Geohydrology and Water Quality of Stratified-DriftAquifers in the Lower Merrimack and Coastal River

Basins, Southeastern New Hampshire

U.S . GEOLOGICAL SURVEY

Water-Resources Investigations Report 91-4025

Prepared in cooperation with the

NEW HAMPSHIRE DEPARTMENT OF ENVIRONMENTAL SERVICESWATER RESOURCES DIVISION

Click here to return to USGS publications

GEOHYDROLOGY OF STRATIFIED-DRIFT AQUIFERS

Stratified drift is subdivided on plates 4-6 intofour categories on the basis of grain size . Thesecategories are (1) coarse-grained material (sandand gravel) with a median particle diameter pre-dominantly larger than 0.0049 in . ; (2) fine-grainedmaterial (very fine sand, silt, and clay) with a medianparticle diameter predominantly smaller than0.0049 in . ; (3) coarse-grained deposits that overliefine-grained deposits ; and (4) fine-grained deposits(including marine) that overlie coarse-graineddeposits .

The coarse-grained deposits in the central Pow-wow River valley, Derry-Island Pond area, the Gold-en Brook valley, and the isolated kame plains anddeltas in Portsmouth-Newington, North Hampton-Hampton, and Kensington (pls . 4-6) constitute someof the most important aquifers in the basin . Of theseareas, the aquifers at Powwow River, Golden Brookvalley, and the Derry-Island Pond area are the onlyones not currently developed at or near their maxi-mum potential yield to wells .

Coarse-grained deposits are part of a valley-train sequence in the Golden Brook valley ofWindham . The stratigraphy of these proglacialdeposits is characterized by coarse sand and graveloverlying very fine to medium sand (fig . 6) . Themaximum thickness of these deposits (60 ft) is deter-mined from seismic-refraction surveys in the area .Present flow regulation at Golden Brook causes thestream to go dry during parts of the year, and, there-fore, the brook cannot be considered a reliablesource of water to recharge the deposits .

Geohydrologic section B-B' (fig . 7), which ex-tends northwest-southeast across the axis of thePowwow River, shows a striking contrast betweenthe level outwash and the irregular bedrock surface .Seismic-refraction data indicate maximum depths tobedrock of 125 ft . The outwash-plain deposits aregenerally well-sorted and permeable ; therefore,water is believed to flow easily between this aquiferand the Powwow River and nearby Great Pond. Theaquifer and surface-water bodies are interdepen-dent and together comprise a water system that iscapable of a high yield to wells completed in theaquifer .

Geohydrologic section C-C' (fig . 8) shows fine-to coarse-grained sand and gravel underlying a kamedelta in North Hampton that is interbedded withdiscontinuous layers of marine sediments . The sandand gravel deposits have excellent water-bearing

17

characteristics and, where saturated thicknessesare large, would yield large supplies of water towells . Kame-delta deposits penetrated by wellNSW-69 consist of exceptionally clean, well-washed coarse-grained sand and pebble gravelthat has a 70-ft saturated thickness . Low rates ofrecharge to the aquifer are considered the primarylimit on the productivity of this and many othercoastal aquifers . This deposit, like similar kameplains and deltas in the eastern part of the studyarea, is small in areal extent ; thus recharge fromprecipitation is limited .

Present at the aquifer margins and within theaquifer section are marine deposits of silt and claythat are relatively impermeable and do not yieldusable quantities of water to wells . Where themarine sediments border the aquifer perimeterand extend the full saturated thickness, as in thenorthwestern end of the section, lateral rechargeto the aquifer as ground-water discharge fromneighboring uplands is restricted . The deposit isnot bordered by land significantly higher in eleva-tion and, therefore, recharge from upland runoff isnegligible . In addition, the hydraulic connectionbetween Cornelius Brook and the aquifer is poorbecause of the presence of marine impermeablesediments in the streambed . Aquifers hydraulical-ly isolated from streams can serve as storage reser-voirs that can be tapped during late summerperiods of drought without causing a seriousdepletion of streamflow that normally would resultif withdrawals are made from aquifers that bordera river and are sustained by induced infiltration ofriver water.

In some localities, sand and gravel is buriedbeneath the marine sediment . Geohydrologic sec-tion D-D' (fig . 9) shows that Seabrook municipalwells SGW-44 and SGW-65 are completed in coarse-grained sand and gravel, which underlies 20 to 40 ftof fine-grained sand and clay . The contact betweenthe base of the marine deposits and the underlyingsand and gravel is gradational at these wells . Con-tacts are typically gradational where clay overliesice-marginal deposits, deposited as submarine fansand deltas (Thompson and others, in press) in ahigh-energy environment that prevailed whenmeltwater currents flowed from the nearby ice mar-gin . To the northwest, the vertical and horizontalcontact between the marine sediments and the sandand gravel is more abrupt and probably is erosional .A 20-ft-thick layer of ice-contact sand and gravelforms an aquifer confined above and below by sedi-ments.

Y0 A'0Lma

0

1000

2000 FEET

EXPLANATION0

200

400

600 METERS

VERTICAL EXAGGERATION X19

DATUM IS SEA LEVEL

COARSE SAND AND GRAVEL

VERY FINE TO MEDIUM SAND

TILL

BEDROCK

--- CONTACT-APPROXIMATELYLOCATED DASHED WHEREINFERRED

WPW-38

W NUMBERIDENTIFICATION

Figure 6.--Geohydrologic section A-A' showing the valley train in southern Windham. Line of section is shown on plate 1 .

F-y

Seismic-refraction dataused for control

a---I!_ Lo

'a

0

100~_0~ -2000FEETI

I

1

17'0

200

400

600 METERSVERTICAL EXAGGERATION X30

Uo

FITT

KT1-61

EXPLANATION

FINE TO COARSE SAND SOME GRAVELCLAY AND SILTY CLAYTILL COVERED BEDROCKCONTACT-APPROXIMATELY LOCATED.DASHED WHERE INFERRED

WELL AND IDENTIFICATION NUMBER

Figure 7 .--Geohydrologic section B-B' showing the outwash plain in Kingston . Line of section is shown on plate 2 .

YOOQ1

y

.

a~

0UI

400

600 METERS

VERTICAL EXAGGERATION X26

Y'D,0Oca (rU1 T

G) Nr. -r

0%_ C)

N

Z

7=7

TILL

Seismic-refraction dataused for control

FINE TO COARSE SANDAND GRAVEL

CLAY AND SILTY CLAYWITH FINE SAND ANDSILT STRINGERS

BEDROCK

ro0

UCO

Figure 8.--Geohydrologic section C-C' showing the kame delta in North Hampton. Line of section is shown on plate 3 .

CONTACT-APPROXIMATELYLOCATED DASHED WHEREINFERRED

NSA-4 WELL AND IDENTIFICATION1 NUMBER

C'

r

Y

00C)

O

Y

VERTICAL EXAGGERATION X30

USeismic-refraction data

cnused for control

Z

HUW

V I~r M

South Hampton,Kensington,Seabrook town line

IO M

JU) (3

0

1000

2000 FEET

0

200

400

600 METERS

N0QJ

N

TILL

EXPLANATION

cu0a=

YCUCYUNCO

BEDROCKFINE TO COARSE SANDAND GRAVEL

CONTACT-APPROXIMATELYCLAY AND SILTY CLAY

LOCATED DASHED WHEREWITH FINE SAND AND

INFERREDSILT STRINGERS

SG1W-1 WELL AND IDENTIFICATIONNUMBER

Figure 9.--Geohydrologic section D-D' showing the kame delta in Kensington and Seabrook . Line of section is shown on plate 3 .

