LONG-TERM CHANGES IN REGIONAL HYDROLOGIC REGIME

FOLLOWING IMPOUNDMENT IN A HUMID-CLIMATE WATERSHED

Francis J. MagilliganDepartment of Geography

Dartmouth CollegeHanover NH 03755

e-mail: magilligan@dartmouth.edu

and

Keith NislowNortheast Research StationUSDA-US Forest Service

Amherst, MA 01003e-mail: nislow@dartmouth.edu

FORTHCOMING IN:

JOURNAL AMERICAN WATER RESOURCES ASSOCIATION

ABSTRACT

We analyzed the type of hydrologic adjustments resulting from flowregulation across a range of dam types, distributed throughout the ConnecticutRiver watershed, using two approaches: 1) the Index of Hydrologic Alteration(IHA) and 2) log-Pearson Type III flood frequency analysis. We applied theseanalyses to seven rivers that have extensive pre- and post-disturbance flowrecords and to six rivers that have only long post-regulation flow records. Lastly,we analyzed six unregulated streams to establish the regional natural flow regimeand to test whether it has changed significantly over time in the context of anincrease in forest cover from < 20 % historically to > 80% at present. We foundsignificant hydrologic adjustments associated with both impoundments and landuse change. On average, maximum peak flows decrease by 32% in impoundedrivers, but the effect decreases with increasing flow duration. 1-day minimumlow flows increase following regulation, except for the hydro-electric facility onthe mainstem. Hydrograph reversals occur more commonly now on themainstem, but the tributary flood control structures experience diminishedreversals. Major shifts in flood frequency occur, with the largest effect occurringdownstream of tributary flood control impoundments, and less so downstream ofthe mainstem's hydro-electric facility. These overall results indicate that thehydrologic impacts of dams in humid environments can be as significant as thosefor large, multiple-purpose reservoirs in more arid environments.

KEYWORDS: dams, floods, low flow, watershed management, aquatic

ecosystems.

INTRODUCTION

Reservoirs, dams, irrigation flowdiversions, and flood control structures havebeen common developments over the pastfive decades to the extent that relatively fewlarge rivers are unaffected by thesealterations (Benke, 1990; Graf, 1999).Reservoirs vary in size and function, sogeneralizations are difficult (Power et al.,1996). Dams, though, generate significanthydro-geomorphic alterations with impactsoccurring both upstream and downstream ofreservoirs (Williams and Wolman, 1984;Collier et al., 1996; Hadley and Emmett,1998), and these hydro-geomorphic impactshave profoundly altered and degraded riverecosystems (cf. Ward and Stanford, 1995;Bovee, 1986; Ligon et. al., 1995; Power etal., 1996). In order to mitigate theenvironmental consequences of flowregulation, numerous instream flowmeasures have been suggested to sustain andmaintain ecological integrity along a streamreach. These measures include establishingand maintaining minimum flow releases(Colby, 1990) or permitting controlled“flushing” releases that establish thenecessary high flows for sediment transport(Williams, 1996; Collier et al., 1996;Stanford et al., 1996; Collier et al., 1997;Rubin et al., 1998). However, controlledflushing flows may not serve to enhance theentire spectrum of ecological orgeomorphological conditions and thereforemay not be a system-wide solution (Kondolfand Wilcox, 1996; Wu, 2000). Instead oftailoring flows for particular species or flowrequirements, Power et al. (1996) suggestthat ecological integrity is best preserved byregulation structures and managementschemes that approximate the natural flowregime inherent prior to flow managementand disturbance (cf. Stanford et al., 1996;

Poff et al., 1997). Determining the effectsof impoundments on this natural regime istherefore critical for restoration andconservation of aquatic ecosystems.

While previous studies havedocumented significant changes inhydrologic and sediment regimes before andafter impoundment (Graf, 1980; Williamsand Wolman, 1984; Chien, 1985; Andrews,1986; Elliott and Parker, 1997; Hadley andEmmett, 1998; and Van Steeter and Pitlick,1998), this literature has some importantlimitations. Specifically, most previousresearch efforts have examined single case-studies of pre- and post-impoundmentconditions in large, water storage reservoirsin semi-arid Western United Stateswatersheds. Impoundment effects underthese conditions are likely to differsignificantly from effects of other reservoirtypes (for example flood-control dams), andfor other, more mesic geographic regions.Focus on single case-studies also precludesintegrated assessment of impoundmenteffects on drainage-wide or region-widescales, where a range of reservoirs typesexist. Finally, focusing on single casestudies of pre-and post impoundment oftenlacks appropriate spatial controls (i.e.physically comparable, unimpoundedrivers). Without these controls, long-termchanges in flow regimes caused by land useor climate change may be incorrectlysingularly attributed to impoundment.

This paper details the hydro-geomorphic adjustments occurring withinthe Upper Connecticut River watershed as afunction of long-standing and widespreadflow regulation. Our basic goal is todetermine the pre-impact and on-going“natural” flow regime for the UpperConnecticut River watershed across a range

of basin sizes, and to determine whathydrologic features have been most affectedby flow regulation. The wide range andmagnitude of flow regulation, and the densenetwork of stream gages with long periodsof record, make the Connecticut basin anexcellent study system to evaluate thesegoals. We applied two techniques, the Indexof Hydrologic Alteration (Richter et al.,1996) and log-Pearson Type III floodfrequency analysis, to nineteen gages withdaily discharge data having an averagelength of record of ~ 50 years. We appliedthese methods to determine if reservoirshave significantly altered the hydrologicregime, and if so, what hydrologic indexbest captures and expresses that disturbance.Secondly, in order to determine if post-reservoir changes resulted singularly fromthe impacts of reservoirs, we examined thehydrologic characteristics for undisturbedstreams to ascertain the “natural”background conditions of those site-specificand regional changes.

STUDY AREA

For the purposes of this study, we aredescribing the Upper Connecticut Riverwatershed as the section upstream of theVermont/Massachusetts border andcontinuing upstream to North Stratford, NH(Fig. 1). This section is an ecologicallysensitive region. Three federally-listedendangered species occur along a 15 kmreach downstream of Wilder Dam in WestLebanon, NH to approximately Hart Island,VT: the dwarf wedge mussel (Alasmidontaheterodon), Jesup's milk vetch (Astragalusrobinsii var. jessupi), and the cobblestonetiger beetle (Cincindela marginipennis).Furthermore, floodplain forests havedeclined in quality and extent in part due toseveral centuries of agricultural disturbance,but also due to reductions in overbankflooding resulting from flow regulation

(Bechtel and Sperduto, 1998). Thoseremaining floodplain forests generallyconsist of single-age stands with few youngtrees (Bechtel, pers. comm.), a patternstrongly associated in other regions withdiminished magnitude and frequency oflarge floods (Molles et al., 1998).

The area is lithologically diverse,and in the West Lebanon, NH section itconsists primarily of intensely folded meta-sedimentary and meta-volcanic rocksranging in metamorphic grade. Lowtemperature biotite-garnet zones occur alongthe mainstem of the Connecticut River withprogressively higher grade metamorphicstrending westward from the ConnecticutRiver (Lyons, 1958). Granitic intrusionsalso commonly exist, and most of the regionis mantled by a relatively thick till coverexcept at the higher elevations or along thesteep hillsides. The region is highlydissected with most of the incision occurringprior to Quaternary glaciation (Denny,1982). The main valley of the ConnecticutRiver is relatively narrow, and thenarrowing is exacerbated by a series of late-Pleistocene and Holocene terraces insetwithin the already constrained valleys.

The region possesses a continentalhumid climate with precipitation averagingapproximately 900 mm/yr with a relativelyeven distribution throughout the year. Theeven precipitation pattern, however, doesnot manifest in an even flood distribution.New England possesses a mixed populationflood regime with several flood producingmechanisms present: rain (sometimes ashurricanes), rain-on-snow/frozen ground,and snowmelt (Magilligan and Graber,1996). Large floods have ravaged NewEngland throughout the early to mid-twentieth century with catastrophic floodsoccurring in 1927, 1936, 1938, 1954, 1955,1969, and 1987 (Jahns, 1947; Patton, 1988).

Because of the intensity and frequency offloods in the region, a series of flood-controldams were constructed in the 1950s and1960s on some of the major tributaries of theConnecticut River. Conversely, most of thelarge dams on the mainstem of theConnecticut River are hydropower dams thatwere constructed in the early and middlepart of the century. These hydropower damsare operated by various public and privateutilities and contain a mixture of storagereservoirs and run-of-the river powergeneration facilities (cf. Fallon-Lambert,1998). The region possesses one of thehighest frequencies of dams per area in theUnited States; however, the storage impactsare not as great as other regions where largerstructures exist and where climates are morearid (Graf, 1999).

