1. Introduction
In coastal regions, the development of sea-breeze circulations (SBCs) is a common occurrence, driven by a landward atmospheric pressure gradient force resulting from differential heating over adjacent land and sea surfaces (e.g., McGowan 1997). Alongside the land–sea temperature contrast, the strength and inland reach of the SBC are influenced by various environmental factors. These factors include the shape of the coastline (e.g., Baker et al. 2001; Gilliam et al. 2004), land surface properties (e.g., Grant and van den Heever 2014; Igel et al. 2018; Park et al. 2020), background wind patterns (e.g., Porson et al. 2007; Steele et al. 2013), coastal topography (e.g., Qian et al. 2012; Davis et al. 2019), and sea surface temperature (e.g., Tang 2012). Note that the nomenclature for SBCs may vary based on the characteristics of the adjacent water-to-land regions, leading to terms such as lake-breeze circulation, gulf-breeze circulation (GBC), or bay-breeze circulation (BBC) (e.g., Sills et al. 2011; Xu et al. 2013; Caicedo et al. 2019).
The SBC significantly influences coastal weather, playing a crucial role in regulating cloud formation and rainfall distribution in these regions (e.g., Golding et al. 2005; Zheng et al. 2022). When acting as a dominant driving force, the SBC encourages the formation of convective clouds along its frontal boundary by enhancing vertical ascent and destabilizing the lower atmosphere (e.g., Kingsmill 1995; Birch et al. 2015). Moreover, the SBC can interact with preexisting clouds, their associated cold pool boundaries, and local circulations, often leading to significant alterations in the life cycle characteristics of convective clouds (e.g., Brummer et al. 1995; Zhang et al. 2022).
Additionally, the SBC has various other notable impacts on coastal environments. First, it influences local wind profiles, with offshore wind recognized as a renewable energy source since the early 1990s due to its substantial energy potential and expansive installation area over the ocean (e.g., Costoya et al. 2020; Xia et al. 2022). Second, the SBC drives the spatiotemporal distribution of primary and secondary air pollutants, holding implications for air quality, particularly in coastal urban regions (e.g., Geddes et al. 2021; Vizuete et al. 2022). Third, the SBC shapes the habitability of coastal areas. It can temporarily reduce the effects of heatwaves and modify the urban heat island (UHI) intensity, which impacts coastal populations (e.g., Hu and Xue 2016; You et al. 2019).
The Houston–Galveston area is a notable region for experiencing SBCs, particularly during the summer months, due to its humid subtropical climate and coastal location (e.g., Darby 2005; Chen et al. 2011; Banta et al. 2011). This susceptibility aligns with the westward movement of the Bermuda high, which extends its influence over the region (e.g., Li et al. 2020; Wang et al. 2022). Additionally, Houston, the fourth-most populous city in the United States (as of 2020), according to data from the U.S. Census Bureau, contributes to a pronounced UHI effect intensified by emissions from the urban center and numerous industrial refineries (e.g., Streutker 2002; Hu and Xue 2016). The UHI circulation interacts with the SBC, shaping the distribution of the aerosol particles and pollutants and altering cloud and precipitation characteristics over the region (e.g., Marinescu et al. 2021; Fan et al. 2020). Furthermore, the SBC in the region is influenced by the land–sea temperature gradients due to the presence of not only the Gulf of Mexico (GBC) but also Galveston Bay (BBC), adding complexity to predictions of coastal weather and environmental conditions (e.g., Banta et al. 2005; Caicedo et al. 2019).
Given the unique characteristics of this region, a series of comprehensive multiagency measurement campaigns were executed recently, as introduced by Jensen et al. (2022). These initiatives, including the Tracking Aerosol Convection Interactions Experiment (TRACER; Jensen 2019), were conducted between 1 July 2021 and 30 September 2022, primarily by the U.S. Department of Energy’s (DOE) Atmospheric Radiation Measurement (ARM) user facility (Ackerman and Stokes 2003; Mather and Voyles 2013). The overarching goal of these field campaigns was to advance our comprehension of the life cycles of convective clouds and aerosols, and their interactions with a specific focus on the influences of GBCs/BBCs (Jensen et al. 2021).
The present study aligns with one of the scientific objectives outlined in the TRACER science plan, namely, the exploration of GBC/BBC characteristics during the summer season and the environmental factors associated with GBC/BBC formation (Jensen 2019). Our focus extends to understanding the impacts of GBC/BBC on local boundary layer dynamics and cloud attributes. Our emphasis is placed on quantifying three pivotal facets: 1) the environmental conditions conducive to GBC/BBC initiation, 2) the timing and intensity of the frontal passage associated with the GBCs/BBCs at TRACER measurement sites, and 3) the variability in associated cloud and precipitation properties. This emphasis is motivated by a need to characterize the mesoscale influences on convective life cycle toward disentangling dynamical and aerosol effects on convective cloud characteristics. To achieve these quantifications, we use a wealth of multiplatform observations collected throughout the TRACER intensive observing period (IOP), spanning June–September 2022.
This research leverages an extended application of machine learning techniques, specifically the self-organizing map (Kohonen 1990), that is used to identify large-scale meteorological regimes favoring GBC/BBC formation. Additionally, a kernel-based changepoint detection method (Celisse et al. 2018) is used for the automated detection of the frontal passage of the GBC/BBC, offering adaptability to diverse observational datasets and high-resolution model outputs. This research represents the first application of the kernel-based changepoint detection method in the context of studies related to mesoscale meteorological phenomena. Finally, we use a convection-centered Lagrangian tracking methodology (Raut et al. 2021) to reveal the characteristics of convective cells associated with GBC/BBC events given their development in an environment that favored the formation of deep convection while being influenced by the meteorological changes associated with the passage of GBC or BBC across the region.
The manuscript is organized as follows: Section 2 provides a comprehensive description of the ARM dataset and supplementary observational datasets. Section 3 outlines the methodology including the selection of GBC/BBC events, classification of synoptic regimes, and the convective cell tracking algorithm. Section 4 offers an in-depth presentation of the results, including an analysis of the large-scale patterns conducive to GBCs/BBCs, the timing of frontal passages at the ARM sites, alterations in surface and boundary layer characteristics along the fronts, and an examination of the associated convective properties. Section 5 serves as a summary, reviewing the key findings derived from this study.
2. Datasets
a. ARM datasets
The primary datasets used in this study originate from the measurements collected at the first ARM mobile facility (AMF1; Miller et al. 2016) deployed at the main TRACER site in La Porte, Texas (M1; coordinates: 29°40′11″N, 95°03′33″W), and an ancillary site in Guy, Texas (S3; coordinates: 29°19′45″N, 95°44′26″W). Both sites are located at a similar distance from the Gulf of Mexico coast, with the M1 site positioned at 51 km and the S3 site at 64 km (Fig. 1). However, a crucial distinction arises as the M1 site sits just 9 km from the western shoreline of Galveston Bay, while the S3 site resides at a distance of about 73 km from the Galveston Bay. Consequently, the M1 site will exhibit a more pronounced influence from the BBC phenomenon.
Domain for the TRACER field campaign and this study. The locations of M1, S3, and radar KHGX are marked as purple dots.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Our investigation relies heavily on measurements obtained from the ARM surface meteorology (MET) systems (Ritsche 2011), as well as data collected by a Doppler lidar (DL; Newsom and Krishnamurthy 2020).
The surface MET instruments (Kyrouac and Shi 2021) record standard meteorological parameters, including surface wind speed and direction at a height of 10 m above the ground, air temperature T, relative humidity (RH) and vapor pressure at a height of 2 m above the ground, and barometric pressure at a height of 1 m above the ground. The data are collected at 1-min intervals and serve the purpose of identifying the timing of frontal passages associated with the GBC/BBC at the ARM sites.
For probing the vertical air motion within the boundary layer, we rely on the Halo Photonics Streamline Pro DL measurements (Flynn and Morris 2021). Operating in the vertically pointing mode and within the near-infrared frequency spectrum, the DL exhibits sensitivity to backscatter signals from atmospheric aerosols, which are treated as passive tracers of atmospheric wind fields. Note that the DL provides particularly accurate measurements of vertical air motion at a temporal resolution of 1 s in regions where aerosol concentrations are sufficiently high to ensure a robust signal-to-noise ratio, with a bias of less than 1 cm s−1 (Newsom 2016). Consequently, our dataset of high-quality DL measurements is primarily constrained to the boundary layer, where aerosols are ubiquitous. Note that signal-to-noise ratio filtering (>0.01) was applied to the vertical velocity field. This study used data from the one that was deployed to the M1 site.
b. Additional datasets
To identify GBC/BBC events that pass over the M1 site, one of the datasets we use is Level-II data from the S-band Doppler weather radar KHGX (Houston, Texas). Positioned at 29°28′19″N, 95°04′45″W (Fig. 1), the KHGX radar is part of the Next Generation Weather Radar (NEXRAD) network, jointly operated by the National Weather Service, the Federal Aviation Administration, and the U.S. Air Force (NOAA/NWS/ROC 1991). This dataset provides radar reflectivity measurements with a spatial resolution of 0.5° azimuthally and 250 m in the radial direction that extend to a range of 460 km. The radar delivers data at 4–6-min intervals. Separately, the processed radar data also serve as input for a tracking algorithm, which will be discussed in greater detail in section 3c.
Our case selection process further uses data from the Geostationary Operational Environmental Satellite 16 (GOES-16; Goodman et al. 2018). Positioned at an altitude of approximately 36 000 km above Earth and stationed at a longitude of 75.2°W, GOES-16 provides varying spatial resolutions, with the visible channel having a higher resolution at 1 km or less than the infrared channels at a resolution of 4 km. Of particular interest to our study are images captured by the visible red band (channel 2, 0.64 μm), which effectively capture the evolution of cumulus clouds along the trajectory of the GBC/BBC as these circulations traverse the region. This channel has the finest spatial resolution of 0.5 km of all 16 bands. The temporal resolution of the data is 5 min.
