1. Introduction
Deep moist convection is a prominent source of vertical mass transport and radiative processes that significantly affect climate, weather, and society directly through severe hazardous conditions and hydrometeorology. However, the predictability of deep convective clouds and precipitation remains a complex and elusive problem plaguing climate models (e.g., Dai 2006; Suhas and Zhang 2014; Covey et al. 2016; Christopoulos and Schneider 2021) and regional weather forecasts (e.g., Roberts et al. 2012; Duda and Gallus 2013; Kain et al. 2013; Burghardt et al. 2014; Stelten and Gallus 2017). One fundamental reason for this uncertainty stems from an incomplete understanding of how cloud-scale updrafts interact with the surrounding environment to support or suppress deep convection initiation (CI) (e.g., Duda and Gallus 2013; Zhang et al. 2015; Hirt et al. 2019; Bachmann et al. 2020; Yano and Ouchtar 2017; Hagos et al. 2020). Many recent cloud-scale simulations have illustrated that the dryness of the free troposphere is a salient governor of cumulus growth because of its role in the dilution of in-cloud buoyancy through dynamic entrainment processes (e.g., de Rooy et al. 2013; Morrison 2017; Rousseau-Rizzi et al. 2017; Nelson et al. 2022; Morrison et al. 2022; Peters et al. 2022a,b). Further, updraft thermals may either be suppressed or promoted while ascending through ambient vertical wind shear (Lee et al. 1991; Ziegler and Rasmussen 1998; Peckham and Wicker 2000; Peters et al. 2019a, 2022a,b). Both shear and entrainment effects can vary significantly depending on the horizontal cross-sectional area of the cloudy updrafts. Relatively wide updrafts are more immune to negative entrainment effects (e.g., Khairoutdinov and Randall 2006; Schlemmer and Hohenegger 2014; Rousseau-Rizzi et al. 2017; Morrison 2017; Peters et al. 2019b, 2020; Morrison et al. 2022) and their growth may be promoted by ambient shear (Peters et al. 2022a,b); whereas, relatively narrow updrafts are comparatively vulnerable to entrainment-driven dilution and are instead suppressed by shear. Several other environmental conditions, such as the vertical distribution of convective available potential energy (CAPE) and terrain–flow interactions, also have been shown to contribute complex and variable impacts on cloud-scale CI processes (e.g., Houston and Niyogi 2007; Kirshbaum 2011; Zhang et al. 2015; Kirshbaum et al. 2018; Bachmann et al. 2020; Singh et al. 2022; Nelson et al. 2022). However, few observational studies exist to verify these findings during real-world CI events.
A second fundamental source of CI uncertainty results from the near-cloud environment being poorly represented on a variety of spatiotemporal scales owing to limited routine observations, and inadequate model configuration or initialization techniques (e.g., Wilson and Mueller 1993; Weckwerth and Parsons 2006; Bodine et al. 2010; Gustafsson et al. 2018; Weckwerth et al. 2019; Degelia et al. 2019). Outside of LES or theoretical formulations, much of what we know about environmental influences upon CI processes has been inferred from targeted observations collected during field campaigns or from regional model analyses (e.g., Mueller et al. 1993; Wilson and Roberts 2006; Lock and Houston 2014; Hagos et al. 2014; Weckwerth and Romatschke 2019; Nelson et al. 2021; Barber et al. 2022). For example, Lock and Houston (2014, hereafter LH14) found that the magnitude of background vertical motion associated with synoptic-scale and mesoscale dynamics, CAPE, convective inhibition (CIN), and the vertical excursion required for a parcel to reach its level of free convection (LFC) corresponded to the probability of CI across the U.S. Great Plains. Mueller et al. (1993) found only a weak correlation between CIN and the probability of CI, noting potentially large sensitivity to unobserved mesoscale moisture variations and a need to observe environments near mesoscale CI triggering mechanisms. However, there are few studies examining a large sample of observed near-cloud environments.
The Cloud, Aerosol, and Complex Terrain Interactions (CACTI; Varble et al. 2021) field campaign sought to better understand convective processes occurring within the Córdoba province of Argentina, South America. Scanning radars, vertical profilers, and balloon radiosondes were deployed as part of the U.S. Department of Energy’s Atmospheric Radiation Measurement (ARM) Mobile Facility (AMF) to study terrain–flow interactions and the conditions supporting CI near the Sierras de Córdoba (SDC) range (Fig. 1). CACTI took place from October 2018 to April 2019, collecting 6.5 months of radar measurements of convective cell structure, as well as nearby ambient meteorological conditions.
Map of terrain elevation (shaded) and geographical location of our analyzed population of CI events (white dots). The location of the AMF containing the CSAPR2 and the radiosonde launch site, the nearby city of Córdoba, and CSAPR2 range rings of 10, 15, and 45 km are shown for reference.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
Feng et al. (2022a) tracked the full radar-detected life cycle of ∼6900 convective storms during CACTI. They examined statistical relationships between certain polarimetric radar precipitation characteristics throughout cell lifetime (e.g., maximum reflectivity, cell area, rainfall rate, and hydrometeor properties) in the context of a limited set of environmental metrics (e.g., CAPE, CIN, LNB, and water vapor mixing ratio and shear at specific ground-relative heights or layers) and interactions among neighboring convective cells. They demonstrated a strong affinity for CI to occur along the SDC, that cells with the greatest precipitation signal contained the moistest low-level environments, and that convective cell area growth rates are more sensitive to low-level moisture and shear than CAPE. However, a thorough examination of physical CI processes involved with cloud–environment interactions was left for future work. Nelson et al. (2021) used radiosonde measurements to characterize environments surrounding CI events during the RELAMPAGO campaign (Nesbitt et al. 2021), which jointly operated with CACTI from 1 November to 15 December 2018. They found that environments that failed to produce CI despite affirmative operational model forecasts contained more CAPE but were overall drier and warmer than conditions yielding CI. However, these statistics were processed using a small sample of manually tracked CI and non-CI events (44 total events across 8 case days), with possible biases introduced from project resource limitations.
In this paper, we utilize the Feng et al. (2022a) automated radar cell tracking algorithm and ambient meteorological observations to examine the correspondence between near-cloud environmental measurements and characteristics of convective cells during their CI period throughout the 6.5-month CACTI campaign. We relate environmental metrics to convective cell width and depth during the earliest portion of the convective life cycle to differentiate the conditions of CI events resulting in a spectrum of robust or marginal storms. A primary goal of this work is to verify hypotheses formulated by past studies simulating and theorizing updraft–environment interaction processes that govern CI, particularly those pertaining to 3D cloud structure. This information can potentially guide the improvement of model parameterizations of important cloud growth processes. Section 2 describes the data and methods used to objectively identify CI events. Sections 3 and 4 describe a statistical analysis characterizing environments differentiating a spectrum of successful and failed CI outcomes. Section 5 discusses these outcomes in the context of physical updraft–environment interaction processes controlling CI hypothesized by recent studies. Finally, our work is summarized in section 6.
2. Method
The fundamental technique behind our analysis involves the pairing of individual precipitating convective cells tracked by a scanning C-band radar (CSAPR2; Hardin et al. 2018) and proximal environmental profiles collected by radiosondes launched at the AMF site (Fig. 1).
a. Identification of CI events during CACTI
To track convective activity within the 6.5-month CSAPR2 CACTI database, we use an objective cell-tracking algorithm called “FLEXTRKR” (Feng et al. 2022a,b). We only briefly summarize this method here and refer the reader to Feng et al. (2022a) for more details on its implementation during the CACTI campaign. FLEXTRKR analyzes gridded 3D volumetric radar reflectivity scans to track convective cells in time. CSAPR2 completed ∼5-min-duration plan position indicator survey volumes at regular 15-min intervals throughout CACTI, with elevation angles spanning 0.5°–32°. Three-dimensional reflectivity data are regridded from their native polar to Cartesian coordinates with a 500-m horizontal and vertical grid spacing, using the Py-ART software (Helmus and Collis 2016). This Cartesian grid spacing was chosen as a compromise for the variable radar gate spacing within the radar domain: ∼150-m radial spacing, ∼785 m azimuthal at 50 km and 500 m at ∼32 km for the CSAPR2 beamwidth of 0.9°. Convective cells are identified using a variety of criteria, including radar echoes that either contain (i) a minimum composite (column maximum) radar reflectivity core of 60 dBZ or (ii) cores defined by the “peakedness” of their composite reflectivity using variable reflectivity threshold differences (maximum 10 dBZ) between points and their surrounding background reflectivity (averaged within a 11-km radius). Convective cores must occupy a minimum 4-km2 horizontal area. Cores are subsequently expanded by a reflectivity based radius to identify individual cells. Individual cells are tracked in time using a first guess advection velocity that is estimated via 2D correlation of composite reflectivity masks from sequential radar volumes. We only consider cells persisting for at least two consecutive radar volumes. Several aspects of the three-dimensional precipitation structure for each tracked cell are saved throughout the cell lifetime, including horizontal shape and area of the precipitation footprint at each height, echo-top height, and magnitude of reflectivity. Though the spatial resolution of radar data and our 4-km2 minimum area filter may be expected to curtail the narrowest tail of the cell area distribution, many of these narrowest cells that do not continue to grow to meet our detection thresholds are removed when filtering those lasting less than two radar scans. Thus, the 15-min radar frequency is one salient limitation of the dataset when examining rapidly evolving deep convection. Finally, although the CI process realistically includes a period of deepening cumulus congestus occurring prior to effective radar detection, for practical purposes, we consider the time and location of CI for each cell to be the first in its radar-tracked history.
