1. Introduction
Deep convection initiation (CI) is a challenging physical problem that depends on a complex array of processes that interact over a wide range of scales, many of which are highly nonlinear. Necessary conditions for CI include convective available potential energy (CAPE) over a deep layer and sufficient parcel ascent to overcome convective inhibition (CIN) (Saucier 1955). In the midlatitudes, synoptic-scale systems often set the conditions needed for CI through large-scale ascent and low-level moisture advection. However, even when necessary conditions for CI are met, mesoscale to microscale processes tend to favor certain regions over others or suppress it entirely. Such processes include, among many others, turbulent mixing (e.g., Derbyshire et al. 2004; Tang and Kirshbaum 2020), vertical wind shear (e.g., Peters et al. 2022), and the morphology and intensity of low-level updrafts (e.g., Marquis et al. 2021).
Mountains often act as loci for CI and greatly modify cloud and precipitation systems that cross them (e.g., Houze 2012; Kirshbaum et al. 2018). As such, they shape global precipitation distributions (e.g., Nesbitt and Anders 2009) and may contribute to extreme precipitation events (e.g., Cancelada et al. 2020; Feng et al. 2022). The mechanisms by which terrain forces CI include forced lifting as incident airflow ascends or detours around terrain (mechanical forcing) as well as daytime heating or cooling along sloping terrain, which drives thermally direct circulations that respectively ascend or descend the terrain (thermal forcing) (Kirshbaum et al. 2018). Thermally forced orographic convection is common in the tropics and midlatitude warm seasons, prevailing under strong insolation and weak winds (e.g., Reisner and Smolarkiewicz 1994; Kirshbaum 2013). During the daytime, this convection usually initiates near or just downwind of mountain crests, where up-mountain flow from different sides of the mountain converges.
Over the Sierras de Córdoba (SDC), an isolated mountain ridge located in central Argentina east of the Andes (Fig. 1), mesoscale circulations driven by diurnal thermal forcing often give rise to summertime CI (e.g., Romatschke and Houze 2010; Nicolini and Skabar 2011; Bueno Repinaldo et al. 2015). In this region, the importance of orography for CI and the modification of existing cloud systems have been highlighted from both observational and modeling perspectives (e.g., Rasmussen and Houze 2016; Mulholland et al. 2019; Cancelada et al. 2020; Feng et al. 2022). Once initiated, orographically induced cloud systems tend to translate eastward in the prevailing westerly upper-level flow, sometimes growing into large mesoscale convective systems (e.g., Anabor et al. 2008; Rasmussen and Houze 2011; Mulholland et al. 2018). The central region of Argentina is known for some of the deepest storms on Earth, often featuring frequent lightning and large hail (e.g., Zipser et al. 2006; Cecil and Blankenship 2012; Cecil et al. 2015; Kumjian et al. 2020; Zipser and Liu 2021).
(a) WRF nested domain configuration, with the outer domain (D1) covering the entire plot window, inner domains D1 and D2 indicated by red dashed lines, and a white inset box over the SDC region. (b) D3 terrain within the white inset box from (a), with the nominal C-SAPR2 radar range (black circle) and stations O1 (black star; the AMF1 site) and O2 (black circle; Villa Dolores) also shown. SM measurement stations are shown in (b) by red plus symbols, and stations used herein are labeled.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
Synoptic-scale processes influencing CI over the SDC include the South American low-level jet (SALLJ), which transports moisture poleward from lower latitudes, specifically from the Amazon rain forest (e.g., Vera et al. 2006; Salio et al. 2007; Saulo et al. 2007; Borque et al. 2010; Oliveira et al. 2018; Montini et al. 2019; Sasaki et al. 2024). This poleward extension of the SALLJ tends to strengthen when midlatitude cyclones approach from the west, which increase low-level meridional pressure gradients (e.g., Salio et al. 2002; Marengo et al. 2004; Nicolini and Saulo 2006; Rasmussen and Houze 2016). Regional surface properties (e.g., soil moisture and albedo) also influence CI by regulating the surface energy balance and the diurnal evolution of the atmospheric boundary layer (ABL) (e.g., Saulo et al. 2010; Ruscica et al. 2015; Spennemann et al. 2018; Yang et al. 2023). For all else being equal, moister soils favor increased evaporation and humidification at the expense of sensible heating. While stronger sensible heating is more effective in eliminating CIN, stronger latent heating is more effective at enhancing CAPE (e.g., Pal and Eltahir 2001).
To investigate the unique climatological environment of the SDC region, including the high frequency of orographically induced clouds and the tendency of these clouds to organize into large thunderstorm complexes, the Remote Sensing of Electrification, Lightning, and Mesoscale/Microscale Processes with Adaptive Ground Observations (RELAMPAGO; Nesbitt et al. 2021) and the Cloud, Aerosol, and Complex Terrain Interactions (CACTI; Varble et al. 2021) joint field campaign took place during the 2018/19 austral warm season. During RELAMPAGO/CACTI, numerous CI events over the SDC were observed using various remote and in situ instruments (see section 2a for details).
Intensive field observations from campaigns like RELAMPAGO/CACTI provide valuable data to gain insight into physical processes of interest, particularly when complemented by simulations from numerical weather prediction (NWP) models. These data can also be used to provide detailed verification of NWP simulations to aid the identification of model biases (e.g., Hanley et al. 2011, 2013). This application is particularly valuable for the numerical forecasting of CI, which is inherently challenged by the short intrinsic predictability limit for convective-scale processes (on the order of 1 h; Lorenz 1969). This challenge is compounded by the highly nonlinear, threshold-like nature of CI. In marginally unstable cases, the presence of even small-amplitude model errors in initial conditions, physical parameterizations, and/or discretization of the governing equations can dictate whether storms develop at all (e.g., Crook 1996).
For NWP models with O(1) km grid spacings, usually called “convective permitting (CP)” models, the ability to represent convective storms with horizontal and vertical scales of O(1–10) km can be problematic because critical processes affecting storm development are not adequately resolved (e.g., Bryan et al. 2003; Petch 2006; Lebo and Morrison 2015; Kirshbaum 2020). As the grid resolution increases, critical processes involved in CI and storm development transition from being parameterized to being explicitly represented. Although CP models can explicitly represent deep convection, they lack the grid resolution to accurately represent boundary layer turbulence, shallow convection prior to CI, and mixing between clouds and their immediate surroundings. These deficiencies may lead to large errors in CI prediction, particularly in weakly synoptically forced cases where poorly resolved surface, boundary layer, and shallow-cumulus processes precondition the local atmosphere for CI (e.g., Stensrud et al. 2000).
Although CP models often fail to accurately predict the timing and location of CI (e.g., Surcel et al. 2015), predictability may be locally enhanced by mountains, which tend to anchor low-level updrafts in favored geographic locations (e.g., Anthes et al. 1985; Hanley et al. 2011). To fully realize the potential predictability enhancements in CI over complex terrain, the terrain itself must be reasonably resolved, which may require even finer grid spacings than those typically used in NWP (e.g., Colle et al. 2013). In addition, errors in the mesoscale environment arising from other sources (e.g., initial conditions and subgrid parameterizations) must be small enough that the environment remains favorable for CI and that the mountain-scale flow regime (e.g., mechanical vs thermal) is reasonably represented.
Atmospheric predictability is often analyzed using NWP ensembles, which provide numerous forecast realizations of a given event (e.g., Bauer et al. 2015). The chaotic nature of the atmospheric system and the numerous sources of NWP error mentioned above can limit the usefulness of deterministic forecasts, as slightly different initial states can diverge quickly for highly nonlinear processes like CI. Convective-scale ensembles are thus useful for sampling uncertainties in model initial state and physical parameterization schemes. NWP ensembles may also be useful for understanding and quantifying detailed physical sensitivities of processes of interest (e.g., Reinecke and Durran 2009; Hanley et al. 2011, 2013; Johnson and Wang 2016). Multiple ensemble types exist, including “initial condition” ensembles that vary the initial and lateral boundary conditions (e.g., Hanley et al. 2011), “model physics” ensembles that vary uncertain parameters and/or assumptions within physical parameterization schemes (e.g., Alvarez Imaz et al. 2021), and ensembles that add multiscale initial condition perturbations representative of unresolved atmospheric variability (e.g., Harnisch and Keil 2015).