N

JN

Q

2Zvco

m

D'

Description of Aquifer Boundaries

Areal extents of stratified-drift aquifers areshown on plates 1-3 . Also shown on the plates arethe locations of marine silt and clay and till-mantledbedrock . The thickness of the solid-line contact onthe 1:24,000-scale U.S . Geological Survey map implies±80-foot horizontal accuracy . Aquifers locatedwithin moderate to steeply sloping uplands and val-ley walls in the western part of the study area aregenerally mapped to this level of accuracy . Delinea-tion of the aquifer is difficult and contact locationsare uncertain in the eastern part of the study area,where a complex geologic history of marine inunda-tion, postglacial erosion, and uplift has producedlow scarps and broad swampy lowlands with few ice-contact slopes or meltwater drainageways . As aresult, several of the aquifer contacts in the coastallowlands are inferred and appear as dashed lines, orthe contacts are concealed by swamps and marshesand appear as dotted lines .

The boundaries of an aquifer are of particularimportance in describing the response of theground-water-flow system to withdrawal stress . Theground-water-flow system has two types of flowboundaries--impermeable (no flow) boundaries andrecharge and discharge (flow) boundaries (Heath,1983) . Low scarps composed of marine sedimentslimit the extent of stratified-drift aquifers in the east-ern part of the study area . Because these geologicunits are virtually impermeable compared tostratified-drift aquifer material, flow across the in-terface is limited . Wells that tap stratified drift inthe proximity of such boundaries yield less waterthan they would if the aquifer had infinite extent . Asshown in figure 10, pumping near an impermeableboundary lowers the water level between the welland the boundary more than pumping does at theopposite side of the well where the aquifer is exten-sive . Drawdown, therefore, increases with decreas-ing distance to an impermeable (no flow) boundaryat a given pumping rate .

Major perennial streams or large pondshydraulically connected to stratified-drift aquifersare considered recharge boundaries because theycan serve as a source of recharge to the aquifer .When a well near a surface-water body is pumped,the water is initially withdrawn from aquifer storage .As the cone of depression continues to spread, waterfrom an increasingly large area flows toward thewell . When the water table in the vicinity of thenearby stream or pond is sufficiently lowered,ground water that would have naturally discharged

22

to the surface water is diverted and captured by thewell. Eventually, ground-water levels can be lowerthan the surface of the surface-water body, at whichtime surface water can recharge the aquifer by in-duced infiltration (fig . 11) .

An understanding of these basic principles ofboundary hydraulics is important for determiningthe optimum sites for ground-water development .Recharge to pumped wells is greatest and the effectson water levels from pumping are the least whenwells are parallel and adjacent to recharge boun-daries and at maximum distance from impermeableboundaries .

Recharge, Discharge, and Directionof Ground-Water Flow

Recharge to the stratified-drift aquifers is byinfiltration from precipitation, seepage losses fromtributary streams, and lateral flow from adjacent tilland bedrock . Discharge from the aquifers is by flowto the rivers and ponds in the basin, by evapotran-spiration in areas where ground water is near theland surface, and by ground-water withdrawals . Thewater table marks the top of the saturated zone inthe unconsolidated deposits and fluctuates con-tinuously in response to changes in recharge anddischarge .

The 10-year hydrograph for well LIW-1, whichis completed in stratified drift in the town of Lee,shows the cyclic seasonal variation of water level thatis common in stratified-drift aquifers (fig . 12) .Ground-water recharge from precipitation exceedsevapotranspiration in late fall and early winter(November-December) and even more in the spring(March-June) when snow melt is an additional com-ponent of recharge . Lowest ground-water levels arein late summer or early fall . The depth to watertable at this well (30 ft) is generally attained only inthe elevated kame plains and deltas of the easternstudy area ; this depth contrasts with the shallowdepths (a few feet) in sand and gravel deposits oflower altitude . The seasonal fluctuation in ground-water levels is typically 3 to 5 ft for wells completedin stratified drift .

Flow

Circulation of ground water is usually confinedwithin topographic basins . The five subbasins thatcomprise distinct ground-water-flow systems of thereport area are shown in figure 1, and include the

Figure 10 .--Ground-water flow and water-level drawdowns at a pumped well near an impermeable boundary .

Figure 11.--Ground-water flow and water-level drawdowns at a pumped well near a recharge boundary .

WU¢

0zJ

OJWmFWWWZJWWJWF-

25

26

27

28

29

30

31

32

33

34

351978 1980 1982 1984 1986 1988

24

165 JW164 wJ163 w162 W>

O161 Q160 wWW159 z

157W

156 Q155

Figure 12.--Water-level fluctuations typical of fluctuations in wells completed in stratified-driftaquifers in elevated kame plains .

Beaver Brook, Spicket River, Little River, PowwowRiver, and Coastal River basins . The generalizedground-water circulation within a subbasin of thestudy area is shown in figure 4. The saturatedstratified drift within each major flow system can bethought of as a large underground reservoir boundedon top by the water table and on the bottom byfractured crystalline bedrock .

Regional ground-water divides commonly coin-cide with topographic divides on ridges of till orbedrock . In some areas, major flow systems extendacross major drainage divides that separate thelower Merrimack basin from adjacent basins . Areaswhere this is most likely to occur include the north-east-southwest-trending faults and stress fracturesthat parallel the regional structural grain and extendbeyond the basin boundaries into the Cocheco andExeter River basins to the north (Lyons and others,1986) .

Ground water in the basin that does not evap-orate or transpire or that is withdrawn by wells even-tually discharges to lakes, streams, or the AtlanticOcean. The transit time of ground water in thesaturated zone differs spatially, depending on thelocations of recharge and discharge zones and thehydrologic characteristics of the aquifer . Generally,the closer to the basin-drainage boundary the waterenters the saturated zone, the deeper and greater thedistance traveled before it discharges .

Recharge

Recharge is the process by which water is addedto the zone of saturation in an aquifer . The amountof water available for development from a stratifieddrift aquifer can be limited by the amount ofrecharge ; therefore, when aquifer potential yield isestimated, the amount of water that recharges thesystem needs to be evaluated . Water pumped fromaquifer storage can be replenished by natural andinduced recharge, and the contributions from bothsources need to be estimated . Neither natural norinduced recharge has been measured directly in thestudy area . The estimates in this report are based onregional information on ground-water discharge andstreamflow .

The sources of recharge to stratified-driftaquifers are infiltration of precipitation that fallsdirectly on the aquifer, surface and subsurfacerunoff from upland hillslopes adjacent to an aquifer,and leakage from streams that cross the aquifer(fig . 4) . Leakage from streams to the ground-water

25

reservoir occurs where the hydraulic head in theaquifer is less than the stream stage .

Under steady-state conditions, the amount ofwater available for natural recharge from precipita-tion on stratified drift is roughly equivalent to theaverage stream runoff (Lyford and Cohen, 1988) .Maps showing contours of average annual runoff forthe glaciated northeast (Knox and Nordenson, 1955)were adjusted to a more recent period of precipita-tion (1951-80) to estimate the average annual runoffin southeastern New Hampshire . From this analysis,the approximate recharge to stratified drift fromprecipitation is 19 in ., or 0 .9 (Mgal/d)/mil .

Natural recharge from tributaries and un-channeled runoff from upland sources can be es-timated from streamflow records . The 1966-77average annual streamflow runoff from till andbedrock uplands was considered a reasonable es-timate for the maximum recharge available from allupland sources .

Recharge from unchanneled runoff from up-land hillsides has not been extensively documented;however, several investigators (Crain, 1974; MacNish and Randall, 1982 ; Randall, 1986) assumed thatthe total runoff from upland hillsides, including sur-face and subsurface flow, will infiltrate a stratified-drift aquifer at the base of a hillside .

