Summary of atmospheric characteristics of days with inland penetrating sea breezes from 2015 to 2021

Sea breezes penetrate inland more than 100 km. Using 7 years of meteorological observations, we have identified 470 cases of deep inland (>100 km) penetrating sea breezes at the Savannah River Site between March and October (27% of days) of 2015–2021. We compared measurements of temperature, dewpoint temperature, incoming solar radiation, cloud fraction, and lightning on days of sea breeze initiation, the day after the sea breeze passage, and all other nonsea breeze (NSB) days for these 8 months over the 7 years. Days of sea breeze initiation were found to have lower cloud fraction, higher temperature, and greater incoming solar radiation compared with NSB days. Variations occurred by time of year as days after the sea breeze passage were found to have higher dewpoint temperature than NSB days in the spring. Lightning density measurements indicated that residual sea breeze conditions could drive earlier initiation of deep convection on days following the sea breeze than normal non sea breeze days. This data set provides a 7‐year record of sea breezes which can be leveraged for future studies.

The Savannah River Site (SRS) is located in the southeastern United States (SEUS) about 150 km inland from the South Carolina coastline (Figure 1).SRS sits on the northwestern edge of the coastal plain bordering the sandhills region just south of the South Carolina-Georgia fall line (Denham, 1999).Elevation at SRS ranges from 40 m above sea level near the Savannah River to 120 m further inland.Due to this low elevation rise from the coast (<1 m per km inland), there is little topography to prevent the inland intrusion of sea breezes which have often been observed at SRS. Buckley andKurzeja (1997a, 1997b) noted that sea breezes unexpectedly modified the transport of chemical tracer during Project STABLE.Wermter et al. (2022) identified that a strong thermal gradient between inland and coastal surface temperature was an additional driver of the deep inland penetration of sea breezes at SRS.This differential heating from near the coast to further inland drives a sufficient mass divergence to generate pressure gradient force large enough for the sea breeze to penetrate farther inland.Viner et al. (2021) identified 177 inland penetrating sea breezes that arrived at SRS through the use of radar reflectivity.Radar coverage between SRS and the coast is provided by radars at Charleston, SC, Columbia, SC, and Peach Tree City, GA, all of which are >70 km from SRS.As such, the radar beam is at least 1 km above the ground at SRS.Thus, one limitation mentioned in their study was that shallow sea breezes reaching SRS might not be identified using this method as the depth of the sea breeze front can range from 300 to 2500 m (Miller et al., 2003).However, with these cases, they noted that when the sea breeze front passed at SRS the dewpoint increased, wind direction turned to the S or SE with a reduction in variability and increased speed, occasionally temperature decreased more rapidly, and there was sharp change in ceilometer attenuated backscatter.These sea breezes typically arrived at SRS in the overnight hours, after 20:00 local time.Elevated dewpoints and elevated ceilometer backscatter often continued into the following day, suggesting a persistent influence from the sea breeze.
Typically, the sea breeze is associated with convection and possible precipitation (Chen et al., 2016;Simpson et al., 2007;Zhu et al., 2017).Fu et al. (2022Fu et al. ( , 2021) ) identified deep convection initiation near the sea breeze front.Sims et al. (2017) noted that in the SEUS, the sea breeze circulation interacts with a circulation from the north near the sandhills region which drives convection generally to the southeast of SRS.However, most sea breeze cases identified by Viner et al. (2021) were not associated with deep convection as thunderstorms or precipitation obscured the radar sea breeze signal.Wermter et al. (2022) noted that sea breeze days were less cloudy than days without a sea breeze for both inland and coastal locations, indicating greater heating but less cloudtopped convection prior to the sea breeze development.However, even though the deep inland penetrating sea breeze does not often trigger deep convection over SRS, it could play a role in cloud development on the subsequent day.