We utilize a subset of the ∼6900 convective cells tracked during the CACTI campaign (Feng et al. 2022a,b). Our subset contains cells occurring between 1200 and 0000 UTC (0900–2100 LT) on each day to focus on convective processes driven by daytime heating and coincident with the daily period when radiosondes were launched. Depending on the daily forecasted potential for deep convection, radiosondes were launched at 1200, 1500, 1800, 2100, and 0000 UTC, or 1200, 1600, 2000, and 0000 UTC. To reduce temporal measurement uncertainty of the near-cloud environment, we only consider CI events occurring during periods when: (i) consecutive radiosondes were launched within a 4-h window, or (ii) within 30 min of an individual radiosonde launch. We also only analyze CI events detected within 45 km of the AMF site and located east of the SDC ridgeline, where radiosonde measurements were made. The 45-km maximum range is chosen to minimize spatial uncertainty in measurements of the near-cloud environment owing to potentially significant mesoscale heterogeneity found in areas of complex terrain (e.g., Zehnder et al. 2009; Behrendt et al. 2011; Kalthoff et al. 2009; Khodayar et al. 2010; Nelson et al. 2021; Marquis et al. 2021), and is based on estimates of temperature, wind, and moisture heterogeneity surrounding the SDC using a coordinated array of radiosonde instruments (Nelson et al. 2021). We exclude cells involved in splitting events and those that end in a merger with neighboring cells within the first two times of their radar-tracked life cycle to limit our analysis to newly forming and relatively isolated CI events. Finally, we exclude events if their environments contain: (i) vertically integrated buoyancy,
b. Classifying CI outcome
We classify various CI outcomes by: (i) overall cell lifetime, (ii) cell width, and (iii) cell depth relative to its predicted MU equilibrium level (EL). To facilitate a simple statistical comparison of a large number of environmental metrics differentiating various CI processes, we discretize each of these CI outcomes into binary categories of: (i) Long- or short-lived cells (total duration ≥ 3 or <3 consecutive radar volumes, respectively, where each volume is separated by 15 min), (ii) deep or shallow CI events relative to their maximum depth potential predicted by parcel theory [10-dBZ echo-top height (ETH) ≥ 66% or <66% of the MU EL height, respectively], and (iii) wide or narrow CI events (horizontal area occupied by the 20-dBZ reflectivity contour for each cell ≥ 50 or <50 km2, respectively). We gauge the CI outcomes using the maximum width and depth achieved by each cell during the first two times of its radar-detected track (hereafter, referred to as the “CI period”). Therefore, these categories can be interpreted as comparing the “relatively rapidly” to the “relatively slowly” deepening and widening CI events. Only ∼16% of slowly deepening cells achieve the 0.66 × ETH/MU EL threshold after the CI period. Further, only 13% of slowly widening events surpass the 50-km2 cell area threshold after the CI period. Thus, ∼84% and 87% of slowly deepening and slowly widening CI events remain shallow or narrow, respectively, during their full cell lifetime. We impose a minimum distance from the AMF site of 10 and 15 km for including CI events in our analysis of cell width and depth, respectively. Thus, we neglect cells initiating within the radar cone-of-silence so that we more confidently characterize 3D reflectivity structure. To maximize population size, we do not apply these minimum range criteria when quantifying cell duration or broadly characterizing environments of all CI events for which cell width and depth are not considered. Approximately 60% of our tracked cells exceed two radar scans. If multiple CI events occur simultaneously, only the most successful one (i.e., the longest-lived, deepest, widest) is included in our population.
Finally, we define a “non-CI” category to identify times during the campaign when precipitating convection is not detected by CSAPR2. Analysis times fitting this category contain no FLEXTRKR-identified convective cells within a running 6-h window and 60-km distance from the AMF site. This 6-h window is designed to exclude brief periods of convective inactivity between CI events. The expansion of the maximum range from 45 to 60 km for this category is chosen to conservatively exclude deep convection while still minimizing uncertainty of sampling free-tropospheric mesoscale heterogeneity in this region (based on measurements by Nelson et al. 2021). As with CI-supporting environments, we only consider times between 1200 and 0000 UTC that are both preceded and succeeded by radiosondes launched within 2 h or times within 30 min of a radiosonde launch. We exclude times when there are no radar observations to confirm a lack of precipitating convection, and impose the same minimum thresholds for MU IB below the LFC, MU CAPE, and MU EL. In total, these constraints yield a population of 212 CI-supporting and 111 non-CI environmental profiles, spanning 33 and 16 different campaign days, respectively (Table 1, Fig. 2).
Sample size of CI and non-CI outcomes examined using detection criteria defined in section 2. Differences in the size of combined “wide” and “narrow” events population (201), and the combined “deep” and “shallow” events population (182) from the total “all CI” events population result from the minimum radar distance criteria imposed when quantifying cell width and depth (10 and 15 km, respectively). No such criteria were imposed when quantifying cell duration.
(top) Number of CI and non-CI events on each day of the CACTI campaign using detection criteria defined in section 2. (bottom) As in the top panel, but for the number of deep and shallow CI events.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
c. Characterizing the near-cloud environment
To characterize the near-cloud environment for each CI event, we utilize the ARM value-added “interpolated sonde” product (“INTERPSONDE”; Jensen et al. 2019), which synthesizes sequential balloon radiosonde profiles (Holdridge et al. 2018) with total column water vapor retrieved by a microwave radiometer (Gaustad et al. 2011) collected at the AMF site. Linear temporal interpolation between launches with moisture scaled by total column water vapor yields full tropospheric temperature, humidity, pressure, and horizontal wind profiles at 1-min intervals at time-constant vertical height levels. Profiles assumed to be representative of the near-cloud conditions for each CI event are the INTERPSONDE profiles contemporaneous with the first time in each FLEXTRKR cell track. Profiles for non-CI events are those valid at times meeting the criteria described in the previous paragraph.
Using these near-cloud CI and non-CI profiles, we calculate an extensive list of sounding metrics, e.g., CAPE, CIN, vertical lapse rates of temperature, moisture, wind speed and direction, vertical wind shear, terrain-relative and cell-relative flow, and numerous derivative quantities. In total, we evaluate 180 parameters (Table S1 in the online supplemental material). This list builds upon foundational metrics examined by past studies (e.g., Johns and Doswell 1992; Mueller et al. 1993; LH14; Peters et al. 2019a; Nelson et al. 2021; Barber et al. 2022) and incorporates additional metrics related to physical processes hypothesized herein to be relevant for CI along the nearby SDC. Metrics are calculated at a variety of standard heights and over vertical layers relevant to various cloud processes; e.g., throughout the depth of the convective boundary layer (objectively determined using a method from Liu and Liang 2010), the free troposphere (between the boundary layer top and EL), the subcloud layer (below the LCL), the active cloud bearing layer (ACBL; the layer located between the height of the LFC and 1.5 km above it; LH14), and an effective inflow layer (EIL; the layer containing CAPE > 100 J kg−1 and CIN > −10 J kg−1).2 Most lifted parcel metrics (e.g., CAPE, CIN, LCL, LFC, EL) are calculated for surface parcels (SFC), MU parcels, and a mean-layer parcel (ML) with traits equal to the mean of the lowest 100 hPa of the sounding that accounts for mixing of an ascending boundary layer thermal with its surroundings in the lowest part of the atmosphere (Craven et al. 2002; Markowski and Richardson 2010). In addition to traditional formulations of CAPE and CIN that represent exclusively integrated positive and negative buoyancy, respectively, we calculate total IB that sums both positive and negative buoyancy within a prescribed layer. This is done to account for net potential buoyancy acceleration in the upper boundary layer and lower free troposphere where temperature profiles sometimes contain fine-scale vertical heterogeneity. Finally, we evaluate background larger-scale ascent surrounding CI and non-CI events throughout various vertical layers by area-averaging profiles of ERA5 model reanalysis of vertical motion within a 2° × 2° region centered at the AMF site.
3. Comparing CI versus non-CI environments
Our first objective is to compare environments that support CI events (of any cell duration, depth, or width) to those of non-CI events. Soundings typifying each event type are shown in Figs. 3a, 4, and 5. Prima facie, mean environments of non-CI events generally have warmer and drier vertical profiles that are coincident with weaker background vertical motion and overall less stable temperature lapse rates throughout the vertical profile than in typical CI-supportive scenarios (Figs. 3a and 4a,b). These findings are qualitatively consistent with the findings of Nelson et al. (2021) among their limited sample of events in this region, and in a variety of other global regions (e.g., Johns and Doswell 1992; Weckwerth and Parsons 2006; Hagos et al. 2014; LH14; Nugent and Smith 2014; Barber et al. 2022). These differences are most evident in the lowest ∼5 km or ∼500 hPa of the atmosphere. The spread among the CI and non-CI thermodynamic profiles contains more overlap above this altitude, suggesting potentially fewer significant differences aloft (e.g., Figs. 3a and 4a,b).
Skew T–logp diagrams of the mean temperature and dewpoint temperature profiles for (a) all CI (teal lines) and non-CI (red lines) events; and (b) deep CI (green lines) and shallow CI (blue lines) events. Shaded areas represent ±1 standard deviation of the temperature and dewpoint temperature profiles at each height.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
Mean vertical profiles of (a) relative humidity, (b) lapse rate of virtual temperature, (c) magnitude of the vertical shear of horizontal wind, and (d) ERA5 vertical motion (“omega” profiles averaged within a 2° × 2° area centered on the AMF site) for the populations of all CI events (teal lines) and non-CI events (red lines). Shaded areas represent ±1 standard deviation in vertical profiles for each population.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
(a) Mean low-level horizontal wind vector (averaged over z = 0–1 km) for all CI events (teal), deep CI events (green), shallow CI events (blue), and non-CI events (red). Shaded areas denote the range of 0–1-km mean wind vectors for deep, wide, and long-lived CI events (green) and shallow, narrow, and short-lived events (blue). Terrain elevation (shaded) is included to illustrate cross-ridge and ridge-parallel flow orientations. (b) Hodographs of the mean wind profile for all CI events (teal), deep CI events (green), shallow CI events (blue), and non-CI events (red). Diamonds indicate the near-surface height along the hodographs and dots indicate z = 1, 3, 6, and 9 km AGL.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
We assess statistically significant differences across CI and non-CI outcomes in each sounding metric from our list of 180 environmental parameters using paired Kolmogorov–Smirnov tests at the 0.05 significance level. We perform these over the population of tracked CI and non-CI events described in section 2b, as well as populations subsampled at 1.5-, 3-, and 24-h3 frequency. These sampling frequency tests evaluate the sensitivity of statistical significance to potential autocorrelation bias arising from radiosonde sampling frequency and clustering of events relative to the temporal scales of evolving background mesoscale and synoptic-scale meteorological features (e.g., Zwiers and von Storch 1995). Though a full list of these test results can be found in supplemental material (Table S2), an abridged version summarizing the statistically significant differences is shown in Table 2. Most of the environmental parameters reported in Table 2 are significantly different for at least three of the four tested sampling frequencies, suggesting relative robustness against potential autocorrelation biases. To summarize these environmental differences:
Abridged list of environmental metrics that are statistically different between the populations of all CI and non-CI events using paired Kolmogorov–Smirnov tests. Metrics are as defined in Table S1. Metrics for which multiple assumed parcel origins show qualitatively similar differences are listed for each entry (e.g., “for SFC and ML parcels”). Statistical significance at various event sampling frequencies (e.g., the full population as described in section 2—“full sample”, and sampling at 1.5, 3, and 24 h) is indicated. Metrics are only included if they are statistically different for at least two of the subsampled populations. The mean of each metric for each CI outcome (for full sample of events) is listed. The p values are listed for the tests on the full sample of events. All heights are valid above ground level. A complete list of differences across all tested metrics and sampling frequencies is shown in Table S2.
