Journal of Coastal Research 413-420 West Palm Beach, Summer 2002

Aquifer Salinization from Storm Overwash William P. Anderson, Jr.

Departm ent of Geology Radford University Campus Box 6939 Radford VA 24141, U.S.A. wpanderso@ra dford.edu ABSTRACT _

ANDERSON, W.P., JR. 2002 . Aquifer salini zat ion from storm overwash. Journal ofCoastal Research, 18(3 ),413-420. .tllllllll:. West Palm Beach (Florida), ISSN 0749-0208. ~ Overwash processes are not only an agent of change to the morphology of barrier islands; they are also a source of instantaneous saltwater intrusion to coastal aquifers., , USA is particularly susceptible euus~~-" to overwash processes because of its geography and the frequency with which tropical and extra-tropical storms strike + S-- the area. Hurricane Emily inundated the island in 1993 with saline water from . The floodwaters recharged the Buxton Woods Aquifer, raising salinity levels from approximately 40 mg/L prior to flooding to nearly 280 mg/L within several weeks of flooding. By 1997, chloride levels still had not returned to pre-storm levels. One-dimensional ana lytical solutions of the advection-dispersion equation are used to simulate chloride transport within the aquifer utilizing a pulse SOUrce implemented with linear superposition. These simulations are matched with chloride breakt hrough curves. Initial simulations show that a pulse duration of five days provides the best fit to the data. Simulation of chloride breakt hrough at two str eamline locat ions demonstrates that higher gradients advect chloride further into the aquifer, causing higher chloride concentrations and increasing the duration ofcontamination. The Cape Hatteras region historically is susceptible to several hurricanes in a single season. In order to analyze the effect of multiple overwash events on water quality, predictive simulations show the effect of two overwash events separated by severa l time lags between storms. Simulations indicate that higher gradients and short time lags be­ tween overwash events result in chloride MCL violations that persist for more than four months.

ADDITIONAL INDEX WORDS: Hurricanes, Hurricane Em ily, North Carolina, Hatteras Island, saltwater intrusion, , ooeruntsh, barrier island, groundwater modeling, analyti cal solution. coastal aqu ifers.

INTRODUCTION lanti c Ocean from Pamlico Sound, a 3100 krn" estuary (Figure l ). Th e Cape Hatteras portion of the island lies where the Water qu ality in coas tal aq ui fers can be threaten ed not island shifts from a southe rly trend to a west-southwesterly only by upconing and saltwate r intrusion, but also by over­ trend. It is also at this point that the island widens from less wash and sto rm surge durin g severe storms. Th ese processes than 1 km in wid th to over 3 km in width. Th e "reversed-L" caus e floodin g that recharges water-table aquifers with sa ­ shape of the island acts as a dam during winter extratropical line wa ter. For exa mple , in August and September, 1993, storms C'Nor'easte rs" ) and hurricanes. Northerly win ds pro­ sound-side flooding during Hurricane Emily in unda te d Hat ­ duced by t hese storms cross as much as 70 km of open water teras Island, North Ca rolina (USA) with sa lt water from in the sound, which, coup led wit h sto rm surge formed by Pamlico Sound, a large estuary on the north side of t he isla nd these intense low-pres sure systems, pr odu ces a rise in Pam­ (Figu re L). Flooding was most severe in the vicinity of the lico Sound of as much as three m (Figure 2). The sa linity of isla nd's wellfield that exploits a shallow aquifer for drinking Pamlico Sound is approximate ly 60% that of seawate r. wa ter. A large qu antity of the floodwater eve ntually re­ charge d the shallow Buxton Woods Aquifer (BWA), raising The Buxton Woods Aquifer salinity and in creasing treatment costs. Chloride levels in the BWA inc reased from ap proximately 40 mglL prior to floodin g The Buxton Woods Aqu ifer (BWA) has histori call y been the to nearly 280 rng/L within sev eral week s of flooding.Chloride sole source of fresh water to residents of southe rn Hatteras levels did not fall below 100 mglL until March, 1994, and still Island. Although water pumped from the fresh water aquifer had not reached pre-floodin g concentrations by J anuary, is now suppleme nted by several moderately-sali ne deep wells 1997 . (- 70 m in depth), which hav e chlori de levels of approximate ­ ly 11,000 mglL and require reverse-osm osis treatmen t, the The Cape Hatteras Region shallow gro undwa ter of the BWA continues to satisfy a sub­ stanti al portio n of the isla nd's freshwater dem and. HEATH Hatteras Island, North Ca rolina, is the largest island of the (1988, 1990 ), ANDERSON (1999), and ANDERSON et al.(2000) , a barrier-island chain th at se pa rates the At- comb in e to provid e a detail ed hydrogeologic framework of the BWA. In general, the aquifer consists of tw o permeable unit s, 01062 received 24 October 2001; accepted in revision 2 January 2002. eac h approxi mately 10 m in thickness, that are se parate d by 414 Anderson

