1. Introduction
Cloud and precipitation processes over mountain ranges are an important, yet incompletely understood and modeled, component of the climate and hydrologic systems. Mountains effectively lift air to condensation and thus can enhance precipitation in transient weather systems (i.e., orographic enhancement; Houze 2012) or generate isolated convective clouds and precipitation over topographic features (Kirshbaum et al. 2018). The resulting enhanced precipitation provides critical water resources to large fractions of Earth’s population (Viviroli et al. 2007).
The processes controlling convection over mountains (orographic convection) differ from those over flat terrain, resulting in differences in the cloud characteristics and organization (e.g., Fig. 1). Over flat and homogeneous surfaces, convective clouds are driven by sensible heating and the resulting buoyant plumes/thermals yield a patchwork of isolated cumulus clouds separated by clear air in subsiding downdrafts (e.g., Fig. 1a). These canonical cumulus processes are typically well resolved in large-eddy simulations (LESs) with spatial resolution of ≤100 m, but are parameterized in coarser-scale models, including most numerical weather prediction and climate models. Parameterization approaches are numerous and include eddy-diffusivity/mass-flux (EDMF) schemes wherein an ensemble of plumes is generated based on the surface sensible heat flux to represent the coherent updrafts that transport mass, momentum, and moisture upward into the cumulus cloud layer (Siebesma et al. 2007; Neggers et al. 2009).
Example of cumulus organization over (a) flat terrain, in Oklahoma, United States, as compared to (b) over an isolated mountain range, the SDC, in Argentina. Data are from the visible channels on the MODIS satellite.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
In contrast, over complex terrain the vertical transport of mass, heat, aerosol, and moisture is accomplished by organized circulations that are incompletely resolved in most global models and unaccounted for in most parameterizations. Up- and down-mountain flows result from both thermal forcing, driven by daytime surface heating (Zardi and Whiteman 2012), and mechanical forcing linked to environmental winds impinging on mountain barriers and the attendant gravity wave response (Smith 1980; Houze 2012; Kirshbaum et al. 2018). Under suitable atmospheric conditions, convective clouds are likely to form, persist, and deepen near ridge crests where there is either enhanced convergence of opposing thermally driven flows or substantial vertical displacement in mechanically forced flows (e.g., Fig. 1b showing clouds over the Sierras de Córdoba in Argentina, the focus of this study). While the basics of thermally and mechanically forced flows are well established (see a review in Kirshbaum et al. 2018), there are relatively few field campaigns that provide long-duration observational datasets allowing for 1) linking variations in upslope flow processes (e.g., depth and strength) to the variations in the forcing mechanisms (thermal versus mechanical) and 2) examining the covariations of these flows with the resulting cumulus cloud fields over mountains. This knowledge gap impedes our ability to parameterize orographic convection in coarse-scale models.
To help fill this gap, we use data from the Cloud, Aerosol, and Complex Terrain Interactions (CACTI) field campaign (Varble et al. 2021) to improve our understanding of orographic and boundary layer processes affecting upslope flows and cumulus convection over the Sierras de Córdoba (SDC) of north-central Argentina (Fig. 1b). This task is well aligned with the overall goals of CACTI, which were to “improve understanding of cloud life cycles and organization in relation to environmental conditions, so that cumulus, microphysics, and aerosol parameterizations in multiscale models can be improved” (Varble et al. 2019). To this end, we use tens of thousands of clear air radar scans from the CACTI period (October 2018–April 2019) to document the structure and variability in thermally and mechanically forced upslope flows over the SDC and then link these flow variations to corresponding variations in the cumulus convection, as observed by geostationary satellites. This second step is accomplished by stratifying our flow and forcing results by the convective outcome (shallow to deep) to examine what factors influence the cloudiness.
2. Background
Similar distinctions between thermal and mechanical forcings were documented during the Dominica Experiment (DOMEX; Smith et al. 2012). Smith et al. (2012) and Nugent et al. (2014) found that the easterly trade wind speed U partially controlled variations in M and the distinction between thermally and mechanically forced convection over the island of Dominica’s mountain ridge. Weaker trade wind flows (small U, big M) favored thermal forcing with clouds forming over the ridgeline, whereas stronger flows (big U, small M) were linked to mechanical forcing and a wave response that shifted cloud development upstream and suppressed cloudiness in the plunging flow down the lee (west facing) slopes. Interestingly, these differences in forcing mechanism were also linked to differences in cloud–aerosol interaction, with thermally forced flows venting aerosol-rich air into the cumulus clouds and mechanically forced flows yielding less-polluted (i.e., more maritime) cloud microphysics.
The analyses of Hagen et al. (2011) and Nugent et al. (2014) demonstrate that M is sometimes useful for distinguishing flow regimes, but there are a number of reasons why M is difficult to define and apply in many real-world settings (e.g., Reinecke and Durran 2008). These limitations include variations in the mixed layer depth Zi relative to the height of the mountain crest H, variations in the strength of capping inversions, and the presence of strongly sheared flows. For example, in the case of the SDC during CACTI, the variations in mixed layer depth, which rarely exceeds the crest height, and nearly omnipresent mean-state shear make M conceptually and practically difficult to apply, so much so that we do not attempt it and instead focus on other approaches at decomposing the flow’s forcing mechanisms.
Indeed, other studies have considered variations in forcing without explicitly evaluating M. For example, convergence of thermally driven circulations and their link to convection were examined during the Cumulus Photogrammetric, In Situ, and Doppler Observations (CuPIDO) experiment in southeastern Arizona (Damiani et al. 2008). Specifically, Geerts et al. (2008) examined the surface, mountain-scale convergent component of the thermally forced circulation and associated hydrostatic pressure perturbations measured around a quasi-circular mountain range. They found that doubling the daytime surface sensible heat flux increased the diurnal amplitude of temperature and horizontal pressure gradient, but it minimally affected the strength of the mean upslope wind such that it was difficult to establish links to convective initiation. Related numerical simulations show that during clear-sky conditions the thermal circulation is contained within the atmospheric boundary layer (Demko and Geerts 2010a), whereas during deeper moist convection, the solenoidal circulation couples to the cloud layer with mass outflow aloft, and that deep convection initiation is also linked to the overall deepening boundary layer Zi over the terrain relative to the level of free convection (LFC; Demko and Geerts 2010b). In the case of CACTI, variations in Zi and the height of the condensation level are likely important to the depth of upslope flows and their link to convective outcomes.