-
Instability: Non-CI environments had 2–3 times greater mean full-tropospheric CAPE contained within 15%–40% deeper layers (i.e., EL–LFC), 2–3 times greater mean concentration of CAPE per kilometer in that layer (i.e., “buoyancy”), and significantly greater CAPE and IB in the lowest parts of the cloud (e.g., ACBL) than CI environments. Further, non-CI environments contained steeper (more unstable) lapse rates in the ACBL and more rapid CAPE tendency than in typical CI environments (Table 2).
-
Inhibition: Non-CI environments had weaker mean SFC CIN but slightly stronger MU CIN, shallower layers containing SFC and MU CIN (thus, net greater IB below the SFC LFC and MU LFC), and more rapidly diminishing CIN (positive CIN tendency) than typical CI environments. However, non-CI environments had greater mean magnitude and depth of ML CIN (and more negative IB below the ML LFC), as well as higher mean LFCs (by ∼200–700 m) than CI-supportive conditions (Table 2).
-
Effective inflow layer: Non-CI environments contained deeper and warmer boundary layers (by ∼400 m) but EILs that were shallower (by ∼200 m) and located slightly closer to the ground than typical CI-supporting conditions. The mean upper bounds of non-CI EILs are ∼400 m lower than the mean boundary layer depth, while mean EILs for CI events generally occupied the mid-to-upper boundary layer and lower free troposphere (Table 2).
-
Thermodynamic properties: Non-CI environments had greater mean water vapor mixing ratio in the EIL (by ∼2 g kg−1) and overall warmer temperatures below midlevels of the troposphere (e.g., by ∼7 K, averaged over the depth of the boundary layer) than CI environments. Non-CI environments also had lower relative humidity throughout the lower half of the troposphere than CI environments (by ∼20%–60% in certain layers). These properties resulted in non-CI environments having higher freezing levels (by ∼500 m) and higher SFC and ML LCLs (by up to 700 m) (Fig. 4, Table 2).
-
Background ascent: Non-CI environments contained slightly weaker mean upslope low-level flow along the east side of the SDC (by ∼1 m s−1, respectively) and 2–3 times weaker mean background large-scale ascent throughout the vertical column than in typical CI environments (Figs. 4d and 5a, Table 2).
-
Horizontal wind: Non-CI environments contained stronger northerly (poleward) low- to midlevel flow along the SDC (by ∼2 m s−1), a slightly weaker mean magnitude of vertical wind shear throughout relatively deep and near-surface layers (e.g., 0–6 km, the subcloud layer, a 1-km-deep layer near the terrain peak), but similar or slightly stronger mean shear in the free-tropospheric layer than in CI environments (Figs. 4c and 5b, Table 2).
These differing conditions are consistent with a variety of physical processes. For one, it is possible that warmer non-CI environments that yielded higher freezing levels may have contributed to suppression of CI because of the additional vertical cloud growth required to yield ice formation, which leads to greater reflectivity and precipitation and possible invigoration of updrafts from the latent heat of fusion (e.g., Arakawa 2004; van den Heever et al. 2011; Storer and van den Heever 2013; Nelson et al. 2021). Another physically plausible explanation involves differing amounts of the dilution of buoyant updrafts from entrained dry ambient air. Relatively dry conditions aloft in non-CI environments are consistent with recent studies pointing to entrainment of free tropospheric air into updrafts as a prominent suppressing effect on growing cumulus (e.g., de Rooy et al. 2013; Morrison 2017; Rousseau-Rizzi et al. 2017). However, entrainment-driven dilution of buoyancy may not be limited to midlevels of the atmosphere. Though non-CI events have more MU CAPE than in CI events, contrary to findings of LH14, MU parcel origins in non-CI environments are closer to the ground and have statistically shallower layers of subcloud moist static energy to tap in to (e.g., shallower EIL depth; Table 2). Thus, lifted non-CI parcels may be exposed to an unfavorable entrainment process over a deeper layer than lifted CI parcels before the updraft can achieve significant in-cloud positive buoyancy (LH14; Mulholland et al. 2021). This echoes the importance of accounting for the mixing of air from the subcloud layer and the lower free troposphere into an ascending thermal, as reflected in the differences in ML CIN across CI and non-CI events (e.g., Weckwerth et al. 1996; Markowski et al. 2006; Markowski and Richardson 2010; LH14; Mulholland et al. 2021).
Our dataset cannot confidently characterize the 3D structure of updrafts accompanying the orographic circulation or other mesoscale dynamic forcings (e.g., air mass boundaries, gravity waves). Hypothetically, even if the depth and strength of such low-level updrafts were similar across CI and non-CI environments, mean low- to midlevel large-scale background ascent in typical non-CI environments is too weak to lift parcels to their LFCs, which are higher for non-CI events than for CI events. Although ascending air on a variety of spatial scales would also be expected to moisten and destabilize the ambient near-cloud conditions (e.g., Ziegler et al. 1997), generally steep lapse rates already present in typical non-CI conditions (Figs. 3 and 4b) suggests that a lack of large-scale lifting of ambient moisture and conditionally unstable parcels to their high LFCs were primary factors preventing CI.
4. Comparing environments across CI outcomes
Having established statistically significant differences between environments associated with CI and non-CI events, we repeat this method to compare environments among the population of 212 CI events leading to different storm durations and relatively rapidly or slowly developing convection (using cell width and depth properties measured by radar during the CI period described in section 2b).
a. Deep or shallow CI events
We start our examination differentiating events by radar-measured cell depth (10-dBZ ETH) relative to the MU EL during the CI period. This metric perhaps best compares to the framework used by some cloud-scale LES studies studying the survival probability of buoyant cloudy thermals ascending through the free troposphere (e.g., LH14; Rousseau-Rizzi et al. 2017; Peters et al. 2022a,b; Nelson et al. 2022). Recall, that ∼84% of slowly deepening cells (ETH/MU EL < 0.66 during the first two radar scans) remain shallow throughout their lifetime; therefore, we can expect conditions associated with slowly deepening CI events to inform upon processes preventing deep convection from reaching its full estimated depth potential in most cases.
Comparisons of environments associated with slowly and rapidly deepening cells (hereafter “shallow” and “deep” CI events, respectively) are shown in Figs. 3b and 5–7, and Table 3. Overall, deep and shallow thermodynamic and wind profiles bear many qualitative similarities, with spread of their temperature and RH profiles overlapping throughout most of their depth (Figs. 3b, 5b, and 6a,b). The same is true for comparisons of long-lived versus short-lived events and wide versus narrow events (not shown). Further, there are far fewer environmental metrics that are statistically significantly different across the robust and marginal CI outcomes for several of the four different population sampling frequencies (the full sample, 1.5-, 3-, and 24-h sampling) than in the analysis comparing CI to non-CI outcomes (cf. Tables 2–5). Though this could indicate a possible vulnerability of the comparison of robust versus marginal CI outcomes to autocorrelation bias, it may also speak to the similarities in their environments. Thus, in general, differences between robust and marginal CI events are comparatively subtle compared to between CI and non-CI events.
As in Fig. 4, but for deep CI events (green), shallow CI events (blue), and non-CI events (red).
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
Histogram of the timing of shallow CI events relative to the first deep event occurring on each day. A negative (positive) time offset indicates that a shallow CI event occurs before (after) the first deep CI event on each day.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
As in Table 2, but for paired statistical tests across deep and shallow CI outcomes (defined by cells with 10-dBZ ETH/MU EL ≥ 0.66 or <0.66, respectively). Metrics with statistical differences at only one sampling frequency are included. A complete list of differences across all tested metrics are shown in Table S2.
As in Table 3, but for statistical tests across long-lived and short-lived cell duration (defined by tracked cell lifetime > 2 or ≤2 consecutive radar scans, respectively). A complete list of differences across all tested metrics is shown in Table S2.
As in Table 3, but for statistical tests across wide and narrow CI events (defined by horizontal area occupied by the 20-dBZ composite reflectivity contour > 50 or ≤50 km2, respectively). A complete list of differences across all tested metrics are shown in Table S2.