In this study, convection-permitting NWP initial condition ensembles are used to study an isolated CI event from the RELAMPAGO/CACTI field campaign. The intensive field observations are first described and then used to provide a detailed characterization of the event (section 2). Then, the ensembles are described and verified in detail to evaluate the practical predictability of the event (section 3). The physical parameters controlling the ensemble variability are then explored (section 4), after which conclusions are reached (section 5).
2. Case overview
a. Field campaign and observations
The observational analysis uses data from CACTI (Varble et al. 2021), which held an extended observing period (EOP) covering 6.5 months (15 October 2018–30 April 2019) with a 1.5-month intensive observing period (IOP) (1 November–15 December 2018), and RELAMPAGO (Nesbitt et al. 2021), which held a 10-month EOP (1 June 2018–30 April 2019) and a 1.5-month IOP (1 November–18 December 2018).
The first Atmospheric Radiation Measurement (ARM) mobile facility (AMF1) deployed in CACTI included around 50 instruments (Mather and Voyles 2013) plus the second-generation C-band Scanning ARM Precipitation Radar (C-SAPR2), both located in a rural region 20 km east of the main SDC ridge at 1141 m above mean sea level (MSL; point O1 in Fig. 1b). The 15-min C-SAPR2 scanning cycle included sequential plan position indicator (PPI) scans at 15 elevation angles between 0.5° and 33°, a vertically pointing, azimuthally rotating ZPPI (vertical PPI), and two six-azimuth hemispheric range–height indicator (HSRHI) scans. A second sounding and meteorological station was deployed at Villa Dolores on the western side of the SDC at 560 m MSL (point O2 in Fig. 1b). The soundings were launched up to every 3–4 h between 1200 and 0000 UTC [0900 and 2100 local time (LT)] at O1 and at 1200 and 1800 UTC (0900 and 1500 LT) at O2. CACTI data used herein include soundings at both sites (Keeler et al. 2018a for O1; Keeler et al. 2018b for O2) and C-SAPR2 data (Hardin et al. 2018), along with the surface meteorological data (Xiao and Shaocheng 2018) and quality-controlled eddy correlation surface flux measurements at O1 (Gaustad 2018).
The RELAMPAGO data used herein include a regional soil moisture network to the east of the SDC, shown by the labeled red plus symbols in Fig. 1b. This network includes NCAR/Earth Observatory Laboratory (EOL) Surface Meteorology and Flux Products (s1–s4; NCAR/EOL 2021) and NCAR/Research Applications Laboratory (RAL) Surface Hydrometeorological Observations (s5–s7; Gochis et al. 2019).
b. Study case
The case of interest featured an isolated thunderstorm that lasted for 2–3 h over the northern SDC during the afternoon of 28 November 2018. Near its peak strength at 1930 UTC (1530 LT), the storm is evident in visible satellite images as a localized cell with a long anvil extending to the east-southeast (Fig. 2). C-SAPR2 composite reflectivity (CR; defined as the largest reflectivity in the vertical column) indicated a transition from mostly ground clutter over the SDC at 1730 UTC to a mature thunderstorm by 1830 UTC (Figs. 3a,b). The storm remained anchored to the western SDC slope as it slowly drifted southward over the next hour, reaching a maximum CR (55 dBZ) at 1800 UTC and maximum echo-top height (11.4 km) at 1930 UTC before mostly dissipating by 2030 UTC (Figs. 3c,d).
GeoColor (Miller et al. 2020) from GOES-16 at 1930 UTC 28 Nov 2018, showing the mature thunderstorm of interest over the SDC (obtained from OHMC 2018).
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
C-SAPR2 CR plotted at (a) 1730, (b) 1830, (c) 1930, and (d) 2030 UTC 28 Nov 2018.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
The synoptic-scale conditions for this case are characterized using ERA5 (Hersbach et al. 2020), a global atmospheric reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF), at 0600 and 1800 UTC 28 November 2018 (Figs. 4 and 5, Hersbach et al. 2023a,b, respectively). Before and during the event, there were two upper-level troughs potentially affecting Argentina, one to the west over the Pacific Ocean and the other located to the southeast of the SDC (denoted T1 and T2 in Fig. 4). Over the 12-h period of interest, T1 moved eastward and narrowed while T2 moved southeastward and weakened slightly.
ERA5 reanalysis 500-hPa height (m; black lines) and 1000–500-hPa thickness (m; filled contours) at (a) 0600 and (b) 1800 UTC 28 Nov 2018. The thick gray line shows the 500-m terrain height contour, and labels T1 and T2 correspond to upper-level troughs.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
ERA5 reanalysis MSL surface pressure (MSLP) (hPa; black contours), θe (K; filled contours), and wind vectors at (a) 0600 and (b) 1800 UTC. The thick gray line shows the 500-m terrain height contour, and labels L1 and L2 denote the presence of two low pressure systems at low levels.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
No obvious lower-level cyclone was associated with T1, but one may have overlapped with a mesoscale mountain-wave-induced pressure minimum along the eastern side of the southern Andes (Fig. 5). As T1 approached the Andes, the mesoscale cross-Andes pressure dipole amplified over the Andes with high pressure to the west and low pressure to the east (L1; Fig. 5b). This pattern, which reflects a blend of synoptic and mesoscale mountain-wave disturbances (e.g., Rasmussen and Houze 2016), induced north-northeasterly geostrophic flow to the west of the SDC region. To the east, there was a more prominent low to the north of T2 (L2; Figs. 5a,b), which produced southeasterly flow to the east and north of the SDC. This flow limited the low-level northerly moisture transport from the Amazon rain forest, which may have inhibited a more widespread deep-convection event.
On this day, the moist instability was marginal, with near-zero mixed layer (lowest 50 hPa) CAPE at 1200 UTC that increased to around 647 (O1) and 852 J kg−1 (O2) by the time of the main cell at 1800 UTC, as shown by the corresponding soundings in Fig. 6. The associated CIN at O1 decreased from 185 J kg−1 at 1500 UTC to 0 J kg−1 at 1800 UTC and then increased back to 104 J kg−1 at 2100 UTC. By contrast, the CIN at O2 was much larger (87 J kg−1 at 1800 UTC) (soundings at O2 were not taken at 1500 or 2100 UTC), suggesting larger inhibition on the western side of the SDC. Despite its larger CIN, the larger CAPE at site O2, combined with the impact of orographic lifting on overcoming CIN, may explain why the observed storm developed just to the west of the SDC crest.
Soundings at 1800 UTC at (a) O2 and (b) O1, from observations and ensemble median values from DE, ME1, and ME2. (c),(d) The lower panels zoom in on the lowest 200 hPa of the respective soundings in (a) and (b).
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
3. Convection-permitting ensemble simulations
a. Model configuration
An initial condition, convective-permitting model ensemble of the above-described event is conducted using the Weather Research and Forecasting Model, version 4.3 (Skamarock et al. 2019), with the Advanced Research version of WRF dynamical solver (ARW-WRF, or simply WRF). The simulations are integrated for 24 h starting at 0000 UTC 28 November 2018. Three two-way nested domains are used, with grid spacings decreasing from 22.5 (D1), to 7.5 (D2), to 2.5 km (D3) (Fig. 1a). D1 covers southern South America and its adjacent ocean basins, D2 focuses on central Chile and Argentina, and D3 zooms into the SDC region. This latter region covers not just the SDC but also the taller Andes to its west, to limit any spurious effects caused by placing domain boundaries over steep terrain and/or very close to the SDC region. The grid spacing of D3 is comparable to modern regional CP models, and hence the ensembles consider the practical predictability of CI in such models.
All three domains use 60 hydrostatic pressure (eta) levels, which follow the terrain near the ground and flatten with height. The first model level is nominally 50 m above ground level, and the model top is at 50 hPa. Scalar advection is monotonic, and both scalar and momentum advections are fifth order in the horizontal and third order in the vertical. The Kain–Fritsch cumulus parameterization (Kain 2004) is used in domains D1 and D2 and disabled in D3. All domains share the same microphysics (Thompson et al. 2008), boundary layer (Janjić 1994), radiation (Iacono et al. 2008), land surface (Unified Noah Land Surface Model; Tewari et al. 2004), and surface layer (Janjić 2001) parameterizations.