Cloud initiation can also be linked to the superposition and interaction of thermally and mechanically forced flows rather than simply attributable to one or the other process. Using observations of summertime airflows over Colorado, Banta (1984) documented thermally driven upslope flows on the lee side of the terrain (relative to the prevailing westerlies) that converge with the prevailing westerly flow crossing the ridge crest, resulting in leeside updrafts capable of initiating cumulus convection to the east of the ridge. Numerical simulations by Kirshbaum and Wang (2014) also show that nonlinear interactions of thermal and mechanical forcing enhance the likelihood of convection initiation over terrain. As will be shown in subsequent sections, these observations and simulations bear some similarity with the superposition of processes affecting the SDC and its cloud cover.
In fact, variations in orographic and environmental forcing mechanisms for convection over the SDC have been documented in a number of recent studies (Rocque and Rasmussen 2022; Singh et al. 2022, among others in the special collection available at https://journals.ametsoc.org/collection/RELAMPAGO-CACTI). Among these studies, we are particularly motivated by Marquis et al. (2021), who identify a convective initiation event during CACTI on 29 November 2018 as being linked to deepening and strengthening thermally driven orographic flow convergence somewhat (∼1–5 km) east of the SDC crest associated with a lowering LFC. They contrast this event with a mechanically forced ascent regime on 4 December 2018 wherein large-scale easterly flow ascends the eastern slopes of the SDC and produces moist convection over the ridge. The distinction between the convective initiation mechanisms on these days motivates the project-long analysis that we embark upon below aimed at further disentangling the contributions of thermally and mechanically forced ascent and their links to convective clouds.
3. Data and methods
a. The CACTI field campaign
The U.S. Department of Energy’s (DOE) Atmospheric Radiation Measurement (ARM) CACTI field campaign occurred over the SDC of north-central Argentina from October 2018 through April 2019 (Varble et al. 2021). A second, closely affiliated National Science Foundation (NSF) campaign entitled Remote Sensing of Electrification, Lightning, and Mesoscale/Microscale Processes with Adaptive Ground Observations (RELAMPAGO) took place over the same region between 1 November and 16 December 2018 (Nesbitt et al. 2021).
The majority of the CACTI instrumentation was located at the ARM Mobile Facility (AMF1; see Fig. 2), including scanning and staring radars, a Doppler wind lidar, surface flux towers, and radiosonde observations. Airborne observations were also obtained but are not used in this study. A map showing the location of AMF1, along with an east–west terrain cross section along the radar scan plane used in our study, is provided in Fig. 2. The SDC crest is ∼22 km west of AMF1 and ∼1500 m taller. The terrain is relatively simple, comprising a nearly north–south zonally asymmetric ridge, with a steep western slope and more gradual eastern slope. The highest part of the ridge is almost due west of AMF1.
Overview of the CACTI sampling domain and SDC topography. (a) Map showing AMF1 and (b) cross section along the radar scan plane.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
The proximity of the CACTI observations to the SDC’s eastern slopes provides long-duration, process-scale, multisensor observations of both the orographic flow dynamics and the clouds initiated by these flows. Among other things, the CACTI dataset provides a rare opportunity to examine how the strength and depth of the upslope flow over the SDC are modulated by thermal and mechanical forcing mechanisms spanning a wide range of environmental conditions encountered over the 7-month campaign and to examine the impact of these flow regimes on the convective outcomes over the SDC.
b. CACTI and ancillary datasets
1) Clear air radar observations
Data from the second-generation C-band scanning ARM precipitation radar (CSAPR2) are used to examine the east–west component of the flow over the SDC. The CSAPR2 location and east–west range–height indicator (RHI) scan plane are shown in Fig. 2b. From these RHI data, we use the radial velocity observations, primarily in nonprecipitating flow where the radar echoes are mostly due to suspended insects (Wilson et al. 1994).
RHI data are regridded to a common polar grid with 0.5°-elevation increments from 0° to 180° and a range resolution of 100 m. This regridding facilitates temporal averaging despite occasional changes in scan parameters. The upper edge of the data is truncated at 4500 m MSL for two reasons: First, clear air returns typically do not exceed this height; and second, we focus our analysis on the horizontal flow components in the lower to midtroposphere where it interacts with the SDC and its convective clouds. These regridded data are further postprocessed to remove spurious velocities and to ignore velocities in precipitating regions using an upper-bound reflectivity threshold of 20 dBZ. Additional filtering includes absolute range limits (40 m s−1), local range filtering in a 3 × 3 pixel stencil, and Laplacian and median smoothing functions. An example of this filtering is provided in Fig. S1 in the online supplemental material.
While we are primarily focused on the radial velocity data, we have also examined the radar reflectivity and its variation in time and height over the SDC domain. This is important in that the number and size of insect scatterers affect the radar’s ability to measure the clear air velocity (e.g., Geerts and Miao 2005; Martin and Shapiro 2007). For the SDC domain, there is a weak diurnal cycle in echo strength (Fig. S2) and an appreciable reduction with height (Fig. S3) such that below ∼3100 m MSL there is typically >50% data availability, whereas above this height the data availability decreases, with less than 25% availability at ∼4000 m MSL. Note that the SDC ridge crest is at ∼2500 m MSL and most of our analysis focuses on the flow in layers below 4000 m. Throughout our results, we limit the analysis to points with >25% availability and also confirm our interpretation with the available radiosonde zonal wind observations, which tend to agree well with the radial velocity composites.
The temporal return interval for the east–west RHI scans for the CSAPR2 was ∼15 min prior to late February 2019 and then changed to ∼45 s due to a change in the radar configuration (associated with a failure of an azimuth scanning motor). To reduce the data volume and favor consistency across the analysis, we subsampled the 45-s data to 15-min intervals. The total number of RHI scans used in the following analyses is ∼76 000, a sampling that, to the best of our knowledge, vastly exceeds previous datasets for orographic flow regimes.
Our results rely on interpretation of the radial velocity components from the RHI scans and their association with the horizontal winds. This interpretation leverages conventional approaches for examining the radial velocity distribution on either size of the “zero isodop,” which is the line of zero flow toward and away from the radar (e.g., Markowski and Richardson 2011). For example, while vertical velocities can contribute to the observed flow field at higher elevation angles, when the zero isodop of our time-averaged RHI data is aligned with the vertical beam (i.e., elevation angle = 90°), and the magnitude of the flow increases/decreases steadily as a function of distance from the isodop, we interpret the flow as being dominated by horizontal components. This assumption is validated by examining the radiosonde zonal wind profiles over AMF1 associated with our composite RHI results, which are indeed consistent in their interpretation, as we show below.