Recharge from tributary seepage losses was notmeasured directly ; however, results from several in-vestigations indicate that rates may vary considerably depending on the location within the basin andstreambed characteristics (Morrissey and others,1988) . Measurements of tributary losses in south-central New York (Randall, 1978) indicate that los-ses were small near the margins of the main valley(average loss of 0 .13 ft3 /s per 1,000 ft of channel) andincreased to 1.0 ft3 /s or more per 1,000 ft of channelfarther downstream . Rates of seepage loss fromstreams can differ spatially and temporally, andupland streams can gain rather than lose water,making predictions of upland recharge rates dif-ficult . Estimates of the maximum upland tributarylosses to aquifers in the study area were assumed toequal the average annual (1966-77) stream runoff forstreams draining till and bedrock uplands.

Few data are available on the amount of runofffrom till-covered uplands because of the absence oflong-term streamflow records . Recharge fromupland runoff was calculated from streamflow datacollected at a long-term station in the nearbyMohawk River basin in the town of Strafford .Average annual runoff from till and bedrock in thisbasin is 1.4 (ft 3 /s)/mi l or 0.9 (Mgal/d)/mi` . Rechargeto the stratified-drift aquifer from upland runoff and

tributary losses was calculated by applying the rateof runoff per unit area of upland to the upland areasbordering the aquifer .

Induced recharge to an aquifer can also occurfrom the pumping of wells near a surface-waterbody. As previously mentioned, sustained withdrawal of wells that tap stratified drift can lower thewater table below an adjacent stream or lake, there-by inducing recharge from the surface-water body tothe aquifer . The quantity of water potentially avail-able to the aquifer through induced recharge duringdry periods is limited by the streamflow . Excessivepumping near streams can result in undesirablereductions in streamflow and perhaps even thedrying up of sections during periods of low flow .

The 7-day, 10-year estimate of low flow 0,10)can be used as a measure of the amount of inducedrecharge available during dry periods . Streamflowstatistics from several streams in New Hampshireindicate that the (Q7, lo ) is about the 99-percent flowduration, which represents the amount of stream-flow equaled or exceeded 99 percent of the time . Ifall the streamflow at 99-percent duration becameinduced recharge, the stream would be dry 1 percentof the time on the average . The (Q,,,o) was selectedbecause (1) flows of this magnitude are a reasonableestimate of the minimum natural ground-water dis-charge to streams during a 180-day period of norecharge and (2) limiting the induced recharge tothis amount reduces the effect of pumping onstreamflow, particularly during dry years .

Discharge

Natural ground-water discharge from theaquifers consists of seepage to streams and pondsand ground-water evapotranspiration .

Ground-water discharge to streams sustainstheir flow in dry weather. After long periods of littleor no precipitation, the rate of discharge per squaremile from coarse-grained stratified drift is manytimes greater than that from till (Thomas, 1966) .Low flow--and, therefore, ground-water discharge--can be estimated for any site on a stream in NewEngland by determining the upstream drainage areaunderlain by coarse-grained stratified drift and till-mantled bedrock (Cervione and others, 1982) and byconsidering certain other watershed properties(Wandle, 1988) .

Streamflow was measured at several sites onunregulated streams in 1986 and 1987 during periodswhen flow was relatively low and consisted entirelyof ground-water discharge . Low-flow measure-

26

ments are summarized in table 3 ; locations ofmeasurement sites are shown on plates 1-3 .

Low-flow measurements at five separate siteson Beaver Brook indicate that for the 2 years ofmeasurements, streamflows increased progressivelydownstream in direct proportion to the upstreamdrainage area . No unusual gains or losses werefound between measured segments for a 4 .2-mireach of the river . Comparison of low-flow yields(discharge per square mile of drainage area) indi-cates that similar proportional relations apply to theother streams measured during the study . Anom-alous sources or sinks of ground water were notfound between measured segments, and moststreamflows increased proportionally to upstreamdrainage area .

Ground-water evapotranspiration is anothersignificant source of discharge from the aquifers andis greatest during the growing season (April-October) when plants use a large amount of water,temperatures are above freezing, and the days arelong . Ground-water evapotranspiration has been es-timated to range from 1 to 9 inches per year in thenortheastern United States (Lyford and others,1984) . Large amounts of moisture are lost to theatmosphere by evapotranspiration from wetlandsand marshes where the water table is within 5 ft ofthe land surface . Streamflow can be reduced wheresuch areas are extensive .

Water Table and Direction of Ground-Water Flow

Generalized water-table altitudes for aquifersare shown on plates 1-3 . These maps are based onavailable water-level data measurements of depth towater at various times during 1986-88 at all observa-tion wells, and surface-water altitudes on topo-graphic maps . Water-table contours represent"average" water levels for a 3-year period (1986-88)and reflect the adjustments made to water-table al-titudes during high- and low-water seasons . Be-cause the seasonal fluctuation in water-table altitudeis typically 3 to 5 ft in the stratified-drift aquifers(fig . 12), an approximate average water-table al-titude was estimated by taking the average of highand low water-table altitudes measured during1986-88 . In addition, topographic maps were usedfor vertical control, such that the uncertainty ofwater levels shown on the plates is approximatelyone-half of the contour interval (10-20 ft) shown onthe map.

Arrows drawn at right angles to the water-tablecontour lines are shown on plates 1-3 to indicate the

Table 3 .--Low-flow measurements at miscellaneous sites

[01073830, station identification number, a unique 8 digit station number assigned by the U .S . Geological Survey to stream-gaging stations or

single-flow measurement sites with numbers ascending in the downstream direction; mi, mile ; mi l , square miles; ft3/s, cubic feet per second ;

(ft3/s)/mil , cubic feet per second per square mile ; --, no data]

Drainage MeasurementsLocation

area

Date

Discharge Basin yield

PICATAQUA RIVER BASIN

Lat 42059 ,25", long 70047 , 48'1, RockinghamCounty, downstream side of bridge atculvert on West Road, 0.15 mi south ofintersection with Garland Road, 0 .36 minorth of intersection with South Road,1 .82 mi southwest of Rye, N .H . (plate 3) .

Lat 42059 , 20", long 70046'37", RockinghamCounty, downstream side of bridge atculvert on Love Lane, 0 .22 mi south-west of intersection with Central Road,0 .60 mi northwest of intersection withSouth Road, 1 .7 mi south of Rye, N .H .(plate 3) .

MERRIMACK RIVER BASIN

Merrimack River

Lat 42050 , 21", long 71 021'00", RockinghamCounty, downstream side of Kendall Pondoutlet exactly on the Windham-Londonderrytown line, 0 .01 mi south of the intersec-tion of South Road and Kendall PondRoad, 3 .45 mi northwest of Windham, N .H .(plate 1) .

Merrimack River

Lat 42049 ,40", long 71 020 ,51", RockinghamCounty, 50 ft behind house numbered 16Pleasant Drive, 0 .06 mi east ofintersection of Pleasant Drive andTranquil Road, measurment site is alsoon Windham-Londonderry town line, 2.9 minorthwest of Windham, N.H . (plate 1) .