Our exploration of the large-scale synoptic conditions conducive to GBC/BBC formation draws upon the fifth major global reanalysis produced by European Centre for Medium-Range Weather Forecasts (ECMWF) (ERA5; Hersbach et al. 2020) dataset. ERA5 emerges from the integration of 4D-Var data assimilation and model forecasts within the ECMWF Integrated Forecasting System (IFS). The ERA5 dataset includes key meteorological parameters across 137 pressure levels within the atmosphere, with the highest level at 0.01 hPa. ERA5 uses a wealth of historical observations to formulate global estimates, including a diverse array of satellite and in situ data. The latter includes records from surface stations, radiosondes, radar retrievals, and aircraft sources. This global gridded dataset exhibits a spatial resolution of 0.25° latitude by 0.25° longitude and a temporal resolution of 1 h. The primary variables used in this study include geopotential heights and wind fields at the 200-, 500-, 700-, and 850-hPa pressure levels.
3. Methodology
a. GBC/BBC event identification
To identify GBC/BBC events and the passage of the gulf-breeze front (GBF) or bay-breeze front (BBF) at the M1 site during the TRACER IOP, we analyze four distinct variables across various data collection platforms. This approach ensures a thorough assessment of GBC/BBC characteristics, including cloud, wind, and water vapor patterns, for each day. These variables include the reflectance from the GOES-16 visible red band, the equivalent radar reflectivity Z from the NEXRAD KHGX-Houston radar, and surface wind direction and water vapor mixing ratio measured at the M1 site using the surface MET instruments. Our criteria for selection are stringent—we only consider days where GBF/BBF signatures are evident in at least three of these four data streams to guarantee the reliability of our choices.
If a case is identified as a GBC/BBC day according to the M1 site, we assume that the S3 site may also have the potential to be influenced by the GBC. However, the front may not extend that far inland on some days, and the maximum distance of front propagation depends strongly on the background conditions and the intensity of the GBF.
Further description of the criteria associated with each variable is presented in the subsequent paragraphs.
1) Inland progression of cumulus clouds on GOES-16 visible imagery
Our initial dataset for case selection derives from the GOES-16 visible red band imagery. In Houston–Galveston, a GBC event is often characterized by a distinct line of cumulus or cumulonimbus clouds, generally aligned parallel to the Gulf of Mexico’s coastline. Over Galveston Bay, the cloud formation zone associated with the BBC typically originates along the bay’s coastline, oftentimes merging with the GBCs (Banta et al. 2005; Caicedo et al. 2019). As the day progresses into the afternoon hours, a cloud-free area at low altitudes often advances inland. This conspicuous line of clouds at the forefront of the advancing marine air mass serves as the delineation for the GBF/BBF boundary.
As an example, on 17 August 2022, we observed a notable instance of an GBC/BBC event, as illustrated in Fig. 2. This event was characterized by a prominent line of cloudiness exhibiting a high visible reflectance factor, effectively marking the GBF along the Gulf coast (red lines) and BBF along the Galveston Bay coast (yellow lines). This distinctive cloud formation first reached the ARM M1 site at approximately 1400 local time (LT) (Fig. 2a). Further inland, we observed the formation of numerous cumulus clouds, a consequence of convective instability induced by the daytime heating of the land surface. These inland clouds exhibit a relatively more uniform distribution than those observed along the GBF and BBF when no other meteorological systems interacted with these circulations.
Reflectance factors from GOES-16 visible red band (0.64 μm) for a GBC/BBC event observed on 17 Aug 2022. Higher reflectance factor values indicate optically thicker clouds. Red triangle marker indicates the location of the M1 site, and the red star marker indicates the location of the S3 site. Red lines highlight the locations of the GBFs, and the yellow lines represent the location of the BBFs. The lines represent the trailing edges of the fronts to avoid overlap between the reflectance signatures and the superimposed colored lines.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
As the day progressed, a cloud-free zone began to expand inland (Figs. 2b–d), closely tracing the contours of the Gulf of Mexico and Galveston Bay coastlines, mirroring the development and northwestward propagation of these fronts. After 1622 LT (Fig. 2d), as the land surface gradually cooled and the GBC/BBC weakened, the widespread cumulus clouds started to dissipate. The duration of this event, from the initial discernment of the GBF/BBF near the coast to the point where a distinct boundary was no longer identifiable, spanned approximately 4 h.
On some days, low-cloud evolution may not be visible due to high clouds that block the satellite view such as on 4 June 2022 (Fig. B1 in appendix B), potentially causing our initial criterion to miss GBC/BBC events. To capture events that may occur under these conditions, three additional criteria are introduced below, ensuring the inclusion of events on days with high clouds if they meet the specified conditions.
2) Radar reflectivity thin lines passing over the ARM site
The second datastream employed in our case selection process originates from the NEXRAD KHGX-Houston radar. GBFs/BBFs are often discernible as “thin lines” within the boundary layer, displaying heightened radar reflectivity values on low-angle plan position indicator scans, a phenomenon also noted in previous studies (e.g., Purba et al. 2021; Li et al. 2023). Across the TRACER study region, these intensified reflectivity thin lines typically run parallel to the gulf coastline, which signifies a GBF, or adopt the configuration of the Galveston Bay coastline, which represents a BBF.
For instance, on 17 August 2022 (Fig. 3), we observed thin lines (red lines) positioned near the gulf coast at 1413 LT, representing GBFs. Note that there were isolated convective clouds with high reflectivity values (>40 dBZ) initiated ahead of this front located at the left bottom corner of the radar domain on this day. We also observed several curvatures mirroring the contours of the north and west segments of Galveston Bay, signifying BBFs (yellow lines). This distinctive reflectivity signature is frequently ascribed to radar returns arising from insects or other passive atmospheric tracers accumulating within the boundary layer. Such accumulation occurs due to the enhanced convergence of marine and continental air masses (Atkins and Wakimoto 1997; Davis et al. 2019).
Radar reflectivity at 0.9° elevation from the KHGX radar for a GBF/BBF passage observed at 1413 LT 17 Aug 2022. Red lines highlight the locations of the GBFs, and the yellow lines represent the location of the BBFs. The lines represent the trailing edges of the fronts to avoid overlap between the reflectivity signatures and the superimposed colored lines.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
It is essential to acknowledge that specific circumstances may obscure the reflectivity thin lines, even when GBC/BBC events are occurring. This obscurity can arise, particularly in environmental conditions (e.g., high winds and postprecipitation) unfavorable for insects such as on 8 June 2022 (Fig. B2 in appendix B). However, if these cases meet the other three selection criteria, they will be incorporated into our statistical analysis. Moreover, on radar images, outflow boundaries originating from cold pools and other meteorological systems can result in a reflectivity thin line (e.g., on 25 August 2022), potentially complicating the detection of GBFs/BBFs. To ensure the precise identification of the frontal passage timing at the M1 site, we exclude such cases from our analysis based on visual inspection of the radar images.
Based on the initiation positions of these thin lines, we can, to some extent, differentiate between BBF and GBF through visual inspection. To ensure the robustness of our differentiation, we also examine the initiation locations of cumulus clouds on satellite images. Cases exhibiting both radar thin lines and cumulus clouds along the coastline of the Galveston Bay around noon or earlier afternoon are determined to have a distinct BBF. In total, nearly half of the cases (23) show clear BBF influences at the M1 site (Table A1 in appendix A).
Note that the satellite and radar signatures can be highly correlated on some days, but not on days when there are high clouds and/or fewer insects. We use both datasets to prevent misidentification in such situations. For instance, on days with high clouds obscuring low-level cumulus clouds but with visible radar reflectivity thin lines, the inclusion of both datasets enables us to avoid missing such cases that would otherwise be overlooked if only satellite data were used.
3) Sudden changes in surface wind direction
The GBF/BBF passage is primarily characterized by a shift in local wind patterns, as documented by Furberg et al. (2002) and Sills et al. (2011). To identify this key feature, we examine the time series of surface wind directions recorded by the surface MET instruments at the ARM sites.
Our approach involves applying a kernel-based changepoint detection method [kernel changepoint (KCP); Celisse et al. 2018; Arlot et al. 2019], which is available as an open-source package “ruptures” (Truong et al. 2020), to the daily time-series data. This method enables us to pinpoint the exact moment (to 1 min, which is the native resolution of the time-series data) when a shift in surface wind direction occurs, directly corresponding to the timing of the frontal passage. The KCP is chosen primarily due to its ability to incorporate characteristic kernels to detect changes in the full distribution of the data instead of single statistical features of time series, such as mean or variance. In addition, it shows high efficiency in identifying chance points when the number of changes is unknown in advance (Arlot et al. 2019). The KCP has been widely used in the changepoint modeling community and has demonstrated success in various fields, providing robust results (e.g., Jones et al. 2021; Lei et al. 2023).
In this analysis, our objective is to partition the wind direction time series into homogeneous segments, ensuring constancy in the statistical properties within each segment. Specifically, we aim to use KCP to identify kernelized mean change, signifying any change in the data distribution that influences the “kernel mean.” The kernel mean, also referred to as the kernel embedding of distributions, represents a probability distribution within a reproducing kernel Hilbert space (Smola et al. 2007). This technique embeds entire distributions into infinite-dimensional feature spaces, preserving all statistical features of diverse distributions. In this particular application, we leverage the KCP with a characteristic kernel, the Gaussian kernel, or Gaussian radial basis function (RBF), so that any change in the data distribution influences the kernel mean. In other words, this configuration estimates consistently and at the maximum rate all changes in the distribution of the data, without any parametric assumption and without prior knowledge about the number of changes.
