Case 2: Deep convection, squall line
This intercomparison study is based on a squall line that occurred on 20 June 2007 in central Oklahoma. The storm was well observed by a combination of dual-polarization radar, surface disdrometer, and mesonetwork of surface meteorological stations. Model simulations at convection-permitting scale (horizontal grid spacing of order 1 km) are to be conducted in two or three dimensions using a quasi-idealized framework, with the environmental sounding based on pre-storm observations and the storm initiated by inserting a cold pool. Sensitivity tests are proposed that vary parameters affecting cold pool strength under different low-level environmental wind shears, given previous studies that have suggested the importance of these interactions to storm structure and intensity. The goals of this study are to characterize the spread of model results for a well-observed squall line, to understand causes of model differences, and to investigate how cold pool-shear interactions vary among simulations.
Background and motivation
Squall lines are quasi-linear systems of organized deep convection. They occur frequently in both mid-latitude and tropical environments and can produce copious precipitation and severe weather. While there have been a number of model intercomparison studies of precipitating convective systems (e.g., Moncrieff et al. 1997; Redelsperger et al. 2000; Xu et al. 2002; Fridlind et al. 2011), there has been limited systematic inter-model comparison for observationally-based cases of mid-latitude squall lines.
The parameterization of cloud microphysics, which has traditionally been a focus of WMO Cloud Modeling Workshops, remains a major challenge for simulating these storms. Several studies have highlighted sensitivity of squall lines to representation of microphysics (e.g., Fovell and Ogura 1988; McCumber et al. 1991; Tao et al. 1995; Ferrier et al. 1996; Liu et al. 1997; Lynn et al 2005; Morrison et al. 2009; Li et al. 2009; Bryan and Morrison 2011; van Weverberg et al. 2011; Morrison et al. 2011). Overall, different microphysics schemes or different parameter settings within a single scheme can produce large differences in squall line characteristics, with often poor simulation relative to observations. Parameterization of both liquid and ice microphysics is important for squall line structure and precipitation. Recent studies have focused particularly on the role of rain microphysics in squall line simulations (Morrison et al. 2011; van Weverberg et al. 2011). These studies showed large impacts of rain microphysics on evaporation and hence cold pool evolution that in turn affect storm structure, precipitation, and propagation.
In the meteorological literature, the impact of cold pools on storm intensity and structure has often been described in relation to the low-level environmental wind shear through "RKW" theory, which explains how density currents evolve in a sheared environment (Rotunno and et al. 1988; Weisman and Rotunno 2004). This theory postulates that maximum system intensity is attained when the vorticity generated by cold pools is in approximate balance with the low-level environmental shear, leading to the deepest lifting of environmental air; hence, changes in cold pool intensity or shear that move the system away from optimal balance will result in weakening of the system. Thus, according to RKW theory, an increase in cold pool strength may or may not increase storm intensity depending upon the strength of the environmental low-level wind shear. However, relevance of RKW theory to squall lines remains an area of active research; it is unclear how various aspects of system strength are related to the intensity of lifting at the cold pool edge (c.f., Bryan et al. 2006). For example, Stensrud et al. (2005) questioned whether there is any relevance of the relative strengths of the cold pool and environmental shear on squall line structure and intensity, especially for observed systems. They noted that several measures of storm intensity in the simulations of Weisman and Rotunno (2004) were greatest for systems relatively far from the optimal cold pool-shear balance.
In a multi-model assessment, Bryan et al. (2006) found that RKW theory could explain many (though not all) quantitative aspects of changes in system intensity with changes in the strength of cold pools and shear. However, they analyzed model simulations utilizing simple liquid-only microphysics schemes that have difficulty producing key squall line features like trailing stratiform precipitation. More recently, Morrison et al. (2011) examined interactions between microphysics, cold pool evolution, and low-level shear using a model with two-moment, mixed-phase bulk microphysics that was able to produce convective, trailing stratiform, and transition regions similar to observations. Their results were qualitatively consistent with some aspects of RKW theory, such as stronger and less tilted convective updrafts when cold pool and shear were more balanced, but found that other aspects of storm intensity such as domain-mean surface precipitation rate and total updraft mass flux were greater in runs with a less optimal cold pool-shear balance. However, they evaluated these interactions over a fairly limited range of wind shears using a single model; hence, the generality of these results is uncertain. To our knowledge, there has not yet been a comprehensive multi-model intercomparison of cold pool-wind shear interactions using models with relatively sophisticated microphysics schemes (mixed-phase bulk and bin schemes). This provides one of the motivations for the current study.
