new a radar-based climatology of july convective … · 2012. 12. 28. · convective initiation in...

1Current affiliation: NOAA, National Weather Service, Rapid City, South Dakota

EASTERN REGION TECHNICAL ATTACHMENTNO. 2000-04

NOVEMBER, 2000

A RADAR-BASED CLIMATOLOGY OF JULYCONVECTIVE INITIATION IN GEORGIA AND

SURROUNDING AREA

Douglas E. OutlawNOAA, National Weather Service

Greenville-Spartanburg, South Carolina

Michael P. Murphy1

NOAA, National Weather ServicePeachtree City, Georgia

1. INTRODUCTION

General thunderstorm climatology andcharacteristics have been extensivelydocumented (e.g., Byers and Braham 1949,U.S. Environmental Data Service 1968,Changery 1981, Doswell 1985). Diurnalvariations of thunderstorms and precipitationin the continental U.S. have been studied(Wallace 1975). A number of more localizedstudies on diurnal convective development inrelation to prevailing wind regimes have alsobeen accomplished (e.g., Smith 1970,Wakimoto and Atkins 1994, Atkins andWakimoto 1997). This current studyexamines specific geographical and diurnalorigins of convection during July in and nearGeorgia. The goal of this study is to compilea detailed climatology of the time, locationand movement of the initial development ofconvection during July for a 10-year periodbased on radar data.

In the month of July, Georgia and neighboringstates are influenced by the subtropicalanticyclone known as the Bermuda High. Thecommon mid-summer weather regimeincludes warm, moist air that flows northwardaround the western periphery of theanticyclone into the southeastern UnitedStates. Rainfall during the summer monthsalmost always occurs without synoptic scaleinfluences and it is almost exclusivelyconvective (i.e., showers and thunderstorms).Although the convection is typically scattered,notable exceptions such as large mesoscaleconvective systems (MCS) and infrequenttropical cyclones produce widespread rainfall.

Showers and thunderstorms typically form inthe afternoon hours when convectivetemperatures are reached. Certain mesoscalecharacteristics in topography, ground cover,soil moisture, etc., result in differentialheating that will cause some locations to bemore prone to producing convective clouds,

2

showers and thunderstorms earlier in the dayand more consistently than others (Rabin et al.1990). The current study was undertaken tolocate such areas of preferred convection inand near Georgia. The focus is on convectioninitiated by mesoscale influences within adefined area. The area is within a 125 n mi(231 km) radius of two designated locations inGeorgia: Athens and Waycross. Allprecipitation that originated outside that areawas excluded from the study. This eliminatedconvection associated with progressivemesoscale and synoptic features whichoriginated outside the 125 n mi radius.

2. DATA

The data used in this study came fromNational Weather Service network WSR-57radars at Athens and Waycross (Fig. 1). Eventhough WSR-88D radars have beenoperational in Georgia since the mid-1990s,WSR-57 data were chosen to provide auniform 10-year data set. WSR-57 hourlyobservations (taken at 30 minutes after thehour and more often as needed) consisted ofUTC date and time, intensity, movement andlocation of observed precipitation. Thelocation of precipitation was based onManually Digitized Radar (MDR) grid boxesfrom the digital section of coded radarobservations. The MDR boxes (Fig. 2) werebased on the gridpoint spacing of the LFMmodel. The dimension of a box wasapproximately 22 n mi by 22 n mi (41 km by41 km). The WSR-57 employed two modesof operation in deriving the intensity of aprecipitating target: logarithmic and linear.The logarithmic mode was used almostexclusively after it was added to the networkradars in the 1970s. In the logarithmic mode,the Digital Video Integrator Processor (DVIP)converted the power that was returned fromthe target into an easy-to-use unit called theVIP. Table 1 provides the VIPs, theirequivalent dBZ values, and the associatedrainfall rates. VIP levels ranged from the

lightest precipitation (VIP1) to extremelyheavy (VIP6). If a grid box contained onlylight rain, that box must be covered by at least20% of the precipitation before it could beassigned a VIP1. For all higher VIP levels,the highest VIP level in the box was the valueassigned to that MDR box regardless of thepercentage covered by precipitation. The datawere collected from both the Athens andWaycross radars for the month of July in the10-year period from 1985 through 1994. Thefollowing criteria were used to determinewhether or not the data qualified as validconvective initiation;

1. The convective initiation must occurafter at least three hours of rain-free radarcoverage within 125 n mi (231 km) of theradar site. (i.e., after three consecutivePPINEs : Plan Position Indicator No Echoes)

2. The convective precipitation must havepersisted for at least two hours after initiation.(i.e., two consecutive hourly observations)

3. Convection that initiated outside 125 nmi and moved into that range was notincluded. This eliminated the inclusion ofconvection by a frontal system or remnants ofa mesoscale convective system (MCS).

4. Radar observations that were takenimmediately after PPIOMs (radar out formaintenance) were not included.

5. The MDR digital data recorded in thefirst observation after PPINEs was used in thestudy.

During the process of manually taking a radarobservation, convection that originated closeto the radar site may have been occasionallymissed. The standard method of taking theobservation required the operator to leave theradar beam rotating at 0.5 degree of elevationas the outer boundaries of the precipitationwere traced. During relatively normal

3

atmospheric conditions in which the radarbeam did not experience extremes ofsubrefraction or superrefraction, long rangeeffectiveness was maximized at 0.5 degree.The radar beam began at the antenna level(105 ft, 32 m). The height of the outgoingbeam increased with increasing distance fromthe radar site. This occurred with theassumption that the beam refracted inaccordance with a near standard atmosphere.At the standard 0.5 degree of elevation, watertowers, mountains, radio and TV antennas, talltrees, etc., would contribute to echo returnsclose to the radar site. This “ground clutter”pattern that was created on the radar scopemade it difficult for the radar operator to findtrue convection within the first 15 to 20 n mi(28 to 37 km) from the radar antenna location.Therefore, convection close to the radar sitewould likely be missed by the radar operatorunless he or she manually raised the radarbeam elevation above the “ground clutter”pattern.