USGS stream monitoring beganearly in New England with numerous gageshaving records beginning in the 1910s. Thislengthy record coincides with other changesin the New England landscape, primarily theshift away from agriculture and grazing(both cattle and sheep). The peak of landclearing occurred in the late 1800s and hasdeclined dramatically over the 20th Century,with many parts of the region currently attheir highest levels of forest cover in wellover one hundred years (Fig. 2). Despitethis widespread reforestation, on-goingurbanization and scattered agriculturalactivity, combined with lag effects of basinrecovery (cf. Magilligan and Stamp, 1997),certainly limit pre-disturbancecharacteristics from being completelyattained. However, modern flowcharacteristics for unregulated and regulatedstreams differ considerably from their early20th Century counterparts due to thisregionally extensive reforestation. In someinstances, the modern regime in manyunregulated streams may approximate theirpre-historical flow conditions.

METHODOLOGY

The dominant focus of this study isto document the impacts of flow regulationby dams, and we have accomplished thisprimarily by analyzing stream gage data forseven rivers that have lengthy pre-impoundment flow data combined withextensive post-impoundment data. One ofthese dams, the Wilder Dam at WestLebanon, NH, is a hydro-electric facility onthe mainstem Connecticut River, while theremaining six sites occur on the tributariesand are primarily flood-control structures(Fig. 1, Table 1). In order to broaden thescope of impoundment impacts, we alsoanalyzed six sites that have USGSmonitoring over the period of impoundment,but that lack a comparative pre-impoundment flow record. These sitescannot be compared to a pre-disturbanceflow regime but can be used as replicates ofpost-impoundment impacts. These sitesgenerally have long gaging records thatcapture long-standing hydrologic changestypical of flow regulation and include abroader range of dam types and contributingbasin area. Two more hydro-electric damson the mainstem of the Connecticut river areincluded: one 60 km upstream of the WilderDam at West Lebanon (the ConnecticutRiver at Wells River), and the Bellows FallsDam at North Walpole, NH approximately60 km downstream of the Wilder Dam.Lastly, in an attempt to characterize long-term hydrologic changes occurringthroughout the 20th Century by climatic andland use controls, we analyzed the gage datafor six unregulated tributary sites. The mainsite for this part of the analysis is the WhiteRiver, VT, which is the largest free flowingmonitored river in the Upper ConnecticutRiver basin (1,757 km2) and also possessesone of the longest USGS gage records,dating back to 1915. It drains the southern

and eastern flanks of the Green Mountainsand empties directly into the ConnecticutRiver, less than 4 km downstream of theWilder Dam in West Lebanon, NH. Wehave analyzed these records for all gagedsites using two distinct methodologies: TheIndex of Hydrologic Alteration and log-Pearson Type III flood frequency analysis.

Index of Hydrologic Alteration

Flow regulation can affect thetiming, magnitude, duration, and frequencyof flows. We have used the Index ofHydrologic Alteration developed by Richteret al. (1996) to quantify these potentialimpacts and applied it to nineteen streamgage records throughout the UpperConnecticut River watershed. This modeluses mean daily discharges and calculates 32indices that describe the hydrologic regimefor that station (Table 2). The thirty-twoindices generated by IHA consist of fivemajor categories : (1) magnitude; (2)magnitude and duration of annual extremeconditions; (3) timing of annual extremeconditions; (4) frequency and duration ofhigh and low pulses; and (5) rate andfrequency of changes in conditions. Inessence, the model evaluates changes inboth minima and maxima, and alsosynthesizes and groups these two extremesover several temporal scales (1-day, 3-day,7-day, 30-day, and 90-day). The methoddoes not characterize the storm hydrographper se, but instead calculates the hydrographrise and fall based on the mean dailydischarges, and determines the number ofrises and falls for a given year. The IHAexpresses hydrologic variability within theyear by calculating the number of times thesequence of daily discharges changes from arising hydrograph to a falling hydrograph,and vice versa. These reversals can occuron daily, weekly, or seasonal scales anddepend upon broad hydro-climatological

conditions reflecting regional flood-producing mechanisms and basincharacteristics. For large basins wheresnowmelt floods dominate the hydrologicregime, spring runoff will be a longcontinuous hydrograph rise with littlefluctuation. In smaller streams or insituations where an interrupted sequence offrontal storms drives runoff, the number ofhydrograph rises and falls will bemaximized. The model also expresses theduration of high and low events anddetermines, for a given year, the number oftimes (i.e. “pulses”) that the 75th percentileis exceeded and the number of times that the25th percentile is not attained. It alsocharacterizes the length of the pulsesexceeding the representative quartiles. Allof the indices of the IHA were alsoregressed against time (in years) todetermine if any significant serial trendsexist due to climatic or land use impacts.Precipitation data came from the Hanover,NH gage which has a lengthy series thatcovers the entire record of streamflow data.

Flood Frequency Analysis

Most of the attention on hydrologicparameters generally addresses in-channelflows (1-day minimum, hydrographreversals, etc.), yet many other ecologicalparameters within the riparian zone dependupon the occurrence of annual maximumevents. For lowland sites in the ConnecticutRiver basin, the condition and maintenanceof floodplain forests is a critical issue.Many of the floodplain forests have beendisturbed or destroyed by logging,agriculture, or urbanization throughout the19th and 20th Century; however, less directcauses may also explain their demise and/ordegradation. Many of the floral and faunalspecies in these riparian-to-terraceenvironments require some modicum ofdisturbance to keep out competitors, to bring

in nutrients, or to enhance seed dispersal(Bayley, 1995; Hughes, 1997). Floods arethe usual mechanism to effect theseconditions, and flow management byimpoundments has greatly reduced themagnitude and frequency of floods in theUpper Connecticut River basin. In order toisolate the changes in flood frequency due toflow regulation, we used a log-Pearson TypeIII analysis for the annual maximum floodseries for each gage and calculated theestimated 2-yr, 5-yr, 10-yr, 20-yr, 100-yr,and 200-yr discharges for all nineteen gages(U.S. Water Resources Council, 1976). Ifthe station had a pre- and post-disturbanceflow record, we calculated these estimatedflood frequencies for the pre- and post-disturbance record to show the hydrologicshifts in out-of-bank discharges as a functionof regulation. These comparisonsdemonstrate, in a probabilistic sense, thechanges that occur in a stream’s ability toinundate overbank areas as well asindicating the shifts in the bankfulldischarge, which is the dominant dischargeassociated with channel formation andsediment transport and strongly correspondsto the 2-yr flood (Wolman and Miller, 1960;Andrews, 1980). In addition to thesecomparisons, data from the flood frequencyanalysis were used to establish a regionalbankfull flood frequency curve forunregulated rivers (Fig. 3). This curve wasthe basis for establishing a critical aspect ofthe natural flow regime, and wassubsequently used to compare bankfull floodfrequencies of impounded rivers.

RESULTS

Hydrologic impacts followingimpoundment

Discharge Variation for High Flows For the Upper Connecticut River

basin the dominant impact of regulation

appears to manifest in the magnitude andduration of extreme conditions. The lack ofimportance of the other IHA parameters maybe a function of the humid climate of thenortheastern US and/or result from the typesof dams present in the region. For example,alterations in the timing of high flows havenot necessarily occurred followingimpoundment in the Upper ConnecticutRiver basin (Fig. 4). This lack of effectcontrasts strongly with impoundmentimpacts in the arid and semi-aridsouthwestern US where snowmelt from themountains is stored for subsequent laterrelease. Furthermore, in five of the sevenbasins, the date of the annual maximum flowis somewhat earlier (2-26 days earlier), butbecause of the large range in the standarddeviations, the trend is not statisticallysignificant.

The most significant changesemerged in the magnitude of high flowsfollowing impoundment. On average, 1-daymaximum flows decrease by approximately32% following flow regulation. Six of theseven sites are flood control structures onmajor tributaries and are therefore designedto contain high flows. However, thedifference in high flows decreases as thetemporal window expands (Fig. 5). At thetemporal scale of the month to the seasonal(i.e. 90-day), no significant differences existfollowing impoundment. The average rangein the annual maximum flood decreases byapproximately 32% and is fairly consistentacross dam type and contributing basin size,except for the Ompompanoosuc River inVermont which experienced only a 20%reduction (Fig. 6). This river has one of thesmallest drainage areas (326 km2) of all oursites and also has a different managementoperation. The dam operates as a run-of-the-river facility from May to November,but maintains a 6 m pool elevation fromNovember to May. Conversely, the North

Hartland Dam on the Ottaquechee River is acombined flood control and recreationalfacility. It has a similar contributingwatershed size (570 km2) as theOmpompanoosuc River, yet it has a greaterreduction in the average of the annualmaximum floods. Because of itsrecreational function, it maintains a constant10.6 m pool elevation throughout the year.