To estimate the number of changepoints, we integrated the RBF with the pruned exact linear time (PELT) algorithm (Killick et al. 2012). PELT introduces a penalty value for each additional changepoint, effectively penalizing the inclusion of superfluous changepoints. Defining an optimal penalty value is crucial for refining PELT’s results, as penalties that are too low may result in detecting many false changepoints, while excessively high penalties may overlook genuine changepoints. In our study, we conducted sensitivity tests on five randomly selected GBC/BBC events, exploring penalty values ranging from 1 to 10. The results indicate that all configurations successfully detect GBF/BBF passages at the M1 site for all events considered, demonstrating very consistent frontal timing (differences within 1 min). However, a penalty value of 5 yielded the most favorable outcome by minimizing unnecessary changepoints, which is easier to locate the GBF/BBF timing, and thus, this value was adopted for the analysis of all events.
Another critical parameter in PELT is the minimum segment length, which defines the minimum number of time steps within each segment. To ensure computational efficiency, we set a minimum segment length of 20 (approximately 20 min), a choice that consistently provided effective results across all cases.
In cases where multiple changepoints are detected in the time series for a single event, we employ a matching procedure to determine the timing of the frontal passage. Initially, we search for an identified distinctive shift point by KCP in the wind direction time series, characterized by a change in wind direction toward southeast or south directions. Subsequently, we compare this timing with the frontal passage timings identified through the radar and satellite images. If the time difference falls within a 15-min window, we consider this changepoint timing as the frontal passage timing, which is then reported in Table A1.
4) Sharp increase in surface water vapor mixing ratio
Another significant feature of the GBF/BBF passage is the notable increase in low-level water vapor, a result of the introduction of moist maritime air into the relatively dry land region (Miller et al. 2003). To capture this critical aspect, we also employ the KCP method but apply it on the time-series data of water vapor mixing ratio measured by the ARM surface MET. We scrutinize the data for a sharp increase occurring within a 20-min window around the timing of the frontal passage over the M1 site identified based on one of the three datasets described above. When such a rapid increase is detected, we consider it a valid feature indicative of the BBF/GBF passage and report it in Table A1. However, note that this feature may not always manifest, and its presence can be influenced by the prevailing background conditions and cloud systems. Note that the parameters utilized in KCP for identifying wind direction changes in section 3a(3) were kept the same in this water vapor mixing ratio analysis for consistency.
We note that the frontal passage times reported in Table A1 should be used as approximate estimates rather than absolute timings of GBF/BBF passage. By definition, a GBF/BBF represents a change in airmass characteristics and can only be estimated based on changes in tracer properties such as mixing ratio and wind direction. For reproducibility, we calculate an uncertainty estimate for the frontal passage time based on the average time difference between the sudden changes in mixing ratio and wind direction across the different GBC/BBC events. Therefore, the passage timings should be interpreted as the timings reported in Table A1 ±12 min.
To illustrate the identification of the surface GBF/BBF, we provide an example that occurred on 9 June 2022, in Fig. 4. This figure displays time-series data of surface wind direction and water vapor mixing ratio at the M1 site, overlaid by the time-series segments identified using KCP (shown in shading). Notably, at 1254 LT, there is a distinct shift in wind direction from south-southwest (with an average wind direction of 206° in the hour prior) to east-southeast (with an average wind direction of 121° in the hour after), indicative of the arrival of a BBF (Fig. 4a). Preceding the arrival of the front, winds remain weak and stagnant for a brief period (1243–1253 LT, not shown). Conversely, the period following the BBF is characterized by sustained, stronger southeasterly (SE) winds reaching up to 5 m s−1. This is accompanied by a significant increase in water vapor mixing ratio, approximately 4 g kg−1 (Fig. 4b), a drop in surface T by 2.3°C, and an increase in RH by 12% (not shown) over the subsequent 50 min. After 1341 LT (not shown), surface T and RH enter a recovery phase over the subsequent hour during which these quantities return to similar conditions experienced prior to the frontal passage, primarily due to diurnal radiative heating. Eventually, surface conditions stabilize with winds at 5 m s−1, T at 32.5°C, and RH at 61.4% after 1423 LT (not shown).
(a) Time series of surface wind direction Wdir and (b) water vapor mixing ratio, for a BBF event that occurred on 9 Jun 2022. The timing of the BBF is at 1254 LT for this case. The blue and pink regions represent time segments identified using the changepoint detection method.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Note that at 1730 LT, another wind direction shift timing is identified, although the changes are gradual (Fig. 4a), potentially indicating a second frontal passage. However, we lack the definitive certainty to classify the second front as a GBF, because it is challenging to verify this timing using other data streams. For instance, on this particular day, clouds were observed propagating inland along with the BBF (Fig. B3 in appendix B); therefore, no clear signature on satellite images can be used to mark the timing or presence of a second front. The thermodynamic conditions have also changed to a marine air mass after the passage of the BBF; therefore, no significant increase can be observed in terms of surface water vapor in Fig. 4b. Therefore, we only document the BBF timing for this day. Similar situations are found for other cases as well. However, in Table A1, we provide an estimate of whether a BBF could be present at the M1 site as discussed in section 3a(2). When the BBC signature is evident or marked as likely, the timing identified at the M1 site is highly likely to correspond to the timing of the BBF.
Throughout the TRACER IOP, we identified a total of 46 days as GBC/BBC days, characterized by clearly meeting a minimum of three out of four predefined criteria (Table A1). Among all of these days, 53% displayed the initiation of convective clouds along the GBF/BBF during its progression inland. This observation highlights the influence of the GBF/BBF in triggering convective activity these days. Note that our methodology is focused on identifying days when we are confident a GBF/BBF has passed over the M1 site and does not identify every GBC/BBC event during the TRACER IOP. For instance, if a day with GBC/BBC has high clouds obscuring low-cloud propagation in satellite images and, at the same time, the environmental conditions are unfavorable for insects, such an event will not be counted in our analysis.
b. Synoptic regime classification
To investigate the influence of large-scale weather patterns on GBC/BBC characteristics and subsequent convective cloud properties, we use the self-organizing map (SOM) approach—an unsupervised machine learning technique (Kohonen 1990). The advantage of using the SOM approach is its ability to autonomously identify continuously distributed synoptic regimes without the need for predefined target outputs. Unlike traditional clustering methods like k-means, the SOM approach provides a more accurate representation of daily weather conditions and their evolution (Mechem et al. 2018; Juliano and Lebo 2020).
In our prior research (Wang et al. 2022), we discerned three primary summertime synoptic regimes within the southeastern Texas region by applying the SOM approach to 10 years (2010–19) of daily 700-hPa geopotential height anomalies (at 0000 UTC) during the summer months (June–September). We opted for the 700-hPa level because it captures the dominant high pressure systems and occasional midlevel troughs that influence the region during the summer.
In our current study, we expand our analysis by including three additional years of data (2020–22), including the year of the TRACER IOP in 2022. We maintain consistency with our previous work by utilizing the same input data for the SOM approach—specifically, the 700-hPa geopotential height anomalies at 0000 UTC. We also maintain the same domain size and initial SOM parameters. Our model performance in terms of error rates and classification results closely matches that of the 10-yr dataset in Wang et al. (2022). A more detailed description of the SOM setup and the related sensitivity tests can be found in Wang et al. (2022).
In alignment with the methodology established in Wang et al. (2022), we identify three dominant synoptic regimes in addition to a transitional regime (Fig. 5). The primary regimes consist of a pretrough regime, characterized by a synoptic trough approaching from the northwest (NW) (Fig. 5a); a posttrough regime, featuring northerly flow following the passage of a trough (Fig. 5b); and an anticyclonic regime linked to a well-established high pressure system over the region (Fig. 5c). A fourth regime is also identified that describes the transitions between the other three regimes (Fig. 5d). A detailed description of each regime can be found in section 3.1 in Wang et al. (2022).
Composites of 700-hPa geopotential heights and wind vectors for each synoptic regime for days during the TRACER IOP.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Among these regimes, the anticyclonic regime stands out as the most prevalent during the summer across the TRACER IOP, accounting for 64% of days. The pretrough regime follows at 19%, while the transitional regime encompasses 11% of days. The posttrough regime is the least common, observed on only 6% of days.
c. Convective cloud cell tracking
Convective cloud cells initiated over the study region on GBC/BBC days are tracked using the TINT algorithm [TINT Is Not TITAN (Thunderstorm Identification, Tracking Analysis, and Nowcasting); Dixon and Wiener 1993], as developed by Raut et al. (2021). TINT is an open-source tracking tool designed to automatically estimate the trajectories of moving cloud objects in sequential images, applicable to various two-dimensional datasets, including both remote sensing measurements and model simulations (Fridlind et al. 2019; Köcher et al. 2022). A more comprehensive description of the TINT algorithm is included in Raut et al. (2021).
The inputs for TINT are derived from the 2-km Z values obtained from gridded NEXRAD KHGX-Houston radar data. The selection of the 2-km level is made to accurately sample precipitation areas, following the methodology outlined in Oue et al. (2022). The position of each cell is determined as the difference-weighted center of the region with Z exceeding a certain threshold value. Tracking is conducted for all GBC/BBC-induced convection days listed in Table A1, starting from 0500 LT and continuing until 0459 LT the following day.
To ensure the optimal performance of TINT, several thresholds are employed:
-
Contiguous object definition: To avoid contamination from signal noise, a single object is defined as a contiguous area comprising at least eight radar grid points (1 km × 1 km) with a minimum Z threshold of 10 dBZ, indicating the presence of precipitation echoes.