The maximum elevation angle that a WSR-57radar could achieve in scanning theatmosphere was 45 degrees above horizontal.However, the quality of the data decreased asthe elevation angle approached the maximum.This meant that any precipitating target abovethat level of elevation would not be detected.It was possible for a shower to developdirectly over the radar site and the radaroperator would not know about it until he orshe noticed that it was raining or was told ofthat developing convection by a co-worker.(The area above and immediately around aradar site is known as “the cone of silence”.)A RADID monitor was available (outside ofthe radar room at both WSR-57 sites) so thatthe radar operator could dial into aneighboring radar. Once the dial-upconnection was made, three scans (taking oneminute) were necessary before the radar imagewould slowly, sector by sector, be displayedon the monitor. There were times when onlyone employee was on duty taking radar

observations, surface aviation observations,making recordings for the radio, answering thephone, etc. The location of the radar consolewas in a windowless room and the consolegenerated a constant high humming soundwhich made it nearly impossible to hear rainfalling outside. Rain had to be manuallystarted on the surface aviation computer(MAPSO), and a special observation mayhave been needed according to changes invisibility or ceiling. When widespreadconvection was observed on radar, the processof taking the radar observation, logging it onthe form and typing it for transmission wouldtake up to 25 minutes. At times, all of theabove contributed to “the cone of silence”being worthy of consideration in this studybecause it was not always practical or possibleto dial up a neighboring radar site. As a resultof “the cone of silence” and the process thatdepended on a radar operator to raise theelevation of the antenna in an attempt to locatenearby convection, the MDR boxes containingand immediately adjacent to the radar sitecould be biased toward lower precipitationdetection values. The disproportionately lowconvective initiation values found close to thenetwork radar offices in this study lendcredence to the bias toward lower values.

During the construction of the convectiveinitiation contour maps (Figs. 5 and 6), MDRvalues 20 to 40 n mi (37 to 74 km) from theradar site were used to interpolate expectedvalues in the low biased “ground clutter” area.The “ground clutter” pattern from the Athensradar also included mountains that werelocated from near 50 n mi (93 km) northwest(just east of Jasper, Georgia) to 70 n mi (130km) north (just west of Cashiers, NorthCarolina) of the radar site. Contouring wasalso interpolated through the mountain“ground clutter” area. A small isolatedfeature, Tanasee Bald (5622 ft, 1714 m), in themountain clutter pattern was located at thejunction of Haywood, Jackson andTransylvania counties in North Carolina.

4

Although this feature may have affected initialprecipitation detection to some small degree,it was not considered to be significant enoughto adjust the contour map. Also, range effectswere not considered during the construction ofthe contour maps. The contour maps weredrawn with as much precision as possibleassuming the convective initiation numberwas centrally located in the MDR box, butalso considering surrounding trends. Thelines indicating the number of convectiveinitiations are in increments of four (fourthrough 24) for both the Athens and Waycrosscontour maps. Significant values withinclosed contours were included for betterillustration.

3. TEMPORAL AND AREALCHARACTERISTICS

a. Time of Convective Initiation

Using the criteria mentioned above, there were187 unique convective precipitation eventsfrom the Athens radar and 177 from theWaycross radar for July during the 10-yearperiod from 1985 to 1994. Figs. 3 and 4 showthe frequency of convective precipitation starttimes for the Athens and Waycross radars,respectively. Convective precipitationdetected by the Athens radar began most oftenat 16 UTC with 46 of the 187 (25%)precipitation events beginning at that time.The southern half of Georgia (Waycross radarcoverage area), had its maximum with 29 ofthe 177 (16%) precipitation events at 18 UTC(Fig. 4). Interestingly, 61% of the events inthe north began within the 3-hour period of 15to 17 UTC, while only 35% of the events inthe south began in that time interval.However, by including one hour at each endof that interval, i.e., from 14 to 18 UTC, thepercentage rose to 60% in the south. Bothradars showed a significant drop in thefrequency of convective initiation immediatelyafter the maximum. Of the 177 events in thesouth, the maximum of 29 occurred at 18

UTC and precipitation only began 10 times at19 UTC. In the north, the maximum of 46convective precipitation initiations occurred at16 UTC. Forty-one convection initiationsoccurred at 17 UTC in the north followed by14 occurrences at 18 UTC. Thus, both radarsdisplayed a sharp decline in convectiveinitiation frequency immediately following thetime of maximum occurrence. These findingsindicate that convective initiation in and nearGeorgia during July most often begins in theearly afternoon and decreases significantly bylate afternoon.

Convection over the Atlantic Ocean within 50n mi (93 km) of the northeast Florida coasthad a tendency to develop in the earlymorning. Eleven of the 15 (73%) convectiveinitiations that were detected off the Floridacoast occurred from 07 to 13 UTC. Theremaining four of 15 (27%) occurred at 14 and15 UTC. Showers and thunderstorms over theAtlantic Ocean within 50 n mi of the Georgiacoast had a tendency to develop in the mid-day hours. Seven of the 27 (26%) eventslocated over Georgia’s coastal waters occurredfrom 05 to 10 UTC. Twenty of 27 (74%)occurred from 14 to 19 UTC. There were 10convective initiations in the only MDR boxthat was completely offshore over the Gulf ofMexico. Six occurred from 04 to 10 UTCand four occurred from 14 to 16 UTC.

