A squall line is most simply defined as any line or narrow band of active thunderstorms. The line can extend hundreds of kilometers in length and last for several hours. Because of their long-lasting and well organized convective nature, squall lines are frequently observed to produce heavy rainfall and severe weather events. The accurate prediction of severe weather events associated with squall lines as well as other mesoscale convective systems (MCS) remains a challenging task. Realistic prediction of the internal structure and evolution of squall lines and other MCS's consitutes an essential step towards significant improvements in quantitative precipitation forecasts and severe weather warnings.
Numerical studies that investigate the structure and evolution of squall lines and other convective systems can be divided into two groups. In the first, the model generally has grid spacings on the order of 20-100 km and parameterizes convection. These studies show that most mesoscale features of squall lines can be reasonably reproduced (e.g., Zhang et al. 1989). These simulations, however, are unable to resolve convective scale processes that may be important to the development, evolution and propagation of squall lines. An additional problem for these studies is that the simulation is often sensitive to the cumulus parameterization scheme used. In studies of the second type, nonhydrostatic models with grid spacings on the order of a few kilometers are used to investigate the internal cloud-scale dynamics of squall lines. Due to the limitation of computing power and the lack of very fine scale observations, most of the early simulations, however, used only idealized initial conditions and are often done in two dimensions (e.g., Thorpe at al. 1978; Hane et al. 1987; Rotunno et al. 1988).
With the rapid increase in computer power, models in the two groups are converging in terms of spatial resolution. Nonhydrostatic models that can resolve multi-scale (from synoptic to cloud scales) flows have been developed and are now capable of explicitly predict storm-scale phenomena from realistic initial conditions.
The purpose of this study is to investigate the propagation and internal structure of a mid-latitude squall line that occurred on 7-8 May 1995 (Wang et al. 1996) during the VORTEX'95 field experiment (Rasmussen et al. 1994) over the south-central plains of the US. Using realistic initial conditions and a 3-D nonhydrostatic model, Xue et al (1998b, hereafter X98b) successfully simulated, on a nested 6 km grid, the initiation and evolution of two convective lines on that day. It is the purpose of this paper to carefully analyze the propagation and the convective and mesoscale structure of the second line as simulated in a numerical model and reported in X98b.
2. THE MODEL AND NUMERICAL SIMULATIONS
The model used is the Advanced Regional Prediction System (ARPS, Xue et al. 1995). It is a multi-scale three-dimensional nonhydrostatic model with comprehensive physics. To take advantage of the availability of various observational data near the squall line initiation time, an intermittent data assimilation (IDA) procedure was used in X98b. With this procedure, the model is integrated forward from an initial analysis for a specified period, and a new analysis is then obtained by combining the model prediction with observations of various scales. The cycle is repeated for several times until an assimilated initial condition is obtained. The model is then integrated forward (see, e.g., Fig. 2 of X98b).
The analysis tool used is the ARPS Data Analysis System (ADAS, Brewster 1996). The data sources include the NCEP RUC (Benjamin et al., 1991) analyses, surface observations from SAO, ASOS and the Oklahoma Mesonet, and additional soundings and wind profiler data collected during the VORTEX experiment. WSR-88D data were used in one of the experiments reported in X98b. It is shown there that three hourly IDA cycles significantly improve the prediction of convective initiation along the dryline, while further improvement can be obtained by using additional cycles and by including radar data at a higher frequency. Xue et al. (1998a) examine the initiation of the second squall line along the dryline, whereas in this paper, we focus on the propagation and structure of this same line during its mature stage.
The results from experiment EX3 of X98b, which uses 6 hourly cycles without radar data, will be examined here. More details on the simulation procedure can be found in X98b.