! N

Pam lico Sound North Carolina Study Area o c e a n CBpePoin t

Figure 3. Approximate exte nt of sound-side flooding in the Cap e Hat­ Figure 1. Location of study site on the Outer Banks barrier island chain ter as region following Hurricane Emi ly. in North Carolina, U.S.A.

contribute to this phenomenon. WILDER et al. (1978) con­ a semi-confining layer of variable thickness. A silty to clayey ducted a study of th e water resources of northeastern North Carolina, including th e Ca pe Hatteras region . WINNER(1978) sand underlying th e lower aquifer unit acts as a semi-confin­ ing layer. BURKETT (1996) performed a 72-hour pumping test pr esents a study of the Ca pe Lookout, North Carolina region which suggests that overwash processes are a likely occur­ at the center of the island (note d with a * in Figure 3) and rence on th e low-lying barrier isla nds and th at weeks to determined the bulk aquifer properties shown in Tabl e 1. These parameters, plus average hydraulic gradients for two months would be required to sufficiently flush the saline wa ­ ter from the freshwater lens. MEW (1999) demonstrated that transect s of th e island that are derived from ANDERSON (1999 ) and MORGAN et al. (1994), are used in subsequent an­ chloride levels at shallow depths within the BWA on the north side of th e isla nd approach the chloride maximum con­ alytical simulations . tam inan t level (MCL), which is th e legal limit for chloride in Th e BWA and oth er coastal aquifers of North Carolina have been th e focus of several papers. KIMREY (1960) pre­ drinking water. All of these studies sugge st that overwash processes play a significant role in controlling th e ground­ sents an early study of the BWA which shows elevated chlo­ water quality within the surficia l aquifers of North Carolina 's ride levels at shallow depths within the aquifer. HARRIS a nd barrier islands. WILDER (1964) and HARRIS(1967) obtained simila r results for the BWA and also suggest that aquifer heterogeneities Coastal Aquife rs Studies of coastal aquifers have mostly been confined to th e development of the freshw ater-saltwater inte rface. Many studies focus on methods of eva luating the freshwater- salt­ , wate r interface in various unconfined settings (VACHER, N 1988; CHENG et al., 1998 ; TAYLOR and PERSON, 1998; BAK­ KER, 1999), whil e othe rs focus on specific coastal settings such as mainland aquifers, carbonate banks, volcanic islands, North Carolina atolls, and isla nds of glacia l origin (WALLIS et al., 1991; UN­ DERWOOD et al., 1992 ; VACHER and WALLIS, 1992; GRIGGS and PETERSON, 1993 ; ROBINSON and GALLAGHER 1999' PERSON et al., 2000 ). Barrier islands have received l e ~ s fo cu ~ th an th e oth er island types, but several reports present case studies of specific barrier islands (COLLINS and EASLEY 1999; ANDERSON et al., 2000; CORBETT et al., 2000 ). Som~ recent work has focused on the effect th at tid al processes

Tabl e 1. Aquifer parameters of the Buxton Woods Aquifer.

Bulk Aquifer Parameter Units Value Storm track adapted from NOAA data Hori zontal hydr aulic conductivity, K; mJd 21.2 Figure 2. Stor m tra ck of Hurricane Emi ly off of the coast of Nor th Ca r­ Vertical hydraulic conductivity, K" mJd 5.6 olin a in Augu st/September 1993. Aquifer th ickness, b m 24.5

Jo urnal of Coastal Resear ch, Vol. 18, No. 3, 2002 Overw ash in Coasta l Aquifer s 415

saltmarsh communities. GUILLEN et al. (1994) present a case 300 study of an over wash event in the Mediterranean coas t of