Following are some key features of thefrequency distributions (Figs. 3 and 4) ofconvective initiations detected by the tworadars:

Athens

• Only two convective initiationsoccurred from 01 through 07 UTC

• 158 (84%) initiations occurred from12 through 20 UTC


• 6 (3%) initiations occurred from 21

5

through 00 UTC

Waycross

• No convective initiations occurredfrom 23 through 03 UTC

• 56 (32%) initiations occurred from 04through 13 UTC


• 76 (43%) initiations occurred from 16through 18 UTC

b. Location of Convective Initiations

This study also evaluated the climatology ofthe location of the first detectable convectiveprecipitation return. The location ofconvective initiation is marked in terms of theMDR grid box spacing of the digital section ofthe radar observation (Fig. 2). By countingthe number of times a MDR grid boxcontained an echo within the first hour of aconvective precipitation event, it is possible toidentify locations that are maxima or minimafor convective precipitation development.

The area containing the highest frequency ofoccurrence in or near north Georgia was in thesouthern Appalachians from near Rome,Georgia to the vicinity of Asheville, NorthCarolina (Fig. 5). The maximum of 25occurrences was located 20 to 25 n mi(approximately 40 km) southwest ofAsheville. Convective initiation decreasedrapidly eastward into the French Broad RiverValley near Asheville. Two MDR grid boxes,each with seven initiations, were noted eastand northeast of Asheville where the Craggyand Black Mountains are located. Thenumber of convective initiations graduallydecreased from the maximum near Ashevillegoing west-southwest down the southernAppalachians toward Rome. Secondarymaxima were located to the southwest andsouth of Atlanta where rainfall initiationoccurred 15 times between Lagrange and

Carrollton and 14 times between Columbusand Thomaston. A less significant maximumof 10 was located at Augusta and wassurrounded by single digit occurrences. Thiselongated area of six to 10 occurrences wassituated along the geographical transition zonefrom the Piedmont to coastal plain. Theminimum of convective initiations extendedacross the Piedmont of South Carolina.During the period of this study there were noinitiations from near Charlotte, North Carolinato Greenwood, South Carolina. This absoluteminimum was separated from the absolutemaximum by only 50 n mi (93 km).

Fig. 6 is a contour map of the number of timeseach Waycross MDR grid box containedprecipitation in the first observation of a validdata set. This shows that precipitation in Julymost often began along the Atlantic and Gulfcoasts and to the west of the OkefenokeeSwamp. The absolute maximum was alongthe Florida Gulf Coast. Twenty-five of the177 precipitation initiation events occurredjust northwest of Cross City, Florida. TheAtlantic Coast maximum consisted of eight to14 convective initiation occurrences from nearSavannah, Georgia to the Daytona Beach,Florida vicinity. Within this maximum, therewere two areas of greater initiationoccurrences. The broadest maximum coveredthe central Georgia coast where 14occurrences were found approximately 10 to15 n mi (19 to 28 km) north-northwest ofBrunswick. A smaller maximum area with 13occurrences was located west of St.Augustine, Florida in central St. JohnsCounty. A maximum with 13 occurrenceswas between the Atlantic and Gulf Coastmaxima. This included the city of Valdosta,Georgia and locations just to the east. An areaof fewer convective initiations extended fromthe southern tip of the Okefenokee Swampextending south along the central Floridapeninsula. There was a distinct minimumlocated northeast of Tallahassee, Florida,mainly in Thomas County, Georgia. Over the

6

remainder of the Waycross radar coveragearea, there were a few minor features ofconvective initiation to note. There were acouple of minor maxima located over theoffshore waters near the Gulf Stream. Therewas a broad inland area featuring relativelyfew convective initiations which coincidedwith the lower terrain under the umbrella ofthe Athens radar.

c. Movement of New Convection

Each radar observation also contained thedirection and speed of movement of theprecipitation. It should be noted that in thefirst radar observation containing echoes, theradar operator was not required to record amovement since there was no previousposition from which a movement could bederived. The cell movement from the secondhour observation was often used in this study.Cell motion on a typical summer day isusually consistent for several hours.

Looking at both the Athens and Waycrossradar data and using an eight-point compass,the predominant movement either had acomponent from the west or was stationary.From the Athens radar (Fig. 7), movementwas from the west (to the east) 33% of thetime (62 of the 188 convective precipitationevents) and was stationary 22% of the time.From the Waycross data (Fig. 8), cells werenearly stationary 27% of the time. In southGeorgia, precipitation movement was mostoften from the southwest, with nearly 52% ofmotion being from either the south, southwest,or west. The movement of the Athensprecipitation was slightly different with nearly56% of the motions from either southwest,west, or northwest. The movement in theremainder of the directions contained singledigit percentages. The general movement ofinitial convection from the Waycross andAthens radars was indicative of the circulationaround the western periphery of the BermudaHigh. Looking at the Waycross and Athens

data together, there is a strong westerlycomponent dominating the movement of theinitial precipitation in Georgia.

4. DISCUSSION

Showers and thunderstorms in the absence ofany frontal lifting generally begin formingaround noon or during the early afternoonhours as convective temperatures are reached.In north Georgia during July, this occurs mostoften at 16 UTC. Across southern Georgia,the time of most frequent occurrence is 18UTC. Convergence of outflow boundarieswill contribute to convective initiation at alltimes of the day. This would account formuch of the variability, particularly outside ofthe times convective temperatures are firstreached. Capping inversions would favor lateday convective development or none at all.Differential heating between land and oceansurfaces would favor convection over the landduring the day and over the water at night.The tendency for convection to developoffshore from Georgia during the mid-dayhours is left to the reader to speculate.