2) Flux data
Eddy correlation (ECOR) data from a flux tower at AMF1 provide the surface sensible and latent heat flux values. These values have been quality controlled (QCECOR) using a standard data procedure described in Tang et al. (2019). These data are used to examine how surface flux processes impact thermally forced flows.
3) Thermodynamic data
4) Satellite data
Cloud cover and depth are characterized using GOES-16 visible (0.5 μm) and infrared (10.3 μm) data. These data are available at 15-min intervals with nominal spatial resolutions of 500 m and 2 km, respectively. These data are regridded to a Cartesian grid over the SDC for averaging. We construct cloud masks from these data using albedo and temperature thresholds for the visible and thermal imagery, respectively. The visible and cold cloud masks use thresholds of 0.4 (albedo) and 273 K, respectively. While more sophisticated adaptive cloud mask approaches are possible (e.g., Tian et al. 2021), our results sufficiently capture the diurnal character of cumulus development and are minimally sensitive to small variations in these thresholds. For example, a 10% change in the albedo threshold produces almost no change in our cloud statistics. The choice of 273 K for cold clouds adequately captures the occurrence of mixed-phase and glaciated clouds linked to deeper convection.
5) Reanalysis data
ERA5 pressure-level reanalysis data are used to examine synoptic-scale variations in the meteorology. We use data on the 750-hPa isobaric surface which is typically near the SDC crest (∼2500 m MSL).
c. Categorizing convective outcomes
The CACTI campaign investigators provided a brief characterization of the convective nature (shallow, congestus, deep, etc.) of each day during the study, the details of which are available in the field campaign final report (see Table 6 in Varble et al. 2019). We expand on this classification using our own manual inspection of available polar-orbiting satellite MODIS and VIIRS daytime visible imagery to examine the qualitative change in the cumulus cloud cover during the diurnal cycle (i.e., morning versus afternoon MODIS and VIIRS overpasses). Days were selected based on synoptic quiescence and a relatively simple diurnal evolution to the clouds to avoid periods where large-scale and transient processes impact the convective evolution. Based on these data, we categorize each day’s deepest convective outcomes as clear, shallow cumulus, cumulus congestus, and deep convection. Days with congestus and deep convection inherently possess a spectrum of cloud depths, with deeper clouds typically occurring later in the day. Table 1 provides these classifications for the 126 days in our study, though only 105 days have CSAPR2 data (days in italics have radar outages). While the Ka- and X-band radar data are available on some of these days, we opted to avoid complexities due to differences in sensitivity and scattering properties and neglect these days from our radar analyses. The days without radar data are still used in the satellite, radiosonde, and flux-based analyses. We show later that the more detailed GOES-16 satellite data support our groupings based on visible and infrared cloud features.
Summary of CACTI days and their convective outcome categories. Radar data are unavailable for days in italic and nonbold font. For each category, we list the total number of days and the subset of days with CSAPR2 radar data, in italics.
4. Results
a. Modes of variability in the upslope flow
1) Mean flow structure
Figure 3a shows the mean radial wind structure
Time mean and leading modes of variability (EOFs) in (a)–(c) the radial velocity from CSAPR RHIs and (d)–(f) corresponding radiosonde observations at AMF1. (a) Time-mean radial velocity. Reds and blues indicate flows toward and away from the radar site (at location zero), respectively. (b) EOF-1, describing ∼17% of the shared variability in these data. (c) EOF-2, describing ∼12% of the shared variability in these data. The data in (b) and (c) have been scaled by the standard deviation of the corresponding PCs, thus representing the amplitude of the EOF in physical units of meters per second. (d) Time-mean zonal components of wind from radiosondes. (e),(f) Mean zonal wind component for positive and negative PCs corresponding to EOF-1 and EOF-2, respectively.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
The upslope flow layer is terrain following but thins as it ascends the slope. It is notable that this upslope flow and shear pattern exist in the mean state, indicating that there is almost always some forcing for air to ascend the eastern slope SDC, though asymmetry in the day length and the PBL depth also may contribute (i.e., more daylight hours than darkness hours in austral summer). Despite the persistence in the mean, we will show below that there is both a strong diurnal variation in the thermally driven components of the SDC flows and a nondiurnally varying mechanical contribution to the flow.
2) EOF-1, mechanically forced flow
The mean state is modulated by the flow structure apparent in EOF-1, which accounts for ∼17% of the shared variability in our data (Fig. 3b). Note these data are scaled by the standard deviations of the principal component, thus representing the amplitude of the signal in physical units (m s−1). EOF-1 features a strong west-to-east signal in the layer centered at the SDC ridge crest at ∼2500 m MSL and thus above the surface-based upslope flow layer. Recalling that this flow superposes on the mean state, the positive mode of EOF-1 corresponds to a strengthening of the westerly flow aloft and a weakening of the upslope flow near the surface. The opposite holds for the negative mode, which corresponds to either a weaker than normal west-to-east flow or, at times, an east-to-west flow that directly ascends the SDC through a deep layer. The radiosonde zonal wind anomalies for the positive and negative modes of EOF-1 are consistent with this interpretation of the RHI-based analysis (Fig. 3e). As we show below, the easterly mode (i.e., negative loading of EOF-1) can be considered the “mechanical” ascent mode over the eastern SDC slope.
To understand the causes of variations associated with EOF-1, we bin the corresponding principal component coefficients by hour of the day and examine the median and interquartile ranges (Fig. 4a). These data show that PC1 has median values close to zero for every hour and interquartile ranges encompassing both positive and negative values. This implies that there is no consistent diurnal preference for positive or negative phases of EOF-1, and thus, it is not linked to variations in surface heating either at the local scale or at the regional scale (e.g., not part of a broader plain-to-mountain circulation associated with the southern Andes). Rather, it is more likely linked to variations in the synoptic-scale flow regimes varying with troughs and ridges that produce either enhanced or diminished westerly flows at ∼2500 m MSL.