J ournal of Coastal Research, Vol. 18, No.3, 2002 416 And erson

Ta ble 2. Simulation parameters-Transects 1 and 2. duration of high chloride concentration will be minimized be­ cau se the high velocities and dispersion values enable a Bulk Aquifer quicker dilution of solute concentrations. Thus, th e simula­ Parameter Units Value tions t hat are discussed below are a minimum estima te of Source CI- Concen tration, Co mg/L 12000 chloride br eakthrough. See the subsequent sensitivity anal­ Avera ge Linear Velocity, v, mid yses for the effects of variable groundwate r velocity and hy­ Transect 1 0.091 Tran sect 2 0.170 drodynamic dispersion on model output. Hyd rodynami c Dispersion, D, m'/d Tr an sect 1 4.55 Transect I Simulatio ns Transect 2 8.85 Figure 5 shows the results of several a nalytical simulations that use different pul se-source durations. Th e data comprise ride transport along Transect 2 to determine the degree of the chloride br eakthrough curve at the center of mass of th e imp act at a proposed wellfield site. Table 2 indicates the sim­ plume for times following the overwash event. The center of ulation parameters (Co, v'" and D) that are used in the cal­ mass of th e plum e wa s used in calibrating the pulse-source ibrated simulati ons of transport along the two transect s. duration because of the uncertainty in th e calibration data. It is important to note that these simulations utilize pa­ Spec ifically, the data are bul k chloride values obtained from rameters that are constant t hroughout the simulations , all of th e pumping wells that were operating in the wellfield . which will be a source of error at late times. Another source It is likely that some of the wells detect th e br eakthrough of of error is the vari ab le density of the groundwater, which is th e center of mass shortly after the beginning of the overwash disregarded in these simula tions . Although the source con­ event; therefore, rather than picking an arbit rary location cen tration does not vary through the length of th e pul se along the streamline for th e calibration, it was decided that source, th e average lin ear velocity along th e streamline and the pul se source calibration should be with the center of mass the re sulting hydrodynamic dispersion would certainly vary locations. Subsequent simulations use the calibrated pul se­ consid erably over time, especially as water levels, and thus source duration for ca libration of the average wellfield loca­ gradients, decrease with th e passing of th e storm. The de­ tion along th e streamline. The left pa nel of Figure 5 shows crease in the water-table gr adient and the average lin ear ve­ simulation results for pulse-source durations of one to six locity is not as significant as it would be at other times of the days. As is indicated in th e detail ed plot shown in the right year, however, becau se the hurricane occurred at th e begin­ panel of th e figure, the five-day pulse source has a pr actically nin g of th e high recharge seas on. Seven years of water-level identical fit based on a linear least-square s regre ssion (r" = data from the Cape Hatteras region show th at water levels 0.819 ) as does the six-day pulse source (r" = 0.818 ). Although begin to rise in the autumn as recharge rates rise in response both pu lse lengths are equa lly reasonable, five-day pulses are to reduced evapotranspiration rates (ANDERS ON , 1999). utilized throughout th e rest of the study. Thus, the water-table gradi ent would be expect ed to be hig h Figure 6 shows breakthrough curves produced by simula­ throughout the duration of the simulations . tions at distances of 25 m, 30 m, and 35 m along the stream­ Because high velociti es have been used in th ese simula­ lin e. These locations alon g the streamline correspond roughly tions, two contradictory aspects of th e simulations must be with the position of th e screen s of the Frisco Wellfield wells. noted. First, mor e ma ss of th e solute per un it time is adv ect ed The best fit to the calibration data occurs at a position of x into the aquifer du e to high groundwater velocities, thus rai s­ = 30 m along th e streamline, where ,.2 = 0.810 . Deviation s ing chloride levels considerably afte r short times. Second , the from the breakthrough curves at late times suggest that th e

500 400

:::1 400 Cl .§. 300 ~ 300 f! C fl 200 c o 200 o, (3 100

50 100 150 200 60 70 80 90 100 110 120 130 140 151. Time since Hurricane Emily (days) Time since Hurricane Emily (days)

Figure 5. Calibration of chloride breakthrough simulations at th e center of ma ss of th e plume at Tr an sect 1.

