David L. Petersonl, David G. Silsbee . > r d o o g d - r ' ' o . l. o o t d , o D d o
Fe d Staton Box 352100.SeattleWashngton98195
Unversty of Washington
. "
o l
o . . - t
( , r . . o T -
-
afd
Kelly T. Redmond WesternFeglonalClirnateCenter,DesertFesearchInstitutePO Bov.64224,Reno.Nevada
89506
DetectingLong-TermHydrologicalPatternsat CraterLake,Oregon
Abstract
Trcc ring chronologiesfbr nounlain hcmlock (I\ugu n.iensiana) werc uscd 1()rcconstruc!the lvrtef level of Craler Lrkc. t
high clcvatio. lake in the southerDCascadcRangc ofOregon. Reconstructionsindicalc that lake le!el since the late 1980shas
beenlowcrthanatan!pointinftela\t300!carscxceprlheearlyl9l0stolnidl9'l0s.Lakclc\elwasconsistentlyhigherdurirg
the Li(lc 1ccAge than during $e late 20rh ccnlliry: during the late 171llceDtur].lakc lcvcl $,asup to I 'n highef th.rnrecent( 1980s
xnd 1990s)lo\\ levels,$hich is consis|cnrwilh paleoclimaticreconstructionsol rcgional precipitationand atnrspheric prcssurc.
Furthernore. in\rrumenral dara availablc ibr rhe 20ft centufy suggestlhar thcrc are strong telecornectionsamong atno\phcric
circulation(e.g.,PacificDecadalOscillaiion).lree grouth. rnd hldrolog) insorrthcmOregon. CfaterLake is sensilivelointcrannual,
inlcrdecadaland intercenteD.r) !ariadon in precipitation rnd atmosphericcirculation. and can be expectedto lrack bo(h shorl
rcmr and long tenn lariation in fegiLrnalclimatic patternsthal mrl occur in thc luturc.
lntroduction
The time seriesof annual growth stored in tree
rings is a repository of intbnnation on regional
climate, as well as on local forest productivity.
techniqueshaveprovenparDendrochronological
ticularly useful in reconstructingpasthydrologi
cal variationby quantifyingtherelationshipamong
interannualvariation in climate, variation in hydrological parameters(c.9.. streamflow) and radial trce growth (Meko et al., l9ll0, Cook and
Jacoby 1983,Blasingand Duvick 198.1.Meko
andSbckton 1985,LandwehrandMatalas1986,
Brinkman 1987,1989, 1992,Michaelsenet al.
1987,Robenson;urdJosza1988.Stahleetal. 1988,
Loaicigaetal. 1992,1993,StahleandCleaveland
1993.Woodhouse1993,Young 199,1.Shenand
quantifybngTabois1995).Thesereconstructions
telm trendsand expectedvariation in hydrological systemsassociatedwith regionalclimate,and
provide valuableinput to policy decisionsabout
waterconsumptionandallttation (Stockton1990).
Reconstructingthe water level oflakes can be
becausemostlakescompriseonly a
challenging.
small lraction of the surf'aceareain a watershed.
Lakcs frequendy have unknown inputs from in
t'low streams.utknown lossesto groundwaterand
lou
n n , , ' r e . p . . l J < n ( (. h u u u h c J J J r ( . e d .
outt'low sffeamsand complex evaporativerela
CraterLake.adeep,
tionshipsu ith theatmosphere.
high-elevationlake located in the southernCascadeRangeof Oregon(Drakeet al. 1990),is a
notablcexceptionto this generalization.Thc surtace areaof Crater Lake comprises78.57r of its
watershcdsudacearea,and therc are no streams
flowing into or out of the lake. BecauscCrater
Lake comprises such a large propofiion of its
r -rl t e r i h e , . lu. u t e r l c r e l i . r e r l r e . p o n . i r e t o
interannual
variationin precipitation(Redmond
1990.Nathenson1992),givingthe lake its rcpu"the world's largestrain gauge"(Larson
tation as
'leak1990),dcspite the fact that there is some
age"duc to seepage.