The boundary and initial conditions were obtained from the Global Ensemble Forecast System (GEFS), version 11.0, operated by the National Centers for Environmental Prediction (NCEP) and described in NOAA (2015). This ensemble has 21 members, of which one is the control and the other 20 are perturbed (Zhou et al. 2016, 2017). The GEFS grid spacing is 0.25° in latitude and longitude, and the lateral boundaries of the outer WRF domain are updated every 3 h.
b. DE
Our first ensemble [the default ensemble (DE)] is created following the above settings. As illustrated by simulated CR in domain D3 for selected ensemble members at 1600 and 1900 UTC, this ensemble exhibits a range of convective outcomes over the SDC (Fig. 7). Although radar shows only ground clutter over the SDC at 1600 UTC (Fig. 7a), some members already develop precipitation by this time (e.g., m07 and m15 in Figs. 7d,f). This difference may be partially owing to beam blocking to the west of the SDC, which may mask shallower observed echoes at earlier times. By 1900 UTC, deep convection is ongoing over the western slope of the northern SDC, with maximum reflectivities around 50 dBZ and a long anvil extending east-southeastward (Fig. 7g). Whereas some members (e.g., m15) qualitatively reproduce this pattern, others (e.g., m18) form little to no precipitation at all (Figs. 7l,i). Even in members that initiate cells at the correct location, the initiation itself occurs too early and the cells are weaker than that observed, particularly at peak intensity at around 1900 UTC.
CR at 1600 UTC for (a) C-SAPR2 observations and DE (b) control case (m00), (c) m18, (d) m07, (e) m12, and (f) m15; and at 1900 UTC for (g) C-SAPR2 observations and DE (h) control case (m00), (i) m18, (j) m07, (k) m12, and (l) m15.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
Table 1 shows ensemble-wide verification of maximum echo-top height (ETH; defined as the maximum height of radar reflectivity exceeding 15 dBZ), CR, and mean-layer (lowest 50 hPa above ground) CAPE (MLCAPE) and CIN (MLCIN) at 1800 UTC at O1 and O2 (Fig. 1b). Amid a wide range of extreme values, the DE simulations exhibit much smaller median MLCAPE and ETH, and moderately smaller CR, than the corresponding observations. Simulated ETH and CR only approach observed values for the most intense ensemble members. The MLCAPE differences are particularly striking: even the DE-maximum values are much less than those observed, and the DE median is an order of magnitude smaller. The MLCIN differences, by contrast, are less systematic and more site-dependent.
Maximum ETH and CR between 1600 and 2300 UTC, along with MLCAPE and MLCIN at 1800 UTC, at locations O1 and O2. Results from the control member, ensemble median, ensemble maximum, and ensemble minimum values of the DE, ME1, and ME2 simulations are compared to corresponding observed values from the C-SAPR2 and radiosondes. Text in parentheses indicates the ensemble member “(mX),” where X is the member number, or the number of members “(Y)” sharing the specified result.
Time series of ensemble ETH indicate that the DE-simulated cells are generally much shallower than the observed cell (Fig. 8a), with the median ETH less than 6 km over the entire storm lifetime. The DE cell onset (we refrain from using the term “CI” due to the shallowness of the cells) generally occurs too early, as evidenced by an approximate 1.5-h lead in maximum simulated ETH (1730 UTC) relative to the observations (1900 UTC). Part of this timing error may be associated with overly rapid simulated precipitation formation, as was previously found in orographic convective cells by Wang and Kirshbaum (2015). Since ETH depends on the maximum height of precipitating echoes, observed precipitation may have only developed once the clouds reached the upper troposphere (∼8 km). By contrast, DE-simulated precipitation existed through most of the storm development, even during its shallow stages. The minimum value of observed ETH never falls below 2.5 km due to the presence of ground clutter near the mountain-top level.
ETH in the C-SAPR2 radar range, based on a 15-dBZ threshold (dotted line), for observations and (a) DE, (b) ME1, and (c) ME2. For each ensemble, the median (green line), range (orange shades), and 50th percentiles (gray shades) are shown. Note that the minimum observed ETH always lies at or above 2.5 km due to ground clutter near the mountaintop level.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
The combination of too-early and too-shallow convection in the default ensemble most likely stems from a common bias: an overly dry surface. Time series of surface latent (LE) and sensible (H) heat fluxes at O1 reveal a strong underestimation of the former and overestimation of the latter (Figs. 9a,b), yielding an overly large simulated Bowen ratio (β = H/LE). As a consequence, the surface water vapor mixing ratio qυ is underestimated by a large margin (Fig. 9c), which limits the moist instability. Although the simulated potential temperature θ verifies well in the near-surface superadiabatic layer (Fig. 9d), it exceeds observed values in the 100 hPa above ground level (Figs. 6c,d), particularly at O2 where a stable inversion atop the ABL is completely absent. This extra low-level warming near O2 expedites the removal of CIN to facilitate CI on the western slope of the SDC, but the lack of low-level qυ limits the buoyancy attained by these clouds.
Comparison of observational and DE-simulated (a) LE, (b) H, (c) water vapor mixing ratio qυ, and (d) potential temperature θ at M1. The ensemble spread is illustrated by the DE median (green line), range (orange shading), and 25th and 75th percentiles (gray shading).
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
The above model bias in the surface energy balance at O1 stems, at least in part, from deficient soil moisture content. Ensemble median, time-averaged (28 November over the 1400–1600 UTC preconvective period) volumetric soil moisture (SM) at 5-cm depth at seven RELAMPAGO soil moisture stations (locations shown in Fig. 1b) exhibits an averaged bias of around −0.1 m3 m−3 in the DE (Fig. 10). With averaged observed SM values of around 0.3 m3 m−3 at these stations, the mean fractional error in DE-simulated SM is around −33%. Substantial spatial variability in both SM and its error is apparent, with site s4 (coinciding with O1) exhibiting the largest negative bias of nearly −0.3 m3 m−3 and site s7 well to the northeast exhibiting a small positive bias. Such pointwise variability is to be expected due to the inherent fine-scale variability of soil conditions (e.g., Qiu et al. 2001) as well as the scale separation of the measurements—while the observed values represent a point location, the simulated values represent a model grid box with an area of 6.25 km2.
Verification of ensemble median volumetric SM at 5-cm depth over 1400–1600 UTC 28 Nov 2018, using both the in situ network from Fig. 1b and SMAP satellite observations. The horizontal bars on the model results indicate the ensemble variability, defined by the mean minimum and maximum values over all times within the time interval.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
Another source of soil moisture information is remote retrievals from satellite observations, which are provided globally at 3-h intervals by NASA’s Soil Moisture Active Passive (SMAP) project (Entekhabi et al. 2010). Because such retrievals are less frequent and more uncertain than in situ point observations from the RELAMPAGO soil moisture network (e.g., Xi et al. 2022), we do not place high confidence in these data. However, they are still useful for qualitatively evaluating the spatial distribution of simulated soil moisture over a broader region than that sampled by the RELAMPAGO network, including over and to the west of the SDC. The satellite soil moisture observations from 28 November, interpolated to RELAMPAGO network stations s1–s7, agree reasonably well with corresponding in situ measurements, with a dry bias of around 0.07 m3 m−3 and a mean absolute error of 0.09 m3 m−3 (Fig. 10).
Major differences in the surface soil moisture distribution are found between the DE control member and the SMAP level 4 soil moisture product (the latter with 9-km resolution). Despite the abovementioned dry bias of the satellite retrieval, the simulated soil is even drier than the satellite product over most of the domain, including over the northwestern SDC where the storm developed (Fig. 11). The agreement between model and satellite observations improves to the east of the SDC except that the simulated soil moisture distribution is much smoother than the satellite-retrieved distribution. This difference results from domain D3 inheriting its initial soil moisture from a coarser global model.
Comparison of 5-cm SM between the (a) control member (m0) of the default ensemble and (b) the SMAP level 4 satellite soil moisture product, averaged over 1400–1600 UTC 28 Nov 2018.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
c. ME1
We begin our verification of ME1 by revisiting surface meteorology time series at O1, which shows a clear improvement over DE in the representation of LE and H. Namely, LE is greatly increased and H is greatly reduced (Figs. 12a,b). Nevertheless, an underestimation of LE and overestimation of H remain, the former limited to the morning (1000–1600 UTC) but the latter prevailing over most of the day (Figs. 10a,b). While qυ shifts closer to observations, it remains about 1–2 g kg−1 too small (Fig. 12c). An underestimation in surface θ is also evident (Fig. 12d), but this deficiency is mostly confined to the superadiabatic surface layer. Above this layer, the simulated θ nearly matches the observed values at O1 (Fig. 6d). However, a warm bias remains at O2, where the positive θ bias of ∼5 K in the convective boundary layer in DE is reduced to ∼2 K in ME1 (Fig. 6c). Thus, despite their increased SM, the ME1 simulations retain a noticeable dry bias in the ABL.