Among all tested environmental metrics, those with the highest degree of confidence against possible serial sampling biases (i.e., significantly different at three or more sampling frequencies) include: (i) low-level northerly flow (greatest for shallow CI events), (ii) background ascent at midlevels of the atmosphere (e.g., 500 hPa; greatest for deep CI events), and (iii) CAPE tendency near the time of CI (greatest for shallow CI events) (Table 3). Additional metrics are statistically significantly different; however, only across two or fewer population sampling frequencies (thus, lower confidence against possible autocorrelation bias):
-
Humidity: Environments of shallow CI events contained slightly greater mean boundary layer moisture (by ∼5%–10%) and lower mean MU LCLs (by ∼700 m), but slightly drier free tropospheric mean RH (by ∼10%) than environments supporting deep CI events (Fig. 6a, Table 3),
-
Instability: Environments of shallow CI events contained greater mean full-tropospheric CAPE (by 20–250 J kg−1, or 4%–30%), higher mean ELs (by ∼900–2500 m) and deeper full-tropospheric CAPE-containing layers (EL-LFC; by ∼1000–2200 m), greater mean CAPE and IB in lowest parts of the cloud (e.g., in the ACBL and near the LCL; by ∼25–50 J kg−1), and higher mean equivalent potential temperature (θe) at low levels (by ∼3.5 K) than environments supporting deep CI (Table 3),
-
Ascent: Shallow CI environments contained stronger low-level ascent (but weaker upper level ascent), slightly greater and deeper low-level mean upslope (easterly) flow along the SDC (by ∼0.4 m s−1 and ∼400 m, respectively), lower mean MU parcel origins (by ∼350 m) and slightly shallower layers for MU parcels to traverse to reach their LFCs (by ∼50 m) than in environments supporting deep CI (Figs. 5a and 6d, Table 3),
-
Terrain- and cell-relative flow: Shallow CI environments contained a slower easterly component of the mean cell motion vector (by ∼4 m s−1), corresponding to a weaker magnitude of the mean low-level cloud-relative inflow winds (by ∼3 m s−1) than deep CI events (Fig. 5, Table 3),
-
Shear: Shallow CI environments contained slightly weaker mean bulk vertical wind shear in the subcloud layer (for MU parcels) and in a layer near the terrain peak (by ∼2–3 m s−1), but greater mean shear spanning the free troposphere for SFC parcels (by ∼5 m s−1) than in deep CI environments (Figs. 5b and 6c, Table 3),
-
Timing of CI: The hour of CI for each event was not statistically different across deep or shallow events. Although shallow CI events sometimes initiate prior to deep CI events, more shallow CI events occur simultaneously with or after deep events (Fig. 7). Thus, it appears that the time distribution of shallow CI events relative to deep ones is not necessarily a result of an environment that is becoming increasingly favorable throughout each day (e.g., through a steady reduction of CIN and increase of CAPE).
This analysis may suggest a few relationships of certain atmospheric metrics across the spectrum of deep, shallow, and non-CI outcomes. For example, the deepest CI events occur when mean background ascent is the deepest (low through midlevels of the atmosphere), while shallow CI events occur when mean background ascent is large only at low-levels and non-CI events occur when mean background ascent is relatively weak at all heights. This ascent profile corresponds to slightly greater midlevel RH for deep CI events than for shallow and non-CI events (Figs. 6a,d) and is consistent with background lift moistening the atmosphere for deep convection (e.g., Ziegler et al. 1997; Markowski and Richardson 2010; Singh et al. 2022). In addition, inflow air may more readily reach the LFC in deep CI environments than in shallow or non-CI environments because MU parcel origins for deep CI events are more elevated, where the background large-scale lift is strong.
Mid- through low-level RH was a strong discriminator between CI and non-CI outcomes (section 3). However, mere modest differences in boundary layer and free tropospheric RH across deep and shallow CI events, as well as lower confidence from potential autocorrelation bias ultimately make it difficult to definitively conclude if there is a monotonic relationship between RH conditions and the full spectrum of deep, shallow, and non-CI outcomes. Free tropospheric RH is greatest for the deepest CI events and is the driest for non-CI events. There is not a similar correlation between the spectrum of CI and non-CI outcomes and boundary layer RH. These relationships between the vertical profiles of RH and the probability of deep CI may be consistent with the relative importance of the entrainment of ambient dry air aloft over the subcloud thermodynamic properties explored for shallow cumulus and applied to deep CI (e.g., Romps and Kuang 2010; Rousseau-Rizzi et al. 2017; Morrison 2017). However, it is important to note that there is a far greater variability of observed near-cloud profiles across our population of environments than in the idealized and controlled conditions prescribed in those past studies, increasing the complexity of comparing the relative importance of individual ACBL and subcloud ingredients in our dataset.
Interestingly, there is an inverse proportionality between the probability of deep CI and both full tropospheric and ACBL CAPE, which differs from some past studies performed outside of orographic forcing regimes (e.g., Houston and Niyogi 2007; LH14; Peters et al. 2022a,b). Additionally, the probability of deep CI is inversely proportional to the depth of the CAPE layer (EL-LFC). Perhaps a simple interpretation of this is that it is more difficult for deepening convection to survive to a high EL if we assume a constant entrainment rate with all other parameters equal. However, if the magnitude of CAPE is the same for events with differing EL-LFC, the vertical concentration of CAPE in that layer will differ. This difference will likely affect the ascent rate and residence time of thermals within the dry free troposphere, altering their exposure period to layers of deleterious entrainment processes before reaching their depth potential and producing precipitation (Houston and Niyogi 2007). Further, as will be discussed in section 5, relatively deep CI events are also relatively wide; therefore, updraft dilution rates are unlikely to be the same across these cases. These findings suggest the importance of entrainment in the lower and middle free troposphere as a prominent governor of CI, even in the presence of substantial CAPE and minimal CIN (e.g., Houston and Niyogi 2007; Morrison et al. 2022), but also indicate enormous complexities involved with disentangling their relative effects on CI outcome. Finally, in addition to potential autocorrelation bias uncertainty, there is substantial sensitivity in the statistical significance of certain full tropospheric and ACBL CAPE metrics and the vertical concentration of full tropospheric CAPE as a discriminator between deep and shallow CI when the ETH/EL threshold of events are varied by 10%–15% (not shown), increasing uncertainty in the interpretation of these entrainment processes in the population of robust and marginal CI events.
Some environmental metrics that differentiated CI and non-CI events did not differentiate shallow and deep CI events. For example, freezing levels are not significantly different across shallow and deep CI events. There are virtually no significant differences in SFC or ML CIN parameters across shallow or deep CI events (among the population with MU CIN > −10 J kg−1). However, expected relationships between reduced CIN and ascent are difficult to interpret because of the dependence of CIN magnitude on the assumed parcel origin and the 2° × 2° area-averaged ERA5 ascent may poorly capture highly localized reduction of CIN and low-level mesoscale flow convergence along the terrain that affects CI probability (e.g., Banta 1984; Demko and Geerts 2010; Kirshbaum 2013; Kirshbaum et al. 2018; Marquis et al. 2021).
Upslope flow represents a localized source of background lift that could prime the near-cloud environment for CI, or in some cases lessen the chances of CI by displacing accumulated boundary layer buoyancy and flow convergence from the terrain peak (e.g., Kirshbaum 2013; Kirshbaum et al. 2018; Nelson et al. 2022). However, there does not appear to be a clear monotonic relationship between the low-level easterly (upslope) component of the near-cloud flow and the probability of deep CI. It is possible that CI may not have occurred in response to mechanical orographic forcing in all cases; rather, other mechanisms (e.g., larger-scale ascent, other mesoscale surface boundaries, terrain-induced waves, etc.) may have played roles in cloud growth. Interestingly, there is a more discernible monotonic relationship between the potential for deep CI and the magnitude of the meridional (along-ridge) wind component. Deep CI events have the least northerly (poleward) low-level mean flow and non-CI events have the strongest (Fig. 5a). It is difficult to confidently generalize differences in vertical wind shear across the full spectrum of deep, shallow, and non-CI outcomes because statistical significance is sensitive to CI outcome, assumed parcel origin, and sampling frequency of each population. For example, though the mean free-tropospheric bulk shear for SFC parcels is weakest for deep CI events and strongest for non-CI events, mean free-tropospheric bulk shear for MU parcels is strongest for shallow CI events, weakest for deep CI events, and intermediate for non-CI events (Tables 2 and 3). Further, certain deep layer bulk shear metrics (e.g., 0–6 and 0–12 km) are significantly different among CI versus non-CI events, but not among shallow versus deep CI events. Possible physical relationships between CI probability and northerly flow, shear, and the magnitude of cloud-relative flow are further discussed in section 5.
b. Long-lived versus short-lived events
Environments measured during the CI period of long-lived convective cells contained few significant differences from those of short-lived cells, and none were significant for three or more of the tested population sampling frequencies (Table 4). Thus, conditions differentiating cell durations were even more difficult to generalize than conditions differentiating the probability of CI or depth of cells during their CI period. Many metrics traditionally seen as fundamentally important ingredients for CI (e.g., low- or midlevel RH, background larger-scale ascent, LFC height, CAPE, and most CIN and shear metrics) during the CI period did not statistically differentiate cell duration. Environments of long-lived cells had slightly warmer, shallower, and less stable boundary layers, and a higher vertical concentration of SFC CAPE, but slightly deeper MU CIN and stronger stable layers above the boundary layer than shorter-lived cells. Longer-lived cells occurred in slightly stronger upslope flow along the SDC and had a slightly slower northerly (poleward) flow than shorter-lived ones. However, with such a short and somewhat disparate list of significant environmental differences, it is difficult to link the ambient conditions during the CI period to processes associated with storm longevity.
c. Wide versus narrow CI events
As with comparisons between long-lived and short-lived CI outcomes, there are no environmental metrics that statistically differ at a 3- or 24-h event sampling frequency, suggesting either uncertainty from potential autocorrelation bias and/or similar conditions for rapidly widening cells (i.e., “wide” CI events) and slowly widening cells (“narrow” CI events).4 Metrics with the highest confidence of discriminating between wide and narrow outcomes are: (i) zonal cell motion and cell-relative inflow between 0 and 4 km, MU LFC, SFC CIN, and bulk shear in the subcloud layer (each greatest for wide CI events); and (ii) depth and strength of the low-level upslope wind (greatest for narrow CI events). Although wide events contained greater CAPE and steeper lapse rates in the ACBL, the importance of this metric exhibited moderate sensitivity to ∼10%–15% changes in the cell area threshold used to distinguish wide from narrow cells (not shown), adding additional uncertainty for it being a discriminator of CI event width. Thus, while wide CI events had the upper hand in terms of cell-relative inflow metrics, they also faced more adverse inhibition metrics, had less ascent from upslope flow, and would have more difficulty lifting parcels to their LFC than narrow events. Based on idealized simulations and theory, complex cloud–scale relationships can exist between updraft width, shear, and the probability for deep cloud growth (e.g., Peters et al. 2019a, 2022a,b), which we will unravel in greater detail in section 5b.