Discharge Variation for Low FlowsThe magnitude, duration, and timing

of low flows control an array of ecologicalfunctions (Poff et al., 1996), and changes inlow flows following impoundment can be assevere as changes in high flows (Hadley andEmmett, 1998). Many of the trends evidentin high flows, though, do not extend todifferences in low flow conditions for thesame time steps, as strong differences existin the response of tributary rivers and themainstem Connecticut River. Tributariesgenerally experienced increases in 1-dayminimum flows following impoundment(Fig. 7), while 1-day minimum flowsstrongly decreased following impoundmentin the mainstem Connecticut (Fig. 7). Thesedifferences in response likely result fromdifferences in dam type and management,and/or basin size. The reduced 1-dayminimum flows may also reflect the lack ofgovernment control on minimum releases inthe early part of the post-impoundmentrecord. Government-mandated minimumflow releases did not occur until the later1970s, and the severely dry years of the late1950s and early 1960s corresponded withunusually low releases from the Wilder Damin West Lebanon, NH (Fig. 8). The need tostore water for hydropower generation andthe occurrence of an unusually dry climateof the 1960s explain the differing pattern forthe Connecticut River at West Lebanonalthough the extensive low flows pre-datethe regional drought. For the tributaries,increased 1-day minimum flows following

impoundment may correspond to the naturalchanges tending to occur over time inunregulated streams. Thus, for minimumflows, the impact of flow regulation on thetributaries appears to mimic currentlyoccurring conditions in free flowing streams(see subsequent section "Hydrologic regimein unregulated rivers").

No trend exists in the average date ofminimum flows as four of the seven basinshave average minimum flows occurringearlier in the year (Fig. 4). The wide rangein the standard deviations precludes anystatistical significance. However, themagnitude of the extreme low flow minima(i.e. 1-day) tends to increase followingimpoundment (Fig. 7), but the trends are lessdecisive as the duration (e.g. 3-day, 7-day,etc.) increases. For the mainstemConnecticut River, low flows for alldurations decrease following impoundment,which may either reflect its basin size,operation function (i.e. hydropower), and/ormanagement history. Post-impoundmentlow flows increased for all durations for theBlack River and West River, but the effect isonly statistically significant for the 90-dayduration for the Black River and for the 3-day minimum flow for the West River. Ingeneral, low flows for all tributary streamstend to increase for all durations followingimpoundment, but the large variance overthe time series limits any definitiveconclusions. For the Connecticut River, lowflows decrease following impoundment forall durations, but the decrease is onlystatistically significant for the 1-day and 3-day low flow durations.

Hydrograph ChangesFor impoundments within the Upper

Connecticut River watershed, hydrographreversals tend to decline, except for theWilder Dam on the Connecticut River whichhas a 30% average increase in the number of

hydrograph reversals (Fig. 9). Trendsevident across the sites again probablyreflect dam type and perhaps, to a lesserdegree, contributing watershed area. Thesummer run-of-the-river managementoperation of the Union Village Dam on theOmpompanoosuc River shows the leastchange in hydrograph reversals. Thismanagement perhaps best mimics theregional natural flow regime as it permitsmore reversals in the summer (typical ofrain-induced runoff punctuating summerstreamflow drawdown) and less reversals inthe late spring (typically a long hydrographrise season) Conversely, the ConnecticutRiver below Wilder Dam has become more“flashy” following impoundment due to itsmanagement as a hydropower facility (withsome flood control). Prior to impoundmentthe Connecticut River at West Lebanon hada typical large watershed, snowmelt-dominated hydrologic regime averaging 108reversals per year (Fig. 9). Subsequent toimpoundment, the number of reversals hasincreased to approximately 160 per year.

Hydrologic Adjustments at OtherImpounded Sites

Hydrologic impacts for the twoadditional hydro-electric sites on theConnecticut River (Table 1) correspond tothe magnitude, type, and direction of shiftsthat exist for the post-impoundment data forthe Wilder Dam on the Connecticut River.Even though the Bellow Falls Site is 60 kmdownstream of the Wilder Dam with anadditional 5,000 km2 of contributing basinarea, its average post-impoundment 1-dayand 3-day minimum flows (Table 3) are wellbelow that of the pre-impoundment 1-dayand 3-day minimum flows for the WilderDam. The trend reverses for low flowdurations of 7-days or greater. Also, if themagnitudes and shifts followingimpoundment of the Wilder Dam are

accurate, then the low flows at the twoadditional sites on the Connecticut River areprobably less than their pre-impoundmentnatural low flows. The post-impoundmentJulian dates for the minimum and maximumflows agree fairly closely amongst the threesites on the Connecticut River, with thedates for the Connecticut River at WellsRiver being almost identical with those atthe downstream Wilder Dam (Table 3 andFig. 4). Furthermore, flow regulation forthese hydropower facilities generates thesame number of reversals evident at theWilder Dam suggesting that the post-impoundment “flashiness” of thehydrograph is a mainstem phenomenon.

Results from the tributaryimpoundments are slightly more difficult tointerpret as they lack specific analogs and/orpre-impoundment flow records. However,when compared to pre- and post-impoundment flow data for similar sizebasins, generally similar impacts result. Thedischarge changes do not necessarily varysequentially with changes in drainage area,which may reflect differences in dammanagement operation. The minimumflows for the West River for the 1-daythrough the 30-day flow durations are lowerthan the releases for the Mascoma Riverwhich has a smaller contributing drainagearea. The Mascoma River reservoir is acombined water supply and recreationfacility while the reservoir on the WestRiver is a U.S. Army Corps of Engineersflood control structure which maintains aconstant pool elevation. Furthermore, theflood control structure on the Ashuelot Riverhas unusually low releases for all low flowdurations.

Mean 1-day maximum flowsfollowing impoundment for these sitesgenerally correspond to post-impoundmentchanges for those sites with a pre- and post-

impoundment flow records. The sites inTable 3 do not have pre-impoundmentrecords for comparison, however data forfree-flowing rivers can be used as surrogatesof pre-impoundment flows if differences indrainage area can be adjusted for. Thequotient of peak flows and drainage area(i.e. specific discharge) for free-flowingrivers and for post-impoundment 1-daymaximum flows (Fig. 10) resembles thesame pattern for those sites with extensivepre- and post impoundment records,indicating that peak discharges per squarekilometer have been reduced on the order ofnine percent.

Impoundment Impacts on FloodFrequency

The reduction in the peak dischargesdue to flow regulation (Fig. 6) can beexpressed by adjustments in the probabilityof achieving a flow of a specific magnitude.Dam type, location, and management cansignificantly modify the natural magnitude-frequency relationships. For alluvialsurfaces containing floodplain forests, theseshifts in magnitude-frequency relationshipsmanifest in a decreased occurrence (in aprobabilistic sense) of inundation. Most ofthe alluvial surfaces containing floodplainforests occur within the main valley of theConnecticut River, and may exist as barsand islands, and/or may be on low-lying latePleistocene and Holocene terraces. Flowregulation by Wilder Dam on the mainstemof the Connecticut River has greatlymodified the natural magnitude-frequencyrelationship which hinders the frequency ofinundation of these riparian surfaces. Log-Pearson Type III analyses for the pre-impoundment flow record generates a 2-yrbankfull discharge of 1,370 m3/s whichcorrelates perfectly with the regional floodfrequency analysis for bankfull dischargesfor free-flowing rivers (Fig. 3). Comparison

of pre- and post-impoundment floodfrequencies indicates that major reductionshave occurred with the impacts becomingespecially acute for floods with a 10-yr orgreater recurrence interval (Fig. 11A). Thepre- and post-impoundment 2-yr bankfulldischarges are relatively close indicatingthat the flow release strategies currently inuse have not affected the magnitude of thebankfull discharge. The dischargedifferences become progressively larger asthe recurrence intervals increase. Many ofthe floodplain forest communities existbeyond the 10-yr floodplain, and thesesurfaces have a post-impoundmentfrequency of inundation that would requirethe post-impoundment 100-yr flood.