-
Convective cell identification: An object is classified as a convective cell if its highest 2-km Z exceeds 30 dBZ. This threshold has been commonly used by radar systems to detect convective clouds, as demonstrated in previous studies such as Petersen et al. (1996), Kumar et al. (2016), and Gupta et al. (2024).
-
Deep convective event definition: A deep convective event is defined as a track with a 2-km Z exceeding 40 dBZ and a 30-dBZ echo top height (ETH) exceeding 5 km at any point during its lifetime, following a similar definition to Dixon and Wiener (1993).
-
Domain exclusion: Cells that move into or out of the radar domain (400 km × 400 km) during their life cycle are excluded from the statistical analysis.
-
Lifetime duration: To minimize potential misidentification resulting from uncertainties in radar data or the tracking method, the analysis primarily focuses on deep convection with convective cores lasting longer than 40 min. This criterion ensures that the cells are detected in at least seven or eight consecutive radar scans.
Numerous key parameters are derived and analyzed for these tracked convective cells, including the maximum Z, cell area, 30-dBZ ETH, convective lifetime, and cell propagation speed. The cell area is calculated by multiplying the number of continuous grid points with Z greater than 30 dBZ by the grid resolution. The 30-dBZ ETH is defined as the maximum height at which the 30-dBZ radar echo is observed. Cell propagation speed is determined as the mean rate at which a cell advances from its first radar-detected location to its final radar-detected position. One example for tracked cells is shown in Fig. 6.
Convective cloud cell tracks identified by TINT on 20 Jun 2022. Dots represent the first identification locations of each cell. The gray shading represents the Houston urban region.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
4. Results
a. Environmental conditions supporting GBC/BBC formation
The formation and evolution of a GBC/BBC are profoundly influenced by synoptic-scale atmospheric weather patterns and surface conditions, as demonstrated for other locations in previous studies (Miller et al. 2003; Sills et al. 2011; Arrillaga et al. 2020). Thus, the first component of our research focuses on assessing the synoptic regimes conducive to these circulations and their associated background conditions for the Houston–Galveston region.
As depicted in Fig. 7, our analysis reveals a notable congruence: 74% of the recorded GBC/BBC events align with anticyclonic conditions, while other regimes account for less than 16% each. This finding highlights a strong association between GBC/BBC occurrences and anticyclonic systems in the region, suggesting that GBC/BBC features are more prominent and identifiable under such conditions than other synoptic conditions.
Donut plot showing the percentage of GBC/BBC days in each SOM weather regime.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Within the anticyclonic regime, the east–west ridge of the Bermuda high plays a key role in shaping wind patterns and moisture availability along the Gulf of Mexico coast during summer months (Hill et al. 2010; Wang et al. 2022).
At 200 and 500 hPa, an anticyclonic system positions itself north of Texas, placing the M1 site south of the high pressure center (not shown). Simultaneously, at lower altitudes, a high pressure system emerges between Texas and Louisiana (not shown). The clear conditions associated with the high pressure systems and subsidence allow for strong solar heating of surface which leads to a rapid increase in land surface temperature. This condition is favorable for creating larger temperature differences with ocean surface (Lyons 1972).
Figure 8a shows a violin plot (Hintze and Nelson 1998) that illustrates a wide range of surface mean temperatures at the M1 site prior to the frontal passage arrivals, with a maximum reaching 38°C and the majority exceeding 30°C. Associated with these warm temperatures, background surface wind speeds tend to be calm, usually below 3.5 m s−1 proceeding GBC/BBC events (Fig. 8c). In other words, no GBC/BBC events were observed when background surface wind speed is higher than 3.5 m s−1 in this study.
Violin plots of (a) 2-h mean values of surface temperature, (b) water vapor mixing ratio, and (c) wind speed prior to the onset of the GBC/BBC as a function of synoptic regimes. A violin plot depicts distributions of data using kernel density estimates that represent a smoothed distribution for probability density estimation. The width of each curve corresponds with the approximate frequency of data points in each region. The blue dots indicate each sample.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Additionally, the M1 site’s southwest position relative to the ridge axis results in predominantly southeasterly winds near the surface and substantial moisture from the Gulf of Mexico (Wang et al. 2022). This is illustrated by surface water vapor mixing ratio w measurements before the GBF/BBF arrival at the M1 site (Fig. 8b), with a mean value of 17.3 g kg−1. These synoptic patterns create tropospheric instability across the TRACER domain, fostering conditions conducive to convective cloud development (Hill et al. 2010).
Regarding GBC/BBC cases in other synoptic regimes (Fig. 8, Table 1), lower prefront surface temperatures and water vapor mixing ratios are observed compared to the anticyclonic regime. Specifically, the mean surface temperature for the pretrough regime is 30°C, significantly lower than the 32°C observed in the anticyclonic regime (according to a t test). Additionally, in terms of surface water vapor mixing ratio, other synoptic regimes exhibit drier conditions, with a mean value below 16 g kg−1, in contrast to the anticyclonic regime. These differences are significant according to a t test.
Mean values of 2-h prefrontal surface temperature 〈T〉 (°C), water vapor mixing ratio 〈w〉 (g kg−1), and wind speed 〈wspd〉 (m s−1) presented as a function of prefrontal wind direction, month of year, time of day, and weather regime, as well as the changes in these variables ΔT (°C), Δw (g kg−1), and Δwspd (m s−1) after the arrival of the GBF/BBF at the M1 site for cases within the anticyclonic regime. The timings of the frontal passage at the M1 and S3 sites are also presented. The last column presents the number of samples (No.) at the M1 site. Categories with a number of samples lower than 3 are not shown. The definitions of each wind direction are as follows: N: 337.5°–360°, 0°–22.5°, S: 157.5°–202.5°, E: 67.5°–112.5°, W: 247.5°–292.5°, NE: 22.5°–67.5°, SE: 112.5°–157.5°, SW: 202.5°–247.5°, and NW: 292.5°–337.5°.
Within the anticyclonic regime, we further break down cases based on 2-h mean prefrontal surface wind direction, month of year, and hour of day (Table 1) to quantify the variability of prefrontal conditions. Here, the prefrontal wind direction is determined by calculating the most frequently occurring surface wind direction within a 2-h window preceding the frontal passage at the M1 site. The results suggest that prefrontal conditions are, to some extent, sensitive to these factors (i.e., prefrontal surface wind direction, month of year, and hour of day). For instance, only two cases developed under northerly wind conditions, as this wind direction opposes the onshore flow, potentially inhibiting the development and/or propagation of the GBF/BBF. Most of the cases were observed under southerly and southeasterly wind conditions, characterized by warm surface temperatures (mean value = 31°–32°C), light winds (mean value = 1.7–2.3 m s−1), and moderate water vapor mixing ratios (mean value = 17 g kg−1; see Table 1).
The prefrontal conditions also exhibit variability over the TRACER summer IOP, with July being the hottest (mean T = 33°C) than June and September. June tends to be drier, with a water vapor mixing ratio value of 16.5 g kg−1, than other summer months, revealing a significant lack of surface water vapor compared to August (based on results from t tests). The trends observed in T and w are consistent with the monthly statistics when considering all days in the 4 months during the summer of 2022. The mean daily sea surface temperature (SST) over the adjacent Gulf of Mexico (at 1200 LT) derived from the ERA5 data exhibits consistent values throughout the summer months of 2022, with the highest values occurring in July and August (30°C), and slightly lower temperatures observed in June and September, approximately half a degree cooler.
In evaluating the influence of the time of day on prefrontal conditions, we find that 1400 LT is the hottest hour at the M1 site, exhibiting higher temperatures but drier conditions, as expected. However, mean prefrontal wind speeds remain similar throughout the afternoon.
b. Timing of GBF/BBF passage at the ARM sites
The ARM M1 site holds a unique position owing to its proximity to the coastlines of both Galveston Bay and the Gulf of Mexico. This geographic location exposes it to the influence of both a BBC and a GBC when environmental conditions are conducive. Notably, it is positioned on the western side of Galveston Bay, a region known for a more pronounced BBC influence compared to areas on the eastern side (Banta et al. 2005; Caicedo et al. 2019). Conversely, the ARM S3 site is located at a greater distance from the Galveston Bay, but slightly closer to the Gulf of Mexico, which may result in a lesser impact from BBC but a major impact from GBC. Consequently, we anticipate different onset timings for GBC/BBC events at these two locations.
Figure 9 presents a violin plot depicting the timing of the GBF/BBF detected at each site using the KCP method. We typically observe a frontal passage at the M1 site, occurring between 1040 and 1630 LT, with a median value of 1310 LT. It is important to clarify that this frontal passage could potentially signify the BBF, GBF, or a combination of the two. We report the likelihood of the presence of the BBF in Table A1 for each case.
Violin plot of GBF/BBF passage timing at the ARM M1 site and the S3 site. In the middle of each violin plot is a small box-and-whisker plot, with the rectangle showing the ends of the first and third quartiles and the central line showing the median value. For the whiskers, the left end represents the 25th percentile − 1.5 × interquartile range, and the right end represents the 75th percentile + 1.5 × interquartile range.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
In addition, a similar analysis was conducted for the S3 site, where frontal passages were identifiable on a subset of days (31 days). These passages occurred between 1250 and 1900 LT, at an average time of 1610 LT, which is approximately 3 h later than those observed at the M1 site. However, it is essential to acknowledge the substantial variability at both sites, potentially driven by factors such as fluctuations in synoptic conditions, diurnal heating patterns, and the presence of clouds and precipitation, as detailed by Miller et al. (2003) and Steele et al. (2013).