3. PROPAGATION AND STRUCTURE OF THE SIMULATED SQUALL LINE
3.1. The propagation of simulated squall-line
As mentioned earlier, two squall lines developed over the south-central plains on 7-8 May 1995. The first was organized from isolated storms that originated from mid-western TX and moved north-northeastward into OK. The second line formed at around 1700 UTC along a well-defined dryline in the TX panhandle. It then propagated eastward away from the dryline and evolved into a classic squall line with a solid leading edge and a trailing stratiform precipitation region. It extended more than 1000 km in length and lasted for more than 10 hours. In this paper, we focus on the second line.
In the model, convective cells started to appear at around 1600 UTC along a north-south axis in the TX and OK panhandle area, just east of the dryline. By 1900 UTC, these cells had organized into a solid line which moved eastward and developed into an intense squall line. Individual cells in the line are found to propagate north-northeastward and the general characteristics agree well with observations.
Fig. 1 shows the three-hourly positions of the simulated squall line compared to observations, from 1800 UTC 7 May to 0600 UTC 8 May 1995. Here, the squall-line position is defined as the leading edge of the near-surface radar echoes. The propagation of the simulated squall line compared well with observations in the first six and the last three hours, but was noticably faster between 00 and 03 UTC.
Fig. 1. Three-hourly positions of the observed (solid lines) and simulated (dashed lines) squall line during the period from 1800 UTC 7 May to 0600 UTC 8 May 1995.
Between 2100 UTC 7 May to 0300 UTC 8 May, the simulated mean propagation speed is about 20 m/s while the observed speed is about 17 m/s. The largest speed discrepancy exists between 00 and 03 UTC, when the northern portion of the simulated line merges (not shown) with the remaining part of the first convective line, resulting in accelerated propagation. In reality, the first line has largely moved out of OK into KS by this time. The simulated line propagated no faster than the real line after 03 UTC. Using the pressure jump across the gust front (Seitter 1986), we found the gust front propagation speed in a 2 m/s opposing flow to be 15.5 m/s at 0300 UTC, which agrees with the average speed of about 15 m/s between 0230 and 0300 UTC estimated from the line locations and only slightly slower than the speed of real line.
3.2. Internal structure of simulated squall-line
3.2.1. Surface features
In Fig. 2, the surface perturbation potential temperature, wind vectors and simulated composite (column maximum) reflectivity at 0300 UTC 8 May 1995 (9 hours into model prediction) are plotted. The most obvious feature is the gust front at the leading edge of the squall line. Across the front are large temperature changes (~ 12K / 30 km) and a sudden shift in wind direction. The gust front propagating into the low-level warm moist flow provides an essential low-level forcing mechanism that sustains this long-lived squall line.
Fig. 2. Simulated surface perturbation potential temperature (1.0K interval), wind vectors and composite reflectivity (shaded, in dBZ) at 0300 UTC 8 May 1995. The think line A-B denotes the location of the vertical cross sections shown in Figs.3 and 4.
Behind the gust front is a cold pool that extends as far as 200 km. Within the cold pools are centers of strong surface divergence associated with intense convective and mesoscale downdrafts. The reflectivity fields exhibit maximum values at the leading edge of the gust front, and secondary maxima about 100 km behind. The leading maxima are obviously associated with the convective cells triggered at the gust front, while the secondary maxima are associated with cells that had moved rearward in the ascending front-to-rear (FTR) flow (Fig.3). The weaker reflectivity further behind resulted from stratiform clouds forming in the ascending FTR flow and is also commonly observed in mature squall lines.
Mesoscale pressure perturbations are also well reproduced in the simulation. They include the presquall mesolow, the squall mesohigh and the wake low (not shown). The presquall mesolow is a local region of relatively low pressure in the warming environment ahead of the gust front. Nonhydrostatic ascent may have also contributed to this low. The squall mesohigh is located behind the convective line and is associated with the hydrostatic effect of cold pool and the nonhydrostatic effect of downdrafts impinging on the ground. The wake low is located behind the mesohigh, and is the low surface pressure region associated with the warming in the area of stratiform precipitation.