In this study,we usedclimatically-sensitive
tree-ring chronologiesand climatic and hydro
logical recordsliom the 20th century to recon
stmctannualprecipitationandwaterlevelat Crater
L r k < . W e q u u n t i l i e t hl c r r t eo l c h a n g ei n u a t e r
levels sincethe late 1600s,which includesmuch
of the Little lce Agc (ca. 1600 1850),a period
u,ith cooler climate in westemNorth America
(Fritts andLough 1985.Graumlich and Brubaker
1986,Folland et al. 1990,Briffa et al. 1992).
Knowledgeaboutpastwaterlevelsoflakes (prior
to measurementrecords) is valuable for understandinghydrological varialion at different temp o r a l . c a l e (. S l i n el q g 0 . l q q 4 rl n d c r n i r n p r o \ (
Vol. 73. No. 2. 1999
NorthwestScienca,
O lq99bI(lr'\onh{ciS.i.nrilicA\n).r!rmn
Allighnr.$^c,l
Il
predictions of how hydrological systemsin the
Pacilic Northwest may respondto various climatic
conditionsin the future.
Methods
Researchwasconductedin CraterLake Natiooal
Park in the southem CascadeRange of Oregon
(Fig. l). CraterLake is a deephyperoligotrophic
lake that coversthc floor of Mt. Mazamacaldera
u'hichformedca.6.800B.P The averageeleva-
-2-
tion of the lake is l,882 m. Area of the water
suface is 53.2 kmr, andthe drainagebasinis 6?.8
km'](Redmond1990,Nathenson1992).As previously noted.the lake occupies78.5'lr ofits own
drainagebasin. Lake volume is 17.3 kmr. with
maximumdepthof 589 m (Collier et al. 1990).
Steepwallsform a high rim (elevationca. 2,100
m) aroundthe lake. inllouing \\ler: originrre
within the calderaand there is no surfaceoutlet.
The only ways for water to leave the basin are
through seepageand evaporation.Seepageis
L =Tsugamertensiana s ites
Figure L Location of Crater Lake in Crater Lake Narional Park. soulhcm CascadcRdnge.Oregon. The weather station. lake
level gauging station and mountain hemlock (IsrA,a m(tten\ieru) \smpling siles are indiciled.
122
Peterson,
Silsbee,and Redmond
approximately0.35 cm/day ( 128cm/yr) anddoes
not appearto vary with lake level. while evaporation is approximatell,0.33 cm/day (120 cm/yr)
(Redmond1990.Nathenson1992.).
M ( ' n t h l )l o t a lp r e c i f i t r t i u nl n d m e u nr i r l c t t l
penture data fbr Crater Lake (measuredat Crater Lake National Park headquarters,elevation
1.973m) areavailablesince193l. Meanannual
precipitation at Park headquartersis 171 cm, of
which 80-90c/r,falls as snow during the winter
(OctoDerthrough April); mean annual snowlall
(in soiidform) is 1,314cm (Redmond1990).Mean
(1961 1990)is 3.1"C.with a
annui,ltemperaturc
meanannualminimum of -2.9"Cand meanannual rnaximum of 8.9"C. Lake level has becn
sincel96l,datahavebeen
measured
since19081
recordedat apemanent gaugelocatedon the west
side of the lake, although measuremcntsprlol to
this time are also consideredto be accurate
(Redmond1990).
Tree ring chronologies were devcloped lbr
mountain hemlock (Tsuga nertensidna) at each
of 7 sites in Cnter Lake National Park. Mountain henrlockis the dominantffee specicsthroughout much of the region adjacentto the dm of the
lake.Shastaredfu (Abiesmagnificuvar.shastensis)
is codominantat somesites.especiallyat lower
elevations.Siteswerelocatedalongan elevational
gradienton the westernsidc (3 sites)and eastcrn
side (3 sites)of the lakc. and on Wizard Island( I
site)within thc lakebasin(Fig. 1). Treeageat I
m heightwasup b,139 years(Tablel). Twenty
tree\ IrLrmrach sitc \\ere corcrlwith incremenl
borers at I m abovethe ground: only larger (and
presumablyolder) treeswithout visible injury to
the cro$,n and bole were sampled. Two cores
1ABLE
were extractedper tree, of which one core was
analyzed.Trec coreswere sandedandcrossdated
(Fritts 1976).and ring widths werc measuredto
the nearest0.01 mm in the laboratory.