(mi l) (ft3/s) ((ft3/s)/mi l)

0 .5 10-21-86 0.15 0 .308-26-87 No flow --

1 .73 10-21-86 .35 .208-26-87 No flow --

30 .8 10-21-86 6 .71 .228-25-87 .97 .031

37 .7 10-21-86 8.88 .248-25-87 1 .33 .035

Location(plateand sitenumber)

Stationident-ificationnumber

Stream Tributary to

Plate 3 01073830 Bailey Atlantic Ocean1 Brook

01073835 Bailey Atlantic Ocean2 Brook

Plate 1 010965844 Beaver3 Brook

4 010965846 BeaverBrook

Table 3 .--Low-flow measurements at miscellaneous sites--Continued

Location Station

Drainage Measurements(plate

ident-

Stream

Tributary to

Location

area

Date

Discharge Basin yieldand site ificationnumber)

number

(mil )

(ft3/s)

((ft3/s)/mil)

Plate 15

Lat 42048'23", long 71 021 , 12", Rockingham

41 .8

10-21-86

11 .3

0.27County, 20 ft upstream from bridge at

8-25-87

1 .26

.030the intersection of State Route128 and Anderson Road, 0 .28 mi north ofthe intersection between StateRoutes 128 and 111, 2 .73 mi westof Windham, N.H . (plate 1) .

Lat 42047 , 25", long 71021'53", RockinghamCounty, upstream from side of bridge onBridle Bridge Road, at the Windham-Hudsontown line, 0 .45 mi west of StateRoute 128, 3 .6 mi southwest ofWindham, N.H . (plate 1) .

Lat 42046'59", long 71 021'14", on theRockingham-Hillsborough county line,100 ft fownstream from bridge onState Route 128 (Mammoth Rd .) .,0 .23 mi south of the intersection withGlance Road, 1 .4 mi south of WestWindham, N .H . (plate 1) .

Lat 42047 ,32 11 , long 71018 1 16", RockinghamCounty, upstream from side of bridge onGolden Brook Road, 0.5 mi northwest ofintersection with State Route 111A, 1 .6 misouth of Windham, N .H . (plate 1) .

Lat 42052'20", tong 71013'47", RockinghamCounty50 ft upstream from bridge onIsland Pond Road, 0 .3 mi northwest ofintersection with North Shore Road,5 .42 mi east of Derry, N .H . (plate 2) .

43.6 10-21-868-25-87

11 .51 .42

.26

.032

47.8 10-21-86 13 .0 .278-26-87 1 .42 .030

-- 8-26-87 .09 .2023

4 .8 10-20-86 1 .07 .228-26-87 .45 .09

01096585 Beaver Merrimack River6 Brook

010965851 Beaver Merrimack River7 Brook

1010965852 Beaver Merrimack River8 Brook

010965905 Golden Beaver Brook9 Brook

Plate 2 011005034 Taylor Spicket River10 Brook ,

MERRI CK RIVER BASIN--Continued

010965848 Beaver Beaver Brook Lat 42049'02", long 71 020'41", Rockingham -- 10-21-86 .99Brook County, 50 ft upstream from mouth of 8-25-87 .06tributary tributary to Beaver Brook, 0.07 mi north

of Sirod Road, 0 .15 mi west of intersec-tion between tributary and Kendall PondRoad, 2 .45 mi northwest of Windham, N .H .(plate 1) .

Table 3 .--Low-flow measurements at miscellaneous sites--Continued

Location Station

Drainage Measurements

(plate

ident-

Stream

Tributary to

Location

area

Date

Discharge Basin yield

and site ificationnumber)

number

(mil )

(ft3/s)

((ft3

l/s)/mi)

MERRIMACK RIVER BASIN--Continued

Plate 2 011005038

Taylor

Spicket River

Lat 42052'10", long 71013'27", Rockingham11

Brook

County, upstream from side of culvert onNorth Shore Road, 0 .12 mi east of inter-section with Island Pond Road, 5 .75 mieast of Derry, N .H . (plate 2) .

01100530

Hi ttytity

Widow Harris

Lat 42048'18", tong 71 013'07", Rockingham12

Brook

Brook

County, downstream 100 ft from culverton Bluff Road, 0 .07 mi west of intersec-tion with Zion's Hill Road, 0 .46 mi eastof intersection with Scotland Avenue,1 .42 mi northwest of Salem, N .H . (plate 2) .

Lat 42047'5811 , long 71011'5811 , RockinghamCounty, at culvert on Bridge Street,0 .23 mi southeast of intersectionwith Bluff Street, 0 .74 mi northof Salem, N .H . (plate 2) .

Lat042 51'12", long 71 004'55", RockinghamCounty, at culvert on Boston and Mainerailroad track, 0 .38 mi southwest ofintersection with Whittier Street Exten-sion, 2.7 mi southwest of Newton, N .H .(plate 2) .

Lat 42051'15",

long 71 006'03", RockinghamCounty, at culvert on State Route 125, 0 .18 misouthwest of intersection with Old CountyRoad and State Route 125, 1 .26 mi northwest ofPlaistow, N.H . (plate 2) .

Lat 42050'37", long 71006'07", RockinghamCounty, downstream side of bridge onNorth Main Street, 0.32 mi southeast ofintersection with State Route 125, 0 .6 minorthwest of Plaistow, N .H . (plate 2) .

1U .S . Geological Survey streamflow-gaging station .

2Low flow is affected by regulation such that the discharge per square mile may not be representative .

01100535 Widow Spicket River13 Harris

Brook

Plate 2 01100674 Little Little River14 River

Tributary

01100675 Kelly Little River15 Brook

01100676 Little Merrimack River16 River

5 .0 10-20-86 .71 .14

9.4

8-26-87

10-20-86

.48

0.94

.09

0 .108-27-87 No flow

10 .8 10-20-86 4 .17 .39

7.95 8-25-87 .42 .18

1 .9 10-20-86 .81 .43

8.8 10-20-86 3 .35 .388-25-87 .54 .06

direction of the horizontal component of ground-water flow . Flow may have a vertical as well ashorizontal component in some localities, such asalong the basin-drainage divides in and adjacent tothe marine-silt and -clay confining layers that are inthe vicinity of pumped wells and beneath ponds andstreams of the basin . Because of the scale of thesemaps and the presence oflocal lenses of fine-grainedmaterial, flow directions at specific sites could differfrom the regional flow directions shown on theplates .

Aquifer Characteristics

The capacity of a stratified-drift aquifer tostore, transmit, and yield water can be partlydescribed by its hydraulic characteristics--saturatedthickness, transmissivity, and storage coefficient--and by its boundary conditions . The estimated yieldto pumped wells and the resulting drawdownsthroughout an aquifer also can be determined ifthese aquifer properties are known.

Saturated Thickness

Saturated thickness of an unconfined stratified-drift aquifer is the vertical distance between thewater table and the bottom of the aquifer . In confined aquifers, saturated thickness is the distancefrom the top or overlying confining layer of theaquifer to the bottom of the aquifer . The stratig-raphy of an unconfined aquifer above a confinedaquifer and water levels in these aquifers is shown infigure 5 . In the western part of the study area, whereextensive fine-grained deposits are absent, the bot-tom of the stratified-drift aquifer is the till orbedrock surface . Some flow is expected to occurbetween stratified drift and bedrock wherever openfractures in bedrock are in contact with the stratifieddrift and there is a hydraulic head difference be-tween these units . Where these conditions exist, thedrift and bedrock form a single water-bearing unit .Given the number of high capacity bedrockmunicipal-supply wells in the study area (8 wells withan average yield of 328 gal/min) recharge frombedrock to stratified-drift aquifers may be sig-nificant in some areas. East of the towns ofKingston, Plaistow, and Newton, the bottom of theaquifer typically is the top of the fine-grained marinedeposits .

30

Saturated thickness was mapped separately forunconfined and confined aquifers (pls . 4-6) .Saturated thickness of unconfined aquifers wasdetermined by plotting the difference between thewater table and the depth to marine sediments, till,or bedrock from interpretation of seismic-refrac-tion profiles (figs . A1-A17, Appendix), seismic-reflection profiles (figs . 13 and 14), logs of wells andtest holes, and bedrock outcrops . It should be em-phasized that the contours on the saturated-thick-ness maps apply only to saturated stratified-driftdeposits .