J ournal of Coasta l Research, Vol. 18, No. 3, 2002 Overw ash in Coastal Aquife rs 417

Table 3. Sensitivity analysis parameters- Tran sect 1. 500 .------, Parameter Units Range of Values Source c i - Conce ntra tion, C mg/L 4000-20000 :::;' 400 Q Average Linear Velocity, V mi d 0.009-0.9 "al y 2 §. x=25m Hydrodynamic Dispersion, D, m /d 0.5- 45.5 l5 300 ~ o Discu ssion ~ 200 8 The simulate d break th rough curves suggest that future Gloo water qu ality scenarios along Transect 2 are problematic. Al­ though the area ap pea rs to be more protected geographically th an the highly exposed area near Tr an sect 1 and the exist­ O ~~~~~----~~---~-~~- ing wellfield (Figure 3), favorable aquifer conditions near a 50 100 150 200 Transect 1 ii.e., lower hydraulic gradient) advect less chloride Time since Hurrtcane Emily [days] into the aquifer. This suggests that less time will be spent in Figure 6. Sim ulated chloride breakthrough at thre e streamline locations violatio n of chloride MCLs with future storms as a resu lt of at Transect 1. pumping from the existing Frisco Wellfield. The following section discusses the sensitivity of th e simulated break­ through curve s to cha nges in model param et ers in an effort dilu tion th at would be expected du rin g higher winter re­ to dem onstrat e th eir influence on model output . cha rge rates is not ta ken into account by the model. SENS ITIVITY ANALYSES Transect 2 Simulati ons Sensiti vity analyses of th e paramet ers involved in simu­ Chloride breakthrough curves were also simulate d for a lating chloride breakthrough were conducted to determine strea mline parallel to Tran sect 2. The simu lations indicate th e level of influence th at each of these paramet ers has on th at the best fit to the break through curve occurs at a dis­ model output . Three paramete rs were va ried: average linear tan ce of 45 m along the st reamline; thus, th e steeper gradient velocity within the aquifer (u,) , hydrodynami c dispersion (D x), at Tran sect 2 advects chloride approxima te ly 15 m further and the sour ce concentration (Co), Both u, and I), were varied into the aquifer tha n at Tran sect 1. This is demonstrated in over two order s of ma gnitude. A range of two orders of mag­ Figure 7, which compares simulation result s for Tran sects 1 nitude was impractical for Co because of th e smaller ran ge an d 2. At a dist an ce of 30 m from th e chlori de source, chloride th at is possible. Sensitivity param eters are shown in Table 3. concentrations along Tr an sect 2 rise to nearly twi ce the level Figu re 8 shows the resu lts of the sensitivity ana lyses at a that they do in Transect 1. In addition, the peak concentra ­ time of 50 days after the overwa sh event . The chloride con­ tion at 30 m arrives nearly ten days earlier along Tr an sect 2 centrations that are plotted in the figur e represent tho se tha n it does along Tr an sect 1. Conditions at a position of 45 found at the center of mass of the chlorid e plum e. As is ex­ m along th e streamline at Transect 2 are similar to Tr an sect pected, the sensiti vity of u, and I), are inversely propo rti onal. 1 at a position of 30 m. This is reflective of the Peclet nu mber, Pe = u;X/Dx' At hig h

1200 .------,--r-,-,------.,----,-- -,---,....,...,...., 500 r------,,' I I , I T2-x=30m ::i' , ,I ll000 I---r-~·~~--I :::;'400 i ;g I l II) 800 l5 300 II --II ­. I l5 600 !c tl 200 ! cs ~ 400 L... c U100 o U (3 o O '----~~~-'--'-'-'--'-'--'---~--'------'------'----'--'--'---u o 50 100 150 200 0.10 1.00 10.00 Time since Hurrtcane Emily [days] Parameter Ratio to Base Case

Figure 7. A comparison of simulate d chloride breakthrough curves at Figure 8. Parameter sens itivity ana lysis results for u, (squa res ), Dy (tr i­ several locations along Transects 1 an d 2. angles), and Co (diamonds).

J ourn al of Coastal Research, Vol. 18, No. 3, 2002 418 Ander son

500 ~------, 7dLag

:r 4oo _ 400 a. '§, S- S- g 300 a 300 ! ! CfMCL 1: 1: fl 200 8 200 c c o o o o - ' - .:. ~ .~ ~ '::T1: : ~ :.::.: ::.:.-':.' o 100 o 100 Single Storm .

O '-~~~~~----~---~-~~--_---J O'--~-~~~----~-~---~~-~~~ o 50 100 150 200 o 50 100 150 200 Time since Initial Hurricane (days] TIme since Initial Hurricane (days]

Figur e 9. Dua l hurricane scena rio simulation results for several lag Figure 10. A comparison of dual hurricane sim ulations for Transect s 1 times -Transect 1. and 2.