Summary of PC1 variability. (a) Hourly PC1 distributions (median in magenta and interquartile range in black bars). Synoptic composites for the strongly (b) positive and (c) negative PC1 periods. Shown are the 750-hPa geopotential height (white contours in decameters), the 750-hPa wind vectors, the terrain (shaded), and a red line indicating the location of the SDC. (d) Distribution of zonal winds over the SDC for each composite. (e) Box-and-whisker plots showing the daytime-averaged zonal winds in the 2500–3500 m MSL layer measured by radiosonde at AMF1 and stratified by PC1 scores.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
We test this presumption by examining the distribution of winds and pressure gradients corresponding to the strongly positive and negative PC1 scores. We use ERA5 data at 750 hPa, which has a representative geopotential height of ∼2500 m MSL, corresponding to the region of highest amplitude in the EOF-1 signal. This layer also resides near the SDC crest and is thus a useful level for diagnosing the synoptic-scale processes impacting the SDC. The PC1 scores are averaged for 1-h periods centered on the ERA5 time steps (e.g., data from 1930 to 2030 UTC are used for the 2000 UTC analysis time). As expected, the composite maps show clear differences between strongly positive and negative PC1 (i.e., |PC1| > 75) periods (Figs. 4b,c). Positive PC1 scores correspond to enhanced northwesterly flow associated with tight north–south pressure gradients over the SDC and a confluent flow in the lee of the upstream Andes Mountains (Fig. 4b). This fits with our interpretation of positive EOF-1 indicating enhanced geostrophic westerly flow aloft. In contrast, the negative PC1 composite features weaker pressure gradients and a southwest (SW)–northeast (NE)-oriented ridge axis to the east of the SDC that generates a north-northeasterly flow across the SDC (Fig. 4c). This fits with our interpretation of the negative EOF-1 phase corresponding to large-scale forcing for an easterly flow component ascending the eastern slopes of the SDC. Histograms of the zonal components of the flow extracted from above the SDC (2500–3500 m MSL) support this interpretation (Fig. 4d), showing a distinct difference in the zonal wind distributions in the positive and negative phases of EOF-1.
The insights from the ERA5 composites are further supported by radiosonde observations at AMF1. Specifically, we examine the distribution of zonal winds in the 2000–4000-m MSL layer as a function of daytime-averaged PC1 scores in three ranges: strongly positive (PC1 ≥ 75), near neutral (|PC1| < 75), and strongly negative (PC1 < −75). The box-and-whisker plots in Fig. 4e indicate that zonal winds vary markedly across the PC1 bins, with the strongest (median 4.5 m s−1) westerly flow aloft associated with the strongly positive PC1 scores. The neutral and negative PC1 bins have progressively smaller zonal winds, with the negative PC1 bin containing a considerable population of negative (i.e., east to west) zonal flows. These independent data further support our interpretation of PC1 being linked to synoptic modification of the east–west flow component and thus a mechanical forcing for ascent or descent of the eastern slopes of the SDC.
3) EOF-2, thermally forced flow
The second EOF, describing ∼12% of the data variance, is linked to thermally driven upslope and downslope flows (Fig. 3c). The structure of EOF-2 is similar to the mean state, simultaneously describing both the strength of the upslope flow and the shear aloft (see annotations in Fig. 3c). Once again, the radiosonde observations support this interpretation of the RHI data (Fig. 3f). The EOF-2 upslope flow layer is somewhat deeper than in the mean state, such that it also modulates the depth of the upslope flow and thus affects the mass flux toward the ridge crest. In other words, when EOF-2 is positive it constructively superimposes on the mean state yielding stronger and deeper upslope flow and stronger shear aloft and the opposite when it is negative.
The temporal variability of EOF-2 (i.e., PC2) possesses a remarkably clear and consistent diurnal cycle (magenta squares and black lines in Fig. 5). The hourly binned PC2 coefficients indicate EOF-2 is dominantly negative overnight and into the early morning and then smoothly switches sign at ∼1000 Argentina standard time (AST; UTC − 3 h) becoming positive in the afternoon. In the morning and afternoon, the interquartile ranges (black vertical bars) do not span zero, indicating consistency in the phase of these flow patterns even as their strength may vary due to modulating processes (e.g., static stability) to be discussed later.
Summary of PC2 diurnal variability. Shown are hourly PC1 distributions showing the median (magenta) and interquartile range (black bars), the hourly mean surface sensible heat flux (red dashed line), and the hourly averaged CBL height Zi (blue dash–dotted line). Times are in AST which is UTC − 3 h.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
The diurnal cycle of EOF-2 is phase shifted with the diurnal cycle of surface sensible heat flux by ∼2 h (Fig. 5). The peak in surface sensible heat flux (red dashed line Fig. 5) occurs at ∼1300 AST, whereas the peak in PC2 scores occurs 2 h later at ∼1500 AST. In contrast, the PC2’s variation is almost exactly in phase with the CBL growth over AMF1 (blue line Fig. 5), which is measured from an independent Doppler lidar dataset based on thresholding the variance of the vertical velocity (Tucker et al. 2009; Lareau et al. 2018). This makes sense in that the CBL depth is typically proportional to the time integral of the sensible heat flux (Stull 1988) and is thus also phase lagged from the flux itself.
To examine EOF-2 in more detail, we also examine the composite flow structure for each hour of the day (Fig. 6). These composites use the full radial velocity data (where data availability > 25%), not the anomaly fields used in the PCA, and clearly show that the variability associated with EOF-2 resembles the mean diurnal cycle (note the corresponding reflectivity data are available in Fig. S2. For example, from 1000 to 1800 AST the upslope flow layer increases in depth and strength, while the westerly shear aloft also increases. The upslope flow is initially (1000–1200 AST) strongly terrain following, but as the layer deepens it becomes less so (e.g., 1600–1800 AST). The upslope flow diminishes overnight, becoming weakly downslope in the 0700–0900 AST window (Fig. 6b). The variability within these composites is summarized by the interquartile range of the flow near the surface (red) and aloft (blue) evaluated over the SDC’s slopes (Fig. 6j; see also boxes in Fig. 6a showing averaging regions), demonstrating that upslope flow and the westerly flow aloft, and thus the shear, increase in concert during the afternoon (i.e., 1000–1700 AST). One potential cause for this joint variation is venting of mass from the upslope flow layer into the return circulation aloft.