As in Fig. 9, but for the ME1 ensemble.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
Like the DE ensemble, the ME1 simulations vary in their realization of the SDC thunderstorm, with a few members appearing highly accurate but others still unable to create a sustained storm (Fig. 13). As a whole, these simulations shift toward a more delayed but more intense thunderstorm over the SDC western slope that better resembles observations (cf. Figs. 7, 8, and 13). Consistently, the median and control (m0) values of ETH, CR, MLCAPE, and MLCIN all shift closer to the observed values (Table 1). The ME1 simulations also produce a slightly more realistic timing of storm onset. Unlike the DE simulations, where the dominant increase in median ETH occurred over 1500–1600 UTC, most of the increase in median ETH in the ME1 simulations occurs after 1600 UTC, in better agreement with observations (Figs. 8a,b). These improvements likely stem from the more accurate surface energy balance in D3, where the moister soil favors increased latent heating at the expense of sensible heating. The former promotes the buildup of MLCAPE while the latter delays CIN removal.
As in Fig. 7, but for the ME1 ensemble.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
Despite the above improvements, the ME1 convective cells over the SDC remain too shallow, with median ETH peaking at ∼8 km, and this convection still leads the radar observations by around 1 h (Fig. 8b). ME1 also aggravates a model bias to the southwest of the SDC, over a shorter and smaller ridge (centered at around 66.1°W and 32.8°S) where weak cell(s) developed in some DE members (Fig. 7). This area is out of C-SAPR2 range and unobserved by other radars. Although visible satellite indicates cumuli over this ridge, these clouds are relatively small and anvil-free at 1900 UTC, contrasting with the mature cell over the SDC (Fig. 2). The ME1 ensemble (particularly m12 and m15) develops stronger convection over this smaller ridge than corresponding DE ensemble members (cf. Figs. 7 and 13). The SMAP-retrieved satellite soil moisture, while uncertain, suggests that these errors may stem from an overly wet soil in the relevant region of the DE ensemble, with simulated soil moisture values of ≈0.26 m3 m−3 exceeding observed values of ≈0.18 m3 m−3 (Fig. 10). This bias is exacerbated in the ME1 ensemble, which may explain the overly strong convection in this area.
d. ME2
The ME2-simulated LE matches observations well before 1600 UTC but exceeds them afterward, particularly after storm initiation at around 1700 UTC (Fig. 14a). In contrast, the ME2-simulated H provides a close match to observations over most of the daytime (Fig. 14b). These improvements give rise to a surface qυ at O1 that more closely matches observations, but a deficiency of around 1 g kg−1 remains (Fig. 14c). The fact that the ME2 ensemble median still significantly underestimates qυ suggests that deficient initial soil moisture only explains part of the persistent qυ bias in the SDC region. The remainder may stem from various sources, including the land surface model, surface-layer and/or boundary layer schemes, or insufficient initial atmospheric water vapor over the region. Returning to the surface-energy balance, a negative bias in simulated θ is again found at O1 (Fig. 14d), but this bias mostly vanishes deeper into the convective boundary layer (Fig. 6d).
As in Fig. 9, but for the ME2 ensemble.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
Due to its improved surface energy balance, the ME2 simulations provide a more reliable representation of the observed SDC thunderstorm than the DE or ME1 simulations. The CR in selected ME2 members again shows cells over the northern SDC in approximately the correct location at 1900 UTC with similar maximum intensities as the corresponding ME1 members (cf. Figs. 13 and 15). Table 1 also indicates increased ME2-median values of ETH and CR that approach the corresponding observed values. Time series of ME2-simulated ETH show intensified deep convection over the SDC and a 1–2-h extension in the presence of deeper clouds (both relative to ME1), both highly consistent with observations (Fig. 8c). On the other hand, ME2 also forms deeper and more extensive spurious deep convection over the smaller ridge to the southwest of the SDC, again likely due to the overmoistening of the soil in that region.
As in Fig. 7, but for the ME2 ensemble.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
4. Physical parameters controlling ensemble variability
Given the success of the ME2 ensemble in representing the timing, location, and intensity of the isolated storm over the western SDC, we henceforth use this ensemble to gain physical insight into the factors controlling the storm development. To that end, we evaluate which environmental and storm-scale parameters correlate the strongest with simulated precipitation P, the latter evaluated over the region A1 in Fig. 16. This region covers the northern SDC and encompasses the entire life cycle of the storm of interest (except for its nonprecipitating anvil). Time series of P show a wide ensemble spread, with members m10 and m18 respectively generating the most and least precipitation (not shown). These results are not verified against radar-derived precipitation because i) terrain blocking of the radar beam over the western SDC makes such estimates highly uncertain and ii) additional verification is irrelevant to the goal of interpreting the ensemble spread.
Areas where the precipitation (A1) and the control variables (A2) are determined.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
a. Identifying control parameters
We first evaluate Pearson linear correlations between selected environmental (and storm scale) “control” parameters and the time-averaged convective precipitation Pm across the ensemble. Although correlations do not indicate causal mechanisms, we chronologically separate the control parameters (evaluated over the preconvective period 1400–1600 UTC) from Pm (evaluated over the precipitating period of 1600–2300 UTC). Thus, the control parameters are time-separated from the storm metric Pm, which avoids any feedbacks of the storm on the control parameters owing to cold pools, gravity waves, local instability release, etc. Also, whereas we spatially average Pm over the region A1 to account for the motion of the storm cells after initiation (Fig. 16), we spatially average the control parameters over the smaller region A2 to focus specifically on the local conditions where CI occurs over the SDC. Unless specified otherwise, all correlations presented below are statistically significant at the 95% confidence level, using the t distribution for the test statistic.
Because the number of potential control parameters is effectively infinite, we limit the parameter space to metrics that might be plausibly expected to affect deep convection. These parameter selections are based on the three key ingredients for CI: i) positive CAPE and parcel buoyancy over a deep layer, ii) sufficient lifting to overcome CIN, and iii) the minimization of relevant processes that may hinder cloud vertical development. To address the first condition, we correlate MLCAPE (calculated at all points within A2 and then averaged) with Pm, which gives a large correlation coefficient of r = 0.88 (Fig. 17a). Thus, as might be expected, larger preconvective MLCAPE is associated with more vigorous convection and heavier precipitation. While other relevant thermodynamic parameters, including A2-averaged surface qυ and θe, also give large correlations with Pm (r = 0.84 and 0.86, respectively), these also correlate strongly with MLCAPE (r = 0.93 and 0.96, respectively), suggesting that only one of these three variables suffices to uniquely characterize the moist instability.
Scatterplots of A2-averaged (a) CAPE, (b) Mbl, and (c) Uml between 1400 and 1600 UTC and A1-averaged Pm between 1600 and 2300 UTC. Annotation indicates corresponding correlation coefficient r and statistical significance p. In (a) and (b), the r and p correspond to the full ensemble, while in (c), they correspond to the 20-member ensemble with m18 removed.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
To address condition ii, we consider two control parameters, both averaged over A2—MLCIN and cumulative boundary layer upward mass flux over all ascending grid points at 2.8 km MSL (Mbl). The latter lies around 800 m above the mean terrain height of A2, which falls within the mountain boundary layer near cloud-base height. Hence, Mbl quantifies the strength and horizontal coverage of subcloud updrafts. While the correlation between MLCIN and Pm is strongly negative (r = −0.72), MLCIN correlates even more negatively with CAPE itself (r = −0.81), and thus it may also not provide unique information beyond that already contained in MLCAPE. A positive correlation is also found between Mbl and Pm (r = 0.65; Fig. 17b), implying that stronger and more widespread preconvective boundary layer ascent over the SDC ridge is associated with larger convective precipitation. The Mbl also correlates positively with MLCAPE (r = 0.72), likely because stronger SDC-crest updrafts imply stronger upslope flows and moisture convergence within A2. Nonetheless, because Mbl provides dynamic information distinct from CAPE, it may still offer unique value.
