, Frank and Kucera (2003) designed their own nomenclature for AOC types for their analysis, which is similar to the nomenclature developed here. Head-on collisions were said to be those OBs that collided with an angle of ≤40°, and they resulted in CI 75% of the time (six out of eight cases). For collisions in which there was an AOC of between 40° and 80°, CI occurred in only one of five cases (20 CI%). The rest of the AOC types were grouped together (AOC of >80°), with a CI% of 28.6 (two out of seven cases). In addition, they found that collisions in which both OBs were moving in the same direction resulted in no CI. If their AOC results from 0° to 80° are compared with the S- and V-type collisions here (combined AOC also from 0° to 80°), a greater chance of CI was found (54% for Frank and Kucera versus 44% for this study). Additionally, when the AOCs between two OBs are greater than 80°, similar CI percentages are found, with 28.6% for Frank and Kucera (2003) and 20% here.

The differences in these comparisons can be attributed to the fact that Frank and Kucera (2003) had only 20 cases, which could have resulted in some skewed statistics because of the small sample size and the season/time frame over which data were gathered. Also, the higher CI% for head-on/near-head-on collisions could also be due to the abundantly moist and highly unstable low levels of the atmosphere in southern Florida. This increased instability can lead to convection occurring more frequently from OB collisions from the strong convergence between the boundaries. This would suggest that the region in which OBs are forming on a particular day is crucial to making good short-term forecasts for potential convection from OB collisions. Also, upper-air soundings should be studied closely to give a good indication as to how unstable the low levels are and if there is a capping inversion that needs to be overcome for convection to occur. If there is a high and/or strong capping inversion, even the strongest of OB collisions may not result in convection. Some similarities between all studies examined include the small CI% for AOCs greater than 80°, higher CI% for AOCs less than 80°, and the fact that Frank and Kucera (2003) also found that the OBs with higher reflectivities resulted in CI more often than when there were collisions between OBs with lower reflectivities. These agreements suggest that extra attention should be paid by forecasters to OB collisions with AOCs less than 80° and involving OBs with higher reflectivities.

Wilson and Schreiber (1986) looked at 49 collision cases over the high plains east of the Rocky Mountains near Denver, Colorado. Part of the study focused on the AOC for the boundary interactions. They divided the AOC types in their study into three different categories: mergers, intersections, and collisions. Mergers were defined as one OB overtaking another OB that is moving in the same direction. An intersection refers to the collision cases in which the two OBs make an angle of >30° at the point of collision. Finally, a collision as defined in their study is when the intersection occurs with an angle of <30° (Wilson and Schreiber 1986). They found that the “collision” (<30°) types resulted in CI at a very high rate of 84%. For the “merger” collisions, the CI% decreased to 63, and for the “intersection” collisions, the CI% was 64. All three of these CI percentages are much higher than the ones that this study has observed in three similar AOC types: 36.2% for S types, 0% for B types, and 40.9% for V and T types combined. Part of the reason for the large difference is because of regional differences between the two areas of study (Colorado vs Alabama). In their area of study, there were high topographic differences throughout, whereas northern Alabama has much less variation in its terrain. However, a much more crucial reason for the differences is most likely the fact that Wilson and Schreiber (1986) did not include boundary data during some of the days in which convection was not expected, days in which they experienced radar difficulties, and days in which operations were halted early. Meanwhile, in this study OB collisions were documented for every summer day over three years, regardless of whether convection was expected. This means that the resulting CI percentages that Wilson and Schreiber (1986) presented may be skewed higher because most of their documentation was done on days on which there was greater instability. On the other hand, lower CI percentages result from this study because of OB collisions being documented on low-instability days when convection is less likely even with a strong convergent collision. However, CI did occur from OB collisions during days in which convection was not expected to occur in northern Alabama, which means that forecasters should take all potential OB collisions seriously, regardless of the stability of the environment. Note that the actual number of OB collisions was not specified in the Wilson and Schreiber (1986) document.

Table 4.

Time of collision vs the W AOC type and the total resulting CI%.