Peclet numbers, advective processes dominate and model out­ with a n ana lytical solution for solute transpor t with two put is sensitive to u, but inse nsitive to D'; At low Peclet num­ pul se sources that are simulate d usin g linear superposition. bers, the opposite is true: disper sive processes domin ate and The an alyti cal solution for two pulse sources of sound water model output is sensitive to D, but insen sitive to u,. Tim e is of duration t / and t2, with hurricane occurrence tim es of tw also an importan t cons tra int in these simula tions . The sen­ and t1l2 , is sitivity of each of th ese param et er s decreases significantly as time increases. c U,tlll] c Ux(tll/ - C (x, t) = -Colenc[x - - eric[x-:.-:r.=;~=- ===t=-l)] Because of the high degree of sens itivity of th e model out­ 2 Y 4D,tHl Y 4D) tm - t, ) put to average groundwate r velocities, it is important that adequate inform ation is used to estima te this param et er. X - Uxtll2] c [x- uX (t1l2 - t2 )]} (2) + erfc [ - en c -:-:;=:::='=:;===:0 Also, becau se of th e high degre e of sensitivity to average Y 4D,t1l2 Y 4Dx(t/l2 - t 2 ) groun dwa ter velocities , the precision of the disper sion esti­ In these simulations, the calibrated effects of Hurrican e Em­ mate is less critical in highl y advective systems . Conversely, ily a re coupled with a second hurricane of equal duration, t , highly disper sive sys te ms will require less precision in av­ 2 lagged a tim e tm afte r the first hurrican e. erage groun dwater velocity estima tes . Model output is also directly prop ortional to th e source concentration, Co' As Co Multiple Hurricane Simulatio ns increases, th e chloride concentration at the cente r of mass of th e plu me also increases; as Codecreases, the chloride con­ Figure 9 shows breakthrough curv es at a position of 30 m centra tion at the center of mass of the plum e also decreases. along a streamline parall el to Tran sect 1 for several du al hur­ ricane scena rios. In these simulations, the time of th e second PREDICT IVE MODE LING-MULTIPLE hurrican e was lagged by values ran ging from 7 to 35 days. HURRICANES AND WATER QUALITY The figu re shows th e resu lts of each lagged hurricane simu­ lation in addition to the calibrated single storm simulation Ther e have been seve ra l in stances in the past decade when tha t was shown in Figure 6. A one-week lag time betw een more than one hurrican e has hit the coast of North Carolina storms shows the most severe effect, with a violation of th e in a single season. Recent exa mples include 1996 with Hur­ chloride MCL for more th an 100 days. The effects diminish rican es Bertha (J uly 12) and Fran (September 5), and 1999 with increas ing lag tim es as chloride levels within th e BWA with Hu rri can es Dennis (August 30) and Floyd (September have a chance to dilute before rece iving th e second pul se. A 16) (Y OU N G et al., 2000). Th e ti me intervals between th ese lag time of 35 days between storms reduces th e duration of histo ric storms are 55 days in 1996 and 17 days in 1999. the violation by at least three weeks. Thus, it is a likely scena rio that two storms would follow a Figure 10 compares multiple hurricane breakthrough track like th at of Hurrican e Emil y in a single season, pro­ curves for Transects 1 and 2. The breakthrough curves rep­ ducing simila r conditions and threats to water quality within resent th e arrival of chloride along streamlines parallel to th e th e BWA but raisin g the threat to water quality. transects at distances of 30 m and 45 m for Tr an sects 1 and 2, respect ively. These locations wer e chosen for compa rison Pul se So urce Analytica l So lutio n with becau se they represent th e location of potential groundwater Second Hurricane withdrawals along the streamlines . The results in Tr an sect The effects on water quality of a second hurricane tha t oc­ 2 a re similar to Tr ansect 1, with two exceptions: the peak curs shortly after a n initial hurrican e can also be calculate d chloride levels within Tr an sect 2 are higher than tho se in

J ourn al of Coastal Research, Vol. 18, No.3, 2002 Overw ash in Coastal Aquifers 4HJ