The shifts in magnitude-frequencyrelationships are even more dramatic forsmaller watersheds, irrespective ofmanagement style. Flow regulation on theBlack River in southern Vermont has sosignificantly altered the flood frequencyrelationships that the pre-impoundment 2-yrbankfull discharge of 158 m3/s exceeds thepost-impoundment 200-yr flood (Fig. 11B).Moreover, the largest discharge releasedfollowing impoundment (117 m3/s) wouldhave only corresponded to the pre-impoundment 2-yr flood. The UnionVillage Dam on the Ompompanoosuc Riveroperates as a run-of the-river facility duringthe summer months and permits some largerflows released. However, the impact ofregulation on flood frequency for thismanagement style (Fig. 11C) mimicsmanagement where a constant poolelevation is maintained (Fig. 11B). As istrue for the Black River, the largest recordedoutflow for the Union Village Dam on theOmpompanoosuc River (65 m3/s) is belowthat of the pre-impoundment bankfulldischarge.

Besides major changes in floodfrequencies, examination of actual releasesfollowing impoundment reveals significantreductions in out-of-bank discharges. The2-yr channel forming discharge rarelyoccurs, especially on the tributaries withflood-control structures (Table 4). The onemajor exception is the Wilder Dam at WestLebanon. This hydro-electric facility has amaximum operating range of 1.5 m withminimal reservoir storage which results innumerous large releases especially duringspring runoff (Fallon-Lambert, 1998). Thepeak discharge of 2,230 m3/s during the1973 flood generated an eventcorresponding to the pre-dam 10-yr flood.Generally, however, post-dam maximumreleases barely achieve pre-dam bankfulldischarges. The data for this analysis can beenhanced by including those gages with onlypost-impoundment records. Although thesesites lack pre-impoundment flow data, it ispossible to estimate the shifts in out-of-bankoccurrences. The regional drainage area-bankfull discharge relationship (Fig. 3) forunregulated rivers was used to estimate thenatural 2-yr flood at each site based on itsdrainage area. The inclusion of these sitesshows that mainstem releases from thehydro-electric facilities commonly achievepre-dam bankfull discharges (Table 4). Thetributary sites exhibiting similarrelationships to pre-dam channel-formingdischarges are those where flood control isnot the sole nor dominant function.

Hydrologic Regime in Unregulated Rivers

Hydrologic shifts have occurredthroughout the 20th Century in all the gaged,unregulated streams in the region (Table 5)and are best demonstrated by the gagerecord of the White River in east-centralVermont (Fig. 1). Peak flows decline overtime, but the trend is not statisticallysignificant (0.10 > α > 0.05). The early part

of the flood series was characterized byextremely large floods, in part due to themoderately high levels of agriculture stillpresent in the watershed, but also due to theoccurrence of an unusual series of climaticevents. The three largest floods in aneighty-year record occurred within an elevenyear period at the beginning of the record.An informal crest-stage recorder on theConnecticut River in Hartford, Connecticutshows that three of the largest four floods ina 200-yr record all occurred in a ten yearwindow in the early to middle part of the20th Century (Patton, 1988).

The changes in land use for theWhite River watershed show up morestrongly in low flow characteristics than inhigh flows. Low flow minima have steadilyrisen over the period of record (α < 0.05), asshown by the changes in the 1-day minimumflow (Fig. 12), although all low flowdurations (3-day, 7-day, etc.) show the sameincreasing trend (Table 5). The increasedlow flow over time probably reflects theincreased re-forestation occurring inVermont over the 20th Century, andcorresponds to the type of hydrologicchanges that occur following land usechange (Potter, 1991). This re-forestationalso influences the timing of low flows. Atthe early part of the record, the Julian date ofthe 1-day minimum flow was, on average,on the 266th day of the year (September23rd); whereas, the date of the 1-dayminimum flow has become progressivelyearlier over the gage record, with the 1-dayminimum flow now occurring, on average,on the 223rd day (August 11th) of the year(Fig. 13). This late summer/early fall timingof the minimum more closely approximatesthe pre-disturbance hydrologic regime thatbiota have evolved to expect for this region.The decrease in the date of the minimumflow reveals the increased regional rechargein conjunction with landscape re-vegetation.

Comparison of the mean annualhydrographs for the beginning and end ofthe record shows the different shapes of therecharge limb during the autumnal recharge,and also the different magnitudes of thesnowmelt signal (Fig. 14). Baseflow beginsearlier in the fall and also remains higherthan in the beginning of the record throughthe fall and into the winter runoff, in partdue to increases in soil infiltration. Therising limb of the snowmelt runoff peaksearlier and larger in the 1920s, potentiallydue to a greater exposure of the snowpack todirect insolation. This early and larger riseoccurs even though the average Marchprecipitation for the 1920s is smaller thanthe 1990s.

Trends expressed by the White Riverdata typify the other five non-regulatedstreams although their shorter gage recordslimit statistical significance. The trends,however, are generally in the same direction.The low flow shifts are especially evidentthroughout the region with all sites showingan increase in low flow over time, with threeof the six sites displaying significantincreases (Table 5). As with the WhiteRiver, basin wide recharge during the latesummer and early fall also occurs in theseunregulated rivers with mean monthlyrunoff for August and September flowsincreasing, with five of the six unregulatedstreams possessing a significant increase ineither August or September flows over time.The date of the minimum 1-day low flowsimilarly decreases over time. Hydrographreversals all show significant changes overtime, but not all the basins behavedsimilarly. Basin size may explain thedifferences. For basins greater than 25 km2

hydrograph reversals decreased significantlyover time indicating that their daily flowsare less flashy. Conversely, smaller basins(Ayers Brook and Mink Brook) havebecome more flashy over time. The

differences result in part because basins ofdiffering size respond differently to eithersnowmelt or rainfall induced events.

DISCUSSION OF RESULTS

These results portray acomprehensive pattern of hydrologicresponses across dam type, function, andcontributing basin size. The hydrologicresponses show up in changes in both lowflows and high flows. Thus, for the UpperConnecticut River watershed disturbance ofthe natural flow regime manifests mostsignificantly in the shifts in the extremes. Insituations where pre-impoundment and post-impoundment flow records exist, the post-impoundment changes vary as a function ofdam type and watershed scale with post-disturbance high flows greatly reduced insize and variance. The average reduction of32% in the 1-day peak flow correspondswith peak reductions noted elsewhere. VanSteeter and Pitlick (1998) noted a 28-39%decrease in the average peak discharge ofthe Colorado River and its major tributary,the Gunnison River. Williams and Wolman(1984) provided mean pre- and post-impoundment peak discharges for 48 sites,with all but one west of the MississippiRiver. A calculation of their meandifferences generates a mean reduction ofpeak discharges of 36% followingimpoundment (standard deviation = 70%),although some sites actually increased inmean peak discharge followingimpoundment. The wide range for theiranalysis results in part from theirincorporation of a wide array of dam types.Other problems emerged as their gages hadvariable flow records and were sometimeslocated long distances from theimpoundment. However, the comparison ofour results with those of others indicatesthat, on average, mean peak discharges tendto decrease by approximately 30-35%

following impoundment. For our sites, thechange also depended on dam management,with a diminished change for the seasonalrun-of-the-river operation of the UnionVillage Dam.

Ecological ImplicationsImpoundment impacts on peak

discharges are probably best represented bychanges in flow frequency. These frequencychanges have important ecologicalimplications in our study reach as themaintenance of floodplain forests dependsgreatly on some intermediate scale ofdisturbance. While decline and extirpationof riparian communities following theelimination of flooding has beendemonstrated in river systems throughoutNorth America (Ward and Stanford, 1995;Molles et al., 1998), almost nothing isknown about the effects of hydrologicalteration on these communities in NewEngland. Part of the problem is that floristicdifferences between upland and riparianforests are much less well defined inNortheastern mesic temperate zones thanthey are in the arid and semi-arid west. Forexample, sugar maple (Acer saccharum)may be an important component of bothforest types (Bechtel and Sperduto, 1998).Also, the long history of intensive land use,often particularly concentrated along rivercorridors, makes it difficult to determine theactual composition and spatial extent ofriparian communities in the absence ofintense human disturbance.