Therefore, we first analyze the timing of the front across different synoptic regimes for both the M1 and S3 sites (Fig. 10). A significant level of variability is observed, particularly in cases occurring during the anticyclonic regime, where the onset time spans from 1000 to 1600 LT at the M1 site (Fig. 10a). A similar distribution is noticed for the pretrough regime, albeit with fewer occurrences. However, in the posttrough and transitional regimes, no fronts arrive at the site earlier than noon, suggesting that on these days, the GBC/BBC either initiates later or progresses inland at a slower pace. The patterns observed at the S3 site parallel those at the M1 site (Fig. 10b), reinforcing the influence of synoptic weather conditions on the timing of sea-breeze front (SBF) passages.
Number of GBF/BBF passages that arrived at (a) M1 and (b) S3 sites at different time intervals (colors).
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
We also investigate the influences of other factors such as background wind direction, time of day, and day of year on the timing of the frontal passages. As shown in Table 1, for cases in the anticyclonic regime, no significant differences are observed according to these breakdowns based on results from t tests.
c. Surface meteorological condition changes associated with GBF/BBF
GBF/BBF intensity, often quantified by the magnitude of thermodynamic or wind speed gradients across the frontal passage, serves as a defining characteristic of this boundary (e.g., Reible et al. 1993). In this section, we assess the strength of the GBF/BBF by assessing changes in surface T, water vapor mixing ratio w, and wind speed wspd within the 30 min following each frontal passage (Fig. 11). These changes (represented by Δ) are calculated by subtracting the mean values of the 5-min period preceding the frontal passage (prefrontal) from the minimum (for T) or maximum (for w and wspd) values observed during the 30 min following the frontal passage. The choice of a 30-min search window is based on the average time it takes for associated quantities to reach a steady value after the arrival of the front at the ARM sites.
Histograms of changes in surface temperature ΔT, water vapor mixing ratio Δw, and wind speed Δwspd after the arrival of the GBF/BBF at the M1 and S3 sites. The densities within each bin are shown in blue lines obtained using a kernel density estimate.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Upon the initial passage of the GBF/BBF, we observe similar changes in surface T at both sites in terms of the median value, which is near −0.5°C (Figs. 11a,d). Note that the distribution at the S3 site is narrower, than the M1 site, with the majority of cases falling in a ΔT range from −0.5° to 0°C. Regarding w, Fig. 11b illustrates an increase of up to 4 g kg−1 at the M1 site during these days, while less significant changes are observed at the S3 site, where the maximum change remains below 1.5 g kg−1 (Fig. 11e). Additionally, surface wind speed experiences a median enhancement of 2.5 m s−1 at the M1 site, slightly surpassing the S3 site (Δwspd = 1.9 m s−1; Figs. 11c,f). Overall, the S3 site exhibits less pronounced variations in surface meteorological conditions, likely attributed to its greater distance from both the bay and gulf coastlines, where the fronts tend to weaken by the time they reach this site.
In addition, no significant differences are identified across various synoptic regimes (not shown). When breaking down the analysis based on the different prefrontal wind directions (Table 1), focusing solely on cases within the anticyclonic regime, it is evident that cases with southwesterly (SW) and westerly (W) prefrontal winds stand out, indicating a stronger front with more substantial changes in all three surface parameters than cases in other wind direction categories. Among all categories, the northeast (NE) category shows the weakest GBF/BBF, showing significantly lower values in ΔT and Δw, than the SW category, as indicated by results from t tests.
In terms of the monthly variation (Table 1), July, the hottest month of the year, shows a stronger GBF/BBF than other months in summer. The ΔT (Δwspd), in particular, is substantially larger for fronts arriving in July than those in September (August), and these differences are statistically significant based on t tests. Fronts arriving around 1400 LT (Table 1), the hottest hour of the day, tend to appear stronger in contrast to fronts arriving at an earlier time such as 1000 LT, showing a larger ΔT. These comparisons also pass the t tests.
d. Boundary layer vertical velocity
In addition to the changes in surface meteorological conditions, the arrival of the GBF/BBF leads to substantial and complex modifications in lower tropospheric dynamics (e.g., Ribeiro and Seluchi 2018). Low-level convergence at the frontal boundary drives localized upward motion that can induce convective initiation and development or sustain existing convection and cloudiness. To investigate the variability in boundary layer forcing, we create a composite of vertical velocity below 2 km, measured with a Doppler lidar at the M1 site, centered around the time of the SBF passage for all cases, as well as cases in different synoptic regimes (Fig. 12). Note that statistics for the posttrough regime are not shown due to the lack of samples in this regime.
Mean values of vertical velocity for GBC/BBC cases identified in different synoptic regimes. Time 0 indicates the timing of the front arrival at the M1 site. The term n is the number of samples included within each composite.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
A pronounced updraft signature is evident at the time of the GBF/BBF arrival due to enhanced low-level wind convergence, as depicted in Fig. 12a, consistent with findings in Finkele et al. (1995). The mean magnitudes of the updraft velocities are as large as 2 m s−1, which is comparable to boundary layer updrafts observed during sea-breeze cases in other regions (Wagner et al. 2022; Liu et al. 2023). The primary updraft core typically initiates approximately 5 min before the GBF/BBF arrival and persists for about 10 min on average, extending to an average height of approximately 1 km. During some cases, the primary updrafts can extend to a maximum height of 1.7 km, such as the case shown in Fig. 13. During this period, updrafts consistently appear across different cases, with a frequency of occurrence exceeding 60% (not shown) and the vertical velocity variances also show a peak, indicating high turbulence in the boundary layer (Fig. 14). These upward motions likely contribute to the formation of convective clouds along the GBC/BBC convergence zone, when other conditions are favorable. For example, the lifting condensation level measured at 0630 LT is around 0.1 km (median) and around 1.2 km (median) at 1230 LT during days of convective clouds, which are favorable conditions for convection to occur.
An example of vertical velocity (shading) for the BBC/GBC case occurred on 11 Jul 2022 along with the vertical velocity variance (dashed line).
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
As in Fig. 12, but for vertical velocity variance.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
When breaking down the analysis into different synoptic regimes, the updraft signature associated with the GBF/BBF is consistently present (Figs. 12b–d). A secondary updraft mode emerges approximately 15–30 min before the arrival of the fronts, particularly pronounced in the pretrough and anticyclonic regimes. Apart from these secondary modes, updrafts dominate throughout the entire 2-h window.
Note that this secondary mode may be attributed to prefrontal gravity waves (e.g., Sha et al. 1993), indicating the initial influence of the GBF/BBF. This is clearly shown in Fig. 13 on 11 July 2022. Additionally, this mode may be associated with the interactions between the preexisting horizontal convective rolls (HCRs; Weckwerth et al. 1997) and the frontal passages (Fovell and Dailey 2001). These HCRs, also known as horizontal roll vortices or cloud trees, were clearly discernible on some GBC/BBC days in the GOES-16 images. For instance, on 8 June 2022 (Fig. B4 in appendix B), we observed the formation of elongated rolls of cumulus clouds aligned in parallel with the prevailing low-level wind, which was predominantly southwesterly, during the morning hours. At the same time, we observed two peaks in the time series of the upward motion in the boundary layer on that day (not shown).
e. Precipitating cloud propagation
In this subsection, we investigate the precipitating core properties for convective clouds associated with the GBC/BBC over the Houston–Galveston area. As shown in previous studies (e.g., Paski et al. 2019), such mesoscale fronts frequently trigger the development of daytime convective clouds along coastal regions, driven by the enhanced convergence of low-level horizontal winds and updrafts along and ahead of the front. To gain deeper insights into the characteristics of the convective cores initiated after the detection of the GBF/BBF at the M1 site, we employ a Lagrangian framework to track convective rainfall echoes using KHGX radar data for the convective clouds associated with these circulations and listed in Table A1. Our focus is on precipitating clouds that initiate over land after 1000 LT and within 120 km of the coastline for our analysis. Statistics for cells initiated beyond a distance of 120 km are not presented due to a significant reduction in the number of available samples.
Our findings reveal that the majority of these precipitating cores tend to develop during the anticyclonic weather regime, predominantly cells with small sizes. The number of cells tends to increase as the day progresses into the afternoon, peaking between 1500 and 1600 LT. These precipitating cores exhibit relatively short lifespans, with an average duration of about 63 min and a mean propagation speed of 5 m s−1.
To delve into how the properties of these precipitating cores evolve as they propagate, we present the cell count, cell area, 30-dBZ ETH, and propagation speed as functions of their distance from the coastline in Fig. 15. Note that each box in Fig. 14 includes convective properties for precipitating cores identified or tracked within the defined distance range, without excluding cells at mature and dissipating stages of their lifetime. In general, there is a declining trend in the number of cells initiated as the distance from the coastline increases (Fig. 15a). The most frequent initiation of cells is typically found within the 20–40-km distance range, with a peak count of 2876 cells. Notably, both the mean and maximum values of precipitating cell area and ETH display an increasing trend as cells mature while advancing inland. These trends are particularly prominent within 80 km from the coastline, with these cells retaining their maximum strength when they reach regions between 80 and 120 km from the coast. Additionally, the enhanced convective strength in this region is attributed to the potential influence of the urban effect, including increased anthropogenic aerosol number concentration and urban heat island effect (Fan et al. 2020; Steensen et al. 2022). Regarding the mean of the cell propagation speed, we observe a slight acceleration from 3.9 to 5.3 m s−1 as we move from the coast to 100 km inland, implying a slight increase in the speed of front propagation.