Although a variety of other metrics statistically differ per relative width of CI events, they do so at only one of the tested population sampling frequencies and sometimes with only small differences. For example, mean values of several low- and midlevel RH and water vapor metrics vary across wide or narrow CI events by less than ∼4%. Several CAPE metrics differ among wide and narrow CI events; however, it is difficult to generalize the conditions favoring each outcome. For example, wide CI events are associated with slightly greater MU ACBL CAPE and deep layer ML CAPE, but it is confined to a shallower EL-LFC layer than for narrow CI events. Lifted MU parcels ascending into wide CI events are slightly more elevated in origin and have a slightly greater vertical distance to travel to reach their LFCs but do so within areas of stronger midlevel background ascent than for narrow CI events. Wide CI events tended to occur in slightly weaker deep layer and free tropospheric shear (e.g., 0–6 km, 0–12 km, free troposphere for MU parcels), but stronger shear confined to the lowest half of the atmosphere (e.g., the subcloud layer and the 1-km-deep layer near the local SDC peak).
To succinctly summarize results across our tested CI outcomes, more environmental metrics statistically differentiate all CI events from non-CI events than differentiate CI events of differing duration, width, or depth relative to the EL. Furthermore, relatively marginal CI events (i.e., shallow, short-lived, narrow) often occur near comparatively robust ones (deep, long-lived, wide) in both space and time (e.g., Fig. 7). This may indicate that differences in unobserved mesoscale forcings, environmental evolution occurring faster than the 3–4-hourly radiosonde launch frequency, or complex intracloud processes (e.g., microphysical properties) differentiate these outcomes. Similarities among the soundings of relatively robust and marginal CI outcomes may also stem from radiosondes sampling portions of preexisting or neighboring clouds that result in nearly saturated layers. Though this may serve to narrow the RH, CAPE and CIN distributions across each CI outcome, moistening of the environment from detrainment of neighboring or prior clouds may be a realistic process that promotes CI (Damiani et al. 2008; Waite and Khouider 2010; Moser and Lasher-Trapp 2017).
5. Discussion
In this section, we further interpret our analyzed environmental parameters in the context of specific physical processes hypothesized by past LES and theoretical studies to play roles in CI, such as mesoscale terrain–flow interactions, entrainment, wind shear, and their relationship with cloud width during CI.
a. Terrain-relative flow during CI
One common theme spanning the spectrum of relatively robust, marginal, and non-CI outcomes is the strength of the low-level meridional flow measured at the AMF site. Non-CI environments had the strongest northerly flow in the lower troposphere, while the most robust CI environments had the weakest (Fig. 5). The reasons for this are not completely clear. We might expect strong northerly low-level flow to increase poleward advection of Amazonian moisture and potentially increase residence time of subcloud inflow within the north–south-oriented mesoscale orographic convergence region for clouds deepening along the ridge, each increasing the chances for deep CI. However, these effects, if occurring, do not appear to instigate CI. Singh et al. (2022) demonstrate that circulations resulting from northerly low-level jet winds impinging on the SDC range alter the mesoscale moisture and CIN distribution to disfavor CI in the area surrounding the AMF site targeted by our study, instead favoring CI in the confluence region near the southern wake of the SDC range (e.g., south of ∼−32.5° latitude; Fig. 1). Though this may occur during certain cases, a qualitative examination of GOES-16 visible imagery indicates that extensive cumulus development generally was not occurring in the southern wake of the SDC ridge on most non-CI days in our population (not shown). Though it is difficult to verify the Singh et al. process without detailed mesoscale flow observations fully surrounding the SDC range, a correlation between northerly flow and weak background ascent may also indicate substantial modification of the near-cloud environment by larger-scale flow that results in unfavorable conditions for CI.
b. Relationships between environment and cell width during CI
Figures 8 and 9 show the correspondence between horizontal cell area and depth during the radar-measured CI period to a variety of ambient conditions tied to specific updraft–environment interaction processes discussed in recent literature. Recall that in the lack of direct updraft observations, we measure cell area using the largest horizontal area occupied by the 20-dBZ reflectivity contour in the vertical. This measurement implicitly assumes that each cell’s precipitation footprint during the CI period correlates with the spatial scales of the updraft generating it. Recent and ongoing work using convection-allowing simulations indicates a correlation between updraft area and the precipitation footprint during the convective life cycle (Nicol et al. 2015; Zhang 2022; Nowotarski et al. 2022). However, there may be conditions that could weaken this assumption, such as a relatively wide precipitation footprint caused by widely dispersed hydrometeor trajectories in relatively strong and/or directionally varying shear.
Two-dimensional scatterplots of select mean environmental metrics among the population of deep CI events (green), shallow CI events (blue), and non-CI events (red). Mean values for each CI outcome (for all sizes) are shown with circles. The population of deep and shallow CI events are subdivided into five size bins of 20-dBZ reflectivity horizontal area (green and blue diamonds). Gray diamonds indicate the mean environmental metrics associated with the combined population of both deep and shallow cells contained in each area size bin. The number of deep and shallow CI events in each cell area size bin are annotated in the legend.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
As in Fig. 8, but for environmental metrics indicating the depth and magnitude of the CIN layer for the populations of shallow and deep CI events that are not required to meet minimum MU CAPE, IB, or EL values.
Citation: Monthly Weather Review 151, 5; 10.1175/MWR-D-22-0243.1
Most of the widest cells (∼84% of events with area > 75 km2) have ETH/EL ≥ 0.66 during their radar-detected CI period, while most of the narrowest cells (∼90% of events with area < 40 km2) fall short of this depth threshold (Fig. 8). Thus, most relatively deep CI events are also relatively wide (and vice versa). We use this generalization to discuss relationships between environmental metrics and these binary distributions of 3D cell size to supplement the statistical analysis conducted in section 4. However, it is worth noting that the explored relationships do not always mimic those among the continuous size distribution of cells (diamonds in Figs. 8 and 9). For example, the mean boundary layer depth is greater for the full population of deep and wide CI events than for the full population of shallow and narrow events (blue and green dots in Fig. 8b). However, cell area across the combined population of deep and shallow CI events exhibits a nonmonotonic relationship with boundary layer depth (gray diamonds in Fig. 8b). These differing relationships likely are a result of an unequal distribution of deep and shallow events across each cell area size bin.
Peters et al. (2019a) found that the vertical growth of narrow clouds can be suppressed in environments containing vertical wind shear owing to the development of vertical pressure gradient accelerations opposing their updrafts. However, relatively wide clouds that are less vulnerable to this shear suppression effect may instead benefit from greater updraft-relative inflow and increased vertical mass influx (Peters et al. 2022a,b). CI events of all widths and depths occur over a relatively small spread of mean ambient deep layer (e.g., 0–9 km) bulk shear values (blue and green dots in Fig. 8a). Though, the narrowest CI events occurred in ∼4 m s−1 greater shear than wide events, which may be evidence of shear suppression effects, most deep layer shear metrics are sensitive to population sampling frequency (Table 5), lowering certainty in this interpretation. However, wide and deep CI events have ∼1.5 m s−1 greater mean cloud-relative flow in the lower half of a typical cloud depth layer (e.g., 0–4 km) than narrow and shallow events (and with slightly less sampling frequency sensitivity; Table 5). Thus, the cloud-relative flow mechanism proposed by Peters et al. (2022a,b) may verify among our population of CI events—bearing in mind, of course, potential measurement uncertainties of cell width using reflectivity observations.
Some studies indicate a positive correlation between updraft width and LCL height (e.g., Williams and Stanfill 2002; McCaul and Cohen 2002; Marion and Trapp 2019), possibly a result of differing expansion rates of ascending dry and saturated thermals and the depth of the conditionally unstable boundary layer air (e.g., Morrison et al. 2021; Mulholland et al. 2021). Most wide and deep CI events occurred in environments with higher mean ML LCLs and deeper boundary layers than narrow and shallow events (blue and green dots in Fig. 8b). Therefore, relatively wide and deep CI events may be a result of a deeper layer over which ascending dry thermals expand (which occurs at a faster rate than saturated thermals) and inherently larger subcloud thermals that spatially scale with deeper boundary layers than in conditions of narrow and shallow CI events (Stull 1988). The mean EIL depth is similar across all CI depth and width outcomes (Tables 3 and 5); therefore, width and depth outcomes do not appear to be dictated by the depth of the conditionally unstable layer throughout which entertainment beneficially preserves the width of ascending thermals at low levels (Mulholland et al. 2021). Non-CI environments have the highest LCLs and deepest boundary layers, but shallowest EILs of all event types (Table 2); thus, it appears that any correlation between thermal width, LCL height, and boundary layer depth is limited when very dry air would be entrained into updrafts over deeper layers.
Buoyant thermals may widen or narrow when ascending through a boundary layer inversion due to a variety of physical processes (e.g., widening by flow divergence, narrowing by reduction of buoyancy from CIN and entrainment-driven dilution of the updraft) that may vary as a function of the strength and depth of the stable layer (e.g., Keene and Lareau 2019; Houston 2022). Among a population of 487 tracked CI events (those meeting all criteria described in section 2 except that the requirements for environments containing MU CAPE > 100 J kg−1, MU IB > −10 J kg−1, and MU EL > 6 km are dropped), on average, narrow and shallow CI events occurred in shallower and slightly weaker MU CIN layers than relatively wide and deep CI events (Fig. 9). This may suggest that deepening clouds benefitted from low-level updrafts that were widened by divergence within their relatively inhibitive boundary layer inversions. Alternatively, strong and deep CIN layers may simply prevent narrow thermals from reaching their LFC; instead, favoring wider and stronger thermals. However, there are obvious limitations to the potential benefits of a substantial CIN layer because ∼92% of the non-CI events among this population lacking the MU IB > −10 J kg−1 requirement occur in environments with substantial inhibition (SFC, MU, and ML CIN < −100 J kg−1).
6. Summary
To improve our understanding of environmental controls on deep convection initiation (CI), we analyzed 180 radiosonde-measured near-cloud thermodynamic and flow parameters collected near the Sierras de Córdoba Mountain range during the 6.5-month CACTI project. These measurements were concurrent with the initiation of hundreds of radar-tracked precipitating clouds occurring within 45 km of the radiosonde launch site, and span a range of precipitation durations and sizes, as well as non-CI events (those failing to yield precipitating convection despite low CIN conditions and the presence of CAPE). Statistical analyses differentiated environmental metrics associated with each CI outcome.