The maps of saturated thickness can be usedin conjunction with other hydrologic data to indi-cate favorable areas for the placement of highyielding production wells . Where all otherhydrologic characteristics are equal, a thick aqui-fer will produce more than a thin aquifer . Strat-ified drift having saturated thicknesses less than20 ft cannot usually provide large water supplies,even where the deposits are coarse grained . Suchthinly saturated, low-yield areas are found alongvalley margins and in upland areas where most ofthe section of stratified drift is dry ; examples in-clude kame-terrace deposits high on valley walls inthe western part of the study area and elevatedkame plains and kame deltas above the marinelowlands in the coastal areas . The saturated thick-ness is less than 20 ft in any areas of stratified driftwhere no saturated-thickness lines are shown onplates 4-6 . In some of these areas, such as betweenthe Powwow River south of Great Pond and the tillcontact east of Powwow Pond, the saturated thick-ness may exceed 20 ft because of discontinuity inthe fine-grained marine sediments .

Saturated thickness ranges from 20 to 100 ft inoutwash in a long north-south-trending valley thatextends from Greenwood Pond to Country Pond inKingston (pl . 5) . This outwash represents the mostextensive, thickly saturated aquifer in the study area .Other aquifers that consist of thickly saturated (60 ftor more) coarse-grained material include : the deltaalong the western shore of Island Pond in Derry ; theice-contact deposits in Newton, south of AmesburyRoad; the kame delta that underlies Pease Air ForceBase in Portsmouth, specifically the center of thekame delta near the Haven supply well (PYW-2) ; thekame delta that crosses Route 95 in Greenland ; andthe kame delta, which includes Knowles Pond, inNorth Hampton . These aquifers contain the largeststored volume of water within stratified-driftaquifers in the lower Merrimack and coastal riverbasins .

W

JW

Q

H

J

WUQ

F-

LU0

0z00W

J_J

z

E

A

1,500

2,000

2,500

3,000

DISTANCE, IN FEETVERTICAL EXAGGERATION X15

Figure 13.--Seismic-reflection profile (A) and cross section (B) of Cobbetts Pond, Windham, interpreted fromseismic-reflection data. Line of section shown on plate 1 .

^ }fi: ;~r

tea, " l, .e.w:1

r ..01

E'

E'

0zOUWJJ

Zw

JWQF_WJQ

WWzuwv

WQ

OJWCO

dW

0

5

10

15

20

25

30

35

A

1,500 2,000 2,500 3,000 3,500

DISTANCE, IN FEET

VERTICAL EXAGGERATION X11

Figure 14.--Seismic-reflection profile (A) and cross section (B) of Island Pond, Derry, interpretedfrom seismic-reflection data . Line of section shown on plate 2 .

32

F'

Water surface

. .,~.w .~. y ",~ E

V

IN -

Saturated thickness was also mapped for con-fined aquifers wherever well data were sufficient tolocate saturated coarse-grained deposits beneathconfining layers . Significant confined aquifershaving saturated thickness of 20 ft or more include :a buried ice-contact deposit tapped by the Sher-burne municipal well field in south Portsmouth; theaquifer tapped by wells PXW-27, PXW-28, PXW-29,and PXW-39; and the buried ice-contact material atthe border of Seabrook, N.H . and Salisbury, Mass.,which is tapped by three municipal wells (SGW-44,SGW-45, and SBW-34) .

Storage Coefficient

The storage coefficient of an aquifer is definedas the volume of water an aquifer releases from ortakes into storage per unit surface area of the aquiferper unit change in head (Theis, 1938) . The storagecoefficient of an unconfined aquifer is equal tospecific yield--the volume of water that can be ob-tained by complete gravity drainage from a unitvolume of the aquifer . Laboratory tests done on 13unconsolidated samples from southern NewHampshire, that ranged from fine-grained lacustrinesands to coarse-grained sands and gravels, indicatethat specific yields range from 0.14 to 0.34 andaverage 0.26 (Weigel and Kranes, 1966) . A value of0.2 is commonly used to estimate specific yield inunconfined aquifers in New England .

The storage coefficient for the confinedaquifers was not measured in this investigation .Typical values for confined aquifers are in the rangeof 0.00005 to 0.005 (Todd, 1980) .

This range indi-cates that much less water is released from storageper unit decline in head from confined aquifers thanfrom unconfined aquifers until pumping lowers thehead in confined aquifers to the point that they areno longer confined.

Transmissivity and Hydraulic Conductivity

Transmissivity, a measure of the ability of anaquifer to transmit a fluid, is calculated by multiply-ing the horizontal hydraulic conductivity (thevolume of water at the existing kinomatic viscositythat moves in unit time through a unit area of aquiferunder a unit hydraulic gradient) by the saturatedthickness (Heath, 1983). The transmissivity dis-tribution in an aquifer reflects the combined effectsof variations in both of these factors . An aquifercomposed of well-sorted, coarse-grained material

33

will have a much higher transmissivity than one withthe same saturated thickness but composed of fine-grained material . For example, although the aquiferin Londonderry immediately south of the Man-chester/Grenier airport and the aquifer at Pease AirForce Base in Portsmouth have similar saturatedthicknesses, the transmissivities of the Portsmouthaquifer are more than double those of the Lon-donderry aquifer because of differences in texture(grain-size distribution) .

Hydraulic conductivities were estimated fromstratigraphic logs of wells and test holes for sites forwhich reliable logs were available . These estimatesare based on an empirical relation developed byOlney (1983) that was derived from regression ofhydraulic conductivity data. The relation expresseshydraulic conductivity (K, in ft/d) as a function ofeffective grain size determined from grain-sizeanalyses (Dio in phi units)--

K = 2,100 (10) -o.655D 10,

Effective size is defined as the grain-size diameter atwhich 10 percent of the sample consists of smallgrains and 90 percent consists of large grains . Arelation between hydraulic conductivity and mediangrain size (table 4) was developed by applying equa-tion 1 to the results of grain-size analyses of 175samples obtained from drilling in southeastern NewHampshire . Hydraulic conductivity values were as-signed to each interval of a stratigraphic log by ap-plying equation 1 if grain-size analyses wereavailable, or by carefully comparing materialdescriptions in the logs to table 4 if grain-size dis-tribution curves were unavailable . Estimates forhorizontal hydraulic conductivity were made for 48test holes by applying equation 1 and for 556 wells onthe basis of lithologic descriptions and the data intable 4 . The latter method is more subjective andless accurate then the method based on grain-sizeanalysis ; however, there is agreement between es-timates derived from descriptive logs and thosederived from quantitative techniques such asspecific-capacity and aquifer tests .

Transmissivity at each well and test hole wasdetermined by multiplying the horizontal hydraulicconductivity by the saturated thickness of the corresponding interval of the stratigraphic log and sum-ming the products . Specific-capacity tests and (or)aquifer tests at nine wells also were used to estimatetransmissivity by methods described by Ferris andothers (1962) and Theis (1963) . Although specific-capacity and aquifer tests provide the most accurateestimates of transmissivity, reliable records of such

Table 4.--Hydraulic conductivity estimatedfor lithologies in stratified drift in southeastern New Hampshire

[mm, millimeters ; ft/d, feet per day]

tests are available for only a few municipal wells ortest wells .

Transmissivity zones on plates 4-6 are based oninterpolation and extrapolation of transmissivity fromindividual wells and auger holes, with consideration ofsaturated thickness and horizontal hydraulic conduc-tivity of the surficial geologic units . Transmissivity canbe extremely variable over short distances because ofthe heterogeneity of stratified-drift deposits . Becausethe distribution and quality of the transmissivity datadiffer from place to place and subjective judgementwas involved in drawing the zones of equal transmis-sivity, the zones shown on plates 4-6 should be con-sidered general estimates of transmissivity .