(a)–(i) Composite radial velocity observations for the diurnal cycle in 2-hourly increments and (j) the mean and interquartile ranges of velocity in the upslope and shear flow layers (red and blue dots, respectively). The upslope and shear flow averaging regions are shown in (a). The sample size N is shown in each panel. Data in (a)–(j) are averaged where there is >25% data availability. The 50% data availability line is shown in white. (k),(l) Potential temperature and zonal wind composites spanning the diurnal cycle.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
We also note that the CBL top over AMF1, as measured by the Doppler lidar (DL) and potential temperature profiles, is typically close to the height of the mean shear layer apparent in the radar radial velocity data in Figs. 6a–j, as shown in the potential temperature and zonal wind composites from radiosonde observations (Figs. 6k,l). As such, when considering the diurnal flow composites in Fig. 6 we can infer that well-mixed CBL does not, on average, grow as deep as the SDC ridge crest. This is also apparent in the mean diurnal cycle of the CBL height shown in Fig. 5, wherein the CBL top (2300 m) is, on average, a few hundred meters less than the SDC ridge crest (2500 m). There are of course important variations in the CBL depth that closely correspond to the depth of the upslope flow layer that we discuss below when inspecting other flow composites based on stability and flow superpositions.
4) Superposition of thermal and mechanical forcing
One useful outcome of the PCA decomposition is that it facilitates examination of how mechanical (EOF-1, PC1) and thermal (EOF-2, PC2) forcings superimpose to affect the strength and depth of the flows over the eastern slopes of the SDC. We do so by examining quintiles of the PC1 scores during periods of only strongly positive PC2 scores (PC2 > 50). This partitioning spans the parameter space of strong westerly (PC1 is positive and large) to easterly flows (PC1 is negative) aloft during periods with otherwise strong thermal forcing. As might be expected, negative and small PC1 values coupled with strong PC2 values yield deep upslope flow layers that readily reach the SDC crest (Figs. 7a–c). These flows reflect a positive superposition of mechanical and thermal forcing for ascent over the eastern SDC slope. In contrast, for increasing PC1 scores (i.e., increasing westerly flow aloft) the superposition is destructive, and the upslope flow becomes progressively shallower and does not reach the SDC crest (Figs. 7d,e), likely creating convergence zones on the eastern slopes of the SDC. These flows are reminiscent of those described by Banta (1984). These flow depth variations are also apparent in the potential temperature and zonal wind composites (Figs. 7g,f) which reveal that the deep upslope flows linked to the positive superposition correspond with deeper, less-stable CBLs, whereas the negative superposition yields a shallower CBL. The radiosonde wind data (Fig. 7f) support the interpretation of the RHI data. Future work will examine elements of these superpositions as they affect nuances in the cloud initiation locations over the eastern slopes of the SDC using additional datasets that are beyond the scope of this study (e.g., Ka-band cloud radar and stereo cameras). In the present analysis, some of these nuanced variations are masked by the compositing approach, whereas in individual cases the data may reveal important details of how flow interactions produce clouds.
(a)–(e) Superposition of EOF-1 and EOF-2 based on quintiles of PC1 and positive (i.e., >50) values of PC2. The number of RHI scans N in each composite is shown. (f),(g) The potential temperature and zonal wind from radiosonde data corresponding to these superpositions. Radial velocity data are only displayed for locations with >25% data availability.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
5) The role of static stability
Next, we examine the impact of static stability on the depth and extent of the daytime upslope flow irrespective of the PC scores (Fig. 8). Using terciles of the SDC potential temperature differences Δθ, we find that for weak stability (Δθ < 5.8 K) the flow more readily reaches ridgecrest (Fig. 8a), whereas for increasing stability bins the flow depth and upslope extent are diminished (Figs. 8b,c). It is interesting to note that the strongest upslope flows reside within the middle stability bin, which exhibits a more terrain following CBL. We also confirm these results using radiosonde dataset, binned in the same way. The potential temperature profiles indicate that the CBL depth decreases as Δθ increases, confirming that our bulk stability metric captures the variation in CBL growth. The zonal wind profiles also agree well with the CSAPR2 data, showing that the depth of the easterly upslope flow layer decreases with decreasing CBL depth and increasing Δθ. Future work using additional datasets or simulation results may provide additional insights into what controls variations in the structure of this upslope flow layer as it varies between more and less terrain following and whether boundary layer air is delivered to cloud base over the mountain crest.
(a)–(c) Variations in daytime flow structure based on terciles of stability Δθ. The number of RHI scans N in each composite is shown. (d),(e) Variations in potential temperature and zonal wind component from the radiosonde data over AMF1 corresponding to the stability terciles.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
b. Linking flow regimes and convective outcomes
1) Composite cloud evolution
Having established the structure and variability of flows over the SDC, we now examine how these flows covary with convective cloud processes. To begin, we examine the four convective outcomes based on satellite analysis over the SDC. As shown in Table 1, these outcomes are clear sky (16 days), shallow cumulus (49 days), cumulus congestus (40 days), and deep convection (21 days).
For each of these convective outcomes, the cumulus cloud fraction and cold cloud fraction are examined as a function of time and longitude over the north–south-oriented SDC (Fig. 9). The cumulus cloud fraction is evaluated as the fraction of cloudy points along a meridian at a given point in time. Meridional fractions are used due to the nearly north–south orientation of the SDC such that large cloud fractions indicate the tendency for clouds to form over a given portion of the mountain range (e.g., large cloud fraction over the crest). For each convective outcome, we examine the composite over all days in the category.
Overview of cloud processes spanning our convective outcome categories. Each row corresponds to a convective outcome progressing from (a),(b) clear sky, (c),(d) shallow cumulus, (e),(f) cumulus congestus, and (g),(h) deep convection. (left) The time-versus-longitude variation in meridional visible cloud fraction. (right) The time-versus-longitude variation in meridional cold cloud fraction. The number of days N in each composition is shown. The bottom panels show a terrain cross section of the SDC.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
As expected, the variation in cloud fraction across our convective outcomes is consistent with our initial (manual) classification. Namely, clear-sky days have, by definition, negligible (<1%) meridional cloud fraction (Fig. 9a). Shallow cumulus days have an increase in cloud fraction over the SDC crest and eastern slopes during the day, reaching a peak of ∼14% (Fig. 9c). Congestus development shows a strong preference for development over and east of the SDC crest, especially after 1300 AST (Fig. 9e). The maximum meridional cloud fraction for the congestus composite is ∼29%. Finally, deep convective days have an appreciably larger cloud fraction, ∼50%, and a notable increase in cloud cover over and far downwind of the SDC in the late afternoon (Fig. 9g). This peak in cloud fraction corresponds with the onset of deep convection.