Although approximately 92% of non-CI events throughout the project contained substantial CIN, the remaining 8% of non-CI events with low CIN typically contained substantially more CAPE than most CI events. The most consistent environmental factors showing monotonic relationships across relatively robust (i.e., deep, wide, long-lived), marginal (i.e., shallow, narrow, short-lived), and non-CI events were: (i) the depth and strength of background large-scale vertical motion, and (ii) the magnitude of the component of the low-level meridional flow. These observations suggest that the inability of the near-cloud environment to sufficiently lift parcels to their LFC likely suppressed CI. The significance of large-scale background ascent from a global model reanalysis that is unlikely to sufficiently resolve terrain–flow interactions in this region suggests that it is generalizable to non-orographic regimes (e.g., LH14; Hagos et al. 2014; Barber et al. 2022). However, the significance of meridional flow may imply important localized modifications to the mesoscale flow surrounding the mountains also played significant roles in the CI process. Low- through midlevel relative humidity was a strong discriminator of the probability of any CI outcome occurring (compared to non-CI events), suggesting the importance of updraft-dilution by entrainment of dry air in the lower-to-middle free troposphere. Though lift and low-level moisture are considered required ingredients in typical convection-supporting environments (e.g., Johns and Doswell 1992), our analysis emphasizes a need to also account for entrainment of dry air in the lower and middle free troposphere, which may not be accurately represented by models employing grid spacing ≥ 1 km that may produce unrealistically wide updrafts (e.g., Bryan and Morrison 2012; Lebo and Morrison 2015; Varble et al. 2020). However, humidity was a much less reliable discriminator between robust and marginal CI events. It was difficult to determine if this analysis is partly a result of bias from radiosonde launch frequency and temporal distribution of events relative to evolving mesoscale and synoptic meteorological times scales, or that many ambient meteorological conditions were realistically quite similar across robust and marginal CI outcomes.
Storm longevity was not easily predicted from the near-cloud environment at the time of CI, suggesting that forecasts must account for complex convective-scale processes occurring throughout the cell lifetime (e.g., cloud microphysics, cold pools) and mesoscale flow heterogeneity to improve the predictability of convective duration. Using the horizontal radar reflectivity footprint as a proxy for cell area during the CI period, we found evidence of broad relationships between cloud area and LCL height and boundary layer depth. Cell width and depth exhibited stronger positive correlations with the depth and strength of the CIN layer, as well as the strength of the low- to midlevel cloud-relative flow. However, direct high-resolution observations of updraft and cloud area, as well as LES of CI events using realistic background conditions, will be necessary to verify these relationships more conclusively.
It is important to note that our radiosonde climatology simplifies the near-cloud environment into individual profiles for each event. These profiles alone cannot adequately map the 3D flow and thermodynamic variability associated with orographic circulations or other mesoscale dynamic forcing mechanisms that play critical roles in triggering CI. Uncertainty may also stem from the logistical difficulty of observing ambient near-cloud conditions for large numbers of CI events from a fixed site with only a few studies documenting realistic mesoscale heterogeneity in the surrounding complex terrain (e.g., Behrendt et al. 2011; Nelson et al. 2021; Marquis et al. 2021). We recommend that future field campaigns employ a mesoscale spatiotemporal array of profiling instruments [observation spacing no larger than O(10) km and O(1–2) h] within a forecasted CI region to minimize this uncertainty. Future work will examine the roles played by the orographic circulation upon CI outcome, such as its control on cloudy updraft width and depth, using LES and dual-Doppler observations of cases contained in our climatology.
The parcel origin level within 4 km of the ground with the greatest full-tropospheric CAPE.
This definition is similar to one by Thompson et al. (2007), but with the CIN criterion eased to be more sensitive to developing convective updrafts.
The profile with the greatest MU IB during each 24-h period is included in the population.
Recall that ∼87% of slowly widening cells have horizontal areas that remain smaller than our prescribed 50-km2 threshold throughout their lifetime; therefore, we can expect conditions associated with them to inform upon processes resulting in narrow CI events.
Acknowledgments.
This work was funded by the U.S. Department of Energy’s (DOE) Office of Science Biological and Environmental Research as part of the Atmospheric System Research program. Additional funding support was provided by NOAA Grant NA19OAR4320073; NSF Grants AGS-1661707, AGS-1928666 and AGS-1841674; and DOE Atmospheric System Research Grant DE-SC0000246356. Pacific Northwest National Laboratory is operated by Battelle for the DOE under Contract DE-AC05-76RLO1830. Thank you to ARM for funding the CACTI project and to all members of the data collection and quality control teams. This work benefited from beneficial discussions with Zhixiao Zhang, Paloma Borque, Andrew Geiss, Enoch Jo, Jerome Fast, and Hugh Morrison. We thank three anonymous reviewers for their very helpful and constructive comments on previous versions of this manuscript.
Data availability statement.
The full FLEXTRKR (∼6900 tracked cells and near-cloud environmental metrics described in Feng et al. 2022b), INTERPSONDE (doi:10.5439/1095316), and CSAPR2 (doi:10.5439/1615604) datasets utilized in this study are available through the DOE ARM data discovery website (https://adc.arm.gov/discovery/#/). Users must register for a free online account to access these datasets. The FLEXTRKR database include the list of INTERPSONDE-derived environmental metrics for each tracked cell that were used in this study.
REFERENCES
Arakawa, A., 2004: The cumulus parameterization problem: Past, present, and future. J. Climate, 17, 2493–2525, https://doi.org/10.1175/1520-0442(2004)017<2493:RATCPP>2.0.CO;2.
Bachmann, K., C. Keil, G. C. Craig, M. Weissmann, and C. A. Welzbacher, 2020: Predictability of deep convection in idealized and operational forecasts: Effects of radar data assimilation, orography, and synoptic weather regime. Mon. Wea. Rev., 148, 63–81, https://doi.org/10.1175/MWR-D-19-0045.1.
Banta, R. M., 1984: Daytime boundary-layer evolution over mountainous terrain. Part I: Observations of the dry circulations. Mon. Wea. Rev., 112, 340–356, https://doi.org/10.1175/1520-0493(1984)112<0340:DBLEOM>2.0.CO;2.
Barber, K. A., C. D. Burleyson, Z. Feng, and S. M. Hagos, 2022: The influence of shallow cloud populations on transitions to deep convection in the Amazon. J. Atmos. Sci., 79, 723–743, https://doi.org/10.1175/JAS-D-21-0141.1.
Behrendt, A., and Coauthors, 2011: Observation of convection initiation processes with a suite of state-of-the-art research instruments during COPS IOP 8b. Quart. J. Roy. Meteor. Soc., 137, 81–100, https://doi.org/10.1002/qj.758.
Bodine, D., P. L. Heinselman, B. L. Cheong, R. D. Palmer, and D. Michaud, 2010: A case study on the impact of moisture variability on convection initiation using radar refractivity retrievals. J. Appl. Meteor. Climatol., 49, 1766–1778, https://doi.org/10.1175/2010JAMC2360.1.
Bryan, G. H., and H. Morrison, 2012: Sensitivity of a simulated squall line to horizontal resolution and parameterization of microphysics. Mon. Wea. Rev., 140, 202–225, https://doi.org/10.1175/MWR-D-11-00046.1.
Burghardt, B. J., C. Evans, and P. J. Roebber, 2014: Assessing the predictability of convection initiation in the high plains using an object-based approach. Wea. Forecasting, 29, 403–418, https://doi.org/10.1175/WAF-D-13-00089.1.
Christopoulos, C., and T. Schneider, 2021: Assessing biases and climate implications of the diurnal precipitation cycle in climate models. Geophys. Res. Lett., 48, e2021GL093017, https://doi.org/10.1029/2021GL093017.
Covey, C., P. J. Gleckler, C. Doutriaux, D. N. Williams, A. Dai, J. Fasullo, K. Trenberth, and A. Berg, 2016: Metrics for the diurnal cycle of precipitation: Toward routine benchmarks for climate models. J. Climate, 29, 4461–4471, https://doi.org/10.1175/JCLI-D-15-0664.1.
Craven, J. P., R. E. Jewell, and H. E. Brooks, 2002: Comparison between observed convective cloud-base heights and lifting condensation level for two different lifted parcels. Wea. Forecasting, 17, 885–890, https://doi.org/10.1175/1520-0434(2002)017<0885:CBOCCB>2.0.CO;2.
Dai, A., 2006: Precipitation characteristics in eighteen coupled climate models. J. Climate, 19, 4605–4630, https://doi.org/10.1175/JCLI3884.1.
Damiani, R., and Coauthors, 2008: The cumulus, photogrammetric, in situ, and Doppler observations experiment of 2006. Bull. Amer. Meteor. Soc., 89, 57–74, https://doi.org/10.1175/BAMS-89-1-57.
Degelia, S. K., X. Wang, and D. J. Stensrud, 2019: An evaluation of the impact of assimilating AERI retrievals, kinematic profilers, rawinsondes, and surface observations on a forecast of a nocturnal convection initiation event during the PECAN field campaign. Mon. Wea. Rev., 147, 2739–2764, https://doi.org/10.1175/MWR-D-18-0423.1.
Demko, J. C., and B. Geerts, 2010: A numerical study of the evolving convective boundary layer and orographic circulation around the Santa Catalina Mountains in Arizona. Part I: Circulation without deep convection. Mon. Wea. Rev., 138, 1902–1922, https://doi.org/10.1175/2009MWR3098.1.
de Rooy, W. C., and Coauthors, 2013: Entrainment and detrainment in cumulus convection: An overview. Quart. J. Roy. Meteor. Soc., 139, 1–19, https://doi.org/10.1002/qj.1959.
Duda, J. D., and W. A. Gallus Jr., 2013: The impact of large-scale forcing on skill of simulated convective initiation and upscale evolution with convection-allowing grid spacings in the WRF. Wea. Forecasting, 28, 994–1018, https://doi.org/10.1175/WAF-D-13-00005.1.