Tra nsect 1, a nd t he ch loride MCL is viola te d for a pproxi­ Tida l effects on sea water intru sion in unconfined aquifers . -lour­ mately t wo more week s along T ransect 2 due to h igh er hy­ nal or Hydrology 216. 17- 31. BA" "I·;li. M., 1999. The size of th e freshwater zone below an elon­ dra u lic gradie nts. Along Tran se ct 2, a lag between hurricanes gated island with infiltration. IVater Resources Research :161 I I. of se ve n day s produces ch loride MCL viola tions for more than 109- 117. four months, while a lag of :35 days prod uces chloride MC L BUH"E'I"r. C.A.. 1996. Est imating the hydrogeologic characterist.ics violation s for nearly f(JUI' months a t lower pea k va lue s. of the Buxton Woods Aquifer . Hatt eras Islan d. North Caro lina. Raleigh. Nort h Carolina: North Caro lina Sta te Universitv. :\I.S. thesis. 86p. DISCUSSION CIII':N( :, .J.-R.: Sruo ur., ItO.: YEll. G,-T.:LI0: . H.-C.. and Cut u, W.H.. Um8. Modeling of 20 density-depe ndent flow and tran sport in t. h« Th e effect of storm overw a sh on t he wa ter quality of coa sta l subsurface. -lourn ttl or Hydrologic Engineering :3(4). 248-2;;7. aq uife rs is a real, yet rela tively un studied , ph enomen on . COI.I.I:'o1S . W.H.. II I and EAs l.EY, D.H.. 1999. Fresh-water lens f(Il '­ Field da ta fro m the Ca pe Hatteras region of No rt h Carolina mati on in an unconfined barrier-islan d aquifer. -lourn al or thv show t hat overwash in t he form of storm surge ha s seriously A merican Water Re sou rces Association 351 11, 1- 21. CllIWETT, D.R.; Dn.r.ox , K.. and BllHNETl', W.. 2000. Tracing a ffected water qu al ity in t he Bu xt on Woods Aquifer. Ea rly groundwater flow on a barrier islan d in the north -east Gulf of st u dies of t he BWA suggest that overwash ha s been a n im­ Mexico. Estuarine. Coasta l and S hel fScienrc 51. 227-242. portant factor in the pa st (HAIWls an d WILIl ER, 196 4; H AR­ COlllin ;l\lA:'o1 CIIE, I\.P.,.JIL; HEsTI·:H. M.W.. and l\1!<:1\ I)J<;I.SSllIIN . HI S, 196 7 ), a nd a st udy a t nearby Cape Look out, No rth Ca r­ LA.. HJ99. Recovery of a Louisiana Barri er Island :'I larsh Plan t olin a sugges ts that overwash cou ld be a se rious wa ter quality Community following Extensive Hurri can e-Induced Overwash. Journal o/ Coastal R esearch 1;;141. 872- 88:3 . th reat to low-lying ba rrier isla nd s (W If' :-.I lm, H)78 1. Bulk GII';SI';, G.L.; Wu.mcn, H.B., and PAl i"I·:li. G.G.. J IL. HJ79. Hydrology grou ndwa ter ch loride levels mon it ored at the F ri sco wa ter of major est ua ries and sounds of North Carolina . U.S. Geologica l treatment plant in 199:3- 94 dem on strate t hat ch loride levels Survey Water Resources Investigations 7 ~) - 4 () . pp.I-1 7:i . violated MCL sta nda rds for more th an one month foll owing GH J( ;(;S, J .E. and PI·:TEHSllN. F.L.. 1 9 9 : ~ . Grou ndwater flow dynam­ th e stor m , with periodi c viol ations for two months. An a lyti cal ics and development strategies at the atoll sca le. ( lrounchrtrtrr :311 21, 209-220. sim u la tions presented in t h is st udy suggest t hat t he effects GUII.I. EN. •J.; Cxxu -. •J.. and PAI.A:'o1!ilJl<:S. A . 1994. Short-timo evo­ of multiple over wa sh eve nts in s uccession cou ld rai se ch loride lut ion of a microtidal barrier-lagoon syste m affpcted by storm and levels in t he Bux to n Wood s aqu ife r a bove t he MCL for more overwas hing: the Tra hucador Bar I Ehro Delta. NW Modiu-rra ­ t han 100 days. nean I. Zeitschri]! [iir Oeo/l/orp!lOlof.( ie :381 :11, 267-281. HAl mls . W.H. and WILllI·;H. H.B., HJ64. Groundwater supply of Ca pe Th e Ca pe Hattera s region now su pp lemen ts its freshwater Hatt eras National Seashore Recreational Area . Part :1 . Haleigb. supply wit h bra ck ish water pu mped from t h ree deep we lls. North Carolina: North Carolina Department of Water Resources Th e Cou nty of Da re , No rth Ca ro lina cur ren tly mixes th e Report of Investigations 4. 