Evaluation of Water Availability andSimulation of Ground-Water Flow

Ground-water availability in the shallowstratified-drift aquifers is generally enhanced wheresurface-water bodies are in direct hydraulic connection with the aquifer . Recharge induced from a sur-face-water body augments well yields, especiallywhere large freshwater ponds or lakes are close topumping centers in coarse stratified drift . In thelower Merrimack River and coastal river basins,however, most public-water-supply wells are instratified-drift aquifers that are several tens ofthousands of feet distant from freshwater bodies ;

34

1 Estimates based on empirical relations assuming isotropic conditions in the sample [From Olney, 1983] .

therefore, induced recharge from non-saline sur-face-water sources is expected to have little to noeffect on well yields for most gravel-pack productionwells in the study. The pond-aquifer system is anunderutilized public-water-supply source that haspotential for further exploration and development .Although it is an untapped water-supply resource, ifpumpage of the pond (and stream) aquifer system ismuch greater than natural recharge, streams couldshow reduced flow or dry up and pond levels couldbe lowered to a point that water resources are adver-sely affected .

Two pond-aquifer systems, the one at CobbettsPond in Windham and another at Powwow Rivervalley in Kingston (pls . 1 and 2), were selected fordetailed hydrologic evaluation because of theirpotential importance as public water supplies .Water from the ponds is an important source ofrecharge because of the large storage volume of theponds and hydraulic connection to the aquifers .Results from drilling and seismic reflection indicatethat the ponds generally have a well-sorted,medium-grained sand bottom in the nearshoreareas and a dense organic mud in the deep parts ofthe ponds . Maximum exchange between surfacewater and ground water is expected where the sandybottom sediments predominate . The objectives ofthe evaluation were to (1) estimate the potentialyield of the stratified-drift aquifer, (2) delineate thecontributing areas for pumped wells withdrawing

Sample description Median grain size (mm) Hydraulic conductivity' (ft/d)

SandVery fine 0.06 - 0.125 3Fine .125 - .25 10Medium .25 - .50 30Coarse .50 - 1.00 130Very Coarse 1.00 - 2.00 190

GravelFine 2.00 - 4.00 250Coarse 4.00 - 16.00 300 or greater

water from the pond-aquifer system, and (3) discussselected aquifer areas in terms of favorability forfuture withdrawal centers .

Description and Conceptualization of Model

The aquifer evaluation involved application of amethod that is analagous to the principle of super-position and incorporates the USGS modular threedimensional finite-difference ground-water-flowmodel (McDonald and Harbaugh, 1988) . The pro-cedure for evaluation involved construction of awater-table map representing a natural (unstressed)aquifer and a numerical model of the aquifer . Themodel was then used to calculate drawdown thatwould result from hypothetical pumping stress .Finally, the composite water-table that would resultfrom the pumping of production wells was generatedby superimposing the drawdowns calculated with themodel on the map of the unstressed water table . Theaquifers near Cobbetts Pond and the Powwow Riverwere simulated as two-dimensional ground-watersystems, and ground-water withdrawals were simu-lated for 180 days without recharge . A stress periodof 180 days approximately represents the maximumduration of annual periods when evapotranspirationis high and recharge is small . It is assumed thatrecharge to the aquifers for the remainder of theyear would supply sufficient water to sustain pump-ing during periods of negligible recharge .

Potential yield was determined from the modelby analyzing the effect of pumping stress on waterlevels in the aquifer . The area contributing water tothe pumped wells was determined from the finalwater-table configuration .

The principle of superposition states that, forsystems governed by linear equations, the solutionsto individual parts of a problem can be added tosolve composite problems (Reilly and others, 1987) .Superposition reduces many ground-water problemsto simpler terms. The most important constraint isthat the governing differential equation and bound-ary conditions must be linear .

The pond aquifers in Windham and Kingstonare unconfined (saturated thickness changes inresponse to pumping) ; therefore, the governing differential equation is nonlinear and superposition isnot strictly applicable . Models can be formulated,however, so that changes in saturated thicknesscaused by pumping and the corollary changes intransmissivity are taken into account in the simula-tion . This enables the problem to be formulated byaccounting only for changes in stress on the system

and calculating drawdowns, which is analagous tosuperposition . Such models calculate drawdown onthe basis of stated initial thicknesses of the aquifer .The water table is specified to start at zero draw-down, and the bottom of the aquifer is set equal tothe negative of the saturated thickness . The draw-down solution is dependent on the initial thicknessand the amount of change in thickness due to theapplied stress . An example of the application of thismethod of prediction of aquifer response to pump-ing stress is described by Moore (1990) for similarstratified-drift aquifers in southeastern New Hamp-shire .

The general procedure for estimating thepotential yield of the aquifers was to introduceproduction wells into the two-dimensional numerical model and vary their discharges . The locationsof hypothetical production wells in the model wereselected on the basis of hydrologic and practicalconsiderations . The maximum well yields were ataquifer locations of maximum transmissivity andnear ponds where induced infiltration could occur.In addition, well sites surrounded by an adequateground-water-protection area were selected . It wasassumed that open space, such as farms, parks, andwetlands, would be the most desirable types of landuse to have in proximity to the wells . Selecting sitesin open areas was done to satisfy the requirementthat wells be surrounded by a minimum 400-footprotection radius, as specified in the New Hamp-shire Water Supply and Pollution Control Commis-sion rules WS 309.04 and 309.5 of the Drinking-WaterRegulations (1984) . This protection radius was notalways possible for the final well site ; however, thechosen area was generally free from conflicting landuses . The maximum possible amounts of groundwater that could be withdrawn were determined byadjusting the number of hypothetical pumped wells,the well locations, and well-discharge rates untilmaximum drawdown within the model approached,but did not exceed, 50 percent of the initial saturatedthickness of the aquifer . A limitation of the numeri-cal model is the inability to simulate the decline inpond-stage that would result from pumping . Themodel assumes the ponds are at constant stage andare unresponsive to changes in pumping stress .

The numerical models developed for theWindham-Cobbetts Pond and Kingston-PowwowRiver aquifers represent approximations of complexnatural systems . Aquifer yields estimated by use ofthe models are based on several assumptions ; there-fore, yields should be considered not as exact quan-tities but rather as reasonable estimates of themaximum amount of water available from wells

having specific locations and construction charac-teristics and tapping a stratified-drift aquifer withspecific hydraulic characteristics and boundaries.Certain basic assumptions and limitations of themodel simulations are as follows :

1 . The aquifer characteristics of transmissivity,saturated thickness, and the water-tableconfiguration shown on plates 1-6 are thesource of input values used in the modelsimulations and are assumed to be repre-sentative of the natural system .

2 . Prevailing ground-water flow in the strat-ified drift is predominantly horizontal, thereis no change in flow to or from the underly-ing bedrock as a result of pumping, and theaquifer is isotropic . Vertical-flow com-ponents, which could be present nearstreams or ponds or near partially penetrat-ing production wells, could cause localdeviations from simulated water levels .

3 . The variable characteristics of the aquifercan be represented by a finite number ofblocks or cells, within which each of theaquifer characteristics is assumed to beuniform . Each block represents a discretearea of the aquifer with a single value forsaturated thickness, horizontal hydraulicconductivity, and specific yield . Thisgeneralization introduces some error be-cause the variability of aquifer charac-teristics in a given area can be greater thanwhat is represented by the correspondingparameters for a discretized cell of themodel .