The cold cloud fraction shows mostly passing cirrus for the clear, shallow, and congestus groups (Figs. 9b,d,f). The congestus group also has some occurrence of cold clouds over and east of the SDC during the day and evening reflecting the generation of deeper cumulus on some afternoons. In contrast, the deep convective days, by definition, have a sudden increase in cold cloud fraction at ∼1500 AST, with appreciable cloud cover extending to the east (e.g., anvil blow off) and throughout the night (Fig. 9h). The nocturnal cold cloud tops likely reflect the upscale growth of some of these systems into MCSs (Feng et al. 2022). Interestingly, there is also an upstream (i.e., to the west) increase in cold cloud fraction at night, reflecting either upstream anvil spreading from systems over and to the east of the SDC, or anvils blowing off convection initiated farther to the west (e.g., over the Andes).
2) Covariation of convective outcomes and upslope flow structure
Using our convective outcome classification, we next examine the composite flow structure and forcing mechanisms associated with variations in cloud depth and cover (Fig. 10). These composites use all RHI scans in the 1400–1700 AST window, reflecting midafternoon conditions.
Mean flow structure stratified by convective outcomes [(a)–(d) from clear sky through deep convection, respectively]. Each panel shows the composite radial velocity for the 1400–1700 AST window, the top of the upslope layer (black dots), and the LCL, shown as a solid green lines with dashed green lines for the interquartile range. Each panel contains annotations with the mean upslope wind speed, upslope layer depth (AGL), mean LCL height, and mean PC1 and PC2 scores. These figures use a 25% data availability threshold. (e),(f) Potential temperature and zonal wind profiles for each convective outcome.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
Comparing across the composites, we find the upslope flow deepens and becomes less terrain following with increasing cloud vertical development (Fig. 10). To be specific, the top of the upslope flow layer (black dots) is at ∼1800, 2100, 2300, and 2500 m MSL at −10 km (i.e., 10 km west of the AMF1 radar site) moving from clear sky to deep convective categories. The strength of the upslope flow increases up through the congestus bin and is approximately equal between the congestus and deep outcomes. Despite the broadly similar upslope flow speeds between the different bins, the upslope mass flux likely increases monotonically from shallower to deeper convective outcomes due to the increasing layer depth. These inferences are supported by the corresponding composite zonal wind profiles from the radiosondes (Fig. 10f).
To interpret the changes in flow depth and strength, we examine the mean PC1 and PC2 coefficient scores for each bin to characterize the comparative contribution of mechanical and thermal forcing (see distributions in Fig. 11; see also text annotations in Fig. 10). In the mean, PC1 is strongly positive for the clear-sky case (154) but decreases for the shallow cumulus (83) and congestus cases (62) and then becomes negative for the deep convective case (−69). These trends apply not only to the mean and median of the distributions but also to the interquartile range of PC1 scores in each convective outcome bin; for example, the interquartile range for deep convective cases is entirely negative (Fig. 11a). In each bin, there are a handful of cases that fall outside of this general interpretation (e.g., a few PC1 < 0 congestus cases and a few PC1 > 0 deep cases). These variations indicate that there is enhanced westerly flow near the SDC crest for the shallow convective cases that negatively superimposes with the mean state yielding weaker and shallower upslope flow, as suggested by the sensitivity to the wind shown in Fig. 7. In contrast, for the deep convective case there is background support for the easterly upslope flow over the SDC in the majority of cases. In other words, there is, on average, some mechanical forcing contributing to the transport of mass to the ridge crest on most days with deep convection. This is consistent with the synoptic composite in Fig. 4.
PC distributions binned by convective outcome (shallow to deep). (a) PC1 representing the mechanical forcing. (b) PC2 representing the thermal forcing.
Citation: Monthly Weather Review 152, 9; 10.1175/MWR-D-23-0254.1
It is important to note that the mechanically forced easterly flow into the SDC on deep convective days also provides critical ingredients for deep convection, namely, moisture and instability. Using the Interpolated Sonde (INTERPSONDE) data averaged over the 1400–1700 AST window for each day, we find days with strongly negative PC1 have high near-surface mixing ratios (∼10 g kg−1), appreciable CAPE (765 J kg−1), low LFCs (3596 m), and high LNBs (10 159 m), whereas strongly positive PC1 days have drier near-surface air (8.2 g kg−1), substantially less moist instability (281 J kg−1), higher LFCs (4883 m), and lower LNBs (8968 m). Thus, while the easterly ascent of the SDC is likely an important trigger for convective initiation, the thermodynamic environment associated with the easterly flow described by PC1 is an important control on the overall occurrence of deep moist convection throughout the region, such as previously documented by Rasmussen and Houze (2016). This also helps explain why mechanical ascent of the western slopes of the SDC (i.e., positive PC1) is not linked to deep convection most of the time, as it fails to supply the critical moisture and instability for deep convection.
The PC2 coefficients are equally insightful (Fig. 11b; see also text annotation in Fig. 10). PC2 increases from the shallow to congestus bins, indicating an increasing contribution of thermal forcing for low-level easterly ascent over the SDC (Fig. 11b). This is apparent in the stronger upslope flows in the congestus bin. PC2 subsequently decreases somewhat for the deep convective bin, which is likely due to the increase in cloud cover on the deep convective days (e.g., Fig. 9g), which diminishes the surface sensible heating from late morning onward (not shown). Indeed, when we consider the daytime average surface sensible heat flux across the bins we find a statistically significant (at 95% confidence using the two-tailed t test) decrease in sensible heat flux on the deep convective days (∼115 W m−2) as compared with the clear, shallow, and congestus fluxes (190, 178, and 181 W m−2), which is due to the increased cloud cover and thus less total flux due to decreased insolation even while the Bowen ratio remains nearly constant across bins (not shown). Note that there are no statistically significant differences in the sensible heat flux among the clear, shallow, and congestus groups.