22p. freshwater fro m t he ex isting Frisco Wellfield with saline wa ­ HAHlos . W.H.. HJ()7. St ratificat ion of fres h and salt water on barri er ters from t he deep wells , a nd t hen t reats the mi xed wa ter islands as a result of differences in sediment perm cabilit v, Water R esources R esea rch 3(1),89-97. with a reverse-osmosis treat ment syste m . Although t hi s HEATII, R.C.. HJ88. Groundu-ater rcso urcc« ofth« Cape Hal/ em s A n 'a makes the over wa sh process less of a prob lem for the Ca pe or North Carolina-s-A report to the Cape Hal/ em s Wat('/' Ass()('ia · Ha ttera s region than it wa s in 199 3, it contin ues to be a con ­ tion, l nc. Buxton, North Carolina: Cape Hatt eras Water Associa­ cern for se ve ral reason s . First, overwa sh wa ters ra ise t he tion, 129p. overa ll salinity of t he in fluent wa ters, leading to in crea sed HEATII, R.C., 1990. Report Oil the prop osed B uxt on n't)()ds u-cllficld or the Ca pe Hal/ era s IVater Aseociotion. Buxton, Nort h Carolina : tre atme nt dem ands , time , a nd costs. Second , coa stal regio ns Cape Hatt eras Wat er Association, 4fip. in th e Unite d Sta tes a re seeing a n in crea se in water dem and KL\lIiEY. .1.0., 1960. Groundw ater supply of Cape Hatt eras Nationa l as population s in t he se areas in crease. The Ca pe Hattera s Seashore Recreational Area. Par t 2. Raleigh. North Carolina : region is transfor m ing from a seasona l tourist desti nation to North Carolina Depar tment of Wat er Resources Report of Inves­ a ye a r-rou nd tou rist destina ti on . The in crea sed water de­ tigations 2. 28p. LI. L.; BAlWY, D.A.: STN;:'o1 ITTI. F., and PAHI.AK(;E. •J.-Y., 2000a. ma nd tha t these year-ro und to u rists brin g to t he a re a m ea ns Groundwate r waves in a coasta l aquifer: A new gover ning equa­ t hat more water is being pumped fro m t he aquifer, thereby tion including vertical effects an d capi llarity. H'ct ter Hesou rcc« He­ shrinking the freshwa ter len s a nd diminishing the aqu ifer's search 36(2). 411- 420, ability to di lute t h e saline overwa sh wa ters . F ina lly, with the LI. L.: BAH HY. D.A.; STA(;:\"I'I"I' I, F.: PAHLA:-i (:E. J .-Y.. and .1E1\(:. D.­ a dvent of globa l wa r ming, ri sing se a levels a n d mor e frequ ent S., 2000b. Beach wate r table fiuctu a t.i ons due to spring-neap : moving boundary effects, Adronce« ill Water Resource» 2:3, 817­ tro pical storm ev ents ex pected on a warmer Earth sugges t 824. that t here will be a n in crea se in the frequency of overwash MEW. H.E. J IL, 1999, Report on the Fre shwater Lens of the Cape eve nts, furth er reducin g the aquifer's water quality. Hatt eras region. Raleigh, Nort h Carolina: Depar tment of Enviro n­ ment and Natura l Resources, Water Quality Division. Ground ­ LITERATU RE CITED water Section, 1:3p. MEW. H.E. JIL; HEATH. R.C.. and EVANS. D.G.. 19!J3. Cape Hatt eras ANllEHSllN. •JH.. W.I'.: EVANS. D.G.. and SNYI)EH, S.W., 2000. The Field Tr ip Notes - Sept ember :~- 4 , 1993. Unpublished. lOp. effects of Holocene barrier-island evolution on wate r-ta ble eleva­ Mll H( ;AN.A S.; GliE(;()!iY, J .. and EVA:'-< s , D.G.. 1994, A barrier is­ tions-Hatteras Island. North Carolina. USA. Hy d rof.(eolog v Jour­ land aquifer response to Hurri cane Emily, Hatt eras Island. Nort h 1/0 1 8141.390- 404. Carolina . Proceeding» 0/ the 211d l nternationol Conference Oil AN1Jl·:HSllN. W.P.. •J11. . HJ!J9 . The Hydrology of Hatt era s Island . Groundwater Ecology. (American Wat er Resources Associat ion and North Carolina. Raleigh. Nort h Carolina: North Carolina Sta te U.S. Environmental Protection Agcncyr. University. Ph.D. dissertati on. 297p. Mll)H;AN, A S.. 1996. Wat er-Table Response to Large Storm Events ATAIE-As IlTIANI . B.: Vor .icr:«. R.E.. and LllC"! :-i(;Tll:'o1 , D.A . 1999. in Freshwater Wetl ands and Adjacent Uplands of a Bar rier Island.