The effectiveness of radar to detect lowtopped convection decreases significantlybeyond 100 n mi from the radar site. Using a0.5 degree elevation scan under near standardatmospheric conditions, the center of the radarbeam is near 12,000 ft (3659 m) at 100 n mi(185 km) and near 17,000 ft (5183 m) at 125n mi (231 km). Convection that is initiallybelow those heights at those ranges would beovershot by the radar beam and would not bedetected. This would result in a low bias inand near the 100 to 125 n mi range aroundeach radar site.

a. Athens Radar Data

A study of cloud-to-ground lightning inGeorgia (Livingston et al. 1996) wasconducted prior to the 1996 SummerOlympics. A flash count vs. local time graph

7

of a cumulative nine year sample (1986-1994)during July and August within 50 km ofAtlanta, indicated cloud-to-ground lightningbeginning to increase during the mid-dayhours. According to their data, flash countswere very low from 09 to 13 UTC and wereslightly higher from 14 to 16 UTC. Flashcounts began to increase at 17 UTC andcontinued to increase until late afternoon. Thetime of the beginning of the flash count risecorrelated very well with the maximum periodof convective initiation from the Athens radardata (Fig. 3).

Convective initiation in and near northGeorgia appears to be strongly influenced byelevation. Days with little synoptic scaleforcing, typical around Georgia in July, werefound to be dominated by mountainconvection (Weisman 1990). Convection ismost likely to begin along the AppalachianMountains and over the higher terrain southand southwest of Atlanta. Mountain ridgeswarm more rapidly than heavily forested orridge shaded valleys during a sunny day. Aswarm air rises from the higher peaks, coolervalley air moves in to replace it. The resultingupdraft can produce convection depending onthe amount of instability available on a givenday. The highest mountains contribute to theformation of unstable lapse rates due to rapidwarming of the summits extending into thecooler air aloft. The differential heatingresults in the highest peaks having convectionsooner and more consistently than mountainswith lower elevation. This explains why themountains southwest of Asheville, whereelevations are around 6000 ft (1829 m), havea relatively high number of convectiveinitiations in July. The values decreaserapidly going east into the lower terrain of theFrench Broad River Valley but increase toseven east and northeast of Asheville in thehigh terrain. Mount Mitchell, the highest peak(6684 ft or 2038 m) east of the MississippiRiver, is located about 20 n mi (37 km)northeast of Asheville just inside the 125 n mi

(231 km) radius around Athens. The GreatSmoky Mountain National Park containsmountains with elevations in excess of 6000 ft(1829 km), but the center of the park is about100 n mi (185 km) from the radar. Rangeresolution is decreased significantly at thatdistance and the mountains of the groundclutter pattern would tend to block initialconvection from being detected because echotops could be low. A valley-to-mountainbreeze develops during the formative stage ofmountain convection. This creates divergencewith sinking motion over the lower terrain andtemporarily suppresses convective initiationover the non-mountainous areas (Gibson andVonder Haar 1990). Initial convection occursto a smaller degree over higher terrain to thesouth and southwest of Atlanta resulting in asecondary maximum. Fig. 9 illustrates thegeneral variation in elevation across thesoutheastern United States.

The less evident maximum of mainly six toten occurrences located from east-centralGeorgia through central South Carolina is inthe region where the Piedmont trough mostcommonly develops. Koch and Ray (1997)conducted a study of central and eastern NorthCarolina summertime convergenceboundaries. Different types of boundary layerconvergence zones were identified anddefined. Sixteen out of 95 identifiedboundaries were originally classified as“unknown (origin) stationary”. Thirteen ofthe 16 boundaries corresponded well with thelocation of the Fall Line which is thetransition between the hard, crystalline rocksof the Piedmont to the soft, sedimentary rocksof the coastal plain. The soil in the coastalplain is primarily sand and sandy loam and thesoil of the Piedmont is primarily clay. Severalstudies (e.g., Ookouchi et al. 1984 andMahfouf et al. 1987) have demonstrated thatcirculations caused by soil and soil moisturediscontinuity can influence convectiveinitiation. Intense solar heating occurs alongthe Fall Line resulting in the formation of the

8

Piedmont trough and thus it is a favoredregion for convection. All of the Piedmonttroughs in the Koch and Ray (1997) studywere found to be autoconvective (producedconvection without the need for interactionwith other features). Nearly all of thesetroughs extended into South Carolina in thevicinity of the Fall Line. This cluster oftrough positions corresponded well with theelongated maximum of convective initiationsin Fig. 5. Weisman (1990) found thePiedmont trough to be present about 40percent of the time between May andSeptember in 1984 and 1985. On days with aPiedmont trough, convergence and ascentwere found to be especially strong in thelowest kilometer of the troposphere during theafternoon. Weisman (1990) also showedcloud-to-ground lightning increasedsignificantly from 16 UTC and continued toincrease through the early afternoon. Thiscorresponded well with the Athens radar’sconvective initiation times (Fig. 3). His studydemonstrated that convective developmentnear the Fall Line during the summer may bemost favored on days with a moderate level ofcyclonic forcing.

The location of the absolute minimum ofconvective initiation was found to be over thenortheast portion of the Piedmont of SouthCarolina. This minimum extended fromGreenwood County northeast towardCharlotte, North Carolina. There are nogeographical features that would promoteconvective initiation because this region hasgentle, rolling hills with a consistent soil type(mostly clay). It is not surprising that aminimum of convective initiation would belocated in that area. What is surprising is thatthere were no initiations in that area duringJuly throughout the 10-year period.

b. Waycross Radar Data

Convective initiation as indicated from theWaycross radar generally had maxima aligned

along the Atlantic and Gulf coasts. Thisresults mainly from the influence of the seabreeze. The terrain in this region is relativelyflat and soil type throughout the coastal plainsis mainly composed of sand or sandy loam.