While the flow structure clearly covaries with convective outcomes, these composites also highlight the importance of the height of the LCL relative to the ridge crest and upslope flow depth. The LCL height broadly indicates moisture content and the degree of lifting required to initiate cumulus clouds. Unsurprisingly, the LCL monotonically decreases in height across our day categories in Fig. 10, eventually descending to just above the SDC ridge crest for the deep convective days. This suggests that on deep convective days, which are characterized by deeper ascent over the SDC, the mean upslope flow layer is very close to the LCL making it very easy to directly initiate clouds due to the orographic ascent. In contrast, on shallow and congestus days, the LCL remains sufficiently elevated that additional processes are needed to bring air to condensation. We note that a similar LCL and LFC dependence was found during CuPIDO (Demko and Geerts 2010a). It is likely that, on the shallower cumulus days, plumes of rising air in the convergence zone near the ridge crest or just to the east, formed between the easterly upslope layer and the opposing westerly flow crossing the ridge, force some air upward to condensation. In a subset of cases, the radar radial velocity data clearly elucidate the location and evolution of these convergence zones, and by using other datasets (e.g., stereo camera imagery), future work can likely establish nuanced variations in cloud initiation location, be it over the crest or over the eastern slopes of the SDC, as it relates to variations in convergence and the superposition of different flow regimes.
We also note that our results only isolate the zonal flow variations associated with different convective outcomes. While this approach is useful for examining the flow of mass toward the ridge, there are also variations in meridional flows that may interact with, or even force, elements of these zonal flows. Other investigators have, for example, noted links between convective development over the SDC region with variations in the South American low-level jet (SALLJ; e.g., Sasaki et al. 2024) and convergence over the terrain forced by along-crest (i.e., meridional) flows (Singh et al. 2022). Our composite analysis of meridional winds (not shown) did not introduce a compelling story linked to the variations in convective outcomes for the subset of days that we examined, though future work might make use of the north–south RHI scans to expand our understanding of these processes.
5. Summary and conclusions
We have analyzed ∼100 days of radar observations and corresponding cloud layer data over the Sierras de Córdoba (SDC) mountain ridge of Argentina to investigate upslope flow strength and convection initiation in this region. Principal component analysis shows that there are two physically distinct modes that describe most of the variation in the observations:
-
The first (PC1, EOF-1) is a mechanically forced mode that is associated with enhanced westerly or easterly flow due to varying synoptic-scale pressure gradients. This mode modulates the ambient airflow’s ascent and descent of the ridge and also impacts the suitability of the thermodynamic environment for deep convection.
-
The second mode (PC2, EOF-2) is thermally forced and diurnally varying and describes the daytime development of enhanced upslope flow in response to surface sensible heating and, to a lesser extent, the nighttime wind reversal owing to radiational cooling. The magnitude of the flow does not vary directly with the variations in diurnal heating, but rather is determined by the nexus of heating, stability, and background winds.
-
The mechanical (EOF-1) and thermal (EOF-2) modes can superimpose either constructively or destructively to modulate the depth and strength of the upslope flow layer. Constructive superposition yields the deepest upslope flows.
We also show that the flow structures covary with convective outcomes, as measured by satellite observations:
-
There is increasing thermally forced ascent (i.e., larger PC2 scores) over the SDC when shifting from clear to congestus outcomes.
-
There is a marked increase in the mechanical forcing (i.e., negative PC1 scores) for ascent that superimposes on the thermal forcing for ascent (positive PC2 scores) yielding a deeper upslope flow (on average) for the deep convective outcomes.
-
These variations in flow properties are also linked to moisture availability, as expressed through the local LCL height, which is much lower and closer to the upslope layer top on days with deep convection, thus making it easier to initiate clouds.
It is important to note, however, that the broad statistical analysis and binning of days mask some of the important day-to-day and hour-to-hour variations in processes impacting cloud development in specific cases. For example, there are an array of possible superpositions of the mechanically and thermally forced flows along with variations in stability and moisture availability that result in nuanced variations in cloud initiation location. One manifestation of this is the apparent bimodal prevalence clouds forming either directly over the SDC crest or displaced somewhat eastward from the crest. The latter category is likely linked to flow convergence occurring over the eastern slopes of the SDC. Additional considerations with respect to transient mesoscale features, interaction with elevated mixed layer from the Andes, and spatially heterogeneous forcings must all be considered when evaluating a given case. Accordingly, ongoing and future work will examine these nuanced processes and their impact on cloud formation.
One of the outcomes of this study is support for the concept of parameterizing orographic forcing for cloud processes over subgrid-scale topography in coarse-scale models. For example, in models that lack the grid resolution to reliably capture the detailed circulations over the SDC, it may be possible to separately represent 1) the mechanical forcing for ascent due to the mean flow impinging on mountains and 2) the thermally forced component of orographic circulations. More work is required, however, to better understand what larger-scale variables can be used to parameterize the subgrid thermally driven component, which varies not only with the heat flux but also other factors such as stability and background winds.
Future work will expand on these analyses to examine in more detail the links between flow structures, including convergence locations, and the resulting clouds using additional data streams, such as ARM’s scanning Ka-band cloud radar and stereographic imaging cameras.
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, DOE.
Data availability statement.
The data used in this study are all freely and publicly available. The details of the ARM data streams can be found in ARM User Facility (2018a,b,c). NOAA Geostationary Operational Environmental Satellite (GOES-16) data are accessible at https://registry.opendata.aws/noaa-goes. ERA5 reanalysis data are available at https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-pressure-levels?tab=overview.
REFERENCES
ARM User Facility, 2018a: C-band scanning ARM precipitation radar (CSAPR2CFRHSRHIQC). ARM Data Center, accessed 6 October 2022, https://doi.org/10.5439/1615607.
ARM User Facility, 2018b: Interpolated Sonde (INTERPOLATEDSONDE). ARM Data Center, accessed 13 October 2022, https://doi.org/10.5439/1095316.
ARM User Facility, 2018c: Quality Controlled Eddy Correlation Flux Measurement (30QCECOR). ARM Data Center, accessed 4 February 2022, https://doi.org/10.5439/1097546.
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.
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.
Demko, J. C., and B. Geerts, 2010a: 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.
Demko, J. C., and B. Geerts, 2010b: A numerical study of the evolving convective boundary layer and orographic circulation around the Santa Catalina Mountains in Arizona. Part II: Interaction with deep convection. Mon. Wea. Rev., 138, 3603–3622, https://doi.org/10.1175/2010MWR3318.1.
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.
Geerts, B., and Q. Miao, 2005: The use of millimeter Doppler radar echoes to estimate vertical air velocities in the fair-weather convective boundary layer. J. Atmos. Oceanic Technol., 22, 225–246, https://doi.org/10.1175/JTECH1699.1.