A multi-cellular structure of the squall line can be deduced from Figs. 2 and 3. New cells are regenerated along the gust front and at the southern end of the squall line (not shown), and move rearward relative to the gust front. Relative to the ground, the cell motion is north-northeastward due to the fast propagation of the gust front (squall line).
3.2.2. Vertical cross sections
The kinematic structure of the simulated squall line is further examined by vertical cross-sections. Fig. 3 shows the total water (liquid and ice) and the east-west component of wind relative to the moving squall line at 0300 UTC 8 May 1995. Fig. 4 shows the relative humidity and the line-relative winds at the same time. Both cross-sections are along the line A-B in Fig. 2.
In Fig. 3, a three-layered flow structure can be seen to the rear of the gust front. Immediately behind the front, beneath the forward propagating cold pool, exists a very thin (~100 m deep) layer of front-to-rear (FTR) flow, which is induced by surface friction. The depth of this flow increases to about 1 km some 100 km behind gust front. This location coincides with the center of divergence at the surface (Fig. 2), and the FTR flow to the rear of the center is in fact the cold pool outflow at its back edge.
A strong rear-to-front (RTF) flow (jet) is observed to extend to a depth of about 7 km from the back edge of the stratiform cloud region several hundred kilometers behind the gust front (Figs. 3 and 4). The 15+ m/s jet is located at 4 km but descends to a much lower level (~ 500 m AGL) 70 km behind the gust front. Much of the low-level RTF results from the cold pool outflow while the mid-level RTF is a result of the mesoscale response to the convective heating according to Pandya and Durran (1996). It is interesting to note that the zones of strongest horizontal convergence exist at mid-levels immediately before the RTF flow descends to a lower level and at the leading edge of the gust front. Directly above both converegence zones exist distinct convective cells.
Fig. 3. Vertical cross section of the east-west component of wind (contours, 3.0 m/s interval) relative to the moving squall line, and the simulated total water (shaded, in g/kg) at 0300 UTC 8 May. The surface gust front location is marked by a cold front symbol.
Fig. 4. Same as Fig. 3, but for relative humidity (0.1 interval) and line-relative wind vectors.
Another dominant feature in Fig. 3 is a rearward-tilting ascending branch of FTR flow that originates from low levels below 4 km ahead of the gust front. This flow jumps in the convective region to the mid-levels and exits at the rear of the system above 8 km . The 20 m/s jet core is located near the tropopause (~11.5 km). Embedded in this FTR flow are two major convective cells mentioned earlier, one at the gust front and one about 70 km behind. The latter is associated with the mid-level convergence between the RTF and the FTR flows. While this mid-level cell can trace its origin to an earlier cell generated at the leading edge of the gust front, the presence of the mid-level convergence does seem to provide a preferred location for the cell intensification. The presence of a zone of relatively weak echo between the two cells along much of the squall line supports this argument (Fig. 2).
Finally, we note the presence of the over-turning branch of RTF flow at the upper levels, which owes much of its origin to convective drafts in the mid-level cells (Fig. 4). This overturning branch and the deep FTR flow spreads at the upper levels in both the forward and backward directions, producing a wide band of anvil clouds commonly observed with mature squall lines.
4. SUMMARY AND DISCUSSION
In this paper, an analysis is performed on a squall line simulated by a 3-D nonhydrostatic model starting from a realistic initial condition. The simulated features include mesoscale cold pools behind a narrow gust front, multi-cellular and three-dimensional structures of the squall line, low and high pressure patterns across the gust front, major branches of front-to-rear and rear-to-front circulations, convective-scale updrafts and downdrafts, proper distribution of rainfall maxima at and behind the gust front, and the stratiform precipitation further behind.
Although we have not verified all simulated features against observations, the results agree qualitatively with earlier studies and accepted conceptual models (e.g., Ogura and Liou, 1980; Smull and Houze, 1987).