In spite of these uncertainties, ourresults point to several areas of concern.Differences in flood magnitude andfrequency associated with impoundmentmean that a given floodplain surface will beinundated less frequently, and that the totalspatial extent of inundated floodplain will beless for an event of a given post-impoundment recurrence interval. The first

effect is likely to have a major impact onriparian communities in New England viaeffects on competitive interactions betweenspecies. The dominant tree species onConnecticut River floodplains, silver maple(Acer saccharinum) is tolerant ofindundation, but intolerant of shading,suggesting that the species would be highlyvulnerable to replacement by shade tolerantspecies if flooding was eliminated (Petersonand Bazzaz, 1984). However, Bechtel andSperduto (1998), in a floodplain forestsurvey, found that silver maple-dominatedforests were found largely on the mainstemConnecticut River, within the 1-2 year post-impoundment floodplain while Metzler andDamman (1985) found silver maple forestswithin the 1-11 year floodplain in the lowerConnecticut River. We found thatimpoundment on the Connecticut Rivermainstem in the vicinity of the WestLebanon, NH stream gage, while having amajor effect on large (greater than or equalto 5-year) floods, had little effect on themagnitude of the 2-year discharge. Thissuggests relatively little effect of hydrologicalteration in the Connecticut basin on thisdominant riparian tree species. However,the extent to which silver maple formerlyoccurred along higher river terraces, oralong major tributaries, is unclear at thepresent time, making this conclusiontentative. Decreased frequency ofinundation is also likely to strongly affectherbaceous plant communities (Barnes,1978) which include a number of rare,threatened, and endangered species (Bechteland Sperduto, 1998). Many uplandherbaceous species in New England areintroduced species which would be expectedto thrive in the high light open canopyconditions of river floodplains in theabsence of frequent flooding. In fact, theabundance of these species has beenincreasing, and has been suggested to be amajor threat to existing populations of native

floodplain plants (Bechtel, pers.comm). Inaddition to changes in flood frequency,reduction in flooded area is likely to have amajor influence on floodplain communitiesalready fragmented and reduced in size as aresult of land use and development.Increased reduction and fragmentation, andreduced habitat connectivity, makesremaining floodplain communities morevulnerable to invasion and randomextinction.

Changes in the natural flow regimecan occur in other metrics besides changesin peak flows, and for some ecologicalparameters, shifts in low flows, timing ofhigh and low flows, or changes acrossdifferent flow durations can be even moresignificant. For the Upper ConnecticutRiver, changes in the timing of either low orhigh flows do not appear to be significant,but this influence may be region specific.Elliott and Parker (1997) found that monthlymean discharge during the April-Julysnowmelt season decreased 63% followingimpoundment, while mean monthly flowsfor the remainder of the year increased170%. This marked shift for semi-aridregions results because of broaderhydroclimatological controls on floodtimings and on dam management whereconsiderable volumes of water are stored forsubsequent release (Graf, 1999). For NewEngland, these shifts are less dramaticbecause of high annual precipitation totals,combined with the lack of a seasonalprecipitation signal, and the minimal waterstorage behind impoundments.

Geomorphic ImplicationsAlthough this work has focused

primarily on the hydrologic shifts due toflow regulation, these results point to anarray of potential geomorphic adjustmentsoccurring in the Upper Connecticut Riverwatershed. We did not analyze any of the

geomorphic impacts followingimpoundment due more to a lack ofavailable data rather than to a lack ofimpacts. Most studies on geomorphicimpacts chronicle changes in stream channelcross-sectional morphology over time (Pettsand Pratts, 1983; Hadley and Emmet, 1998);changes in channel planform or bedelevation downstream of reservoirs (Chien,1985); or sedimentological changesdownstream of a dam (Graf, 1980; Petts,1984; Chien, 1985; Andrews, 1986; ASCE,1992; Schmidt et al., 1995; Pitlick and vanSteeter, 1998; Brandt, 2000). Unfortunately,the Upper Connecticut River basin lacks thenecessary data sources to appropriatelyevaluate geomorphic changes. AlthoughNew England has one of the densestnetworks of USGS stream gages, thedistribution of sediment gages is chronicallylimited. Similarly, it lacks the critical pre-impoundment channel surveys necessary forcomparison to post-impoundment channelchanges.

Because of regional geologicconditions, many of the potentialgeomorphic shifts that tend to occurfollowing impoundment may not necessarilyaffect the mainstem of the ConnecticutRiver, or affect it minimally. Beddegradation generally occurs immediatelydownstream of dams (cf. Williams andWolman, 1984), yet conditions present inthe Connecticut River basin may act todiminish the role of channel incision.Significant Holocene isostatic reboundfollowing deglaciation (Koteff and Larson,1989) initiated an incisional episode, and themainstem flows on, or near, bedrockthroughout the study reach. Furthermore,the geologic history and setting of themainstem also limits other geomorphicadjustments. Changes in channel planformand/or sinuosity may also occur followingimpoundment (Williams and Wolman,

1984). However, the Connecticut Riverflows down strike of a narrow belt ofresistant meta-volcanic and meta-sedimentary rocks. This structural andlithologic control generates a constrainedvalley throughout most of the Vermont andNew Hampshire border, thereby inhibitingsignificant lateral channel migration.Geomorphic adjustments of the tributariesmay not be as limited as those on themainstem as the tributaries included in thisanalysis flow across a diverse array oflithologic and structural environments.

The greatest potential adjustmentsappear to be associated withsedimentological differences or channelnarrowing. The flux of suspendedsediments from the tributaries combinedwith diminished magnitude of large floods,especially in the impounded tributaries, maylead to enhanced fine-grained deposition andembeddedness (cf. Andrews 1986). Also,because of the abundant coarse-grained tillthroughout the region and the locallyavailable coarse-grained bed material on themainstem, impacts on bedload could besignificant. Thus impoundment is likely tohave strong, but complex, effects onsubstratum composition, an importantdeterminant of instream benthic invertebratecommunity structure (Minshall, 1984).Channel bed particle size distributions tendto change significantly immediatelydownstream of dams (Williams andWolman, 1984). In the Connecticut River,where threatened habitats and speciesfrequently occur on low-lying islandsformed and maintained by river-transportedsediments, this effect may be significant.Because the abundance and diversity ofbenthic organisms generally increases withincreasing particle size, changes in substratesize may negatively impact thesecommunities.

CONCLUSIONS

Our analysis provides a frameworkfor anticipating potential ecologicalconsequences of hydrologic alteration in theregion, and can also help set priorities forconservation and research. Lack of strongeffects on discharge seasonality, or on totaldischarge volume provide a major contrastto the effects of large dams in western semi-arid regions (Graf, 1980; Williams andWolman, 1984; Chien, 1985; Andrews,1986; Elliott and Parker, 1997; Hadley andEmmet, 1998; and Van Steeter and Pitlick,1998). As a consequence, we do not expectecological consequences arising fromreductions in total instream habitat area(Bovee 1986), major shifts in watertemperature regime (Ward and Stanford,1995), or mismatch between the timing offlow events and life history events(Montgomery et al., 1983; Puckridge et al.,1998), to be major factors in the ConnecticutRiver Basin. However, impoundment in theConnecticut basin has significantly alteredthe natural flow regime by eliminatingextreme flows and changing hydrographvariability. These alterations havepreviously been associated with majorecological change in other systems, affectingboth instream and riparian habitats.

The effects of hydrologic alterationon ecological communities must be put inthe context of other sources of long-termchange in both the physical and bioticenvironment of the New England region.We found that long-term gage recordsrevealed significant change over timeunrelated to the effects of impoundment,particularly in the increased low flow andbaseflow over time and the earlier timing of1-day minimum flows. Without thisknowledge, we could have mistakenlyattributed these phenomena to the effects ofimpoundment. Similarly, we caution that

the use of “snapshot” associations betweenecological and physical characteristics insuch a dynamic environment may bemisleading. We suggest that betterunderstanding of both the mechanisms ofinteraction between ecological communitiesand hydrologic alteration, along with moredata on historical ecological characteristicsand channel morphology, are necessary toadequately assess the ecological andgeomorphic impacts of hydrologic alterationin the Connecticut River basin.

FIGURE CAPTIONS

Figure 1 Map of Study Area (adopted fromFallon-Lambert, 1998)

Figure 2. Time Series of VermontAgricultural Activity (data from theVermont County Census).

Figure 3. Regional bankfull discharge (m3/s) vs. drainage area (km2). The pointfor the largest drainage area is for thepre-impoundment West Lebanongage. We have included this datumto extend this relationship to largerwatersheds. Neither the trend linenor the explained variance is affectedby the inclusion of the West Lebanondatum. Additional data forunregulated streams comes fromMagilligan and Graber (1996)

Figure 4. Mean Julian date for dailyminimum flows (upper diagram) andmaximum flows (lower diagram) forpre- and post-impoundmentconditions. The bars represent thestandard deviation.

Figure 5. Mean difference in dischargefollowing impoundment for differentflow durations.