(a) Number of precipitating cells in each distance range bin in units of kilometers. (b) Box-and-whisker plots of precipitating cell area, (c) ETH, and (d) moving speed as a function of their distance from the coastline. The rectangle shows the ends of the first and third quartiles, and the central line shows the median value. For the whiskers, the bottom end represents the 25th percentile − 1.5 × interquartile range, and the top end represents the 75th percentile + 1.5 × interquartile range. The outliers are not shown. The white dots are mean values, while the white line in the middle of each box is the median value.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
However, it is crucial to acknowledge the presence of a large spread for each parameter within each distance bin. This variability may arise from various factors, including the interactions of the fronts with other circulations such as the UHI effect and aerosol particles present in the region.
Note that the use of radar data for cell tracking allows us to characterize the life cycle of small, short-lived, slow-moving afternoon convective cores. However, due to the limitation of the pencil point measurement of boundary layer properties at the M1 site, the tracked convective cells can only be associated with an environment conducive to the development and maintenance of GBC/BBC events. To confidently attribute not only the development but also the initiation of these convective cells to the GBC/BBC, future work will involve large-eddy simulations, which will provide a more comprehensive understanding of the role of GBC/BBC events in convective cell initiation processes.
5. Conclusions
Situated in a humid subtropical climate region with diverse aerosol sources, Houston–Galveston has garnered significant attention in recent decades for its research on GBC/BBC characteristics, convective cloud evolution, air pollution transport, and aerosol–convection interactions. This region was also the focal point for the TRACER field campaign, which aimed to investigate the life cycles of clouds and aerosols, particularly in association with GBC/BBC. Leveraging the extensive datasets collected during TRACER, we have endeavored to enhance our understanding of GBC/BBC characteristics, their surrounding environments, and the associated convective clouds. Our analysis has identified a total of 46 GBC/BBC cases during the IOP, drawing several key conclusions from these cases:
-
Within the Houston–Galveston region, 74% of the GBC/BBC cases observed during the TRACER IOP occurred under anticyclonic conditions. Notably, a predominant occurrence is noted when the background surface wind direction emanates from the southwest. This synoptic regime is conducive to GBC/BBC formation because it favors a notable temperature contrast between land and ocean while maintaining light surface winds, facilitating the genesis of such circulations. Additionally, it provides the sufficient moisture content for convective clouds to initiate along the length of the GBF/BBF.
-
Along the western shoreline of Galveston Bay, the M1 site is markedly influenced by the BBC (nearly half of the cases), resulting in an earlier onset of frontal boundaries, typically occurring around 1310 ± 0012 LT on average, compared to the S3 site, which witnesses these boundaries around 1610 ± 0012 LT on average. Variability in the timing of GBF/BBF onset is discernibly influenced by synoptic regimes but does not exhibit a statistically significant sensitivity to prefrontal surface wind speed and month of year when analyzing cases in the anticyclonic regime. In addition, these fronts exhibit greater intensity at the M1 site, particularly when the surface background wind direction originates from the southwest or west. In such scenarios, a significant increase in water vapor mixing ratio and wind speed is observed, accompanied by a pronounced drop in surface T after the arrival of the GBF/BBF.
-
The arrival of the front triggers an increase in upward motions within the boundary layer, featuring mean velocities of up to 2 m s−1 up to 1 km at the moment of GBF/BBF passage. For a fraction of cases, these regions of enhanced updrafts give rise to isolated convective cores and subsequent precipitation. Significantly, the majority of these convective cores materialize within 100 km from the coast, peaking in occurrence within the region where the distance from the coast spans between 20 and 40 km. These precipitating cores exhibit a relatively short lifespan, averaging about 63 min, and maintain a sedate pace of movement, with a mean propagation speed of 5 m s−1.
In summation, this study offers a comprehensive assessment encompassing large-scale synoptic regimes, surface conditions, and the intricacies of boundary layer updrafts underlying the genesis and evolution of GBC/BBC. Furthermore, it sheds light on the variability of GBC/BBC features within the Houston–Galveston region during the summer months, thereby providing invaluable insights essential for the refinement of numerical atmospheric models aimed at simulating such mesoscale circulations and their associated convective clouds.
Acknowledgments.
This project was supported by the U.S. Department of Energy (DOE) Atmospheric System Research (ASR) program and the Office of Workforce Development for Teachers and Scientists (WDTS) under the Science Undergraduate Laboratory Internship (SULI) program. This paper has been authored by employees of Brookhaven Science Associates, LLC, under Contract DE-SC0012704 with the U.S. DOE. We would like to acknowledge the DOE Early Career Research Program and the ARM TRACER operation and science teams. We also acknowledge the Atmospheric Radiation Measurement (ARM) program, a user facility of the U.S. DOE, Office of Science, sponsored by the Office of Biological and Environmental Research, and support from the ASR program of that office.
Data availability statement.
ARM data can be downloaded from https://adc.arm.gov/discovery/#/. GOES-16 data Knapp and Wilikins (2018) are available at https://data.nodc.noaa.gov/cgi-bin/iso?id=gov.noaa.ncdc:C00993#. NEXRAD data are accessible at https://registry.opendata.aws/noaa-nexrad/. ERA5 data are available at https://cds.climate.copernicus.eu/#!/home. The ruptures package is available at https://centre-borelli.github.io/ruptures-docs/. The TINT package is available at https://github.com/openradar/TINT.
APPENDIX A
Identified GBC/BBC Cases
Table A1. shows the identified GBC/BBC days, the associated thermodynamic and dynamic properties, and the SOM weather regimes.
GBC/BBC days and timing (LT) of the identified GBF or BBF boundaries over the ARM M1 and S3 sites based on wind direction (Wdir) and water vapor mixing ratio (w, in parentheses), the probability of BBF influence at the M1 site, the type of the associated convective cloud (LC = locally forced, LS = large scale), the SOM weather regime, and invalid criterion at the M1 site when selecting GBC/BBC cases. The background wind includes the averaged 850-hPa wind speed (wspd; m s−1) and the most frequent wind direction within an inland region adjacent to the ARM sites during the hours from 0500 to 0900 LT. The boundaries of this region are defined by the coordinates of four corners: (29.66°N, 96.54°W), (30.27°N, 95.39°W), (29.83°N, 95.15°W), and (29.22°N, 96.30°W).
APPENDIX B
Supporting Plots
Figures B1–B4 show the satellite and radar images for particular days during the TRACER IOP.
Reflectance factors from GOES-16 visible red band for a GBC/BBC event observed on 4 Jun 2022. Higher reflectance values indicate optically thicker clouds. Red triangle marker indicates the location of the M1 site, and the red star marker indicates the location of the S3 site.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
Radar reflectivity at 1.5° elevation from the KHGX radar for a GBF/BBF passage observed at 1422 LT 8 Jun 2022.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
As in Fig. B1, but for 10 Jun 2022.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
As in Fig. B1, but for 8 Jun 2022.
Citation: Monthly Weather Review 152, 10; 10.1175/MWR-D-23-0292.1
REFERENCES
Ackerman, T. P., and G. M. Stokes, 2003: The atmospheric radiation measurement program. Phys. Today, 56, 38–44, https://doi.org/10.1063/1.1554135.
Arlot, S., A. Celisse, and Z. Harchaoui, 2019: A kernel multiple change-point algorithm via model selection. J. Mach. Learn. Res., 20, 1–56.
Arrillaga, J. A., P. Jiménez, J. V.-G. de Arellano, M. A. Jiménez, C. Román-Cascón, M. Sastre, and C. Yagüe, 2020: Analyzing the synoptic-, meso- and local-scale involved in sea breeze formation and frontal characteristics. J. Geophys. Res. Atmos., 125, e2019JD031302, https://doi.org/10.1029/2019JD031302.
Atkins, N. T., and R. M. Wakimoto, 1997: Influence of the synoptic-scale flow on sea breezes observed during cape. Mon. Wea. Rev., 125, 2112–2130, https://doi.org/10.1175/1520-0493(1997)125<2112:IOTSSF>2.0.CO;2.
Baker, R. D., B. H. Lynn, A. Boone, W.-K. Tao, and J. Simpson, 2001: The influence of soil moisture, coastline curvature, and land-breeze circulations on sea-breeze-initiated precipitation. J. Hydrometeor., 2, 193–211, https://doi.org/10.1175/1525-7541(2001)002<0193:TIOSMC>2.0.CO;2.
Banta, R. M., and Coauthors, 2005: A bad Air day in Houston. Bull. Amer. Meteor. Soc., 86, 657–670, https://doi.org/10.1175/BAMS-86-5-657.
Banta, R. M., and Coauthors, 2011: Dependence of daily peak O3 concentrations near Houston, Texas on environmental factors: Wind speed, temperature, and boundary-layer depth. Atmos. Environ., 45, 162–173, https://doi.org/10.1016/j.atmosenv.2010.09.030.
Birch, C. E., M. J. Roberts, L. Garcia-Carreras, D. Ackerley, M. J. Reeder, A. P. Lock, and R. Schiemann, 2015: Sea-breeze dynamics and convection initiation: The influence of convective parameterization in weather and climate model biases. J. Climate, 28, 8093–8108, https://doi.org/10.1175/JCLI-D-14-00850.1.
Brummer, B., B. Hennemuth, A. Rhodin, and S. Thiemann, 1995: Interaction of a cold front with a sea-breeze front observations. Tellus, 47A, 383–402, https://doi.org/10.1034/j.1600-0870.1995.t01-2-00001.x.
Caicedo, V., and Coauthors, 2019: Bay breeze and sea breeze circulation impacts on the planetary boundary layer and air quality from an observed and modeled DISCOVER-AQ Texas case study. J. Geophys. Res. Atmos., 124, 7359–7378, https://doi.org/10.1029/2019JD030523.