The final condition iii relates to a range of processes that may regulate the growth of cumuli over the ridge. These include adverse vertical perturbation pressure gradients, entrainment/detrainment, and elevated downdrafts, to name a few. For this analysis, we consider two parameters, one the averaged midlevel (4–8 km) qυ and the other (qυ)ml. In general, drier environments suppress deep convection by inducing more evaporative cooling and buoyancy loss per unit entrainment of surrounding air (e.g., Derbyshire et al. 2004; Rousseau-Rizzi et al. 2017). However, the correlation between (qυ)ml and Pm is found to be weakly negative (r = −0.04) and statistically insignificant, suggesting that it does not regulate Pm in this ensemble, or that its effects are offset by other processes.
Another parameter relevant to condition iii is the midlevel (4–8 km) zonal wind speed (Uml), which has multiple potential impacts. First, it controls the zonal transport of moisture ventilated to midlevels by the thermally forced circulation over the SDC (Kirshbaum 2011). However, given the lack of correlation between (qυ)ml and Pm just obtained, this effect is likely unimportant in this case. Second, Uml may control internal gravity waves aloft. Despite the minimal low-level cross-barrier flow in this case, which limits the formation of mountain waves, midlevel flow past a quasi-stationary elevated updraft may also generate standing internal gravity waves (e.g., Clark et al. 1986). However, the correlation between Uml and Pm is only weakly negative (r = −0.318; Fig. 17c) and not statistically significant.
The weak correlation between Uml and Pm is caused in part by a clear outlier member that bucks the generally decreasing trend (m18; green hexagon in Fig. 17c). This member is characterized by minimal Pm and marks the lower extreme of mean MLCAPE (<100 J kg−1; Fig. 17a) and maximum CR (37 dBZ; Table 1). We speculate that, in this case, CAPE is too small to permit a sustained cumulonimbus to develop, regardless of the value of Uml. If this member is excluded from the correlation analysis, the negative correlation between Uml and Pm strengthens to r = −0.57 (Fig. 17c) and becomes statistically significant.
To better understand the impact of Uml on CI, we create two five-member composites that average over members with the largest Uml (the “strong” composite) and the smallest Uml (the “weak” composite). Zonal cross sections, temporally averaged over 1700–1800 UTC and meridionally averaged over the northern SDC (from 31.85 to 31.30°S), reveal a similar low-level thermally driven circulation in both composites, with air ascending the SDC from both sides and converging just to the west of the ridge crest (Figs. 18a,b). By definition, the strong ensemble is characterized by stronger westerly winds above 4 km. While both ensembles exhibit a prominent thermally forced updraft above the surface-based convergence zone, the updraft in the strong composite is shallower and topped by a relatively deep downdraft above 4 km (Fig. 18c). In contrast, the downdraft in the weak composite is displaced to the east of the main updraft and poses less resistance to vertical cloud development (Fig. 18d).
Meridionally averaged zonal–vertical cross section across the northern SDC, showing (a),(b) zonal wind overlaid by contours of potential temperature (θ; K) for five-member composites with the largest Uml in (a) and the smallest Uml in (b). (c),(d) Corresponding vertical velocities for each composite are shown. Data are averaged in time (1700–1800 UTC) and in latitude (31.85°–31.30°S), and the brown region at the bottom of each panel is the terrain.
Citation: Monthly Weather Review 153, 2; 10.1175/MWR-D-24-0125.1
The vertically alternating, upstream-tilted updrafts and downdrafts in the strong ensemble resemble vertically propagating mountain waves (e.g., Smith 1979) but are more likely owing to midlevel flow past the quasi-stationary thermally forced updraft over the SDC. Whereas the westerly winds near the updraft top at approximately 4 km are around 5 m s−1 in the strong composite, they are only about 2 m s−1 in the weak composite. The stronger zonal winds past this updraft in the strong composite give rise to a larger-amplitude gravity wave aloft. This wave contains a stronger downdraft immediately above the updraft in the strong composite, which creates a more hostile environment for incipient clouds ascending through it. Another factor suppressing the moist convection in the strong composite is its stronger vertical shear of the zonal wind, which enhances adverse vertical perturbation pressure gradients (e.g., Peters et al. 2019).
b. Multilinear regressions
Table 2 shows r and p for different combinations of control parameters, evaluated over all 21 ensemble members. As already found, a large and statistically significant correlation coefficient of r = 0.879 is obtained using X1 = MLCAPE as the sole control parameter, which greatly exceeds the correlations obtained for X1 = Mbl or Uml. Adding a second control variable X2 = Mbl does not change r, suggesting that Mbl does not provide unique information beyond that already provided by MLCAPE. If X2 is alternatively set to Uml, r increases, but only modestly to r = 0.884. Combining all three variables yields the largest correlation (r = 0.885), but it remains only marginally stronger than that using MLCAPE alone. Thus, the preconvective MLCAPE (or other metrics reflective of low-level moisture content) appears to exert the largest influence on the simulated Pm.
Summary of Pearson correlations between Pm and different combinations of control parameters MLCAPE, Mbl, and Uml, where r denotes correlation coefficient and p denotes statistical significance.
The above finding may be surprising given the clear correlations with Pm exhibited by Uml and Mbl individually in Figs. 17b and 17c. However, Mbl also correlated strongly with MLCAPE, a relationship that may relate to the strength of the mountain thermal circulation. Stronger circulations tend to enhance both Mbl due to stronger low-level convergence and MLCAPE due to stronger low-level moisture transport. Thus, the effects of MLCAPE and Mbl may be largely inextricable in this analysis. Also, the correlation between Uml and Pm was insignificant over the whole ensemble, which explains why the impacts of Uml are relatively weak in this analysis.
Only by removing the outlying ensemble member m18 was a significant correlation between Uml and Pm obtained. Hence, we repeat the above analysis with m18 removed. In this case, with X1 = MLCAPE as the sole predictor, r remains large (0.854) but slightly smaller than that for the full ensemble. Adding X2 = Mbl yields a larger r enhancement than before (r = 0.859), but this increase is again outpaced by setting X2 = Uml (r = 0.877). Combining all three predictors increases r to 0.880, similar to but just smaller than the value obtained with all three control parameters over the entire ensemble. Thus, the marginal values of both Mbl and, in particular, Uml, in explaining the ensemble spread increase when the dry outlier member (m18) is omitted.
Altogether, the above statistical analyses suggest that variability in MLCAPE (and other measures of low-level moisture) dominates the ensemble spread, with Uml of secondary importance and Mbl even less influential. This finding may reflect that CI is generally limited by a lack of low-level moisture and moist instability in this marginally unstable case. Although the correlation coefficients obtained by considering all three independent parameters are large (r = 0.88–0.89) and statistically significant, there remains a substantial degree of scatter around the corresponding best-fit lines (not shown), similar to those in the single-variable regressions in Fig. 17. Thus, either additional preconvective (or convective) control parameters beyond those considered herein may be important and/or that the inherently chaotic nature of convection initiation produces some degree of scatter regardless of which control parameters are considered.
5. Conclusions
This study has investigated the convection-scale predictability of an isolated afternoon thunderstorm over the Sierras de Córdoba (SDC) mountain ridge of central Argentina during the joint CACTI/RELAMPAGO field campaign on 28 November 2018. Under largely undisturbed synoptic conditions, the storm initiated at around 1700 UTC (1400 LT) along the northern half of the SDC and remained anchored to the ridge over 2–3 h, reaching a peak radar echo-top height of 11.4 km. A 21-member WRF-ARW initial condition ensemble, using a minimum horizontal grid spacing of 2.5 km on the finest domain encompassing the SDC (representative of modern regional NWP models), was conducted for this case. This ensemble formed cells close to the observed storm, but they initiated 1–2 h too early and failed to reach the depth or intensity of the observed cell.
Exploiting the wealth of observations for this case, the above model errors were traced to a local deficiency in soil moisture and a surface energy balance dominated by sensible heating. In situ soil moisture measurements indicated a dry bias at stations to the east of the SDC, and satellite measurements suggested an even stronger dry bias over the northern SDC where the observed storm developed. Two additional ensembles, ME1 and ME2, were performed with progressively enhanced soil moisture over the finest model domain to improve the model representation of the storm. Of these, the ME2, which overcompensated for the soil moisture bias to also mitigate a stubborn dry bias in the surface energy budget and low-level humidity, provided the best match to the observed surface energy balance and the SDC storm of interest. However, both ensembles aggravated a positive soil moisture bias to the southwest of the SDC, leading to the development of spurious cells there in most members.