Changery (1991) shows July thunderstormsare more frequent along Florida’s west coastthan at the same latitude on the east coast. Ananalysis of the mean number of thunderstormsin July indicates 24 to 30 occur from CrossCity to Tallahassee compared with 20 to 22from Jacksonville to near Daytona Beach.Another analysis shows the primary hour ofthunderstorm initiation in July to be around 18UTC for those same coastal locations. Also,inland areas of north Florida generally havethunderstorm initiation from 19 to 20 UTC.The findings in Changery (1981) wereconsistent with the locations and generallyconsistent with the timing of convectiveinitiation in the current study.

Henry et al. (1994) state that the watertemperature of the Gulf of Mexico is generallywarmer than the Atlantic Ocean in July. Seawater temperatures in the Gulf at that time ofyear are well into the 80s

°F, while Atlantictemperatures are in the lower 80s

°F and aremore commonly subjected to upwellingcausing localized cooler temperatures. Windtrajectories over the Gulf will impart moremoisture to the air than trajectories over theAtlantic. A southwest wind tends to createmore showers and thunderstorms over Floridaand adjacent areas in Georgia than other windpatterns. The development of showers andthunderstorms is dependent on the lack of asufficiently strong high pressure system whichwould produce enough subsidence to suppressconvection.

Much of the July convective initiation overnorth Florida and adjacent areas in southGeorgia is dependent on convergence ofmesoscale boundaries, especially sea breezefronts. The development of sea breeze fronts

9

is strongly dependent on diurnal temperaturedifferences between land and sea caused bythe different response of the two surfaces toinsolation. If there is sufficient solar radiationto warm the ground, the land will heat rapidlyduring the day. Widespread cloudy skies willprevent sea breeze front formation. A seabreeze typically begins when the land is 5 to6°F warmer than the adjacent ocean. Seabreezes most often begin at 10 to 11 a.m.Speeds of inland movement are typically 5 to15 mi/h (4 to 13 km/h) (Henry et al. 1994).Normal high temperatures in July across northFlorida and south Georgia are generally in thelower 90s°F and early morning lows are in thelower 70s°F. The diurnal air temperaturevariation above the sea surface is on the orderof one degree (Wallace and Hobbs 1977). Ona typical mostly sunny morning, the landgradually warms creating rising air which isreplaced by cooler air from the sea. The airthat has risen over the land is then redirectedaloft toward the sea to replace the air that hasmoved inland from the sea. Cumulus cloudsform over the land while the sky over theocean surface becomes relatively cloud-freearound noon and generally remains that waythroughout the remainder of the daylighthours. Enhanced vertical motion in thevicinity of the sea breeze front occurs becauseof the convergent wind flow and solenoidalcirculation. Sea breeze fronts usually formaround 16 UTC. Visible satellite imagery aidsin following the formation and evolution of asea breeze front. On Doppler radar, 1.1 km,0.5 degree reflectivity and radial velocityimages are usually very helpful in tracking theprogress of a sea breeze front (Gould et al.1996).

The land cools rapidly near sunset. Around0100 UTC on a typical July evening, thetemperature of the land and ocean are aboutthe same. The land gradually becomes coolerthan the sea after this time. During the earlymorning hours, when this difference is mostsignificant, a land breeze occurs as air rises

over the sea and is replaced by air from overthe cooler land. This circulation is weakerthan a typical sea breeze because thedifference between land and sea temperatureis usually less than 5°F . Turbulent mixingwith a land breeze may be just as vigorousnear the coast as a typical sea breeze, althoughland breezes generally produce lessconvection over the sea surface (Henry et al.1994). Southwest to westerly winds tend tocreate convection over the eastern part ofApalachee Bay as the land breeze convergeswith the prevailing wind. A southerly windhelps to focus convection south of St. Marks,FL due to convergence of the land breezesfrom the southwest and southeast facingshores.

The locations where convective developmentis most favored are determined by the low-level (below 700 mb) wind regime present onany particular day (Smith 1970; Gould et al.1996). Light and variable low-level windstend to favor convective initiation along thesouthwest and southeast facing coasts of theBig Bend area of Florida. Sea breeze frontsgenerally do not extend very far inland duringthe light and variable wind regime. During asoutherly regime, sea breeze fronts are pushedfurther inland from Apalachee Bay with thecombined effects of the sea breeze and thewind pattern of the day. During a westerlywind regime, the sea breeze front along thewest coast of the peninsula is pushed wellinland. Also with this type of regime,convergence is most significant along theAtlantic coast as that sea breeze front movesinland against the westerly wind flow.Northerly and easterly wind regimes aregenerally drier although a flow from the northwould tend to converge with a northwardmoving sea breeze front from the ApalacheeBay area. The latter regimes are alsoconsidered to be somewhat rare in July.

The configuration of the coastline alsocontributes to convective initiation. Places

10

located near the landward tip of bays andinlets tend to favor the development ofconvection before other land areas. This isdue to the combined effects of the sea breezeand bay/inlet breeze which results in astronger flow within the boundary (Gould etal. 1996). The increased flow near the tip ofthe bays/inlets also creates divergencebetween bays/inlets which results in lessconvection. The varying configuration of theGeorgia coastline provides many locations ofconverging and diverging sea breezes whichalso combines its effect with bay, marsh orsound breezes. The portion of the Georgiacoastal area from Brunswick to Savannah islargely composed of numerous small riversand streams flowing into marshland and alsoincludes part of the Altamaha River justbefore it flows into the Atlantic. The marshesare a large expanse of somewhat shallowwater that provides a good moisture source forconvection. This would help to create ahigher frequency of convective initiationalong the Georgia coast.

The amount of thermodynamic instability hasbeen found to play a significant role inwhether convection occurs along a sea breezefront. A study by Fuelberg and Biggar (1994)focused on ‘The preconvective environment ofsummer thunderstorms over the Floridapanhandle.’ A study by Gould et al. (1996)further emphasized the importance ofexamining the local upper air soundings todetermine the likelihood of sea breezeprecipitation. As the latter study states, “Thethermodynamic variability within each flowregime holds the key to successful sea breezeforecasting.”