Celisse, A., G. Marot, M. Pierre-Jean, and G. J. Rigaill, 2018: New efficient algorithms for multiple change-point detection with reproducing kernels. Comput. Stat. Data Anal., 128, 200–220, https://doi.org/10.1016/j.csda.2018.07.002.
Chen, F., S. Miao, M. Tewari, J.-W. Bao, and H. Kusaka, 2011: A numerical study of interactions between surface forcing and sea breeze circulations and their effects on stagnation in the greater Houston area. J. Geophys. Res., 116, D12105, https://doi.org/10.1029/2010JD015533.
Costoya, X., M. deCastro, D. Carvalho, and M. Gómez-Gesteira, 2020: On the suitability of offshore wind energy resource in the United States of America for the 21st century. Appl. Energy, 262, 114537, https://doi.org/10.1016/j.apenergy.2020.114537.
Darby, L. S., 2005: Cluster analysis of surface winds in Houston, Texas, and the impact of wind patterns on ozone. J. Appl. Meteor., 44, 1788–1806, https://doi.org/10.1175/JAM2320.1.
Davis, S. R., J. T. Farrar, R. A. Weller, H. Jiang, and L. J. Pratt, 2019: The land-sea breeze of the red sea: Observations, simulations, and relationships to regional moisture transport. J. Geophys. Res. Atmos., 124, 13 803–13 825, https://doi.org/10.1029/2019JD031007.
Dixon, M., and G. Wiener, 1993: TITAN: Thunderstorm identification, tracking, analysis, and nowcasting—A radar-based methodology. J. Atmos. Oceanic Technol., 10, 785–797, https://doi.org/10.1175/1520-0426(1993)010<0785:TTITAA>2.0.CO;2.
Fan, J., Y. Zhang, Z. Li, J. Hu, and D. Rosenfeld, 2020: Urbanization-induced land and aerosol impacts on sea-breeze circulation and convective precipitation. Atmos. Chem. Phys., 20, 14 163–14 182, https://doi.org/10.5194/acp-20-14163-2020.
Finkele, K., J. M. Hacker, H. Kraus, and R. A. D. Byron-Scott, 1995: A complete sea-breeze circulation cell derived from aircraft observations. Bound.-Layer Meteor., 73, 299–317, https://doi.org/10.1007/BF00711261.
Flynn, D., and V. Morris, 2021: Total sky imager (tsiskycover). Atmospheric Radiation Measurement (ARM) user facility, accessed 2 October 2023, https://doi.org/10.5439/1025308.
Fovell, R. G., and P. S. Dailey, 2001: Numerical simulation of the interaction between the sea-breeze front and horizontal convective rolls. Part II: Alongshore ambient flow. Mon. Wea. Rev., 129, 2057–2072, https://doi.org/10.1175/1520-0493(2001)129<2057:NSOTIB>2.0.CO;2.
Fridlind, A. M., and Coauthors, 2019: Use of polarimetric radar measurements to constrain simulated convective cell evolution: A pilot study with Lagrangian tracking. Atmos. Meas. Tech., 12, 2979–3000, https://doi.org/10.5194/amt-12-2979-2019.
Furberg, M., D. G. Steyn, and M. Baldi, 2002: The climatology of sea breezes on Sardinia. Int. J. Climatol., 22, 917–932, https://doi.org/10.1002/joc.780.
Geddes, J. A., B. Wang, and D. Li, 2021: Ozone and nitrogen dioxide pollution in a coastal urban environment: The role of sea breezes, and implications of their representation for remote sensing of local air quality. J. Geophys. Res. Atmos., 126, e2021JD035314, https://doi.org/10.1029/2021JD035314.
Gilliam, R. C., S. Raman, and D. D. S. Niyogi, 2004: Observational and numerical study on the influence of large-scale flow direction and coastline shape on sea-breeze evolution. Bound.-Layer Meteor., 111, 275–300, https://doi.org/10.1023/B:BOUN.0000016494.99539.5a.
Golding, B., P. Clark, and B. May, 2005: The boscastle flood: Meteorological analysis of the conditions leading to flooding on 16 August 2004. Weather, 60, 230–235, https://doi.org/10.1256/wea.71.05.
Goodman, S. J., T. J. Schmit, J. Daniels, W. Denig, and K. Metcalf, 2018: 1.05-GOES: Past, present, and future. Comprehensive Remote Sens., 1, 119–149, https://doi.org/10.1016/B978-0-12-409548-9.10315-X.
Grant, L. D., and S. C. van den Heever, 2014: Aerosol-cloud-land surface interactions within tropical sea breeze convection. J. Geophys. Res. Atmos., 119, 8340–8361, https://doi.org/10.1002/2014JD021912.
Gupta, S., D. Wang, S. E. Giangrande, T. S. Biscaro, and M. P. Jensen, 2024: Lifecycle of updrafts and mass flux in isolated deep convection over the Amazon rainforest: insights from cell tracking. Atmos. Chem. Phys., 24, 4487–4510, https://doi.org/10.5194/acp-24-4487-2024.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hill, C. M., P. J. Fitzpatrick, J. H. Corbin, Y. H. Lau, and S. K. Bhate, 2010: Summertime precipitation regimes associated with the sea breeze and land breeze in southern Mississippi and eastern Louisiana. Wea. Forecasting, 25, 1755–1779, https://doi.org/10.1175/2010WAF2222340.1.
Hintze, J. L., and R. D. Nelson, 1998: Violin plots: A box plot-density trace synergism. Amer. Stat., 52, 181–184, https://doi.org/10.1080/00031305.1998.10480559.
Hu, X.-M., and M. Xue, 2016: Influence of synoptic sea-breeze fronts on the urban heat island intensity in Dallas–Fort Worth, Texas. Mon. Wea. Rev., 144, 1487–1507, https://doi.org/10.1175/MWR-D-15-0201.1.
Igel, A. L., S. C. van den Heever, and J. S. Johnson, 2018: Meteorological and land surface properties impacting sea breeze extent and aerosol distribution in a dry environment. J. Geophys. Res. Atmos., 123, 22–37, https://doi.org/10.1002/2017JD027339.
Jensen, M. P., 2019: Tracking Aerosol Convection Interactions ExpeRiment (TRACER) science plan. Tech. Rep. BNL-212068-2019-INRE, 41 pp., https://doi.org/10.2172/1561242.
Jensen, M. P., F. Fan, S. Collis, E. Bruning, S. E. Giangrande, A. Miltenberg, and D. Wang, 2021: The Aerosol-Cloud-Precipitation-Climate (ACPC) initiative deep convective cloud roadmap: Tracer and follow-on activities. Tech. Rep., 20 pp., http://acpcinitiative.org/Docs/ACPC_DCC_Roadmap_2021.pdf.
Jensen, M. P., and Coauthors, 2022: A succession of cloud, precipitation, aerosol, and air quality field experiments in the coastal urban environment. Bull. Amer. Meteor. Soc., 103, 103–105, https://doi.org/10.1175/BAMS-D-21-0104.1.
Jones, C., S. Clayton, F. Ribalet, E. V. Armbrust, and Z. Harchaoui, 2021: A kernel-based change detection method to map shifts in phytoplankton communities measured by flow cytometry. Methods Ecol. Evol., 12, 1687–1698, https://doi.org/10.1111/2041-210X.13647.
Juliano, T. W., and Z. J. Lebo, 2020: Linking large-scale circulation patterns to low-cloud properties. Atmos. Chem. Phys., 20, 7125–7138, https://doi.org/10.5194/acp-20-7125-2020.
Killick, R., P. Fearnhead, and I. A. Eckley, 2012: Optimal detection of changepoints with a linear computational cost. J. Amer. Stat. Assoc., 107, 1590–1598, https://doi.org/10.1080/01621459.2012.737745.
Kingsmill, D. E., 1995: Convection initiation associated with a sea-breeze front, a gust front, and their collision. Mon. Wea. Rev., 123, 2913–2933, https://doi.org/10.1175/1520-0493(1995)123<2913:CIAWAS>2.0.CO;2.
Knapp, K. R., and S. L. Wilkins, 2018: Gridded Satellite (GridSat) GOES and CONUS data. Earth Syst. Sci. Data, 10, 1417–1425, https://doi.org/10.5194/essd-10-1417-2018.
Köcher, G., T. Zinner, C. Knote, E. Tetoni, F. Ewald, and M. Hagen, 2022: Evaluation of convective cloud microphysics in numerical weather prediction models with dual-wavelength polarimetric radar observations: Methods and examples. Atmos. Meas. Tech., 15, 1033–1054, https://doi.org/10.5194/amt-15-1033-2022.
Kohonen, T., 1990: The self-organizing map. Proc. IEEE, 78, 1464–1480, https://doi.org/10.1109/5.58325.
Kumar, V. V., A. Protat, C. Jakob, C. R. Williams, S. Rauniyar, G. L. Stephens, and P. T. May, 2016: The estimation of convective mass flux from radar reflectivities. J. Appl. Meteor. Climatol., 55, 1239–1257, https://doi.org/10.1175/JAMC-D-15-0193.1.
Kyrouac, J., and Y. Shi, 2021: Surface Meteorological Instrumentation (MET). Atmospheric Radiation Measurement (ARM) user facility, accessed 2 January 2023, https://doi.org/10.5439/1786358.
Lei, X., Z. Chen, H. Li, and S. Wei, 2023: A change-point detection method for detecting and locating the abrupt changes in distributions of damage-sensitive features of SHM data, with application to structural condition assessment. Struct. Health Monit., 22, 1161–1179, https://doi.org/10.1177/14759217221101320.