4 . The use of a numerical model to solve theground-water-flow equation provides onlyan approximation of the true solution . Thenumerical model arrives at a solution in-directly through successive iterations ofsolving the ground-water-flow equationuntil the maximum head difference betweensuccessive iterations is less then a settolerance level . Errors introduced are min-imal as compared with errors involved inestimating aquifer characteristics .

5 . All hypothesized production wells are con-sidered to be screened through the fullsaturated thickness of the aquifer and to be100-percent efficient . To compensate forthe fact that few wells actually meet theseideal well-construction criteria, the maxi-

36

mum allowable drawdown at wells was 50percent of the initial saturated thickness .Because drawdowns in the model representthe average drawdown over the area of amodel cell, an adjustment is necessary toestimate drawdown at a pumped well . Thetotal drawdown at well locations was es-timated by adding the average drawdowncomputed by the model for the cell to theadditional drawdown in the well calculatedfrom solving the following equation (Tres-cott and others, 1976):

where SA

S _ 2.3Q

aA

2rrT log 4.81(rw) '

(2)

is an adjustment to drawdown cal-culated in the cellcontainingthepumped well ;

Q is discharge at the well, in cubicfeet per day;

T

is transmissivity, in ft/d ;a

is length of a model cell, in ft ; andrWis radius of pumped well, in ft.

The equation for estimating the total draw-down at a well location is

ST = Sm + SA,

(3)

where

ST is the theoretical total drawdown,in feetin the production well ; and

Sm is drawdown, in feet, computedbythe finite-difference flow modelin the model cell containing thepumped production well.

The theoretical total drawdown representedby this sum, which is always more than theaverage drawdown computed by the model,more accurately reflects the actual draw-down at the production well than the draw-down for the model cell .

6 . The simulation of ponds as head-dependentflux boundaries allows a reasonable estimatefor the potential yield of a pond-aquifer sys-tem . In the model simulation, the ponds aretreated as constant-head boundaries suchthat surface-water elevations are arbitrarilyheld at a constant level . In actuality,ground-water withdrawals in an adjacentaquifer may lower pond levels . If the

amount of discharge at the production wellswere greater than the amount of water thatnaturally flows out of the system, pond levelsas well as ground-water levels would belowered throughout the modeled area . Be-cause the model could not simulate this oc-currence, an independent calculation of theamount of water that naturally flows out ofthe system was needed to check the reason-ableness of the model solution .

Description of Analytical Method

An analytical method was applied to check thereasonableness of the total amount of waterwithdrawn from pumped wells compared with theavailability of water in the aquifer . Estimates of theamount of water available to an aquifer in hydraulicconnection with a surface-water body require infor-mation on the quantity of water obtained from thesurface reservoir . The low flow of streams, specifi-cally the 7-day, 10-year low flow 0,io), can be usedas an index for evaluating basin-wide potential yieldin stream-aquifer systems . The Q7,io is a statisticallyderived value ; the minimum mean discharge for 7consecutive days in a given year will be equal to orless than the 07,10 average for once in 10 years . Thisindex was chosen as a reasonable approximation ofaquifer potential yield after considering the following :

1 . Any withdrawal from wells will be balancedby an equal reduction in discharge tostreams or ground-water evapotranspira-tion. As long as the amount of waterwithdrawn from wells is small in relation tothe natural ground-water discharge tostreams, the average decline in the watertable from pumping should not differ fromwhat would occur naturally during a non-pumping period .

2 . The Q,,lo is a reasonable estimate of the min-imum natural ground-water discharge tostreams during a 180-day period withoutrecharge .

3 . Verification that the simulated model dis-charge does not exceed the Q7,io is importantbecause the model is based on the assump-tion of constant stage in several ponds. Ifthe average decline in the water table duringa stress period was no greater than wouldoccur naturally, pond level would not declinesignificantly during the stress period . Typical-ly, pond levels in these aquifers remain rela-

37

tively constant throughout the summer ;therefore, the model would be a close ap-proximation to reality . If a significantdecline in water table were allowed, themodel-simulated recharge from pondswould be too large and the results would beinvalid .

The aquifer at Cobbetts Pond is drained by asmall intermittent stream regulated by sluice gates atthe western shore of Cobbetts Pond. Because of thisregulation, an accurate measure of the range innatural streamflow is not possible ; however, theQ7,1o can be estimated by applying an analyticalmethod that was developed for ungaged basins (Cer-vione and others, 1982) . The method, based on theassumption that low flows are sustained by the dis-charge of water from adjacent aquifers, has beensuccessfully applied to estimate potential yield forstratified-drift aquifers in Connecticut and Mas-sachusetts . The analytical method, developed byThomas (1966) and later modified by Cervione andothers (1982), makes use of a regression equation torelate the Q7,lo at any site on a stream to the propor-tion of upstream drainage area underlain bystratified drift and the proportion underlain by till-mantled bedrock . This regression equation is

Q7,io = 0 .67Asd + 0 .01A, ; � ,

where

Q7,,o is the 7-day, 10-year low flow, inft'/s ;

A,d

is the drainage area underlain bycoarse-grained stratified drift,in mi'; and

Aml is the drainage area underlain bytill-mantled bedrock, in mil .

Application of the method is as follows :

1 . The basin drainage divide upstream from asite on the master stream of the basin is drawnon the map by use of the topographic con-tours .

2 . The area enclosed by the drainage divide ismeasured in square miles .

3 . The area of stratified drift contained withinthe drainage divide is measured in squaremiles . (The area of till-mantled bedrock isequal to the total drainage area less the areaof stratified drift) .

4 . Q7,1o is estimated by use of equation 4 .

Windham-Cobbetts Pond Aquifer

Water-laid deposits of sand and gravel fill anarrow, linear valley immediately west and to thesouth of Cobbetts Pond in Windham . This stratified-drift aquifer is currently undeveloped and wasselected for evaluation because of its potential im-portance for public water supply. The location ofthis valley aquifer is shown in figure 16 . The finite-difference grid used to model this area consisted of67 columns and 26 rows with uniform cell dimensionsof 200 ft on a side (fig . 15) . Only the cells on theaquifer were considered "active" and were involvedin the numerical computations . The total area ofaquifer represented by active cells was 1.14 mil.

Boundary conditions

The eastern and western borders of the aquiferare defined by the contact between till or bedrockand stratified drift . This geologic contact was represented in the model by a "no-flow" boundary . Al-though termed a "no-flow" boundary, some flow doesoccur between the bedrock and stratified-driftaquifers and it may be significant in places . Resultsfrom an inventory of public-supply wells indicatethat bedrock wells account for approximately one-fifth of the total water pumped from the study area(table 2) . Because the bedrock aquifer can be asignificant water producer, the flux across thebedrock stratified-drift boundary can be an impor-tant component of the total recharge to thestratified-drift aquifer . The exact location, direc-tion, and quantity of flow between these aquifers willvary depending, in part, on the prevalence of water-bearing fractures in the rock and the absence of tillsemi-confining layers between aquifers . Becausethis investigation focused on the geohydrology ofstratified-drift aquifers little is known about thehydraulic properties of bedrock and till and the ef-fect on water availability in the stratified-driftaquifer . In a superposition model, the no-flowboundary assumes that pumping will not create addi-tional flow across the boundary. This condition wasnot strictly met in the model because slight (3 ft orless) drawdowns produced by pumping reached themodel "no-flow" boundaries in places . The modelresults, therefore, are slightly conservative in theestimate of aquifer potential yield and the calculatedcontributing area is slightly larger than what wouldbe estimated if the model were simulated such that

pumping could create additional flow across theboundaries .