The above findings highlight the importance of accurate initial soil moisture distributions for simulating weakly forced orographic convection events like this one (and a similar European event studied by Hanley et al. 2011). Although satellite-based soil moisture estimates are inherently uncertain, their global availability may be useful for improving the initial representation of soil moisture on high-resolution nested domains, which otherwise inherit very coarse soil moisture fields from global models with parameterized convection. Alternatively, inflating the spread of initial soil moisture distributions across ensemble members may improve model forecasts by accounting for potentially large soil moisture biases like those seen in the present case.
To understand the physical processes controlling the simulated convection over the SDC, multilinear regression analysis was conducted. Preconvective control parameters, and linear combinations thereof, were correlated against mean cumulative precipitation Pm over the SDC during the subsequent convective phase. Of these parameters, measures of low-level moisture over the high SDC terrain, including mean-layer CAPE (MLCAPE), surface water vapor mixing ratio, and equivalent potential temperature, correlated very strongly with the mean cumulative precipitation over the SDC. A secondary factor influencing Pm was found to be the zonal midlevel winds over the SDC (Uml), which controlled the formation of midlevel gravity waves over the quasi-stationary SDC thermally forced updraft. Larger Uml tended to reduce Pm due to the suppressive effects of elevated downdrafts on incipient clouds. Although boundary layer vertical mass flux over the SDC (Mbl) correlated with Pm individually, it did not provide significant value beyond that already contained in MLCAPE. This null result may stem from the combined control of Mbl and MLCAPE by the diurnal thermal circulation over the SDC.
An important question stemming from the above analysis is whether the parameter sensitivities are generalizable to other observed convection initiation events during CACTI/RELAMPAGO. The dominance of MLCAPE in explaining ensemble spread may stem from the marginally supportive and moisture-limited environment for deep convection in this case. In cases with richer moisture supplies coinciding with synoptic-scale forcing for ascent, MLCAPE may be more than sufficient to support deep convection in all ensemble members, and other factors (e.g., Uml or others not considered herein) may play stronger roles in dictating the growth rates of individual cells. Future work will examine the controls on CI across a broader sampling of CACTI/RELAMPAGO cases.
Acknowledgments.
This work was supported by a Grant (DE-SC0022279) from the Atmospheric Systems Research (ASR) program in the Office of Biological and Environmental Research, Office of Science, U.S. Department of Energy. The numerical simulations were performed on the Béluga supercomputer, supported by the Digital Research Alliance of Canada (alliancecan.ca). The authors are grateful to scientific insights and links to relevant data sources provided by Adam Varble and James Marquis.
Data availability statement.
The model simulations conducted herein were driven by GEFS analyses freely provided by the NOAA Operational Model Archive and Distribution System (https://nomads.ncep.noaa.gov/). The WRF code used for the numerical simulations is freely provided by the Mesoscale and Microscale Meteorology Division of NSF NCAR (https://www.mmm.ucar.edu/models/wrf). CACTI observations used for model verification were freely downloaded from the ARM data archive (https://adc.arm.gov/discovery/; the datasets used are cited in the main text), and RELAMPAGO observations were freely downloaded from the NSF NCAR Earth Observing Laboratory (https://www.eol.ucar.edu/field_projects/relampago; the datasets used are cited in the main text). Scripts used to analyze the observational and model data will be made freely available via https://web.meteo.mcgill.ca/∼dkirshbaum/Research/codes.html. Data packages from the WRF simulations are available upon request to the corresponding author.
REFERENCES
Alvarez Imaz, M. A., P. Salio, M. E. Dillon, and L. Fita, 2021: The role of atmospheric forcings and WRF physical set-up on convective initiation over Córdoba, Argentina. Atmos. Res., 250, 105335, https://doi.org/10.1016/j.atmosres.2020.105335.
Anabor, V., D. J. Stensrud, and O. L. L. De Moraes, 2008: Serial upstream-propagating mesoscale convective system events over southeastern South America. Mon. Wea. Rev., 136, 3087–3105, https://doi.org/10.1175/2007MWR2334.1.
Anthes, R. A., Y.-H. Kuo, D. P. Baumhefner, R. M. Errico, and T. W. Bettge, 1985: Predictability of mesoscale atmospheric motions. Advances in Geophysics, Vol. 28, Academic Press, 159–202, https://doi.org/10.1016/S0065-2687(08)60188-0.
Bauer, P., A. Thorpe, and G. Brunet, 2015: The quiet revolution of numerical weather prediction. Nature, 525, 47–55, https://doi.org/10.1038/nature14956.
Borque, P., P. Salio, M. Nicolini, and Y. G. Skabar, 2010: Environment associated with deep moist convection under SALLJ conditions: A case study. Wea. Forecasting, 25, 970–984, https://doi.org/10.1175/2010WAF2222352.1.
Bryan, G. H., J. C. Wyngaard, and J. M. Fritsch, 2003: Resolution requirements for the simulation of deep moist convection. Mon. Wea. Rev., 131, 2394–2416, https://doi.org/10.1175/1520-0493(2003)131<2394:RRFTSO>2.0.CO;2.
Bueno Repinaldo, H. F., M. Nicolini, and Y. G. Skabar, 2015: Characterizing the diurnal cycle of low-level circulation and convergence using CFSR data in southeastern South America. J. Appl. Meteor. Climatol., 54, 671–690, https://doi.org/10.1175/JAMC-D-14-0114.1.
Cancelada, M., P. Salio, D. Vila, S. W. Nesbitt, and L. Vidal, 2020: Backward adaptive brightness temperature threshold technique (BAB3T): A methodology to determine extreme convective initiation regions using satellite infrared imagery. Remote Sens., 12, 337, https://doi.org/10.3390/rs12020337.
Cecil, D. J., and C. B. Blankenship, 2012: Toward a global climatology of severe hailstorms as estimated by satellite passive microwave imagers. J. Climate, 25, 687–703, https://doi.org/10.1175/JCLI-D-11-00130.1.
Cecil, D. J., D. E. Buechler, and R. J. Blakeslee, 2015: TRMM LIS climatology of thunderstorm occurrence and conditional lightning flash rates. J. Climate, 28, 6536–6547, https://doi.org/10.1175/JCLI-D-15-0124.1.
Clark, T. L., T. Hauf, and J. P. Kuettner, 1986: Convectively forced internal gravity waves: Results from two-dimensional numerical experiments. Quart. J. Roy. Meteor. Soc., 112, 899–925, https://doi.org/10.1002/qj.49711247402.
Colle, B. A., R. B. Smith, and D. A. Wesley, 2013: Theory, observations, and predictions of orographic precipitation. Mountain Weather Research and Forecasting: Recent Progress and Current Challenges, F. Chow, S. De Wekker, and B. Snyder, Eds., Springer, 291–344.
Crook, N. A., 1996: Sensitivity of moist convection forced by boundary layer processes to low-level thermodynamic fields. Mon. Wea. Rev., 124, 1767–1785, https://doi.org/10.1175/1520-0493(1996)124<1767:SOMCFB>2.0.CO;2.
Derbyshire, S. H., I. Beau, P. Bechtold, J.-Y. Grandpeix, J.-M. Piriou, J.-L. Redelsperger, and P. M. M. Soares, 2004: Sensitivity of moist convection to environmental humidity. Quart. J. Roy. Meteor. Soc., 130, 3055–3079, https://doi.org/10.1256/qj.03.130.
Entekhabi, D., and Coauthors, 2010: The Soil Moisture Active Passive (SMAP) mission. Proc. IEEE, 98, 704–716, https://doi.org/10.1109/JPROC.2010.2043918.
Feng, Z., A. Varble, J. Hardin, J. Marquis, A. Hunzinger, Z. Zhang, and M. Thieman, 2022: 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.
Gaustad, K., 2018: Quality controlled eddy correlation flux measurement (30qcecor). Atmospheric Radiation Measurement (ARM) user facility, 2018-11-28 to 2018-11-28, ARM Mobile Facility (COR) Cordoba, Argentina; AMF1 (main site for CACTI) (M1). ARM Data Center, accessed 16 October 2024, https://doi.org/10.5439/1097546.
Gochis, D., J. Lentz, F. Dominguez, and C. M. Garcia, 2019: NCAR-RAL surface hydrometeorological observation network data, version 1.0. UCAR/NCAR – Earth Observing Laboratory, accessed 16 October 2024, https://doi.org/10.26023/KW8Z-F2WX-H0Y.