This intercomparison focuses on squall line simulations using models run at convection-permitting scales (i.e., horizontal grid spacing of order 1 km or less), without utilization of deep convection parameterizations. Understanding, evaluating, and improving such simulations is increasingly important given that many centers now run numerical weather prediction models at these scales, or plan to do so within the next few years. The performance of models at these scales is also relevant to global cloud models and "superparameterized" models.
The broader goals of this intercomparison are:
- to characterize the spread of model results for a well-observed mid-latitude squall line case, focusing especially on a comparison of simulations using microphysics schemes of varying complexity
- to understand causes of differences between model simulations and spur improvements in model parameterizations
- to evaluate interactions between microphysics, cold pools, and ambient wind shear among different models
Specific tasks for addressing these goals are:
- to develop a squall line case study for model intercomparison by synthesizing a suite of observations
- to perform and compare baseline model simulations using different models in the context of these observation
- to perform and analyze sensitivity tests, focusing on interactions between microphysics, cold pools, ambient wind shear, and storm dynamics, using different models
We note that the experimental design for this intercomparison study is such that it can be easily adapted to other case studies of deep convection, requiring only small modifications (namely, environmental sounding and shear). Therefore, we anticipate this framework will be useful for additional observationally-based case studies once the data become available.
The 20 June 2007 squall line in central Oklahoma was part of a large MCS (Fig. 1) that resulted from three separate systems that formed during the afternoon of 19 June 2007 and merged together later that evening. This storm was the subject of previous studies on electrification by Lang et al. (2010) and sensitivity of a squall line to treatment of raindrop breakup by Morrison et al. (2011); the development of this case study follows closely from Morrison et al. (2011). Southeasterly low-level flow had set up on 19 June, advecting moisture northward from the Gulf of Mexico into the central plains to the east of a dry line across eastern New Mexico. Extremely moist and unstable conditions prevailed over Oklahoma and Texas, with surface dew point temperatures exceeding 25°C and convective available potential energy (CAPE) exceeding 4000 J kg−1 (shown by the sounding at 00 UTC on 20 June in Fig. 2). The first storms developed by 1700 UTC on 19 June in northwestern Kansas along a stationary front. This cluster of storms moved southeastward toward Oklahoma and merged with another set of storms that formed by 2100 UTC in northern Oklahoma, forming the portion of the squall line that passed over the Norman, OK (KOUN) dual-polarization radar domain (Fig. 1) that is the focus of the present intercomparison study.
This squall line was well observed using a combination of remote sensors and ground-based in situ instrumentation. The observational infrastructure in central Oklahoma for this case included two two-dimensional video disdrometers (2DVDs; Kruger and Krajewski 2002), a dual-polarization S-band WSR-88D radar at KOUN, and a high-resolution network of surface observations from the Oklahoma Mesonet. A detailed description of the observations for this case is provided by Morrison et al. (2011); a brief description is given here. The Oklahoma Mesonet (Brock et al. 1995) provided 5-min surface measurements of temperature, humidity, wind, and precipitation, at more than 100 sites in the state of Oklahoma. The 2DVDs were deployed within the KOUN radar domain during the 2007 wet season: one operated and owned by NCAR (17 km to the east of KOUN) and the other by the University of Oklahoma (OU; 65 km to the southwest of KOUN). The 2DVDs provide detailed information on surface rain DSDs. An example of the disdrometer data (scatterplots of the derived raindrop size distribution intercept parameter, N0, versus rainwater content, RWC) for the June 20 squall line compared to other convective systems for the 2007 warm season is shown in Fig. 3.
Basic radar analysis procedures were performed and the data were gridded to a Cartesian coordinate system. Radar analyses provide 3D fields of reflectivity, rain rates, and raindrop median volume diameter. Analyses for this case focus on a 50 km wide region 25-75 km east of the radar to avoid influence from the cone of silence, and between 0740 and 0800 UTC when most of the storm (in a direction perpendicular to the line) was within the KOUN radar domain. Fig. 4 shows horizontal and line-average vertical contour plots of grided radar reflectivity at three different time slices.