Natural flux across the till or bedrock, strat-ified-drift no-flow boundary was accounted for inthe model because the map of initial water-tablealtitudes (fig . 16), upon which drawdowns are super-imposed, indicates the natural lateral flow from thetill-covered valley walls . The model boundary wasarbitrarily terminated southwest of Simpson Pond.This termination corresponds to a southeastern edgeof the aquifer that is beyond the reach of drawdownscalculated by the model and, therefore, was simu-lated as a no-flow boundary.

Three ponds, Cobbetts Pond in the north-eastern corner and Rock and Simpson Ponds in thesoutheastern corner of the model, overlie part of theaquifer . The ponds were simulated in the model ashead-dependent flux cells . This type of simulationallows leakage across a semipermeable pond bottomin response to a head gradient between water in thepond and in the aquifer .

The underlying till-bedrock surface was simu-lated as a no-flow boundary . Some leakage probablyoccurs across this boundary, but it was assumed to besmall enough to be considered negligible in a waterbalance for the aquifer .

Aquifer parameters

The parameters of hydraulic conductivity,water-table altitude, aquifer-bottom elevation, andspecific yield were assigned to each cell in themodel . Appropriate values of hydraulic conduc-tivity were determined from the maps of transmis-sivity and saturated thickness (pl . 4) . The gridnetwork is superimposed over these maps, andparameter values were calculated for the respectivecells . For example, hydraulic conductivity was com-puted by dividing transmissivity by saturated thick-ness . Altitudes of the water-table and pond surfacewere set to an arbitrary initial value of zero feet,which created a continuous flat boundary suitablefor analysis by superposition . Aquifer-bottom eleva-tion was equivalent to the negative of the saturatedthickness at each cell .

Parameters describing the hydrologic proper-ties of the pond bottom were assigned to cells in themodel that correspond to locations of Cobbetts andSimpson Ponds. Grid cells in which leakage betweenthe pond and aquifer was simulated are shown infigure 16 . Depths of the ponds range from 10 to 20 ftand were determined from bathymetric data ob-tained from seismic-reflection traverses across the

42°47'30"

42°45'

71°20'

////0/ lV?//'//J

.""k;

4 `/

r. zBase from U .S . Geological Survey

CONTOUR INTERVAL 20 FEETWindham, N.H . 1 :24,000, 1953,photorevised 1985

NATIONAL GEODETIC VERTICAL DATUM OF 1929

0

71 ° 17'30"

MODEL CELL

MODEL BOUNDARY

39

12 KILOMETERS

2 MILES

Figure 15 .--Finite-difference grid used to discretize the Windham-Cobbetts Pond aquifer .

0 1EXPLANATION

STRATIFIED DRIFT Q

7-7-71// TILL AND BEDROCK LL

POND CELLS

Yv"'yY/

iR~?IX/ A,

gd6fn'(////~F/!J1y

fry

,x,cx3s° v

,rV

Y//r/;

/rr/ ii°.

fQ6~~

/f//' ~r

10/

'ARCr

%,~// fpoi>~fl

y,0r

f/

ilskfr -

Base from U .S . Geological Survey

CONTOUR INTERVAL 20 FEETWindham, N .B . 1 :24,000, 1953,photore

/O/Z.

Figure 16.--Initial water table for the Windham-Cobbetts Pond simulation .

ised 1985

0

0

NATIONAL GEODETIC VERTICAL DATUM OF 1929

1 2 MILES

1 2 KILOMETERS

EXPLANATION

STRATIFIED DRIFT -11 MODEL BOUNDARY

TILL AND BEDROCK -170- WATER-TABLE CONTOUR--Shows approximateaverage altitude of the water table under naturalconditionS in 1987

VerticalCqnt ur interval 10 feet .

POND CELLS National Geodetic Datum of 1929

ponds . Figure 13 shows a typical seismic-reflectionprofile for a traverse across the western shore ofCobbetts Pond in a region of hydrologic connectionbetween pond and aquifer . Hydraulic conductivityof the pond bottom was assumed to be 2 ft/d), andthe thickness of the material was assumed to be 2 ft .These estimates were based on values calculated forriverbed deposits (Haeni, 1978; de Lima, 1989), andwere assumed to be representative of the pond-bot-tom hydrology .

The final parameter, specific yield, was es-timated to be 0.2 for this unconfined aquifer and wasassigned to the entire model area .

Application of numerical model

Five wells were simulated in the model andpumped at various rates until the limits for draw-down were attained at the end of 180 days . Fivewells were needed because the model cells that con-tained the pumped wells went dry if the total dis-charge simulated was distributed among fewer wells .The locations of the wells and resultant drawdownsare shown in figure 17 . The combined pumping rateof the five wells, 0.64 Mgal/d, represents the es-timated potential yield of the basin . The potentialyield as used in this report describes the rate atwhich water can be withdrawn from a basin underconditions of pumping for 180 days without rechargeand without having adverse impacts on the waterresource (see glossary) .

Sensitivity analysis

A sensitivity analysis of some of the modelparameters was done to show the effect theseparameters have on estimates of potential yield . Theanalysis determined the relative importance of inputdata values on calculations of potential yield andprovided a basis for assessing uncertainty in thesimulations given the likely range in each value .

The principal input parameters of aquiferhydraulic conductivity, pond-bottom conductance,specific yield, and duration of pumping were independently increased and decreased by a constantfactor throughout the modeled area while otherparameters were left unchanged . A referencesimulation was selected to represent the best es-timate of the hydrologic properties of the aquifer .This reference serves as a standard from which sub-sequent simulations with different input values canbe compared . The amount of adjustment of each

41

parameter differed according to the likely range ofeach parameter .

The results of the analysis of each change inparameter value are shown in figures 18 and 19 foran east-west profile along column 44 (fig. 15) . Theprofile includes the hypothetical production well oflargest simulated discharge (0.135 Mgal/d) and rep-resents an area of the model (column 44, row 11)where the greatest difference in drawdowns oc-curred between simulations . The computed per-centage changes in storage and leakage and thedrawdowns for each simulation are listed for com-parison in table 5 . It should be emphasized thatthese data represent changes in two of three com-ponents of recharge, aquifer storage, and pond andriver leakage . Not represented is the change inrecharge from till and bedrock uplands, which iscommonly simulated in numerical ground-water flowmodels . The superposition technique, however,simplifies the flow regime by simulating the till-aquifer boundary as "no flow", where additional flowis not induced across this boundary . The tabulatedresults, therefore, are intended only to show relativedifferences in the simulations .

Aquifer hydraulic-conductivity (Ka) valueswere multiplied by factors of 1 (model standard), 2,and 0.5 in three separate simulations . Doubling thehydraulic conductivity of the aquifer resulted in a32-percent overall reduction in drawdown atproduction wells and a decrease in the naturalground-water flow to the ponds . The reduction indrawdown extended generally to a distance of1,500 ft from well locations, beyond which draw-downs were slightly greater than at correspondingcells in the model standard.

Reducing the hydraulic conductivity by 0.5 hadthe greatest effect on drawdown at well locations forthese simulations . Overall, drawdown increased by57 percent at production wells, and a greater per-centage of water came from aquifer storage . Draw-downs decreased slightly relative to the modelstandard near the till-aquifer boundaries .

Pond-bottom conductance (Kb) values weremultiplied by factors of 1 (model standard), 10, and0.1 in three separate simulations . These changesresulted in computed heads that were identical tothose of the model standard . No change in theproportions of water derived from the two sources(table 5) was observed . The insensitivity of themodel shows that pond leakage is not the primarycontrol on the availability of water to the simulatedproduction wells, even though one simulatedproduction well was within 600 ft of Cobbetts Pond.The results are probably a reflection of (1) limited