Geerts, B., Q. Miao, and J. C. Demko, 2008: Pressure perturbations and upslope flow over a heated, isolated mountain. Mon. Wea. Rev., 136, 4272–4288, https://doi.org/10.1175/2008MWR2546.1.
Hagen, M., J. van Baelen, and E. Richard, 2011: Influence of the wind profile on the initiation of convection in mountainous terrain. Quart. J. Roy. Meteor. Soc., 137, 224–235, https://doi.org/10.1002/qj.784.
Houze, R. A., Jr., 2012: Orographic effects on precipitating clouds. Rev. Geophys., 50, RG1001, https://doi.org/10.1029/2011RG000365.
Jensen, M. P., and T. Toto, 2016: Interpolated sounding and gridded sounding value-added products. DOE/SC-ARM-TR-183., 12 pp.
Kirshbaum, D. J., and C. Wang, 2014: Boundary layer updrafts driven by airflow over heated terrain. J. Atmos. Sci., 71, 1425–1442, https://doi.org/10.1175/JAS-D-13-0287.1.
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.
Lareau, N. P., Y. Zhang, and S. A. Klein, 2018: Observed boundary layer controls on shallow cumulus at the ARM Southern Great Plains site. J. Atmos. Sci., 75, 2235–2255, https://doi.org/10.1175/JAS-D-17-0244.1.
Ludwig, F. L., J. Horel, and C. D. Whiteman, 2004: Using EOF analysis to identify important surface wind patterns in mountain valleys. J. Appl. Meteor., 43, 969–983, https://doi.org/10.1175/1520-0450(2004)043<0969:UEATII>2.0.CO;2.
Markowski, P., and Y. Richardson, 2011: Mesoscale Meteorology in Midlatitudes, John Wiley and Sons.
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.
Martin, W. J., and A. Shapiro, 2007: Discrimination of bird and insect radar echoes in clear air using high-resolution radars. J. Atmos. Oceanic Technol., 24, 1215–1230, https://doi.org/10.1175/JTECH2038.1.
Neggers, R. A. J., 2009: A dual mass flux framework for boundary layer convection. Part II: Clouds. J. Atmos. Sci., 66, 1489–1506, https://doi.org/10.1175/2008JAS2636.1.
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.
Nugent, A. D., R. B. Smith, and J. R. Minder, 2014: Wind speed control of tropical orographic convection. J. Atmos. Sci., 71, 2695–2712, https://doi.org/10.1175/JAS-D-13-0399.1.
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, 2008: Estimating topographic blocking using a Froude Number when the static stability is nonuniform. J. Atmos. Sci., 65, 1035–1048, https://doi.org/10.1175/2007JAS2100.1.
Rocque, M. N., and K. L. Rasmussen, 2022: The impact of topography on the environment and life cycle of weakly and strongly forced MCSs during RELAMPAGO. Mon. Wea. Rev., 150, 2317–2338, https://doi.org/10.1175/MWR-D-22-0049.1.
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.
Siebesma, A. P., P. M. M. Soares, and J. Teixeira, 2007: A combined eddy-diffusivity mass-flux approach for the convective boundary layer. J. Atmos. Sci., 64, 1230–1248, https://doi.org/10.1175/JAS3888.1.
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.
Smith, R. B., 1980: Linear theory of stratified hydrostatic flow past an isolated mountain. Tellus, 32A, 348–364, https://doi.org/10.3402/tellusa.v32i4.10590.
Smith, R. B., and Coauthors, 2012: Orographic precipitation in the tropics: The Dominica Experiment. Bull. Amer. Meteor. Soc., 93, 1567–1579, https://doi.org/10.1175/BAMS-D-11-00194.1.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Atmospheric and Oceanographic Sciences Library, Vol. 13, Springer Science & Business Media, 670 pp.
Tang, S., S. Xie, Y. Zhang, and D. R. Cook, 2019: The QCECOR value-added product: Quality-controlled eddy correlation flux measurements. Rep. DOE/SC-ARM-TR-223, 16 pp., https://www.arm.gov/publications/tech_reports/doe-sc-arm-tr-223.pdf.
The MathWorks Inc, 2022: MATLAB version: 9.13.0 (R2022b). The MathWorks Inc., https://www.mathworks.com.
Tian, J., Y. Zhang, S. A. Klein, L. Wang, R. Öktem, and D. M. Romps, 2021: Summertime continental shallow cumulus cloud detection using GOES-16 satellite and ground-based stereo cameras at the DOE ARM Southern Great Plains site. Remote Sens., 13, 2309, https://doi.org/10.3390/rs13122309.
Tucker, S. C., C. J. Senff, A. M. Weickmann, W. A. Brewer, R. M. Banta, S. P. Sandberg, D. C. Law, and R. M. Hardesty, 2009: Doppler lidar estimation of mixing height using turbulence, shear, and aerosol profiles. J. Atmos. Oceanic Technol., 26, 673–688, https://doi.org/10.1175/2008JTECHA1157.1.
Varble, A., and Coauthors, 2019: Cloud, Aerosol, and Complex Terrain Interactions (CACTI) field campaign report. Rep. DOE/SC-ARM-19-028, 31 pp., https://doi.org/10.2172/1574024.
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.
Viviroli, D., H. H. Dürr, B. Messerli, M. Meybeck, and R. Weingartner, 2007: Mountains of the world, water towers for humanity: Typology, mapping, and global significance. Water Resour. Res., 43, W07447, https://doi.org/10.1029/2006WR005653.
Wilson, J. W., T. M. Weckwerth, J. Vivekanandan, R. M. Wakimoto, and R. W. Russell, 1994: Boundary layer clear-air radar echoes: Origin of echoes and accuracy of derived winds. J. Atmos. Oceanic Technol., 11, 1184–1206, https://doi.org/10.1175/1520-0426(1994)011<1184:BLCARE>2.0.CO;2.
Wulfmeyer, V., and Coauthors, 2011: The Convective and Orographically-induced Precipitation Study (COPS): The scientific strategy, the field phase, and research highlights. Quart. J. Roy. Meteor. Soc., 137, 3–30, https://doi.org/10.1002/qj.752.
Zardi, D., and C. D. Whiteman, 2012: Diurnal mountain wind systems. Mountain Weather Research and Forecasting: Recent Progress and Current Challenges, F. Chow, S. De Wekker, and B. Snyder, Eds., Springer, 35–119.