The Convection and PrecipitationElectrification Experiment (CAPE) wasconducted during July and August 1991 nearCape Canaveral, Florida by a multi-agencygroup (Wilson and Megenhardt 1997). TheCAPE project demonstrated the importance ofmesoscale boundaries, such as sea breeze

fronts, moving in a similar direction and speedwith pre-existing cumulus clouds (a pre-existing updraft). It was determined fromprevious studies (e.g., Moncrieff and Miller1976 and Weisman and Klemp 1982) thatstorm merger, organization, and lifetime weregreatly enhanced when the clouds weremoving at a similar velocity as theconvergence line. This project also exhibiteda case where the portion of a boundary in anunstable environment produced thunderstormswhen the boundary intercepted pre-existingcumulus clouds. The other portion of thisboundary was moving through a more stableatmosphere with no pre-existing cumulus andno convective initiation occurred. Windsduring July over north Florida at the 2 to 4 kmlevel generally have a westerly component.The average prevailing wind direction in Julyat both Tallahassee and Jacksonville is fromthe southwest (Henry et al. 1994). This wouldtend to contribute to greater convectiveinitiation along Florida’s west coast becausethe sea breeze front and the clouds wouldlikely be moving along a similar vector.

A location with a significant maximum alongthe East Coast was located between the SaintJohns River and Saint Augustine, Florida. Asthe St. Johns River flows north from LakeMonroe (at Sanford, Florida) the closestapproach to the Atlantic Ocean beforereaching Jacksonville is near St. Augustine. A20 n mi (37 km) portion of the river near St.Augustine is approximately parallel to theAtlantic Coast. Also at this point, the St.Johns River is located only about 15 n mi (28km) from the beach. The nearly parallelorientation of the river and coastline creates afavorable situation for mesoscale boundaryconvergence. A river breeze that convergesdirectly with a sea breeze produces themaximum forcing for upward motion asopposed to these boundaries meeting at anangle.

The convective initiation minimum located

11

over Thomas County, Georgia could beexplained by divergence of the sea breezearound Apalachee Bay. The shape of the Gulfcoast of northwest Florida, specifically thearea between Apalachicola and Cross City, isconcave. A study conducted by Smith (1970)examined low-level prevailing winds in thisregion. In the light and variable low-levelwind regime [average of 5 knots (10 km/h) orless in the lowest 5000 ft (1524 m)], whichcontained from nearly a third to as much as ahalf of the data in that study, a sea breezedevelops at mid-day around Apalachee Bayand moves a few miles inland to the northwestand northeast. Local divergence results in thearea from the northern point of Apalachee Bayto northeast of Tallahassee extending to nearthe Georgia/Florida border (Fig. 9).Convection from the sea breezes on light andvariable wind days was generally confined towithin 20 n mi (37 km) inland from the coast.In the light and variable prevailing low-levelwind conditions, convective development isminimized in the area of sea breezedivergence throughout the afternoon. Whenthe prevailing flow is from the south, thediverging effects may be projected furtherinland. On occasions with a prevailing low-level wind other than from the south, localdivergence would not be favored and mayexplain why the maximum along the Gulfcoast separates the minimum over ThomasCounty from Apalachee Bay. Although theeffects of this local divergence resulting fromthe sea breezes have been previouslyconsidered to not extend into Georgia, there isclearly some mesoscale process at work tolimit convective initiation. Velocity of theprevailing wind was not considered in thisstudy and was not a factor during theconstruction of the contour map (Fig. 6).

The very light wind at the various levels of thetroposphere was evident on many occasionswhile watching the ascent of radiosondeballoons at Waycross. There were some daysduring the warm season with relatively clear

skies when the balloon would be observed togo straight up over the launch site for much ofthe duration of the flight. This observancewas best in the evenings as the setting sunwould brighten the brownish, red balloonagainst an increasingly dim sky. The balloonswould usually reach a height of over 100,000ft (30,488 m) in about two hours.

Comparing the Smith (1970) study to theGould and Fuelberg (1996) study, areas ofconvective initiation are similar for the givenwind regimes. Early morning convectiontends to cease over the water and then beginover land with the ‘light and variable’ windregime. This contrasts with the ‘southerly’and ‘westerly’ regime where morningconvection over the water propagates onto theland by late morning (Smith 1970).

A relative maximum between Jacksonvilleand Tallahassee connected the east coast seabreeze maximum to the west coast sea breezemaximum. This maximum bridgeencompassed part of the Suwannee RiverBasin. The highest number of convectiveinitiation occurrences in this area was 13which occurred from Valdosta, Georgia to afew miles west of the Okefenokee Swamp.The location of this maximum was betweenthe Wahlacoochee and Suwannee Rivers. TheAlapaha River flows through the center of thismaximum. The Suwannee River flows fromthe west side of the Okefenokee Swamp. Theswamp’s northern point is around 10 n mi (19km) south of Waycross and is approximately50 n mi (93 km) long from north to south and20 to 30 n mi (37 to 56 km) wide from east towest. It has been the experience of the authorsof this study to observe (by way of the WSR-57 radar) a shower or storm to move over theOkefenokee Swamp from the west and buildvertically several thousands of feet in heightas it came under the influence of the swampwaters. On many occasions, the weatherobserver at the Waycross office could look tothe south and witness a vivid lightning display

12

over the swamp. As this convection movedeast of the swamp it would then decrease invertical extent to around its previous level. Itis possible that towering cumulus couldexperience increased development and beginto precipitate as they move into this area ofincreased moisture availability. The influenceof the warm, shallow swamp waters of theOkefenokee combined with the moisturesupply from the rest of the Suwannee RiverBasin may account for the convectiveinitiation maximum in this area.