Models should use a quasi-idealized setup for this case, similar to many previous studies (Weisman et al. 1997; Bryan and Morrison 2011; Morrison et al. 2011). In this setup, models are initialized with the observed sounding, and the storm is initiated by inserting a cold pool into a portion of the domain. More details on the initial conditions are given below. We are currently working on a model setup for real-world simulations of this case using 3D lateral boundary and initial conditions from analyses; we will update participants as this work progresses.
For the quasi-idealized setup, a set of two-dimensional (2D) simulations is requested. Lateral boundary conditions should be open, and the model top should be high enough to allow a sufficiently deep sponge layer that will not interfere with tropospheric circulations (therefore 20 km or higher). 9-h model integrations should be performed, although modelers should feel free to run longer simulations. The model domain in the horizontal direction (i.e., perpendicular to the squall line) should be sufficiently large to allow propagation of the squall line while staying sufficiently far away from the boundary within the 9 h integrations (likely to be at least 450 km). For simplicity, models should neglect radiative transfer and surface fluxes and assume a free slip lower boundary. The choice of horizontal and vertical grid spacings is left to the modeler, but should be sufficient to resolve deep convective motion and cold pool evolution (i.e., horizontal grid spacing of order 1 km or less and 350 m and vertical grid spacing below the freezing level of 350 m or less).
For modelers who are able to do so, a set of three dimensional (3D) simulations is also requested. The proposed 3D setup is the same as the 2D setup described above, except for periodic lateral boundaries with a domain length of at least 100 km in the third dimension (i.e., in the horizontal direction parallel to the squall line). For models unable to use periodic lateral boundaries, open boundary conditions should be used but with the domain made sufficiently large so that the central portion of the line (at least 100 km in length) is relatively unaffected by the boundaries. Coriolis acceleration should be neglected.
All simulations should be initialized with the sounding indicated by the gray lines in Fig. 2. These profiles are based on the observed KOUN sounding at 00 UTC 20 June, smoothed with a 1.25 km running average. For baseline simulations, the low-level environmental horizontal wind (u) profile is idealized to give a constant shear of 0.0048 s−1 between a height 0 and 2.5 km, with zero u above. This shear corresponds with a difference in the 0 to 2.5 km horizontal wind of 12 m s−1. For 3D simulations, the environmental wind in the along-line (v) direction is zero. A simple ASCII text file of the initial sounding is provided here.
Convection in both 2D and 3D simulations should be initiated by inserting a low-level cold pool into a portion of the domain. For the initial cold pool, it is recommended that modelers apply a perturbation to the initial θ of −5 K at a height of 0 m linearly decreasing to 0 K at a height of 4.5 km. This θ perturbation should be applied to the portion of the domain between 0 and 200 km in the x-direction (perpendicular to the squall line). For the 3D simulations, modelers should apply small (less than approx 0.1 K) random perturbations to the θ field below 4.5 km and between 175 < X < 200 km to initiate motion in the along-line direction. After initiation of the squall line, it is expected that the storm will propagate toward increasing X. See Fig. 5 for a schematic diagram of the model setup.
As stated in section 1, a primary goal of this intercomparison is to quantify how cold pool-shear interactions vary among models. To this end, we recommend sensitivity tests to investigate further this aspect. In these tests, parameters that are believed to be critical in cold pool evolution are varied, as well as low-level wind shear of the environment. We recommend performing sensitivity tests in both 2D and 3D, although we understand that the large computational burden of 3D simulations may be an issue for some modelers.
Specifically, the following sensitivity tests are proposed. For sensitivity to 0 to 2.5 km environmental wind shear:
- SHEAR8 → As in baseline, but low-level shear decreased to 0.0032 s−1
- SHEAR16 → As in baseline, but low-level shear increased to 0.0064 s−1
- SHEAR20 → As in baseline, but low-level shear increased to 0.008 s−1
- SHEAR24 → As in baseline, but low-level shear increased to 0.0096 s−1
Reminder: the baseline low-level shear is 0.0048 s−1. These tests correspond with 0 to 2.5 km differences in horizontal wind of 8, 16, 20, and 24 m s−1, respectively, compared to the baseline of 12 m s−1.