In this study, we will utilize characteristics in atmospheric data determined by Viner et al. (2021) to independently identify sea breezes reaching SRS from 2015 to 2021 and then confirm them using the previous radar method.With that expanded data set, we will compare the day before the sea breeze arrival (as the sea breeze usually arrives between 23:00 and 5:00 UTC, 19:00 to 1:00 local time, BSB), the day after the sea breeze (ASB), and any day that is not included in these sea breeze related days (nonsea breeze [NSB]).By characterizing these differences, we will better understand the influence F I G U R E 1 Location of the Savannah River Site in the southeastern US.Zoomed inset show the location of the meteorological towers (MT1, MT2, MT3), Central Climatology Facility (CCF), Augusta Regional Airport (AGS), Allendale airport (AQX), and the ceilometer (CEI).
of these inland penetrating sea breezes at the SRS.Additionally, this data set of dual verified sea breezes will form a rigorous catalogue of events that can be used by other researchers interested in sea breezes in South Carolina.

| MEASUREMENTS AND METHODS
The same three meteorological towers and the ceilometer (Figure 1) used by Viner et al. (2021) were used to identify sea breeze cases based on the indicators.The meteorological towers are aligned quasiperpendicular to the coast with the first tower (MT1) about 140 km from the coast, the second tower (MT2) about 150 km from the coast, and the third tower (MT3) about 160 km from the coast.The ceilometer is located about 155 km from the coast and 9 km west of MT2.Tower measurements are made at 61 m above ground level, which is above the treetops and limits terrain, frictional, and vegetation influence.Sea breezes were identified when changes in dewpoint and wind direction occurred in temporal succession from south to north (MT1 to MT3) during the afternoon to early morning hours.These were then compared with ceilometer attenuated backscatter that identified changes in the airmass (aerosol properties) due to the sea breeze front.Our research encompasses the months of March to October, when penetrating sea breezes are prevalent in the SEUS, during the years 2015 to 2021.All new instruments were installed on the meteorological towers at SRS in 2013 and 2014.While data for previous years exist, data in these more recent years are more consistent and limit any need to compare across instrument types in our analysis.
We focused on the best sea breeze indicators of dewpoint, wind direction, and ceilometer backscatter, but also examined wind speed and temperature.We included as many indicators as possible because not every sea breeze had all indicators.In a typical case, the sea breeze passage would cause the dewpoint to rise in succession at the three towers and the wind direction would turn in succession to the S or SE.Additionally, the wind direction would become less variable after the sea breeze passage which acted as a good indicator if the winds were already from the S or SE.Temperature would generally drop more sharply with the sea breeze than with nighttime cooling and was not always obvious, especially with weaker sea breezes.The sharper drop in succession at the towers would appear occasionally and was only used as a secondary indicator for confirmation if one of the other data sets was unavailable.As a complement to these measurements, the ceilometer indicated sea breezes by sharp changes in the attenuated backscatter.Generally, there was an increase in backscatter at the low levels, below 1500 m.This increase was likely due to larger particles or particles swelling with increased humidity/dewpoint.Occasionally, backscatter was lower which could have indicated cleaner maritime air brought in by the sea breeze.Another occasional backscatter indicator was a clear undercutting of the backscatter return aloft as the sea breeze cut off the evening convection which seemed to reduce cloud formation along the sea breeze front at SRS.Based on these indicators, each case was given an approximate time of arrival at SRS.This time was not exact due to uncertainty caused by travel across the site towers and smoothing of the ceilometer backscatter to reduce noise, but arrival times ranged from 15:30 to 3:30 EDT (19:30-7:30 UTC).
All cases with indications were considered plausible, but some were identified as having a stronger signal than others.To assure a rigorous sea breeze identification, radar information was then used to confirm each case.In some instances, the radar directly confirmed the sea breezes.In others, sea breeze lines disappeared on the radar before reaching SRS but with plausible continuation of the sea breeze after lost from view of the radar.This was often confirmed by sea breeze lines closer to the radar that seemed to be able to be extrapolated to SRS.Finally, others showed no indication that a sea breeze occurred that day or the radar indicated that the sea breeze stalled before being lost from view on radar imagery.These cases, without dual verification, are not included in this sea breeze data set.