Range resolution should also be consideredwhen evaluating the Waycross radar data. Adeveloping shower with a top of 10,000 ft(3049 m) would be detected over Brunswick,GA but not detected over Ocala, FL.Brunswick is 45 n mi (83 km) from theWaycross radar site and Ocala is 122 n mi(226 km) from that point. Using a radar beamelevation of 0.5 degrees, the mid-beam heightabove ground would be 3623 ft (1105 m) overBrunswick and 16,227 ft (4947 m) over Ocala.Flagler Beach, on Florida’s east coast, is at arange of 124 n mi (230 km) from Waycross.

The sea breeze fronts from Florida’s east andwest coasts often converge after 19 UTC.This usually produces showers andthunderstorms. However, for the purpose ofthis study, showers and thunderstormsnormally would have already occurred prior tothat time period within the range of theWaycross radar. Therefore, according to thisstudy’s definition of valid convectiveinitiation, late afternoon convection wouldusually be excluded.

c. Overlapping Radar Coverage

In comparing the overlap region between theAthens and Waycross radars, there were acouple of common features of interest.According to the Waycross radar, there was aslight increase of convective initiation locatednear Augusta, Georgia. This minor maximum

was coincident with the Piedmont troughmaximum that was clearly evident from theAthens radar data. Augusta is approximately75 n mi (86 km) from the Athens radar siteand approximately 125 n mi (144 km) fromWaycross. Comparing this location betweenthe two radars illustrates the limitation onradar due to range effects beyond 100 n mi(185 km). Also, the minimum observed in theWaycross data over central Georgia wascoincident with the minimum observed fromthe Athens radar. These locations do not haveany known geographical features that wouldpromote convective initiation.

Since this study only dealt with MDR digitaldata in which each box has an area roughly484 sq n mi (896 sq km), some precision waslacking in determining exact locations ofinitiation. Therefore, the exact mountainswhere convective initiation occurred could notbe determined within the Athens radar servicearea. Likewise, the exact location within theWaycross radar service area whereconvergence from sea breeze or river breezeinteraction took place could not bedetermined.

The movement of precipitation was generallystationary or slow. The speed rarely exceeded20 kt (37 km/h). The average speed derivedfrom both radars was around 9 kt (16 km/h).The high percentage of stationary convectionand the low average speed is indicative of thelight wind field through the lower atmosphereduring the month of July.

5. SUMMARY

This study produced a detailed climatology ofthe time, location and movement of the initialdevelopment of convection during July for a10-year period based on conventional radardata. The data were obtained from WSR-57radar observations taken at Athens andWaycross, Georgia from 1985 to 1994. Aneffort was made to eliminate the inclusion of

13

convection that originated outside the 125 nmi radius of each radar. Only convectiveinitiations during the first hour after at leastthree consecutive hours of rainfree radarcoverage were considered. Contour mapswere drawn indicating the number ofconvective initiations for locations within a125 n mi radius around each radar site.

The times of the initial convection in Julyshow a maximum occurrence at 16 UTC in theAthens coverage area and at 18 UTC in theWaycross coverage area. The number ofinitiations decreases rapidly after those hours.Convective initiation in the absence ofsynoptic scale influences typically occurs as aresult of convective temperatures beingreached or convergence of mesoscaleboundaries such as a sea breeze, river breeze,or outflow from previous thunderstorms.

According to the Athens radar data, thegreatest number of convective initiationsoccurred over the higher terrain of thesouthern Appalachian Mountains. The highestmountains are more efficient at producingconvection because they induce upslope flowand their summits extend into the cooler airaloft creating lapse rates that are steeper thanthose found over adjacent lower terrain. Lesssignificant maxima were found over a veryhilly area south and southwest of Atlanta. Anelongated maximum was in the vicinity of theFall Line. Koch and Ray (1997) found theFall Line to be a favorable location forautoconvective development. No convectiveinitiations occurred during the 10-year periodin the Piedmont from Greenwood, SouthCarolina to near Charlotte, North Carolina.

According to the Waycross radar data, thegreatest number of convective initiationsoccurred along the Gulf Coast of Florida nearCross City. A long, less significant maximumextended along the Atlantic Coast fromsouthern South Carolina to near DaytonaBeach, Florida. The maxima along the coasts

were the result of convective initiation alongsea breeze fronts.

Data from both radars showed the movementof the initial convection was nearly stationaryabout one fourth of the time. When there wasmovement, it was most often from thesouthwest in southern Georgia and from thewest in northern Georgia. The predominantmovement of initial convection was indicativeof flow around the western periphery of theBermuda High.

This study utilized conventional radar datafrom the last decade of the era whenobservations were taken manually. Dataavailable from the WSR-57 radars consists ofthe hand written observations and poorresolution reflectivity photographs. Thephotographs were taken every 5 minutes innon-severe situations and every other 20-second scan during severe weather. Theability of the WSR-88D to archive largeamounts of high resolution reflectivity andvelocity data makes the newer radar muchmore useful for scientific research. Perhapsthis project will encourage future studies ofconvective initiation using WSR-88Dtechnology.

ACKNOWLEDGMENTS

A special thanks to Lans Rothfusz forcomments and guidance in the initial stages ofthis study. Thanks to Gary Beeley, SOO atPeachtree City (FFC), for comments and aninitial local review. Thanks to Dan Smith,Southern Region SSD Chief, for the initialregional review and helpful comments.Thanks to Ben Moyer at Greenville-Spartanburg (GSP) for creating the maps forFigures 1, 5 and 6 using ArcView GISVersion 3.1. Thanks to Larry Lee, SOO atGSP, for his input and final local reviews.Thanks to Julie Gaddy, Eastern Region SSD,for the final review process. Thanks to thecomments and suggestions from the staffs at

14

Peachtree City and Greenville-Spartanburg.