For each of the above wind shear settings (including baseline), we recommend the following sensitivity tests with altered rain evaporation and hence cold pool strength in addition to runs with unmodified evaporation:
- EVAPL → Decrease rainwater evaporation rate by a factor of 4
- EVAPH → Increase rainwater evaporation rate by a factor of 4
Thus, a total of 14 sensitivity tests plus the baseline simulation is proposed. We anticipate recommending additional sensitivity tests as this project progresses. We also encourage any additional sensitivity tests that modelers may be interested in sharing at the workshop. We especially encourage sensitivity tests of parameters affecting the raindrop size distribution and other aspects of microphysics, as well as horizontal and vertical resolution.
Example of results
Results from a simulation applying the configuration described above are shown in Figs. 6-8. These simulations apply the Weather Research and Forecasting model (WRF) with the two-moment microphysics scheme of Morrison et al. (2009). Figs. 6 and 7 show evolution of the horizontal and line-average vertical cross sections of radar reflectivity, while Fig. 8 shows time-series of various quantities.
As in previous Cloud Modeling Workshop case studies, we recommend modelers bring a series of plots for each simulation to compare at the workshop. A good set to start is the following:
- Horizontal contour of radar reflectivity (dBZ) at 1 km height (nearest model vertical level is sufficient)
- Line-average vertical contour plots of reflectivity (dBZ)
- Line-average vertical contour plots of horizontal wind (m s−1)
- Line-average vertical contour plots of buoyancy (m s−2)
- Line-average vertical contour plots of equivalent potential temperature (K)
- Line-average vertical contour plots of perturbation potential temperature (K)
- Line-average plots of surface precipitation rate (mm h−1) and rate at approx 1 km
- Line-average vertical contour plots of raindrop median volume diameter (mm; averaged only within regions containing rain)
- Scatterplots of raindrop median volume diameter (mm) as a function of rainwater content at the lowest model level (g m−3)
- Line-average vertical contour plots of rain, cloud water, ice, snow, graupel, and hail mixing ratios (g kg−1)
- • surface precipitation rate (mm h−1)
- • total condensation rate (g m−2 s−1)
- • total evaporation rate (g m−2 s−1)
- • total melting rate (g m−2 s−1)
- • maximum updraft/downdraft velocity within the domain (m s−1)
- • average lowest model level perturbation θ within the cold pool (K)
- • fraction of the domain covered by cold pool at the lowest model level
- • cold pool intensity (°C) averaged over the region 30 km behind the surface gust front
- • domain-mean updraft mass flux (only including regions with vertical velocity w > 0m s−1)
- • fraction of the domain with convective updrafts (regions with w > 2m s−1)
- • mean convective updraft mass flux (i.e., conditionally-averaged over regions with w > 2m s−1)
In the above, total condensation, evaporation, and melting rates correspond to vertically-integrated,
domain-averaged values. The cold pool area is defined by the −2 K perturbation θ
isotherm and buoyancy is given by
where g is acceleration of gravity, qv is the water vapor mixing ratio, and qh is the total (from all species) hydrometeor mixing ratio. Overbars denote the model base state, which is assumed equal to the initial values from the environmental sounding.
Cold pool intensity C is given by (Weisman and Rotunno 2004; Bryan et al. 2006):
where H is the cold pool height, defined as the height of the level of neutral buoyancy.
The development of this case study was supported by the USWRP Short-Term Explicit Prediction (STEP) program at NCAR. KOUN radar is operated by NSSL and data was provided by Terry Schuur (CIMMS). Mike Dixon (NCAR) provided guidance on the radar and mesonet data processing. Oklahoma Mesonet data are a cooperative venture between Oklahoma State University and The University of Oklahoma and supported by the taxpayers of Oklahoma.
Brock, F. V., K. C. Crawford, R. L. Elliott, G. W. Cuperus, S. J. Stadler, H. L. Johnson, and M. D. Eilts, 1995: The Oklahoma Mesonet: A technical overview. J. Atmos. Oceanic Technol., 12, 5–19.
Bryan, G., J. C. Knievel, and M. D. Parker, 2006: A multimodel assessment of RKW theory's relevance to squall-line characteristics. Mon. Wea. Rev., 134, 2772–2792.
Bryan, G., and H. Morrison, 2011: Sensitivity of a simulated squall line to horizontal resolution and parameterization of microphysics. Mon. Wea. Rev. (in press).
Ferrier, B. S., J., Simpson, and W.-K. Tao, 1996: Factors responsible for precipitation efficiencies in midlatitude and tropical squall simulations. Mon. Wea. Rev., 124, 2100–2125.