The identified sea breeze cases were used to extract data for comparisons of commonalities of the 7-year data set.Since the average estimated sea breeze arrival time was about 21:30 EDT (1:30 UTC), we divided data into three sets: BSB days, ASB days, and NSB days.BSB days were days (0:00-23:59 EDT) when the sea breeze initiated at the coast and traveled inland.ASB days are the days (0:00-23:59 EDT) following BSB days.And NSB days are all other days (not associated with sea breezes) between March and October for 2015-2021.We then analyzed surface-based measurements of temperature and dewpoint temperature that were made at 2 m above ground level at the SRS Central Climatology Facility (CCF; 5 km southwest of MT2) and surface measurements from meteorological stations at Augusta Regional Airport (AGS) and Allendale airport (AQX).We also analyzed incoming solar radiation measurements at CCF; cloud fraction measurements (from Vaisala CL31 ceilometer; CEI in Figure 1) that were taken from a location 9 km west of MT2; and lightning data from the National Lightning Detection Network (NLDN).
We identified and verified 470 sea breezes that reached SRS over 7 years (2015)(2016)(2017)(2018)(2019)(2020)(2021) during the typical sea breeze season (March to October), which accounts for about 27% of days on average (all dates are listed in Table A1).May had the most sea breezes on average but April, the second most by average, had a maximum of 19 sea breezes in 2017 (Table 1).Overall, the peak for sea breezes occurs from late spring (April) to late summer (Talbot et al., 2007), 78% of sea breezes occurred during these 5 months.Only 22% of sea breezes occurred in the earlier and later parts of the season.The years 2019 and 2020 had 70% fewer sea breezes on average than the average of the other 5 years.In these years, April and June were generally lower than other years but May was nearly the same.

| Temperature and dewpoint temperature
Differences in surface temperatures from BSB days, ASB days, and NSB days for CCF, AGS, and AQX are shown in Figure 2. The difference in surface temperature at CCF shows warmer temperatures occur on the afternoon of the BSB days compared with the ASB day.Additionally, both the BSB and ASB days have warmer temperatures during the day than NSB days, especially in the spring and fall.The clear demarcation in these differences at 08:00 EDT indicates that temperatures on BSB days and ASB days are warmer even at the beginning of the day after the sun rises above the treetops.Temperature differences indicate warmer overnight temperatures following the sea breeze on ASB nights than on BSB nights.This seems counter to Viner et al. (2021) except those temperatures were taken at 61 m above ground.AGS and AQX show similar patterns in temperature differences.
Dewpoint temperature differences at CCF show higher dewpoints on the ASB days (and overnight hours) than on BSB days (Figure 3a), particularly during the spring months.Much higher dewpoints exist in the spring on the ASB days than on NSB days.Springtime months in the SEUS are usually dry, thus the sea breeze brings in uncharacteristically moist air from the coast.Sea breezes in the fall seem to occur on days (BSB) with higher dewpoints than NSB days and these higher dewpoints persist into the following day (ASB), but there are much fewer sea breezes in fall (especially October) than spring.Additionally, the increase in dewpoints and possible resulting fog act to slow radiative cooling overnight after the sea breeze and are likely the cause of the warmer overnight temperatures on ASB nights compared with BSB nights.Dewpoint temperature differences for AGS (Figure 3b) and AQX (Figure 3c) show similar patterns.