Figure 6. Pre- and post-impoundmentmagnitudes of 1-day peak flow forall sites. The bars represent thestandard deviation.

Figure 7. Mean pre- and post-impoundment1-day low flow minima for all sites.The bars represent the standarddeviation.

Figure 8. Time series of the ConnecticutRiver at West Lebanon, NH 1-dayminimum flows.

Figure 9. Mean number of hydrographreversals following impoundment.The bars represent the standarddeviation.

Figure 10. Mean 1-day maximum specificdischarge (per km2) for unregulatedstreams vs. streams with only a post-impoundment record (streams listedin Table 1).

Figure 11. Results of log-Pearson Type IIIflood frequency analysis for (A) theConnecticut River at the WilderDam, (B) for the Black River, VT,and (C) the Ompompanoosuc Riverat the Union Village Dam. Thedashed upper line in each plot is forthe pre-dam flood frequency analysisand the solid lower line is for thepost-impoundment record.

Figure 12. Time series of 1-day minimumover time for the White River, VT

Figure 13. White River: Time Series of theJulian Date of the 1-day minimumflow.

Figure 14. Mean annual hydrograph forthe White River, VT. The dashedline is for the period 1920-29, andthe solid line is for the most recentten years (1988-1997).

TABLE CAPTIONS

Table 1: Basins included for the analysisshowing dam name, function, date ofdam, drainage area upstream of gage,the beginning date of the gagerecord, and the gage ID number. Fordam function, "f" represents floodcontrol, "p" represents hydro-electricpower generation, and "r" representsrecreation. Dam functioninformation from Fallon-Lambert(1998).

Table 2: Indices for Index of HydrologicAlteration. Adapted from Richter etal. (1996)

Table 3: Results from the IHA analysis forrivers with only a post-impoundmentflow record.

Table 4: Frequency counts of number offlows released after impoundmentthat equal or exceed the 2-yr bankfulldischarge, and the recurrenceinterval following impoundment.For those gages lacking pre-impoundment flow data, the 2-yrbankfull discharge was estimatedfrom the regional relationship inFigure 3.

Table 5: Results from the IHA analysis forunregulated rivers (** indicatessignificant correlation (at α < 0.05)for regression trend against time(yrs)).

LITERATURE CITED

Andrews, E. D., 1980. Effective andBankfull Discharges of Streams inthe Yampa River Basin, Coloradoand Wyoming. Journal of Hydrology46(3-4):311-330.

Andrews, E. D., 1986. Downstream Effectsof Flaming Gorge Reservoir on theGreen River, Colorado and Utah.Geological Society of AmericaBulletin 97(8):1012-1023.

ASCE, 1992. Sediment and Aquatic Habitatin River Systems. Journal ofHydraulic Engineering, AmericanSociety of Civil Engineers118(5):669-687.

Barnes, W. J., 1978. The Distribution ofFloodplain Herbs as Influenced byan Annual Flood Elevation.Wisconsin Academy of Sciences,Arts and Letters 66():255-265.

Bechtel, D. A. and D. D. Sperduto, 1998.Floodplain Forest NaturalCommunities Along Major Rivers inNew Hampshire. New HampshireNatural Heritage Inventory.Concord, NH, 56 pp.

Benke, A., 1990. Perspective on America'sVanishing Streams. Journal of theNorth American BenthologicalSociety 9(1):77-88.

Bovee, K. D., 1986. Development andEvaluation of Habitat SuitabilityCriteria For Use in the InstreamIncremental Methodology. U.S. Fish& Wildlife Service BiologicalReport. 86(7).

Brandt, S. A., 2000. Classification ofGeomorphological EffectsDownstream of Dams. Catena 40(2000): 375-401.

Chien, N., 1985. Changes in River RegimeAfter the Construction of UpstreamReservoirs. Earth Surface Processesand Landforms 10(2):143-159.

Colby, B. G., 1990. Enhancing InstreamFlow Benefits in an Era of WaterMarketing. Water ResourcesResearch 26(6):1113-1120.

Collier, M. P., R. H. Webb, and J. C.Schmidt, 1996. Dams and Rivers.U.S. Geological Survey Circular1126, 94 pp.

Collier, M. P., R. H. Webb, and E. D.Andrews, 1997. ExperimentalFlooding in the Grand Canyon,Scientific American 276(1):82-89.

Denny, C. A., 1982. The Geomorphology ofNew England. U.S. GeologicalSurvey Professional Paper 1208, 18pp.

Elliott, J. G. and R. S. Parker, 1997. AlteredStreamflow and SedimentEntrainment in the Gunnison Gorge.Journal of the American WaterResources Association 33(5):1041-1054.

Fallon-Lambert, K., 1998. Instream FlowUses, Values and Policies in theUpper Connecticut River Watershed.Connecticut River JointCommissions Report, 178 pp.

Graf, W. L., 1980. The Effect of DamClosure on Downstream Rapids.Water Resources Research16(1):129-136.

Graf, W. L., 1999. Dam Nation: AGeographic Census of AmericanDams and Their Large-ScaleHydrologic Impacts. WaterResources Research 35(4):1305-1311.

Hadley, R. F. and W. W. Emmett, 1998.Channel Changes Downstream froma Dam. Journal of the AmericanWater Resources Association34(3):629-637.

Jahns, R. H., 1947. Geological Features ofthe Connecticut River Valley,Massachusetts as Related to RecentFloods. U.S. Geological SurveyWater Supply Paper 996, 158 pp.

Kondolf, G. M. and P. R. Wilcock, 1996.The Flushing Flow Problem:Defining and Evaluating Objectives.Water Resources Research32(8):2589-2599.

Koteff, C. and F. D. Larsen, 1989.Postglacial Uplift in Western NewEngland: Geologic Evidence forDelayed Rebound. In: Earthquakes atNorth-Atlantic Passive Margins:Neotectonics and PostglacialRebound, S. Gregersen and P.W.Basham (Editors). Kluwer AcademicPublishers, pp. 105-123.

Ligon, F. K., W. E. Dietrich, and W. J.Trush, 1995. Downstream EcologicalEffects of Dams BioScience45(3):183-192.

Lyons, J. B., 1958. Geology of the HanoverQuadrangle. New Hampshire StatePlanning and DevelopmentCommission 43 pp.

Magilligan, F. J. and B. E. Graber, 1996.Hydroclimatological andGeomorphic Controls on the Timingand Spatial Variability of Floods inNew England, USA. Journal ofHydrology 178(1-4):159-180.

Magilligan, F. J and M. Stamp, 1997.Historical Land-cover Changes andHydrogeomorphic Adjustment in aSmall Georgia Watershed. Annals ofthe Association of AmericanGeographers 87(4):614-635.

Metzler, K. J. and A. W. H. Damman, 1985.Vegetation Patterns in theConnecticut River Flood Plain inRelation to Frequency and Durationof Flooding. Naturaliste Canadien112(4):535-547.

Minshall, G. W., 1984. Aquatic Insect -Substratum Relationships. In: TheEcology of Aquatic Insects, V. H.Resh and D. M. Rosengerg (Editors)Praeger Publishers, New York, NewYork, pp. 358-401.

Molles, M. C., C. S. Crawford, L, M. Ellis,H. M. Valett, and C. N. Dahm, 1998.Managed Flooding for RiparianEcosystem Restoration - ManagedFlooding Reorganizes RiparianForest Ecosystems Along the MiddleRio Grande in New Mexico.BioScience 48(9):749-756.

Montgomery, W. L., S. D. McCormick, R. J.Naiman, F. G. Whoriskey, and G. A.Black, 1983. Spring MigratorySynchrony of Salmonid, Catostomid,and Cyprinid Fishes in Rivière à laTruite, Québec. Canadian Journal ofZoology. 61(11): 2495-2502.

Patton. P. C., 1988. Geomorphic Responseof Streams to Floods in the GlaciatedTerrain of Southern New England.In: Flood Geomorphology, V. R.Baker, G. Kochel, and P. C. Patton(Editors). John Wiley and Sons, NewYork, New York, pp. 261-277.

Peterson, D. L., and F. A. Bazzaz, 1984.Photosynthetic and Growth-responses of Silver Maple (Acer-saccharinum) Seedlings to Flooding.American Midland Naturalist112(2):261-272.

Petts, G. E., 1984. Sedimentation Within aRegulated River. Earth SurfaceProcesses and Landforms 9(2):125-134.

Petts, G. E. and J. D. Pratts, 1983. ChannelChanges Following ReservoirConstruction on a Lowland River.Catena 10(1-2):77-85.