Li, J., Y. Pan, Q. Li, D. H. Lenschow, M. Zhou, X. Xiao, Y. Wang, and Z. Cheng, 2023: Observational analyses of a penetrating sea-breeze front in the Beijing-Tianjin-Hebei urban agglomeration. Urban Climate, 47, 101353, https://doi.org/10.1016/j.uclim.2022.101353.
Li, W., Y. Wang, C. Bernier, and M. Estes, 2020: Identification of sea breeze recirculation and its effects on ozone in Houston, TX, during DISCOVER-AQ 2013. J. Geophys. Res. Atmos., 125, e2020JD033165, https://doi.org/10.1029/2020JD033165.
Liu, S., K. D. Sindhu, and D. J. Kirshbaum, 2023: Observations of boundary layer convergence lines and associated updrafts in the U.S. southern great plains. J. Atmos. Sci., 80, 2947–2968, https://doi.org/10.1175/JAS-D-23-0089.1.
Lyons, W. A., 1972: The climatology and prediction of the Chicago lake breeze. J. Appl. Meteor., 11, 1259–1270, https://doi.org/10.1175/1520-0450(1972)011<1259:TCAPOT>2.0.CO;2.
Marinescu, P. J., and Coauthors, 2021: Impacts of varying concentrations of cloud condensation nuclei on deep convective cloud updrafts–A multimodel assessment. J. Atmos. Sci., 78, 1147–1172, https://doi.org/10.1175/JAS-D-20-0200.1.
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.
McGowan, H. A., 1997: Book reviews: Simpson, J.E. 1994: Sea breeze and local winds. Cambridge: Cambridge University Press. xiv + 234 pp. £29.95, US$39.95 cloth. ISBN: 0 521 45211 2. Prog. Phys. Geogr., 21, 622–624, https://doi.org/10.1177/030913339702100419.
Mechem, D. B., C. S. Wittman, M. A. Miller, S. E. Yuter, and S. P. de Szoeke, 2018: Joint synoptic and cloud variability over the northeast Atlantic near the Azores. J. Appl. Meteor. Climatol., 57, 1273–1290, https://doi.org/10.1175/JAMC-D-17-0211.1.
Miller, M. A., K. Nitschke, T. P. Ackerman, W. R. Ferrell, N. Hickmon, and M. Ivey, 2016: The arm mobile facilities. The Atmospheric Radiation Measurement (ARM) Program: The First 20 Years, Meteor. Monogr., No. 57, Amer. Meteor. Soc., https://doi.org/10.1175/AMSMONOGRAPHS-D-15-0051.1.
Miller, S. T. K., B. D. Keim, R. W. Talbot, and H. Mao, 2003: Sea breeze: Structure, forecasting, and impacts. Rev. Geophys., 41, 1011, https://doi.org/10.1029/2003RG000124.
Newsom, R., 2016: Analysis of the uncertainty in wind measurements from the atmospheric radiation measurement Doppler Lidar during XPIA: Field campaign report. Tech. Rep. DOE/SC-ARM-15-091, 16 pp., https://www.arm.gov/publications/programdocs/doe-sc-arm-15-091.pdf.
Newsom, R. K., and R. Krishnamurthy, 2020: Doppler Lidar (DL) instrument handbook. Tech. Rep. DOE/SC-ARM-TR-101, 59 pp., https://www.arm.gov/publications/tech_reports/handbooks/dl_handbook.pdf.
NOAA/NWS/ROC, 1991: NOAA Next Generation Radar (NEXRAD) level 2 base data. NOAA, accessed 2 October 2023, https://doi.org/10.7289/V5W9574V.
Oue, M., S. M. Saleeby, P. J. Marinescu, P. Kollias, and S. C. van den Heever, 2022: Optimizing radar scan strategies for tracking isolated deep convection using observing system simulation experiments. Atmos. Meas. Tech., 15, 4931–4950, https://doi.org/10.5194/amt-15-4931-2022.
Park, J. M., S. C. van den Heever, A. L. Igel, L. D. Grant, J. S. Johnson, S. M. Saleeby, S. D. Miller, and J. S. Reid, 2020: Environmental controls on tropical sea breeze convection and resulting aerosol redistribution. J. Geophys. Res. Atmos., 125, e2019JD031699, https://doi.org/10.1029/2019JD031699.
Paski, J. A. I., I. J. A. Saragih, D. S. Permana, M. I. Hastuti, A. Kristianto, and E. E. S. Makmur, 2019: Simulation of land-sea breeze effect on the diurnal cycle of convective activity in the eastern coast of North Sumatra using WRF model. 2019 IEEE Asia-Pacific Conf. on Geoscience, Electronics and Remote Sensing Technology (AGERS), Jakarta, Indonesia, Institute of Electrical and Electronics Engineers, 67–71, https://doi.org/10.1109/AGERS48446.2019.9034301.
Petersen, W. A., S. A. Rutledge, and R. E. Orville, 1996: Cloud-to-ground lightning observations from TOGA COARE: Selected results and lightning location algorithms. Mon. Wea. Rev., 124, 602–620, https://doi.org/10.1175/1520-0493(1996)124<0602:CTGLOF>2.0.CO;2.
Porson, A., D. G. Steyn, and G. Schayes, 2007: Sea-breeze scaling from numerical model simulations, Part II: Interaction between the sea breeze and slope flows. Bound.-Layer Meteor., 122, 31–41, https://doi.org/10.1007/s10546-006-9092-2.
Purba, D. L. A., I. J. A. Saragih, and D. S. Sosaidi, 2021: Utilization of weather-radar data to observe the Sea Breeze Front (SBF) over the North Coast of Banten-Jakarta (case study in 2018). IOP Conf. Ser. Earth Environ. Sci., 893, 012055, https://doi.org/10.1088/1755-1315/893/1/012055.
Qian, T., C. C. Epifanio, and F. Zhang, 2012: Topographic effects on the tropical land and sea breeze. J. Atmos. Sci., 69, 130–149, https://doi.org/10.1175/JAS-D-11-011.1.
Raut, B. A., R. Jackson, M. Picel, S. M. Collis, M. Bergemann, and C. Jakob, 2021: An adaptive tracking algorithm for convection in simulated and remote sensing data. J. Appl. Meteor. Climatol., 60, 513–526, https://doi.org/10.1175/JAMC-D-20-0119.1.
Reible, D. D., J. E. Simpson, and P. F. Linden, 1993: The sea breeze and gravity-current frontogenesis. Quart. J. Roy. Meteor. Soc., 119 (509), 1–16, https://doi.org/10.1002/qj.49711950902.
Ribeiro, B. Z., and M. E. Seluchi, 2018: A climatology of quasi-linear convective systems and associated synoptic-scale environments in southern Brazil. Int. J. Climatol., 39, 857–877, https://doi.org/10.1002/joc.5847.
Ritsche, M., 2011: ARM surface meteorology systems instrument handbook. Rep. DOE/SC-ARM/TR-086, 20 pp., https://doi.org/10.2172/1007926.
Sha, W., T. Kawamura, and H. Ueda, 1993: A numerical study of nocturnal sea breezes: Prefrontal gravity waves in the compensating flow and inland penetration of the sea-breeze cutoff vortex. J. Atmos. Sci., 50, 1076–1088, https://doi.org/10.1175/1520-0469(1993)050<1076:ANSONS>2.0.CO;2.
Sills, D. M. L., J. R. Brook, I. Levy, P. A. Makar, J. Zhang, and P. A. Taylor, 2011: Lake breezes in the southern great lakes region and their influence during BAQS-Met 2007. Atmos. Chem. Phys., 11, 7955–7973, https://doi.org/10.5194/acp-11-7955-2011.
Smola, A., A. Gretton, L. Song, and B. Schölkopf, 2007: A Hilbert space embedding for distributions. Algorithmic Learning Theory, M. Hutter, R. A. Servedio, and E. Takimoto, Eds., Springer, 13–31.
Steele, C. J., S. R. Dorling, R. von Glasow, and J. Bacon, 2013: Idealized WRF model sensitivity simulations of sea breeze types and their effects on offshore windfields. Atmos. Chem. Phys., 13, 443–461, https://doi.org/10.5194/acp-13-443-2013.
Steensen, B. M., L. Marelle, Ø. Hodnebrog, and G. Myhre, 2022: Future urban heat island influence on precipitation. Clim. Dyn., 58, 3393–3403, https://doi.org/10.1007/s00382-021-06105-z.
Streutker, D. R., 2002: A remote sensing study of the urban heat island of Houston, Texas. Int. J. Remote Sens., 23, 2595–2608, https://doi.org/10.1080/01431160110115023.
Tang, Y., 2012: The effect of variable sea surface temperature on forecasting sea fog and sea breezes: A case study. J. Appl. Meteor. Climatol., 51, 986–990, https://doi.org/10.1175/JAMC-D-11-0253.1.
Truong, C., L. Oudre, and N. Vayatis, 2020: Selective review of offline change point detection methods. Signal Process., 167, 107299, https://doi.org/10.1016/j.sigpro.2019.107299.
Vizuete, W., J. Nielsen-Gammon, J. Dickey, E. Couzo, C. Blanchard, P. Breitenbach, Q. Z. Rasool, and D. Byun, 2022: Meteorological based parameters and ozone exceedances in Houston and other cities in Texas. J. Air Waste Manage. Assoc., 72, 969–984, https://doi.org/10.1080/10962247.2022.2064004.
Wagner, T. J., A. C. Czarnetzki, M. Christiansen, R. B. Pierce, C. O. Stanier, A. F. Dickens, and E. W. Eloranta, 2022: Observations of the development and vertical structure of the lake-breeze circulation during the 2017 Lake Michigan Ozone Study. J. Atmos. Sci., 79, 1005–1020, https://doi.org/10.1175/JAS-D-20-0297.1.