| Solar radiation, cloud fraction, and lightning
Differences in solar radiation are apparent between BSB and ASB, especially in the springtime (Figure 4a).More solar radiation is present in BSB afternoons than the ASB afternoons.March in particular has much more solar radiation the entire BSB day than ASB day.Compared with NSB days, the BSB days have more incoming solar radiation in all months.Again, the clear demarcation near 08:00 EDT is an indication of increases heating once the sun rises on these BSB days.ASB days mostly have more solar radiation when compared with NSB days.The only exception is in March where NSB days have more solar radiation.In March, the area surrounding SRS tends to have drier days (less precipitation, Qian et al. 2021)  Cloud fraction differences show that generally (Figure 4b), NSB days have a greater cloud fraction than BSB days or ASB days.The exceptions to this include days in March with the greater solar radiation and in October for NSB days compared with ASB days.These ASB days also have more clouds than the BSB days which reduce the daytime heating, limiting the temperatures as indicated in Figure 2. Thus, in these drier months, the sea breeze tends to bring in additional moisture and greater cloud cover, which is supported by higher dewpoints (Figure 3).Furthermore, BSB days are generally less cloudy than other NSB days, especially earlier in the day, which increases solar heating (Wermter et al., 2022).
Lightning acts as an indicator of deep convection.The most lightning at SRS on BSB days occurred in afternoons in June to July (Figure 5a).There was secondary lightning density peak that also occurred in April.Additionally, there is some lightning that could be from sea breeze convection in the early morning of ASB days in June.Generally, the inland penetrating sea breeze does not seem to form strong convective storms that drive lightning production at SRS (Figure 5a).On ASB days, there is more lightning in the late afternoon in April and June through September with the strongest signal at 18:00 EDT in July and 15:00 EDT in August (Figure 5a).On NSB days, the strongest signals for lightning occur at 20:00-21:00 EDT in July and 18:00 EDT in August.There is also hardly any lightning in April in the afternoons on NSB days.

| DISCUSSION
The sea breeze affects the environment at SRS differently at different times of the year.Temperatures tended to be higher on both BSB and ASB days compared with NSB days, but dewpoint temperatures were much greater on F I G U R E 2 Differences by month and hour between days when the sea breeze initiated (BSB), days after or following the sea breeze (ASB), and nonsea breeze days (NSB) in temperature for the Savannah River Site (SRS) (a); Augusta Regional Airport (AGS) (b); and Allendale airport (AQX) (c).
ASB days in the early spring when drier air is more persistent.NSB days tended to have more clouds than BSB and ASB days, except in March and October when ASB days had more clouds than the normal NSB days.The higher dewpoints on ASB days during these normally dry times of year could be driving higher cloud fractions for these ASB days.When clouds did occur, we noted that cloud bases were higher on both BSB and ASB days compared with NSB days.With the change in temperature and dewpoint temperature associated with the sea breeze, we expected to find variations in calculated lifting condensation level (Lawrence, 2005).However, there did not seem to be much difference and the calculated lifting condensation level tracked poorly with the ceilometer measured cloud base.Nevertheless, fewer clouds, and higher temperatures with greater solar radiation on BSB days generate the mass divergence that creates the inland pressure gradient force that drives the inland penetration of the sea breeze to SRS, as indicated by Wermter et al. (2022).
Lightning patterns were different between BSB, ASB, and NSB days.Here, we use lightning as a proxy for convection and the differences in lightning timing provides insight into convection timing since the formation of lightning requires a strong updraft and mixed-phase clouds.Not much lightning occurred at SRS on BSB days which indicate that the sea breeze arrival at SRS did not drive much strong convection.This is in contrast with Fu et al. (2021Fu et al. ( , 2022)); however, there were some weaker lightning signals in June and July, but the timing was not much different from NSB days.Differences were more apparent in the ASB days.In April and September, lightning occurred mostly on ASB days in the afternoon (15:00-16:00 EDT) where NSB days have limited lightning occurrences in these months.Most markedly, in July, the highest lightning density occurred at 18:00 EDT F I G U R E 3 Differences by month and hour between days when the sea breeze initiated (BSB), days after or following the sea breeze (ASB), and nonsea breeze days (NSB) in dewpoint temperature for the Savannah River Site (SRS) (a); Augusta Regional Airport (AGS) (b); and Allendale airport (AQX) (c).(Figure 5a,b), which was 4 h before the peak lightning densities for NSB days in July at 22:00 EDT (Figure 5a,b).In August, the ASB peak lightning density is much less but does occur about 3 h earlier (15:00 EDT) than the peak for NSB days (18:00 EDT).