Hanley, K. E., D. J. Kirshbaum, S. E. Belcher, N. M. Roberts, and G. Leoncini, 2011: Ensemble predictability of an isolated mountain thunderstorm in a high-resolution model. Quart. J. Roy. Meteor. Soc., 137, 2124–2137, https://doi.org/10.1002/qj.877.
Hanley, K. E., D. J. Kirshbaum, N. M. Roberts, and G. Leoncini, 2013: Sensitivities of a squall line over central Europe in a convective-scale ensemble. Mon. Wea. Rev., 141, 112–133, https://doi.org/10.1175/MWR-D-12-00013.1.
Hardin, J., and Coauthors, 2018: C-band scanning arm precipitation radar (csapr2cfrppiqc). Atmospheric Radiation Measurement (ARM) user facility, 2018-11-28 to 2018-11-28, ARM Mobile Facility (COR) Cordoba, Argentina; AMF1 (main site for CACTI) (M1). ARM Data Center, accessed 16 October 2024, https://doi.org/10.5439/1615604.
Harnisch, F., and C. Keil, 2015: Initial conditions for convective-scale ensemble forecasting provided by ensemble data assimilation. Mon. Wea. Rev., 143, 1583–1600, https://doi.org/10.1175/MWR-D-14-00209.1.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hersbach, H., and Coauthors, 2023a: ERA5 post-processed daily statistics on pressure levels from 1940 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), accessed 16 October 2024, https://doi.org/10.24381/cds.50314f4c.
Hersbach, H., and Coauthors, 2023b: ERA5 post-processed daily statistics on single levels from 1940 to present. Copernicus Climate Change Service (C3S) Climate Data Store (CDS), accessed 16 October 2024, https://doi.org/10.24381/cds.4991cf48.
Houze, R. A., Jr., 2012: Orographic effects on precipitating clouds. Rev. Geophys., 50, RG1001, https://doi.org/10.1029/2011RG000365.
Iacono, M. J., J. S. Delamere, E. J. Mlawer, M. W. Shephard, S. A. Clough, and W. D. Collins, 2008: Radiative forcing by long-lived greenhouse gases: Calculations with the AER radiative transfer models. J. Geophys. Res., 113, D13103, https://doi.org/10.1029/2008JD009944.
Janjić, Z. I., 1994: The step-mountain Eta coordinate model: Further developments of the convection, viscous sublayer, and turbulence closure schemes. Mon. Wea. Rev., 122, 927–945, https://doi.org/10.1175/1520-0493(1994)122<0927:TSMECM>2.0.CO;2.
Janjić, Z. I., 2001: Nonsingular implementation of the Mellor-Yamada level 2.5 scheme in the NCEP Meso model. NCEP Office Note 437, 61 pp., https://repository.library.noaa.gov/view/noaa/11409.
Johnson, A., and X. Wang, 2016: A study of multiscale initial condition perturbation methods for convection-permitting ensemble forecasts. Mon. Wea. Rev., 144, 2579–2604, https://doi.org/10.1175/MWR-D-16-0056.1.
Kain, J. S., 2004: The Kain–Fritsch convective parameterization: An update. J. Appl. Meteor., 43, 170–181, https://doi.org/10.1175/1520-0450(2004)043<0170:TKCPAU>2.0.CO;2.
Keeler, E., K. Burk, and J. Kyrouac, 2018a: Balloon-borne sounding system (sondewnpn). Atmospheric Radiation Measurement (ARM) user facility, 2018-11-28 to 2018-11-28, ARM Mobile Facility (COR) Cordoba, Argentina; AMF1 (main site for CACTI) (M1). ARM Data Center, accessed 17 October 2024, https://doi.org/10.5439/1595321.
Keeler, E., K. Burk, and J. Kyrouac, 2018b: Balloon-borne sounding system (sondewnpn). Atmospheric Radiation Measurement (ARM) user facility, 2018-11-28 to 2018-11-28, ARM Mobile Facility (COR) Cordoba, Argentina; BBSS sonde (at airport) (S1), ARM Data Center, accessed 16 October 2024, https://doi.org/10.5439/1595321.
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., 2020: Numerical simulations of orographic convection across multiple gray zones. J. Atmos. Sci., 77, 3301–3320, https://doi.org/10.1175/JAS-D-20-0035.1.
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.
Kumjian, M. R., and Coauthors, 2020: Gargantuan hail in Argentina. Bull. Amer. Meteor. Soc., 101, E1241–E1258, https://doi.org/10.1175/BAMS-D-19-0012.1.
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.
Lorenz, E. N., 1969: The predictability of a flow which possesses many scales of motion. Tellus, 21A, 289–307, https://doi.org/10.3402/tellusa.v21i3.10086.
Marengo, J. A., W. R. Soares, C. Saulo, and M. Nicolini, 2004: Climatology of the low-level jet east of the Andes as derived from the NCEP–NCAR reanalyses: Characteristics and temporal variability. J. Climate, 17, 2261–2280, https://doi.org/10.1175/1520-0442(2004)017<2261:COTLJE>2.0.CO;2.
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.
Mather, J. H., and J. W. Voyles, 2013: The ARM climate research facility: A review of structure and capabilities. Bull. Amer. Meteor. Soc., 94, 377–392, https://doi.org/10.1175/BAMS-D-11-00218.1.
Miller, S. D., D. T. Lindsey, C. J. Seaman, and J. E. Solbrig, 2020: GeoColor: A blending technique for satellite imagery. J. Atmos. Oceanic Technol., 37, 429–448, https://doi.org/10.1175/JTECH-D-19-0134.1.
Montini, T. L., C. Jones, and L. M. V. Carvalho, 2019: The South American low-level jet: A new climatology, variability, and changes. J. Geophys. Res. Atmos., 124, 1200–1218, https://doi.org/10.1029/2018JD029634.
Mulholland, J. P., S. W. Nesbitt, R. J. Trapp, K. L. Rasmussen, and P. V. Salio, 2018: Convective storm life cycle and environments near the Sierras de Córdoba, Argentina. Mon. Wea. Rev., 146, 2541–2557, https://doi.org/10.1175/MWR-D-18-0081.1.
Mulholland, J. P., S. W. Nesbitt, and R. J. Trapp, 2019: A case study of terrain influences on upscale convective growth of a supercell. Mon. Wea. Rev., 147, 4305–4324, https://doi.org/10.1175/MWR-D-19-0099.1.
NCAR/EOL, 2021: NCAR/EOL ISFS surface meteorology and flux products, 5-minute, version 2.0. UCAR/NCAR—Earth Observing Laboratory, accessed 16 October 2024, https://data.ucar.edu/dataset/ncar-eol-isfs-surface-meteorology-and-flux-products-5-minute.
Nesbitt, S. W., and A. M. Anders, 2009: Very high resolution precipitation climatologies from the Tropical Rainfall Measuring Mission precipitation radar. Geophys. Res. Lett., 36, L15815, https://doi.org/10.1029/2009GL038026.
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.
Nicolini, M., and A. C. Saulo, 2006: Modeled Chaco low-level jets and related precipitation patterns during the 1997–1998 warm season. Meteor. Atmos. Phys., 94, 129–143, https://doi.org/10.1007/s00703-006-0186-7.
Nicolini, M., and Y. G. Skabar, 2011: Diurnal cycle in convergence patterns in the boundary layer east of the Andes and convection. Atmos. Res., 100, 377–390, https://doi.org/10.1016/j.atmosres.2010.09.019.
NOAA, 2015: Global Ensemble Forecast System. https://www.emc.ncep.noaa.gov/emc/pages/numerical_forecast_systems/gefs.php.
OHMC, 2018: GOES-16 GeoColor. https://wrf.ohmc.ar/goes/GeoColor/.
Oliveira, M. I., E. L. Nascimento, and C. Kannenberg, 2018: A new look at the identification of low-level jets in South America. Mon. Wea. Rev., 146, 2315–2334, https://doi.org/10.1175/MWR-D-17-0237.1.
Pal, J. S., and E. A. B. Eltahir, 2001: Pathways relating soil moisture conditions to future summer rainfall within a model of the land–atmosphere system. J. Climate, 14, 1227–1242, https://doi.org/10.1175/1520-0442(2001)014<1227:PRSMCT>2.0.CO;2.