Feng, Z., A. Varble, J. Hardin, J. Marquis, A. Hunzinger, Z. Zhang, and M. Thieman, 2022a: Deep convection initiation, growth, and environments in the complex terrain of Central Argentina during CACTI. Mon. Wea. Rev., 150, 1135–1155, https://doi.org/10.1175/MWR-D-21-0237.1.
Feng, Z., A. Varble, J. Hardin, J. Marquis, A. Hunzinger, Z. Zhang, and M. Thieman, 2022b: flextrkrcsapr2: C-CSAPR2 convective cell tracking database during CACTI. Atmospheric Radiation Measurement, accessed 1 January 2022, https://doi.org/10.5439/1844991.
Gaustad, K. L., D. D. Turner, and S. A. McFarlane, 2011: MWRRET value-added product: The retrieval of liquid water path and precipitable water vapor from microwave radiometer (MWR) data sets. Tech. Rep. DOE/SC-ARM/TR-081.2, 19 pp., https://www.arm.gov/publications/tech_reports/doe-sc-arm-tr-081.2.pdf.
Gustafsson, N., and Coauthors, 2018: Survey of data assimilation methods for convective-scale numerical weather prediction at operational centres. Quart. J. Roy. Meteor. Soc., 144, 1218–1256, https://doi.org/10.1002/qj.3179.
Hagos, S., Z. Feng, K. Landu, and C. N. Long, 2014: Advection, moistening, and shallow-to-deep convection transitions during the initiation and propagation of Madden-Julian Oscillation. J. Adv. Model. Earth Syst., 6, 938–949, https://doi.org/10.1002/2014MS000335.
Hagos, S., Z. Feng, R. S. Plant, and A. Protat, 2020: A machine learning assisted development of a model for the populations of convective and stratiform clouds. J. Adv. Model. Earth Syst., 12, e2019MS001798, https://doi.org/10.1029/2019MS001798.
Hardin, J., A. Hunzinger, E. Schuman, A. Matthews, N. Bharadwaj, A. Varble, K. Johnson, and S. Giangrande, 2018: C-band scanning ARM precipitation radar, CF-radial, quality-controlled plan position indicator scans (csapr2cfrppiqc). Atmospheric Radiation Measurement (ARM) User Facility, accessed 9 March 2021, https://doi.org/10.5439/1615604.
Helmus, J. J., and S. M. Collis, 2016: The python ARM radar toolkit (Py-ART), a library for working with weather radar data in the python programming language. J. Open Res. Software, 4, e25, https://doi.org/10.5334/jors.119.
Hirt, M., S. Rasp, U. Blahak, and G. C. Craig, 2019: Stochastic parameterization of processes leading to convective initiation in kilometer-scale models. Mon. Wea. Rev., 147, 3917–3934, https://doi.org/10.1175/MWR-D-19-0060.1.
Holdridge, D., E. Keeler, and J. Kyrouac, 2018: Balloon-borne sounding system (SONDEWNPN). ARM User Facility, accessed 26 July 2019, https://doi.org/10.5439/1021460.
Houston, A. L., 2022: Convective inhibition and deep convection initiation. 31st Conf. on Weather Analysis and Forecasting/27th Conf. on Numerical Weather Prediction, Houston, TX, Amer. Meteor. Soc., J2.4, https://ams.confex.com/ams/102ANNUAL/meetingapp.cgi/Paper/398820.
Houston, A. L., and D. Niyogi, 2007: The sensitivity of convective initiation to the lapse rate of the active cloud-bearing layer. Mon. Wea. Rev., 135, 3013–3032, https://doi.org/10.1175/MWR3449.1.
Jensen, M., S. Giangrande, T. Fairless, and A. Zhou, 2019: Interpolated Sonde (INTERPOLATEDSONDE). Atmospheric Radiation Measurement (ARM) User Facility, accessed 24 November 2020, https://doi.org/10.5439/1095316.
Johns, R. H., and C. A. Doswell III, 1992: Severe local storms forecasting. Wea. Forecasting, 7, 588–612, https://doi.org/10.1175/1520-0434(1992)007<0588:SLSF>2.0.CO;2.
Kain, J. S., and Coauthors, 2013: A feasibility study for probabilistic convection initiation forecasts based on explicit numerical guidance. Bull. Amer. Meteor. Soc., 94, 1213–1225, https://doi.org/10.1175/BAMS-D-11-00264.1.
Kalthoff, N., and Coauthors, 2009: The impact of convergence zones on the initiation of deep convection: A case study from COPS. Atmos. Res., 93, 680–694, https://doi.org/10.1016/j.atmosres.2009.02.010.
Keene, C., and N. Lareau, 2019: The observed variation of updrafts with height in the cumulus topped boundary layer. 2019 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract A41L-2745, https://agu.confex.com/agu/fm19/meetingapp.cgi/Paper/631586.
Khairoutdinov, M., and D. Randall, 2006: High-resolution simulation of shallow-to-deep convection transition over land. J. Atmos. Sci., 63, 3421–3436, https://doi.org/10.1175/JAS3810.1.
Khodayar, S., N. Kalthoff, J. Wickert, U. Corsmeier, C. J. Morcrette, and C. Kottmeier, 2010: The increase of spatial data resolution for the detection of the initiation of convection: A case study from CSIP. Meteor. Z., 19, 179–198, https://doi.org/10.1127/0941-2948/2010/0439.
Kirshbaum, D. J., 2011: Cloud-resolving simulations of deep convection over a heated mountain. J. Atmos. Sci., 68, 361–378, https://doi.org/10.1175/2010JAS3642.1.
Kirshbaum, D. J., 2013: On thermally forced circulations over heated terrain. J. Atmos. Sci., 70, 1690–1709, https://doi.org/10.1175/JAS-D-12-0199.1.
Kirshbaum, D. J., B. Adler, N. Kalthoff, C. Barthlott, and S. Serafin, 2018: Moist orographic convection: Physical mechanisms and links to surface-exchange processes. Atmosphere, 9, 80, https://doi.org/10.3390/atmos9030080.
Lebo, Z. J., and H. Morrison, 2015: Effects of horizontal and vertical grid spacing on mixing in simulated squall lines and implications for convective strength and structure. Mon. Wea. Rev., 143, 4355–4375, https://doi.org/10.1175/MWR-D-15-0154.1.
Lee, B. D., R. D. Farley, and M. R. Hjelmfelt, 1991: A numerical case study of convection initiation along colliding convergence boundaries in northeast Colorado. J. Atmos. Sci., 48, 2350–2366, https://doi.org/10.1175/1520-0469(1991)048<2350:ANCSOC>2.0.CO;2.
Liu, S., and X.-Z. Liang, 2010: Observed diurnal cycle climatology of planetary boundary layer height. J. Climate, 23, 5790–5809, https://doi.org/10.1175/2010JCLI3552.1.
Lock, N. A., and A. L. Houston, 2014: Empirical examination of the factors regulating thunderstorm initiation. Mon. Wea. Rev., 142, 240–258, https://doi.org/10.1175/MWR-D-13-00082.1.
Marion, G. R., and R. J. Trapp, 2019: The dynamical coupling of convective updrafts, downdrafts, and cold pools in simulated supercell thunderstorms. J. Geophys. Res. Atmos., 124, 664–683, https://doi.org/10.1029/2018JD029055.
Markowski, P., and Y. Richardson, 2010: Mesoscale Meteorology in Midlatitudes. Wiley-Blackwell, 407 pp.
Markowski, P., C. Hannon, and E. Rasmussen, 2006: Observations of convection initiation “failure” from the 12 June 2002 IHOP deployment. Mon. Wea. Rev., 134, 375–405, https://doi.org/10.1175/MWR3059.1.
Marquis, J. N., A. C. Varble, P. Robinson, T. C. Nelson, and K. Friedrich, 2021: Low-level mesoscale and cloud-scale interactions promoting deep convection initiation. Mon. Wea. Rev., 149, 2473–2495, https://doi.org/10.1175/MWR-D-20-0391.1.
McCaul, E. W., Jr., and C. Cohen, 2002: The impact on simulated storm structure and intensity of variations in the mixed layer and moist layer depths. Mon. Wea. Rev., 130, 1722–1748, https://doi.org/10.1175/1520-0493(2002)130<1722:TIOSSS>2.0.CO;2.
Morrison, H., 2017: An analytic description of the structure and evolution of growing deep cumulus updrafts. J. Atmos. Sci., 74, 809–834, https://doi.org/10.1175/JAS-D-16-0234.1.
Morrison, H., J. M. Peters, and S. C. Sherwood, 2021: Comparing growth rates of simulated moist and dry convective thermals. J. Atmos. Sci., 78, 797–816, https://doi.org/10.1175/JAS-D-20-0166.1.
Morrison, H., J. M. Peters, K. K. Chandrakar, and S. C. Sherwood, 2022: Influences of environmental relative humidity and horizontal scale of subcloud ascent on deep convective initiation. J. Atmos. Sci., 79, 337–359, https://doi.org/10.1175/JAS-D-21-0056.1.
Moser, D. H., and S. Lasher-Trapp, 2017: The influence of successive thermals on entrainment and dilution in a simulated cumulus congestus. J. Atmos. Sci., 74, 375–392, https://doi.org/10.1175/JAS-D-16-0144.1.
Mueller, C. K., J. W. Wilson, and N. A. Crook, 1993: The utility of sounding and mesonet data to nowcast thunderstorm initiation. Wea. Forecasting, 8, 132–146, https://doi.org/10.1175/1520-0434(1993)008<0132:TUOSAM>2.0.CO;2.
Mulholland, J. P., J. M. Peters, and H. Morrison, 2021: How does LCL height influence deep convective updraft width? Geophys. Res. Lett., 48, e2021GL093316, https://doi.org/10.1029/2021GL093316.
Nelson, T. C., J. Marquis, A. Varble, and K. Friedrich, 2021: Radiosonde observations of environments supporting deep moist convection initiation during RELAMPAGO-CACTI. Mon. Wea. Rev., 149, 289–309, https://doi.org/10.1175/MWR-D-20-0148.1.
Nelson, T. C., J. Marquis, J. M. Peters, and K. Friedrich, 2022: Environmental controls on simulated deep moist convection initiation occurring during RELAMPAGO-CACTI. J. Atmos. Sci., 79, 1941–1964, https://doi.org/10.1175/JAS-D-21-0226.1.