The sea breeze travels more than 150 km inland to reach SRS by late in the evening.This timing allows for the overnight stability to "trap" sea breeze conditions in place at various times of the year.This likely leaves a residual of the sea breeze conditions within the boundary layer.Indeed, lower cloud fractions (Figures 4b and 5b) give rise to higher temperatures (Figure 2) on ASB days compared with NSB days.Higher dewpoints in the morning hours (6:00-12:00 EDT) of ASB days compared with NSB days (Figure 3) also indicate this residual layer.However, dewpoints in the afternoon (12:00-18:00 EDT), especially in July, are lower on ASB days than NSB.Increased mixing from convection on ASB days could play a role in this reversal.Of note, cloud fractions in July have a sharp increase during the time of peak lightning densities on ASB days, which could indicate cloud growth.On the other hand, NSB days have a sharp increase in cloud fraction before the peak lightning.This points to a rapid transition from shallow to deep convection on ASB days compared with NSB.Thus, especially in July, the sea breeze residual on ASB days drives the earlier (4 h) onset or transition to deep convection July that drives lightning production.Of note, measured cloud bases in July for BSB, ASB, and NSB at 18:00 EDT (peak ASB lightning) were not notably different.
There is one caveat that should be noted with this data.Occasionally, sea breezes occurred on subsequent days.This occurred most often during the months of May (57%) and July (50%).All other months were <50%.However, none of the BSB and ASB figures look the same, therefore, this did not appear to affect these comparisons.Rather, it is possible that subsequent sea breezes interacted more with residuals left by the sea breeze on the previous day, but this is outside the scope of this current research.

| CONCLUSION
Using a dual-verification method of measurements at the SRS and local radar data, we identified 470 deep inland penetrating sea breezes in the SEUS for the months of March through October from 2015 to 2021.Temperature, dewpoint temperature, clouds, and solar radiation associated with sea breeze days varied by month and differed from those on NSB days.Differences were also apparent between days when the sea breeze was initiated and days following the passage of the sea breeze.This highlights the residual influence of the sea breeze.
Lightning strike densities at SRS were used as proxy for deep convection.Days on which the sea breeze initiated were found to have lower overall lightning densities than the other days.This seems to indicate less convection on sea breeze days.But, in some months during the summer, lightning densities for days following a sea breeze were found to reach a maximum 2-4 h earlier than on NSB days.Thus, the sea breeze residual on these days impacted the earlier onset of deep convection.
This research was conducted to build a catalogue of deep inland sea breeze events and to define novel methods to do so.The dual-verification technique used was able to detect an ample amount of sea breezes (27% of days from March to October on average) that penetrated upwards of 100 km inland.These methods could be adapted towards other locations depending on the availability of measurements.This catalogue of sea breeze events (shown in Table A1) can also be used in future research such as in training for machine learning methods of detecting sea breeze signatures.Note: These dates are when the sea breeze initiated, but the sea breeze may not have arrived at SRS until after 0:00 UTC on the following date.

F
I G U R E 4 Monthly and hourly differences in (a) incoming shortwave radiation and (b) cloud fraction at Savannah River Site (SRS).F I G U R E 5 (a) Monthly and hourly means of National Lightning Detection Network (NLDN) lightning density for the three sea breeze regimes.(b) Hourly averages of lightning density and cloud fraction at Savannah River Site (SRS) during the month of July.
dominated by clearer skies.Additionally, in South Carolina, March may have higher sensible heat flux because leaves at times do not fully emerge until April.
Dates for sea breezes identified and verified to arrive at the Savannah River Site.