REFERENCES

Atkins, N.T., and R.M. Wakimoto, 1997:Influence of the synoptic-scale flow on seabreezes observed during CaPE, Mon. Wea.Rev., 125, 2112-2130.

Byers, H. R., and R. R. Braham, Jr., 1949:The Thunderstorm. Supt. of Documents,U.S. Government Printing Office,Washington, D.C., 287 pp.

Changery, M.J., 1981: National ThunderstormFrequencies for the Contiguous UnitedStates. National Climatic Center, NationalOceanic and Atmospheric Administration,Asheville, NC, 80 pp.

Doswell, C. A., 1985: The operationalmeteorology of convective weather vol.II: storm scale analysis. NOAA TechnicalMemorandum ERL ESG-15, NationalOceanic and Atmospheric Administration,U.S. Department of Commerce, 5-29.

Fuelberg, H. E., and D. G. Biggar, 1994: Thepreconvective environment of summerthunderstorms over the Florida panhandle.Weather and Forecasting, 9, 316-326.

Gibson, H. M., T. H. Vonder Haar, 1990:Cloud and convection frequencies over theSoutheast United States as related tosmall-scale geographic features. Mon.Wea. Rev., 118, 2215-2227.

Gould, K. J., and H. E. Fuelberg, 1996: Theuse of GOES-8 imagery and RAMSDIS todevelop a sea breeze climatology over theFlorida Panhandle. 8th Conf. on SatelliteMeteorology and Oceanography, Atlanta,GA, 100-104.

Gould, K. J., C. G. Herbster, J. D. Korotky,

and P. H. Ruscher, 1996: WSR-88D,GOES-8, and MM5 mesoscale modelobservations of the Florida panhandle seabreeze circulation under different synopticflow regimes. Conf. on Coastal Oceanicand Atmospheric Prediction, Atlanta, GA,364-369.

Henry, J. A., K. M. Portier, and J. Coyne,1994: The Climate and Weather ofFlorida. Pineapple Press, Inc., 76-79 and103-110.

Koch, S. and C. Ray, 1997: Mesoanalysis ofsummertime convergence zones in centraland eastern North Carolina. Weather andForecasting, 12, 66-67.

Livingston, E. S., J. W. Nielsen-Gammon, andR. E. Orville, 1996: A climatology,synoptic assessment, and thermodynamicevaluation for cloud-to-ground lightningIn Georgia: a study for the 1996 SummerOlympics. Bull. Amer. Meteor. Soc., 77,1483-1495.

Mahfouf, J., E. Richard, and P. Mascart,1987: The influence of soil and vegetationon the development of mesoscalecirculations. J. Climate Appl. Meteor.,26, 1483-1495.

Moncrieff, M. W., and M. J. Miller, 1976:The dynamics and simulation of tropicalcumulonimbus and squall lines. Quart. J.Roy. Meteor. Soc., 102, 373-394.

Moore, R. L., A. D. Cummings, and D. L.Smith, 1974: The National WeatherService manually digitized radar programand some applications. NOAA TechnicalMemorandum NWS SR-75, NationalOceanic and Atmospheric Administration,U.S. Department of Commerce, 21 pp.

Ookouchi, Y., M. Segal, R. C. Kessler, and R.A. Pielke, 1984: Evaluation of soil

15

moisture effects on the generation andmodification of mesoscale circulations.Mon. Wea. Rev., 112, 2281-2292.

Rabin, R. M., S. Stadler, P. J. Wetzel, D. J.Stensrud, and M. Gregory, 1990:Observed effects of landscape variabilityon convective clouds. Bull. Amer. Meteor.Soc., 71, 272-280.

Smith, D. L., 1970: The application ofdigitized radar data to the prediction ofsummertime convective activity in coastalregions. Preprints 14th Weather RadarConf., Tucson, AZ, Amer. Meteor. Soc.,347-352.

U.S. Environmental Data Service, 1968:Climatic Atlas of the United States,Washington, D.C., 80 pp.

Wakimoto, R.M., and N.T. Atkins, 1994:Observations of the sea-breeze frontduring CaPE. Part I: Single-Doppler,satellite, and cloud photogrammetryanalysis. Mon. Wea. Rev., 122, 1092-1114.

Wallace, J. M., 1975: Diurnal variations inprecipitation and thunderstorm frequencyover the conterminous United States,Mon. Wea. Rev., 103, 406-419.

Wallace, J. M., and P. V. Hobbs, 1977: Atmospheric Science An IntroductorySurvey. Academic Press Inc., 345.

Weisman, M. L., and J. B. Klemp, 1982: Thedependence of numerically simulatedconvective storms on vertical wind shearand buoyancy. Mon. Wea. Rev., 110,504-520.

Weisman, R. A., 1990: An operational studyof warm season southern Appalachian leetroughs. part II: thunderstorm genesiszones. Mon. Wea. Rev., 118, 2020-2024.

Wilson, J., and D. Megenhardt, 1997:

Thunderstorm initiation, organization, andlifetime associated with Florida boundarylayer convergence lines. Mon. Wea. Rev.,125, 1507-1525.

16

Table 1. VIPs, their equivalent dBZ values and the associated convective rainfall rates.

VIP 1 = 18-30 dBZ (convective rainfall rates of less than 0.2 in/h (0.5 cm/h))

VIP 2 = 30-41 dBZ (convective rainfall rates of 0.2 to less than 1.1 in/h (2.8 cm/h))




VIP 6 = any dBZ greater than or equal to 57 (convective rainfall rates of 7.1 in/h or greater)