Fovell, R. G., and Y. Ogura, 1988: Numerical simulation of a mid-latitude squall line in two dimensions. J. Atmos. Sci., 45, 3846–3879.
Fridlind, A. M., and Coatuhors: A comparison of TWP-ICE observational data with cloud-resolving model results. J. Geophys. Res. (submitted).
Kruger, A. and W. F. Krajewski. 2002. Two-dimensional video disdrometer: A description. J. Atmos. Oceanic Technol. 19, 602–617.
Lang, T. J., W. A. Lyons, S. A. Rutledge, J. D. Meyer, D. R. MacGorman, and S. A. Cummer, 2010: Transient luminous events above two mesoscale convective systems: Storm structure and evolution. J. Geophys. Res., 115, A00E22, doi:10.1029/2009JA014500.
Li, Xiaowen, Wei-Kuo Tao, Alexander P. Khain, Joanne Simpson, Daniel E. Johnson, 2009: Sensitivity of a Cloud-Resolving Model to Bulk and Explicit Bin Microphysical Schemes. Part I: Comparisons. J. Atmos. Sci., 66, 3–21.
Liu, C., M. W. Moncrieff, and E. J. Zipser, 1997: Dynamical influence of microphysics in a tropical squall line: A numerical study. Mon. Wea. Rev., 125, 2193–2210.
Lynn, B. H., A. P. Khain, J. dudhia, D. Rosenfeld, A. Pokrovsky, and A. Seifert, 2005: Spectral (bin) microphysics coupled with a mesoscale model (MM5). Part II: Simulation of a CaPE rain event with a squall line. Mon. Wea. Rev., 133, 59–71.
McCumber, M., W.-K. Tao, J. Simpson, R. Penc, and S.-T. Soong, 1991: Comparison of ice-phase microphysical parameterization schemes using numerical simulations of tropical convection. J. App. Meteor., 30, 985–1004.
Moncrieff, M. W., S. K. Krueger, D. Gregory, J-L. Redelsperger, and W.-K. Tao, 1997: GEWEX Cloud System Study (GCSS) Working Group 4: Precipitating convective cloud systems. Bull. Amer. Meteor. Soc., 78, 831–845.
Morrison, H., G. Thompson, and V. Tatarskii, 2009: Impact of cloud microphysics on the development of trailing stratiform precipitation in a simulated squall line: Comparison of one- and two-moment schemes. Mon. Wea. Rev., 137, 991–1007.
Morrison, H., S. Tessendorf, K. Ikeda, and G. Thompson, 2011: Sensitivity of a simulated mid-latitude squall line to parameterization of raindrop breakup. Mon. Wea. Rev. (submitted)
Redelsperger, J.-L., and Coatuhors, 2000: A GCSS model intercomparison for a tropical squall line observed during TOGA-COARE. I: Cloud-resolving models. Quart. J. Roy. Meteor. Soc., 126, 823–863.
Rotunno, R., J. B. Klemp, and M. L. Weisman, 1988: A theory for strong, long-lived squall lines. J. Atmos. Sci., 45, 463–485.
Stensrud, D. J., M. Coniglio, R. P. Davies-Jones, and J. S. Evans, 2005: Comments on "A theory for strong long-lived squall lines revisited." J. Atmos. Sci., 62, 2989–2996.
Tao, W.-K., J. R. Scala, B. Ferrier, and J. Simpson, 1995: The effect of melting processes on the development of a tropical and midlatitude squall line. J. Atmos. Sci., 52, 1934–1948.
Van Weverberg, K., A. M. Vogelmann, H. Morrison, and J. A. Milbrandt, 2011: Sensitivity of idealized squall line simulations to the level of complexity used in two two-moment bulk microphysics schemes. Mon. Wea. Rev. (accepted).
Weisman, M. L., W. C. Skamarock, and J. B. Klemp, 1997: The resolution dependence of explicitly modeled convective systems. Mon. Wea. Rev., 125, 527–548.
Weisman, M. L., and R. Rotunno, 2004: A theory for strong long-lived squall lines revisited. J. Atmos. Sci., 61, 361–382.
Xu, K.-M., and Coauthors, 2002: An intercomparison of cloud-resolving models with the Atmospheric Radiation Measurement summer 1997 Intensive Operation Period data. Quart. J. Roy. Meteor. Soc., 128, 593–624.