Petch, J. C., 2006: Sensitivity studies of developing convection in a cloud-resolving model. Quart. J. Roy. Meteor. Soc., 132, 345–358, https://doi.org/10.1256/qj.05.71.
Peters, J. M., W. Hannah, and H. Morrison, 2019: 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., H. Morrison, T. C. Nelson, J. N. Marquis, J. P. Mulholland, and C. J. Nowotarski, 2022: 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.
Qiu, Y., B. Fu, J. Wang, and L. Chen, 2001: Spatial variability of soil moisture content and its relation to environmental indices in a semi-arid gully catchment of the Loess Plateau, China. J. Arid Environ., 49, 723–750, https://doi.org/10.1006/jare.2001.0828.
Rasmussen, K. L., and R. A. Houze Jr., 2011: Orogenic convection in subtropical South America as seen by the TRMM satellite. Mon. Wea. Rev., 139, 2399–2420, https://doi.org/10.1175/MWR-D-10-05006.1.
Rasmussen, K. L., and R. A. Houze Jr., 2016: Convective initiation near the Andes in subtropical South America. Mon. Wea. Rev., 144, 2351–2374, https://doi.org/10.1175/MWR-D-15-0058.1.
Reinecke, P. A., and D. R. Durran, 2009: Initial-condition sensitivities and the predictability of downslope winds. J. Atmos. Sci., 66, 3401–3418, https://doi.org/10.1175/2009JAS3023.1.
Reisner, J. M., and P. K. Smolarkiewicz, 1994: Thermally forced low Froude number flow past three-dimensional obstacles. J. Atmos. Sci., 51, 117–133, https://doi.org/10.1175/1520-0469(1994)051<0117:TFLFNF>2.0.CO;2.
Romatschke, U., and R. A. Houze Jr., 2010: Extreme summer convection in South America. J. Climate, 23, 3761–3791, https://doi.org/10.1175/2010JCLI3465.1.
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.
Ruscica, R. C., A. A. Sörensson, and C. G. Menéndez, 2015: Pathways between soil moisture and precipitation in southeastern South America. Atmos. Sci. Lett., 16, 267–272, https://doi.org/10.1002/asl2.552.
Salio, P., M. Nicolini, and A. C. Saulo, 2002: Chaco low-level jet events characterization during the austral summer season. J. Geophys. Res., 107, 4816, https://doi.org/10.1029/2001JD001315.
Salio, P., M. Nicolini, and E. J. Zipser, 2007: Mesoscale convective systems over southeastern South America and their relationship with the South American low-level jet. Mon. Wea. Rev., 135, 1290–1309, https://doi.org/10.1175/MWR3305.1.
Sasaki, C. R. S., A. K. Rowe, L. A. McMurdie, A. C. Varble, and Z. Zhang, 2024: Influences of the South American low-level jet on the convective environment in central Argentina using a convection-permitting simulation. Mon. Wea. Rev., 152, 629–648, https://doi.org/10.1175/MWR-D-23-0122.1.
Saucier, W. J., 1955: Principles of Meteorological Analysis. University of Chicago Press, 438 pp.
Saulo, C., J. Ruiz, and Y. G. Skabar, 2007: Synergism between the low-level jet and organized convection at its exit region. Mon. Wea. Rev., 135, 1310–1326, https://doi.org/10.1175/MWR3317.1.
Saulo, C., L. Ferreira, J. Nogués-Paegle, M. Seluchi, and J. Ruiz, 2010: Land–atmosphere interactions during a northwestern Argentina low event. Mon. Wea. Rev., 138, 2481–2498, https://doi.org/10.1175/2010MWR3227.1.
Skamarock, W. C., and Coauthors, 2019: A description of the Advanced Research WRF Model version 4. NCAR Tech. Note NCAR/TN-556+STR, 145 pp., https://doi.org/10.5065/1dfh-6p97.
Smith, R. B., 1979: The influence of mountains on the atmosphere. Advances in Geophysics, Vol. 21, Academic Press, 87–230, https://doi.org/10.1016/S0065-2687(08)60262-9.
Spennemann, P. C., M. Salvia, R. C. Ruscica, A. A. Sörensson, F. Grings, and H. Karszenbaum, 2018: Land-atmosphere interaction patterns in southeastern South America using satellite products and climate models. Int. J. Appl. Earth Obs. Geoinf., 64, 96–103, https://doi.org/10.1016/j.jag.2017.08.016.
Stensrud, D. J., J.-W. Bao, and T. T. Warner, 2000: Using initial condition and model physics perturbations in short-range ensemble simulations of mesoscale convective systems. Mon. Wea. Rev., 128, 2077–2107, https://doi.org/10.1175/1520-0493(2000)128<2077:UICAMP>2.0.CO;2.
Surcel, M., I. Zawadzki, and M. K. Yau, 2015: A study on the scale dependence of the predictability of precipitation patterns. J. Atmos. Sci., 72, 216–235, https://doi.org/10.1175/JAS-D-14-0071.1.
Tang, S. L., and D. J. Kirshbaum, 2020: On the sensitivity of deep-convection initiation to horizontal grid resolution. Quart. J. Roy. Meteor. Soc., 146, 1085–1105, https://doi.org/10.1002/qj.3726.
Tewari, M., and Coauthors, 2004: Implementation and verification of the unified NOAH land surface model in the WRF model. 20th Conf. on Weather Analysis and Forecasting/16th Conf. on Numerical Weather Prediction, Seattle, WA, Amer. Meteor. Soc., 14.2a, https://ams.confex.com/ams/84Annual/techprogram/paper_69061.htm.
Thompson, G., P. R. Field, R. M. Rasmussen, and W. D. Hall, 2008: Explicit forecasts of winter precipitation using an improved bulk microphysics scheme. Part II: Implementation of a new snow parameterization. Mon. Wea. Rev., 136, 5095–5115, https://doi.org/10.1175/2008MWR2387.1.
Varble, A. C., 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.
Vera, C., and Coauthors, 2006: The South American Low-Level Jet Experiment. Bull. Amer. Meteor. Soc., 87, 63–78, https://doi.org/10.1175/BAMS-87-1-63.
Wang, C.-C., and D. J. Kirshbaum, 2015: Thermally forced convection over a mountainous tropical island. J. Atmos. Sci., 72, 2484–2506, https://doi.org/10.1175/JAS-D-14-0325.1.
Xi, X., P. Gentine, Q. Zhuang, and S. Kim, 2022: Evaluating the variability of surface soil moisture simulated within CMIP5 using SMAP data. J. Geophys. Res. Atmos., 127, e2021JD035363, https://doi.org/10.1029/2021JD035363.
Xiao, C., and X. Shaocheng, 2018: ARM best estimate data products (ARMBEATM). Atmospheric Radiation Measurement (ARM) user facility, 2018-11-28 to 2018-11-28, ARM Mobile Facility (COR) Cordoba, Argentina; AMF1 (main site for CACTI) (M1). ARM Data Center, accessed 16 October 2024, https://doi.org/10.5439/1333748.
Yang, Z., and Coauthors, 2023: Sensitivity of precipitation displacement of a simulated MCS to changes in land surface conditions. J. Geophys. Res. Atmos., 128, e2022JD037642, https://doi.org/10.1029/2022JD037642.
Zhou, X., Y. Zhu, D. Hou, and D. Kleist, 2016: A comparison of perturbations from an ensemble transform and an ensemble Kalman filter for the NCEP Global Ensemble Forecast System. Wea. Forecasting, 31, 2057–2074, https://doi.org/10.1175/WAF-D-16-0109.1.
Zhou, X., Y. Zhu, D. Hou, Y. Luo, J. Peng, and R. Wobus, 2017: Performance of the new NCEP Global Ensemble Forecast System in a parallel experiment. Wea. Forecasting, 32, 1989–2004, https://doi.org/10.1175/WAF-D-17-0023.1.
Zipser, E. J., and C. Liu, 2021: Extreme convection vs. extreme rainfall: A global view. Curr. Climate Change Rep., 7, 121–130, https://doi.org/10.1007/s40641-021-00176-0.
Zipser, E. J., D. J. Cecil, C. Liu, S. W. Nesbitt, and D. P. Yorty, 2006: Where are the most intense thunderstorms on Earth? Bull. Amer. Meteor. Soc., 87, 1057–1072, https://doi.org/10.1175/BAMS-87-8-1057.