The 6 km horizontal resolution is admittedly too coarse to accurately resolve the detailed convective-scale structures. This may account for part of the quantitative difference between the simulation and observations. Additional higher resolution nested grid simulations are being conducted that may lead to improved results.
This research was supported by NSF grant ATM91-20009 to CAPS and by a supplemental grant to CAPS through NSF from FAA. The simulations were made on the Cray C90 and T3D at the Pittsburgh Supercomputing Center.
Benjamin, S. G., K. A. Brewster, R. L. Brummer, B. F. Jewett, T. W. Schlatter, T. L. Smith, and P. A. Stamus, 1991: An isentropic three-hourly data assimilation system using ACARS aircraft observations. Mon. Wea. Rev., 119, 888-906.
Brewster, K., 1996: Application of a bratseth analysis scheme including Doppler radar data. 15th Conf. on Weather Analysis and Forecasting, Amer. Meteor. Soc., Norfold, VA, 92-95.
Hane, C. E. C. J. Kessinger, and P. S. Ray, 1987: The Oklahoma squall line of 19 May 1977. Part II: Mechanisms for maintenance of the region of strong convection. J. Atmos. Sci., 44, 2866-2883.
Ogura, Y., and M. T. Liou, 1980: The structure of a mid-latitude squall line: A case study. J. Atmos. Sci., 37, 553-567.
Pandya, R. E., and D. R. Durran, 1996: The influence of convectively generated thermal forcing on the mesoscale circulation around squall lines. J. Atmos. Sci., 53, 2924-2951.
Rasmussen, E. N., R. P. Davies-Jones, C. A. D. III, J. M. Straka, F. H. Carr, M. D. Eilts, and D. R. MacGorman, 1994: The Verifications of the Origins of Rotation in Tornadoes Experiment: VORTEX. Bull. Amer. Meteor. Soc., 75, 995-1006.
Rotunno, R., J. B. Klemp, and M. L. Weisman, 1988: A theory for strong, long-lived squall lines. J. Atmos. Sci., 45, 463-485.
Seitter, K. L., 1986: A numerical study of atmospheric density current motion including the effects of condensation. J. Atmos. Sci., 43, 3068-3076.
Smull, B. F., and R. A. Houze, 1987: Rear inflow in squall lines with stratiform precipitation. Mon. Wea. Rev., 115, 2869-2889.
Thorpe, A.J. and Miller, M.J., 1978: Numerical simulations showing the role of the downdraught in cumulonimbus motion and splitting. Quart. J. Roy. Met. Soc., 104, 873-893.
Wang, D. H., M. Xue, V. Wong, and K. K. Droegemeier, 1996: Prediction and simulation of convective storms during VORTEX95. 11th Conf. on Num. Wea. Prediction, Amer. Meteor. Soc., Norfold, VA, 301-303.
Xue, M., D. C. Hou, D. H. Wang, and K. K. Droegemeier, 1998a: Analysis and prediction of convective initiation along a dryline. 16th Conf. on Wea. Anal. and Forecasting, Amer. Meteor. Soc., Phoenix, AZ
Xue, M., D. H. Wang, D. C. Hou, K. Brewster, and K. K. Droegemeier, 1998b: Prediction of the 7 May 1995 squall lines over the central US with intermittent data assimilation. 12th Conf. on Num. Wea. Prediction, Amer. Meteor. Soc., Phoenix, AZ
Xue, M., K. K. Droegemeier, V. Wong, A. Shapiro, K. Brewster, 1995: ARPS Version 4.0 User's Guide. 380pp. Available from CAPS, University of Oklahoma, Norman, OK 73019.
Zhang, D. -L., K. Gao, and D. B. Parsons, 1989: Numerical simulation of an intense squall line during 10-11 June 1985 PRE-STORM. Part I: Model verification. Mon. Wea. Rev., 117, 960-994.