Nesbitt, S. W., and Coauthors, 2021: A storm safari in subtropical South America: Proyecto RELAMPAGO. Bull. Amer. Meteor. Soc., 102, E1621–E1644, https://doi.org/10.1175/BAMS-D-20-0029.1.
Nicol, J. C., R. J. Hogan, T. H. M. Stein, K. E. Hanley, P. A. Clark, C. E. Halliwell, H. W. Leancand, and R. S. Plant, 2015: Convective updraught evaluation in high-resolution NWP simulations using single-Doppler radar measurements. Quart. J. Roy. Meteor. Soc., 141, 3177–3189, https://doi.org/10.1002/qj.2602.
Nowotarski, C. J., J. M. Mulholland, J. P. Peters, and E. R. Nielsen, 2022: Effects of vertical wind shear on updraft hydrometeor loading and downdraft origin heights. 19th Conf. on Mesoscale Processes, Houston, TX, Amer. Meteor. Soc., 710, https://ams.confex.com/ams/102ANNUAL/meetingapp.cgi/Paper/397525.
Nugent, A. D., and R. B. Smith, 2014: Initiating moist convection in an inhomogeneous layer by uniform ascent. J. Atmos. Sci., 71, 4597–4610, https://doi.org/10.1175/JAS-D-14-0089.1.
Peckham, S. E., and L. J. Wicker, 2000: The influence of topography and lower-tropospheric winds on dryline morphology. Mon. Wea. Rev., 128, 2165–2189, https://doi.org/10.1175/1520-0493(2000)128<2165:TIOTAL>2.0.CO;2.
Peters, J. M., W. Hannah, and H. Morrison, 2019a: The influence of vertical wind shear on moist thermals. J. Atmos. Sci., 76, 1645–1659, https://doi.org/10.1175/JAS-D-18-0296.1.
Peters, J. M., C. J. Nowotarski, and H. Morrison, 2019b: The role of vertical wind shear in modulating maximum supercell updraft velocities. J. Atmos. Sci., 76, 3169–3189, https://doi.org/10.1175/JAS-D-19-0096.1.
Peters, J. M., C. J. Nowotarski, and G. L. Mullendore, 2020: Are supercells resistant to entrainment because of their rotation? J. Atmos. Sci., 77, 1475–1495, https://doi.org/10.1175/JAS-D-19-0316.1.
Peters, J. M., H. Morrison, T. C. Nelson, J. N. Marquis, J. P. Mulholland, and C. J. Nowotarski, 2022a: The influence of shear on deep convection initiation. Part I: Theory. J. Atmos. Sci., 79, 1669–1690, https://doi.org/10.1175/JAS-D-21-0145.1.
Peters, J. M., H. Morrison, T. C. Nelson, J. N. Marquis, J. P. Mulholland, and C. J. Nowotarski, 2022b: The influence of shear on deep convection initiation. Part II: Simulations. J. Atmos. Sci., 79, 1691–1711, https://doi.org/10.1175/JAS-D-21-0144.1.
Roberts, R. D., A. R. S. Anderson, E. Nelson, B. G. Brown, J. W. Wilson, M. Pocernich, and T. Saxen, 2012: Impacts of forecaster involvement on convective storm initiation and evolution nowcasting. Wea. Forecasting, 27, 1061–1089, https://doi.org/10.1175/WAF-D-11-00087.1.
Romps, D. M., and Z. Kuang, 2010: Do undiluted convective plumes exist in the upper tropical troposphere? J. Atmos. Sci., 67, 468–484, https://doi.org/10.1175/2009JAS3184.1.
Rousseau-Rizzi, R., D. J. Kirshbaum, and M. K. Yau, 2017: Initiation of deep convection over an idealized mesoscale convergence line. J. Atmos. Sci., 74, 835–853, https://doi.org/10.1175/JAS-D-16-0221.1.
Schlemmer, L., and C. Hohenegger, 2014: The formation of wider and deeper clouds as a result of cold-pool dynamics. J. Atmos. Sci., 71, 2842–2858, https://doi.org/10.1175/JAS-D-13-0170.1.
Singh, I., S. W. Nesbitt, and C. A. Davis, 2022: Quasi-idealized numerical simulations of processes involved in orogenic convection initiation over the Sierras de Córdoba. J. Atmos. Sci., 79, 1127–1149, https://doi.org/10.1175/JAS-D-21-0007.1.
Stelten, S., and W. A. Gallus Jr., 2017: Pristine nocturnal convective initiation: A climatology and preliminary examination of predictability. Wea. Forecasting, 32, 1613–1635, https://doi.org/10.1175/WAF-D-16-0222.1.
Storer, R. L., and S. C. van den Heever, 2013: Microphysical processes evident in aerosol forcing of tropical deep convective clouds. J. Atmos. Sci., 70, 430–446, https://doi.org/10.1175/JAS-D-12-076.1.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic Publishers, 666 pp.
Suhas, E., and G. J. Zhang, 2014: Evaluation of trigger functions for convective parameterization schemes using observations. J. Climate, 27, 7647–7666, https://doi.org/10.1175/JCLI-D-13-00718.1.
Thompson, R. L., C. M. Mead, and R. Edwards, 2007: Effective storm-relative helicity and bulk shear in supercell thunderstorm environments. Wea. Forecasting, 22, 102–115, https://doi.org/10.1175/WAF969.1.
van den Heever, S. C., G. L. Stephens, and N. B. Wood, 2011: Aerosol indirect effects on tropical convection characteristics under conditions of radiative–convective equilibrium. J. Atmos. Sci., 68, 699–718, https://doi.org/10.1175/2010JAS3603.1.
Varble, A., H. Morrison, and E. Zipser, 2020: Effects of under-resolved convective dynamics on the evolution of a squall line. Mon. Wea. Rev., 148, 289–311, https://doi.org/10.1175/MWR-D-19-0187.1.
Varble, A., and Coauthors, 2021: Utilizing a storm-generating hotspot to study convective cloud transitions: The CACTI experiment. Bull. Amer. Meteor. Soc., 102, E1597–E1620, https://doi.org/10.1175/BAMS-D-20-0030.1.
Waite, M. L., and B. Khouider, 2010: The deepening of tropical convection by congestus preconditioning. J. Atmos. Sci., 67, 2601–2615, https://doi.org/10.1175/2010JAS3357.1.
Weckwerth, T. M., and D. B. Parsons, 2006: A review of convection initiation and motivation for IHOP_2002. Mon. Wea. Rev., 134, 5–22, https://doi.org/10.1175/MWR3067.1.
Weckwerth, T. M., and U. Romatschke, 2019: Where, when, and why did it rain during PECAN? Mon. Wea. Rev., 147, 3557–3573, https://doi.org/10.1175/MWR-D-18-0458.1.
Weckwerth, T. M., J. W. Wilson, and R. M. Wakimoto, 1996: Thermodynamic variability within the convective boundary layer due to horizontal convective rolls. Mon. Wea. Rev., 124, 769–784, https://doi.org/10.1175/1520-0493(1996)124<0769:TVWTCB>2.0.CO;2.
Weckwerth, T. M., J. Hanesiak, J. W. Wilson, S. B. Trier, S. K. Degelia, W. A. Gallus Jr., R. D. Roberts, and X. Wang, 2019: Nocturnal convection initiation during PECAN 2015. Bull. Amer. Meteor. Soc., 100, 2223–2239, https://doi.org/10.1175/BAMS-D-18-0299.1.
Williams, E., and S. Stanfill, 2002: The physical origin of the land–ocean contrast in lightning activity. C. R. Phys., 3, 1277–1292, https://doi.org/10.1016/S1631-0705(02)01407-X.
Wilson, J. W., and C. K. Mueller, 1993: Nowcasts of thunderstorm initiation and evolution. Wea. Forecasting, 8, 113–131, https://doi.org/10.1175/1520-0434(1993)008<0113:NOTIAE>2.0.CO;2.
Wilson, J. W., and R. D. Roberts, 2006: Summary of convective storm initiation and evolution during IHOP: Observational and modeling perspective. Mon. Wea. Rev., 134, 23–47, https://doi.org/10.1175/MWR3069.1.
Yano, J.-I., and E. Ouchtar, 2017: Convective initiation uncertainties without trigger or stochasticity: Probabilistic description by the Liouville equation and Bayes’ theorem. Quart. J. Roy. Meteor. Soc., 143, 2025–2035, https://doi.org/10.1002/qj.3064.
Zehnder, J. A., J. Hu, and A. Radzan, 2009: Evolution of the vertical thermodynamic profile during the transition from shallow to deep convection during CuPIDO 2006. Mon. Wea. Rev., 137, 937–953, https://doi.org/10.1175/2008MWR2829.1.
Zhang, Y., F. Zhang, D. J. Stensrud, and Z. Meng, 2015: Practical predictability of the 20 May 2013 tornadic thunderstorm event in Oklahoma: Sensitivity to synoptic timing and topographical influence. Mon. Wea. Rev., 143, 2973–2997, https://doi.org/10.1175/MWR-D-14-00394.1.
Zhang, Z., 2022: Evaluation of deep convective upscale growth in multiscale simulations using RELAMPAGO-CACTI observations. Ph.D. dissertation, Dept. of Atmospheric Sciences, University of Utah, 188 pp.
Ziegler, C. L., and E. N. Rasmussen, 1998: The initiation of moist convection at the dryline: Forecasting issues from a case study perspective. Wea. Forecasting, 13, 1106–1131, https://doi.org/10.1175/1520-0434(1998)013<1106:TIOMCA>2.0.CO;2.
Ziegler, C. L., T. J. Lee, and R. A. Pielke Sr., 1997: Convective initiation at the dryline: A modeling study. Mon. Wea. Rev., 125, 1001–1026, https://doi.org/10.1175/1520-0493(1997)125<1001:CIATDA>2.0.CO;2.
Zwiers, F. W., and H. von Storch, 1995: Taking serial correlation into account in tests of the mean. J. Climate, 8, 336–351, https://doi.org/10.1175/1520-0442(1995)008<0336:TSCIAI>2.0.CO;2.