.Journal of Coastal Research, Vol. 18. No.3. 2002 420 Anderson

Hatteras Island, North Carolina. Raleigh, North Carolina: North and the super-elevation of groundwater at the coast. Journal o] Carolina State University, M.S. thesis, 150p. Coastal Research 1:3( 1), 46-60. N"':LsEN, P., 1990. Tidal dynamics of the watertable in beaches. Wa­ UNDEI{WOOD, M.R.; PETEI{SON, F.L., and Voss, c.r., 1992. Ground­ tcr RC,'Wllrces Hesearch 26( 9), 2127-2134. water lens dynamics of atoll islands. Water Resources Research NATIONAL OCEANIC and ATMOSPHEHIC ASSOCIATION, 1993. Hurri­ 28( 11), 2889-2902. cane Emily preliminary report. Unpublished. VAcHI·:n, H.L., 1988. Dupuit-Ghyben-Herzberg analysis of strip-is­ PI':I{SON, M.; TAYLOI{, J.Z., and DIN(;MAN, S.L., 1998. Sharp inter­ land lenses. GSA Bulletin 100, G80-G91. face models of salt water intrusion and wellhead delineation on VACHEH, H.L. and WALLIS, T.N., 1992. Comparative hydrogeology Nantucket Island, Massachusetts. Groundumter :36(5), 731-742. of fresh-water lenses of and Grent Exum» Island, Ba­ hamas. ;30, 1f)-20. ROBINSON, M.A. and GALLACHEH, D.L., 1999. A model of ground­ Grou nduatcr WALLIS, T.N.; VAClIl<:H, H.L., and STI':WI\I{T, M.T., 1992. Hydrogeol­ water discharge from an unconfined coastal aquifer. Groundwater ogy of freshwater lens beneath a Holocene strandplain, Great Ex­ ;~7( 1), HO-H7. uma, Bahamas. Journal oj' Hydrology 12G, ~):3-1()9. SALINAS, L.M.: DELAllNI<:, R.D., and PATI{ICK, W.H., JI{., 1986. WILDEI{, H.B.; ROBISON, T.M., and LINI)SI~OV, K.L., 1978. Water Chang"es occurring along a rapidly submerging coastal area: Lou­ resources of northeast North Carolina. U.S. G(~ologica] Survey Wa­ isiana, USA. •lournal o] Coastal Research 2(3),269-284. ter Resources Investigations 77-HI, pp. 1-11:3. ~J.M., DIN(~LEI{, ~J.R., BT< >NI':, G.W.; GHYl\H:S, III; and PEPPEI{, D.A., WINNE/{, M.D., -Iu., 1978. Groundwater resources of the Cape Look­ 1997. Overview and significance of hurricanes on the Louisiana out National Seashore, North Carolina. U.S. Geological Survey coast, USA. •Iournal ol Coastal Research 13(3),656-669. Water Resources Investigations 78-G2, pp. 1-24. 1':\ YLOj{, ~J.Z. and PEHSON, M., 1998. Capture zone delineations on YOUN(;, R.S.; BUSH, D.M.; COBlll{N, A.S.; PILKEY, O.H., and island aquifer systems. Groundu.ater :36(5),722-730. CLI':AI{Y, W.~J., 1999. Hurricanes Dennis and Floyd: Coastal effects 1'1'I{NEH, LL.; COATES, B.P., and Acwol{TH, R.I., 1997. Tides, waves and policy implications. GSA Today 9( 12), 1-6.

Journal of Coastal Research, Vol. 18, No.3, 2002