Pitlick J. and M. M. Van Steeter, 1998.Geomorphology and EndangeredFish Habitats of the Upper ColoradoRiver 2. Linking Sediment Transportto Habitat Maintenance. WaterResources Research 34(2):303-316.

Poff, N. L., J. D. Allan, M. B. Bain, J. R.Karr, K. L. Prestegaard, B. D.Richter, R. E. Sparks, and J. C.Stromberg, 1997. The Natural FlowRegime. BioScience 47(11):769-784.

Potter, K. W., 1991. Hydrological Impactsof Changing Land ManagementPractices in a Moderate-SizedAgricultural Catchment. WaterResources Research 27(4):845-855.

Power, M. E., W. E. Dietrich, and J. C.Finlay, 1996. Dams and DownstreamAquatic Biodiversity: Potential FoodWeb Consequences of Hydrologicand Geomorphic Change.Environmental Management20(6):887-895.

Puckridge, J. T., F. Sheldon, K. F. Walker,and A. J. Boulton, 1998. FlowVariability and the Ecology of LargeRivers. Marine and FreshwaterResearch 49(1):55-72.

Richter, B. D., J. V. Baumgartner, J. Powell,and D. P. Braun, 1996. A Method forAssessing Hydrologic AlterationWithin Ecosystems. ConservationBiology 10(4):1163-1174.

Rubin, D. M., J. M. Nelson, and D. J.Topping, 1998. Relation of InverselyGraded Deposits to Suspended-sediment Grain-size EvolutionDuring the 1996 Flood Experimentin Grand Canyon, Geology 26(2):99-102.

Schmidt, J. C., P. E. Grams, and R. H.Webb, 1995. Comparison of theMagnitude of Erosion Along 2 LargeRegulated Rivers. Water ResourcesBulletin 31(4):617-631.

Stanford, J. A., J. V. Ward, W. J. Liss, C. A.Frissell, R. N. Williams, J. A.Lichatowich, and C. C. Coutant,1996. A General Protocol forRestoration of Regulated Rivers.Regulated Rivers: Research andManagement 12(4-5):391-413.

U.S. Water Resources Council, 1976.Guidelines for Determining FloodFlow Frequency. Bulletin 17, UnitedStates Government Printing Office,Washington, D.C., 77 pp.

Van Steeter, M. M. and J. Pitlick, 1998.Geomorphology and EndangeredFish Habitats of the Upper ColoradoRiver 1. Historic Changes inStreamflow, Sediment Load, andChannel Morphology. WaterResources Research 34(2):287-302.

Ward, J.V. and J.A. Stanford, 1995. TheSerial Discontinuity Concept -Extending The Model To FloodplainRivers. Regulated Rivers Research& Management 10 (2-4): 159-168.

Wiele, S. M., J. B. Graf, and J. D. Smith,1996. Sand Deposition in theColorado River in the Grand Canyonfrom Flooding of the Little ColoradoRiver. Water Resources Research32(12):3579-3596.

Williams, C. W., 1996. Flooding theColorado River. Geotimes 41(8):6-7.

Williams, G. P. and M. G. Wolman, 1984,Downstream Effects of Dams onAlluvial Rivers. U.S. GeologicalSurvey Professional Paper 1286, 83pp.

Wolman M. G. and J. P. Miller, 1960.Magnitude and Frequency of Forcesin Geomorphic Processes. Journal ofGeology 68(1):54-74.

Wooton, J. T., M. S. Parker, M. E. Power,1996. Effects of Disturbance onRiver Food Webs. Science273(5281):1558-1561.

Wu, F. C., 2000. Modeling Embryo SurvivalAffected By Sediment DepositionInto Salmonid Spawning Gravels:Application To Flushing FlowPrescriptions. Water ResourcesResearch 36(6): 1595-1603.

Figure 1:

Figure 2:

0

10

20

30

40

50

60

70

80

90

1840 1880 1920 1960 2000

YEAR

VE

RM

ON

T: P

ER

CE

NT

LA

ND

IN

FA

RM

S

Figure 3:

Figure 4:

0

100

200

300

400

Day

of

Yea

r

Om

pom

p. R

.

Bla

ck R

.

Ott

aque

ech.

R.

Wes

t R.

Ash

uelo

t R.

Ott

er B

rook

Con

n. R

.

Post-Impoundment

Pre-Impoundment

a) Date of Minimum Flow

0

50

100

150

200

250

300

Day

of

Yea

r

Om

pom

p. R

.

Bla

ck R

.

Ott

aque

ech.

R.

Wes

t R.

Ash

uelo

t R.

Ott

er B

rook

Con

n. R

.

b) Date of Maximum Flow

-50

-40

-30

-20

-10

0

10

20Pr

e-Po

st D

iffe

renc

e

1-D

ay

3-D

ay

7-D

ay

30-D

ay

90-D

ay

Figure 5:

Figure 6:

a)

0

100

200

300

400

500

Q (

m3 /

s)

Om

pom

p. R

.

Bla

ck R

.

Otta

quee

ch. R

.

Wes

t R.

Ash

uelo

t R.

Otte

r Bro

ok

Post-Impoundment

Pre-Impoundment

0

500

1000

1500

2000

2500

Q (

m3 /

s)

Con

n. R

.

b)

Figure 7:

(0)

0

0

1

2

2

2

3

Q (

m 3/s

)

Om

pom

p. R

.

Bla

ck R

.

Otta

quee

ch. R

.

Wes

t R.

Ash

uelo

t R.

Ott

er B

rook

Post-Impoundment

Pre-Impoundment

0

10

20

30

40

Q (

m 3/s

)

Con

n. R

.

a)

b)

Figure 8:

Figure 9:

0

50

100

150

200

Num

ber

of R

ever

sals

Om

pom

p. R

.

Bla

ck R

.

Otta

quee

ch. R

.

Wes

t R.

Ash

uelo

t R.

Otte

r B

rook

Post-Impoundment

Pre-Impoundment

0

50

100

150

200

Num

ber

of R

ever

sals

Con

n. R

.

a)

b)

Figure 10:

0

5

10

15

20

25

301-

Day

Max

/Dra

inag

e A

rea

Free Impounded

Figure 11:

1000

10000

1 10 100 1000RECURRENCE INTERVAL (yrs)

A.

10

100

1000

1 10 100 1000

RECURRENCE INTERVAL (yrs)

10

100

1000

1 10 100 1000

RECURRENCE INTERVAL (yrs)

B.

C.

Figure 12:

y = 0.023x - 41.5R2 = 0.10p <0.05

0

2

4

6

8

10

12

1910 1930 1950 1970 1990

WATER YEAR

DISCHARGEm3/s

Figure 13:

y = -0.51x + 1229.9R2 = 0.11

0

50

100

150

200

250

300

350

1910 1930 1950 1970 1990

YEAR

JUL

IAN

DA

TE

Figure 14:

Table 1:

STATION MAJOR DAM(S) DATE GAGE GAGE GAGINGOF INFLUENCE OF DRAINAGE START STATIONAND FUNCTION(S) DAM AREA (km2) DATE ID #

I. IMPOUNDED R.S WITH PRE- AND POST IMPOUNDMENT DATA

Connecticut R. @ West Lebanon, NH Wilder Dam: p, f 1951 10598.3 1911 1144500Ashuelot R.@ Hinsdale, NH Surry Mt. Dam: f, r 1942 1087.8 1914 1161000West R.@ Newfane, VT Townshend Dam: f, r 1961 797.7 1928 1156000Ottaqueechee R. @ N. Hartland, VT N. Hartland Dam: f, r 1961 572.4 1930 1151500Black R. @ N. Springfield, VT N. Springfield Dam: f 1960 409.2 1929 1153000Ompomp. R. @ Union Village, VT Union Village Dam: f 1949 336.7 1940 1141500Otter Brook near Keene, NH Otter Brook Dam: f, r 1958 122.2 1923 (pre-dam) 1158500

(post-dam)1158600

IMPOUNDED R.S WITH POST-IMPOUNDMENT DATA ONLY

Connecticut R. @ N. Walpole, NH Bellows Falls Dam: p, r 1928 14226.9 1942 1154500Connecticut R. @ Wells R., VT Comerford Dam: p, r 1931 6848.0 1949 1138500McIndoes Falls Dam: p 1931Moore Dam: p, r, f 1957Sugar R. @ West Claremont, NH Sunapee Dam: p 1920s 696.7 1928 1152500West R. @ Jamaica, VT Ball Mt . Dam: f, r 1961 463.6 1946 1155500Mascoma R. @ Mascoma, NH Lake Mascoma Dam: 1920s 396.3 1923 1150500Ashuelot R. nr. Keene, NH Surry Mt. Dam: f, r 1942 261.6 1945 1158000