Time of collision vs the W AOC type and the total resulting CI%.
Time of collision vs the W AOC type and the total resulting CI%.

Acknowledgments

This research was supported by National Atmospheric and Oceanic Science Grants NA07OAR4600493 and NA080AR4600896 and NSF Grant ATM-0533596. We are grateful for reviews provided by Tammy Weckwerth, Jim Wilson, and one other anonymous reviewer; their comments greatly improved the quality of this manuscript.

REFERENCES

REFERENCES
Clarke
,
R. H.
,
1984
:
Colliding sea breezes and atmospheric bores: Two-dimensional numerical studies.
Aust. Meteor. Mag.
,
32
,
207
226
.
Droegemeier
,
K. K.
, and
R. B.
Wilhelmson
,
1985
:
Three-dimensional numerical modeling of convection produced by interacting thunderstorm outflows. Part I: Control simulation and low-level moisture variations.
J. Atmos. Sci.
,
42
,
2381
2403
.
Frank
,
P. J.
, and
P. A.
Kucera
,
2003
:
Radar characteristics of convection along colliding outflow boundaries observed during CRYSTAL-FACE.
Preprints, 31st Int. Conf. on Radar Meteorology, Seattle, WA, Amer. Meteor. Soc., 12A.8. [Available online at http://ams.confex.com/ams/pdfpapers/64232.pdf]
.
Guldner
,
J.
, and
D.
Spänkuch
,
2001
:
Remote sensing of the thermodynamic state of the atmospheric boundary layer by ground-based microwave radiometry.
J. Atmos. Oceanic Technol.
,
18
,
925
933
.
Intrieri
,
J. M.
,
A. J.
Bedard
Jr.
, and
R. M.
Hardesty
,
1990
:
Details of colliding thunderstorm outflows as observed by Doppler lidar.
J. Atmos. Sci.
,
47
,
1081
1099
.
Karan
,
H.
, and
K.
Knupp
,
2006
:
Mobile Integrated Profiler System (MIPS) observations of low-level convergent boundaries during IHOP.
Mon. Wea. Rev.
,
134
,
92
112
.
Karan
,
H.
, and
K.
Knupp
,
2009
:
Radar and profiler analysis of colliding boundaries: A case study.
Mon. Wea. Rev.
,
137
,
2203
2222
.
Kingsmill
,
D. E.
,
1995
:
Convection initiation associated with a sea-breeze front, a gust front, and their collision.
Mon. Wea. Rev.
,
123
,
2913
2933
.
Koch
,
S. E.
, and
C. A.
Ray
,
1997
:
Mesoanalysis of summertime convergence zones in central and eastern North Carolina.
Wea. Forecasting
,
12
,
56
77
.
Mahoney
III,
P. W.
,
1988
:
Gust front characteristics and the kinematics associated with interacting thunderstorm outflows.
Mon. Wea. Rev.
,
116
,
1474
1491
.
Mueller
,
C. K.
,
J. W.
Wilson
, and
N. A.
Crook
,
1993
:
The utility of soundings and mesonets to nowcast thunderstorm initiation.
Wea. Forecasting
,
8
,
132
146
.
Weckwerth
,
T. M.
, and
D. B.
Parsons
,
2006
:
A review of convection initiation and motivation for IHOP_2002.
Mon. Wea. Rev.
,
134
,
5
22
.
Wilson
,
J. W.
, and
W. E.
Schreiber
,
1986
:
Initiation of convective storms by radar-observed boundary layer convergent lines.
Mon. Wea. Rev.
,
114
,
2516
2536
.
Wilson
,
J. W.
, and
C. K.
Mueller
,
1993
:
Nowcasts of thunderstorm initiation and evolution.
Wea. Forecasting
,
8
,
113
131
.

Footnotes

Corresponding author address: John R. Mecikalski, Atmospheric Science Dept., University of Alabama in Huntsville, National Space Science and Technology Center, 320 Sparkman Dr., Huntsville, AL 35805-1912. Email: johnm@nsstc.uah.edu