UNIMPOUNDED R.S:

White R. @ West Hartford, VT N/A N/A 1787.1 1915 1144000Williams R. nr. Rockingham, VT N/A N/A 290.1 1940 1153500 &

1153550Wells R. @ Wells R., VT N/A N/A 254.9 1942 1139000Ayers Brook @ Randolph, VT N/A N/A 79.0 1939 1142500East Orange Branch @ E. Orange VT N/A N/A 23.1 1959 1139800Mink Brook nr. Etna, NH N/A N/A 11.9 1962 1141800

Table 2:

RegimeIHA statistics characteristics Hydrologic parameters

Group 1: Magnitude of monthly water Magnitude Mean value for each calendar month conditions Timing

Group 2: Magnitude and duration of Magnitude Annual minima 1-day meansannual extreme water conditions Duration Annual maxima 1-day means

Annual minima 3-day meansAnnual maxima 3-day meansAnnual maxima 7-day meansAnnual minima 7-day meansAnnual maxima 30-day meansAnnual maxima 30-day meansAnnual minima 90-day meansAnnual maxima 90-day means

Group 3: Timing of annual extreme Timing Julian date of each annual 1 day maximum water conditions Julian date of each annual 1-day minimum

Group 4: Frequency and duration of high Magnitude No. of high pulses each year and low pulses Frequency No. of low pulses each year

Duration Mean duration of high pulses within each yearMean duration of low pulses within each year

Group 5: Rate and frequency of water Frequency Means of all positive differences betweenconsecutive daily means condition Rate of Change Means of all negative differences between changeconsecutive daily values No. of rises

No. of falls

Table 3:

1-day 3-day 7-day 30-day 90-day 1-day 3-day 7-day 30-day 90-day Base- Date Date # ofmin. min. min. min. min. max. max. max. max. max. flow min. max. Rev.(m3/s) (m3/s) (m3/s) (m3/s) (m3/s) (m3/s) (m3/s) (m3/s) (m3/s) (m3/s)

1. Connecticut River@N. Walpole 21.0 32.8 47.8 63.2 94.4 1689.4 1546.5 1343.6 914.7 567.0 0.18 223 120 163

2. Connecticut River@Wells R iver 13.5 21.3 30.8 41.0 60.4 866.7 772.7 641.2 423.7 266.8 0.22 221 136 158

3. Sugar River@West Cl aremont 1.4 1.5 1.6 2.0 3.1 117.4 92.8 71.9 44.1 26.2 0.14 253 125 113

4. West River@Jamaica 0.6 0.6 0.7 1.0 2.3 123.6 96.1 73.4 43.7 24.2 0.06 240 145 103

5. Mascoma River@Mascoma 0.9 1.3 1.5 1.6 2.1 54.6 47.8 37.6 21.8 12.9 0.25 241 134 72

6. Ashuelot Rivernear Keene 0.2 0.2 0.3 0.4 0.9 28.2 27.3 25.4 18.6 11.2 0.05 239 115 85

Table 4:

IMPOUNDED R.S WITH PRE- AND POST-IMPOUNDMENT DATANumber of Timesthat Calculated Pre-Dam Frequency

Drainage Calculated Pre-Dam 2-yr Q is Exceeded of Largest Post-Area (km2) 2-yr Q (m3/s) by Pos t-Impoundment Outflow Dam Even

Connecticut R. @ West Lebanon, NH 10598.3 1376.4 18 5-10 yr floodAshuelot R. @ Hinsdale, NH 1087.8 180.1 3 10 yr floodWest R. @ Newfane, VT 797.7 356.8 0 ~ 2-yr floodOttaqueechee R. @ N. Hartland, VT 572.4 145.9 7 2-5 yr floodBlack R. @ N. Springfield, VT 409.2 156.6 0 < 2 yr floodOmpompanoosuc R. @ Union Village, VT 336.7 76.2 0 2 yr floodOtter Brook n ear Keene, NH 122.2 32.3 0 < 2 yr flood

MPOUNDED R.S WITH POST-IMPOUNDMENT DATA ONLY Number of Timesthat Calculated Pre-Dam2-yr Q is exceeded Pre-Dam Frequency

Drainage Calculated By Post-Impoundment of Largest Post-Area (km2) 2-yr Q (m3/s) Outflow Dam Event

Connect icut R. @ N. Walpole, NH 14226.9 2519.7 8 N/AConnecticut R. @ Wells R., VT 6848.0 1393.6 9 N/ASugar R. @ Wes t Cl aremont , N H 696.7 218.9 9 N/AWest R. @ Jamaica, VT 463.6 157.4 16 N/AMascoma R. @ Mascoma, NH 396.3 138.6 6 N/AAshuelot R. nr. Keene, NH 261.6 99.0 0 N/A

Table 5:

1-day 3-day 7-day 30-day 90-day 1-day 3-day 7-day 30-day 90-day Baseflow Date Date Number ofmin. min. min. min. min. max. max. max. max. max. min. max. Reversalsm3/s m3/s m3/s m3/s m3/s m3/s m3/s m3/s m3/s m3/s

1. Mink Brook (Drainage Area = 11.9 km2; n = 33)Mean 0.20 0.23 0.31 0.60 1.34 115.03 80.94 56.96 30.60 17.32 0.04 229 116 99Std. Dev. 0.01 0.01 0.01 0.02 0.03 1.72 1.05 0.65 0.33 0.16 0.00 23.11 69.16 12.62Corr. Coeff. 0.24 0.24 0.22 0.34** 0.34** 0.22 0.20 0.06 -0.04 -0.04 0.27 -0.20 -0.02 0.41**

2. East Orange River (Drainage Area = 23.1 km2; n = 37)Mean 0.04 0.05 0.05 0.08 0.12 3.95 3.06 2.56 1.72 1.02 0.00 237 139 115Std. Dev. 0.03 0.04 0.04 0.05 0.07 1.62 1.22 0.98 0.52 0.25 0.00 43.89 80.01 10.67Corr. Coeff. 0.49** 0.46** 0.45** 0.44** 0.41** -0.11 -0.21 -0.25 -0.28 -0.09 0.55** -0.06 0.08 0.36**

3. Ayers Brook (Drainage Area = 78.9 km2; n = 54)Mean 0.13 0.14 0.15 0.20 0.33 14.54 10.79 8.39 5.36 3.13 0.00 231 131 115Std. Dev. 0.09 0.09 0.10 0.13 0.20 5.35 3.39 2.47 1.50 0.76 0.00 45.35 78.71 11.15Corr. Coeff. 0.18 0.22 0.28** 0.25 0.24 -0.07 -0.05 -0.04 -0.05 0.03 0.29** -0.37** -0.08 -0.39**

4. Wells River (Drainage Area = 259.4 km2; n = 54 )Mean 0.57 0.61 0.67 0.91 1.37 38.92 30.90 24.04 14.99 8.95 0.00 223 135 113Std. Dev. 0.21 0.23 0.26 0.42 0.59 13.11 9.49 6.91 3.94 2.13 0.00 62.99 73.71 11.80Corr. Coeff. 0.12 0.11 0.17 0.28** 0.29** 0.19 0.15 0.03 -0.11 -0.05 0.02 -0.13 -0.05 -0.51**

5. Williams River (Drainage Area = 290.1 km2; n = 54)Mean 0.35 0.37 0.40 0.55 1.00 77.65 52.81 38.37 21.91 12.49 0.00 242 130 108Std. Dev. 0.19 0.20 0.23 0.34 0.65 33.31 20.26 13.36 6.73 3.09 0.00 37.34 84.96 10.70Corr. Coeff. 0.38** 0.40** 0.39** 0.35** 0.36** 0.47** 0.45** 0.35** 0.28** 0.33** 0.32** -0.06 -0.04 -0.14

6. White River (Drainage Area = 1787.1 km2; n = 78 )Mean 3.71 3.96 4.31 5.64 8.95 394.84 287.39 217.02 132.94 76.85 0.00 244 128 118Std. Dev. 1.66 1.69 1.88 2.82 4.75 142.06 97.38 68.60 36.29 16.58 0.00 34.77 80.28 12.94Corr. Coeff. 0.32** 0.24** 0.24** 0.30** 0.23** -0.14 -0.13 -0.12 -0.19 -0.17 0.27** -0.33** 0.03 -0.62**