(through
wereci,rlculated
Tree-ringchronologies
standardization)with the program ARSTAN
(Holmcs 198,1)
usinga stiffcubic smoothingspline
with a 507cfrcquencyresponsecutoffof 200years
(CookandPeters1981,Holmes1983)appliedto
the ring-width seriesofeach tree. The ring width
for eachyearin the serieswasdividedby the spline
value to give a time sedesof index values. An
autoregressivemodel was fit to each index. and
the residual sedes tbr all trecs in a stand were
averagedto give a site chronology. More flex
ible smoothingsplines(507ofrequencyresponse
cutoff of 25, 50 and 100 years) were tested to
detemine the potcntialimpact of standardization
on reconstructions
of precipitationand lake level.
Theseother curve fits had minimal effect or the
i h r o n o l o g i eosr u l l i m r t cr e c o n s t r u c l i o n: o\ . $ (
retained the stiffer spline which would tend to
retain low-frequency variancein the caloulated
chronologies.
Correlationanalysiswasusedto determinethe
relationshipbetweenannual tree gfowth and
monthly and annual precipitation (bascdon hy
drologicalycars,OctoberthroughSeptember).The
periodofconcurrenttrec-growthandclimaticdatil
usedin thc analysiswas 1931throughl99l .
Principalcomponentanalysis(PCA) wasconducted on the 7 site chronologiesas \\,ell as the
chronologieslagged forward I ycar (total of 14
variablesin thc PCA). PCA is cornmonly used
to definethecommonvarianceamongchronologles
L sunmary of charlcrerisrics ofsampled mountair hemlock (I!ugu ne rrensiantit n' Crtter Lake Nrtional P.rrk. Trees
on thc wesrernand eirsternrim ofthe lake (20 treespcr si!e) \lere sanpled acros\ an elelational gradient (high. mid.
low): onh one sitc was samDledon $'izdrd I$and.
Earliest tear ir chrcnologv
Mean
Ele\.'ali(rn
0n)
Wesr high
E a r t- h r g h
Easr mid
Flart ' low
Wizad lsland
(cnr.llSD)
ll90
2060
8 2 . 6( 1 4 . 1 )
70.li(r0.8)
66.7( l,l.l)
89.5(20.2)
73.6( 9.6)
l9t)t-)
2000
6 t . 6( 1 2 . 7 )
5 8 . 6( l 1 . 5 )
ll00
2060
19.10
Mcan
( y r .t l S D )
l l r 5( 5 7 )
(36)
17.15
1757(68)
1766(6.1)
r796(,19)
l716(60)
1726(83)
Rrnge
(calendaryr)
Chronologv statistics
Co'nmon
t'arlance
N,iean
Auto
sen\ilivit)
coreladon
l'la
16071787
1687-1809
1597-1856
| 6 , 1 l1- 8 8 9
0.201
0.06r
51.0
0.230
0.t73
0.196
0.052
0.193
0.076
50.7
16,10
1887
1 5 7 61 8 l l
0.193
0.1,+4
0.201i
1 5 5 31 8 5 6
0.189
56.8
.11.8
.11.6
0.215
0.157
,19.,1
Hydrological Vadation at Cratcr Lake
123
(Graumlich 1993), rcducethe nurnberof variables
usedin regressionand reduccmulticollinearity
betweenvariables(Graumlich1987.Loaicigaet
al. 1993,Pctcrsonet al. 1993). PCA was con
ducted for the entirc time seriesof tree growth,
and 6 principal componcnts(PCs) since l93l
(u'henclimaticdataarc available)were incJuded
in a stepwiseregressionprocedureto predict annual precipitationchangc. The PCs were not rotated. Coeftrcients of determiDation(ri) were
calculated1brthc regressionmodel. Obsenations
u e r el e l to u to l t h el c c r e . s i o m
n o r l eol n cJ I l r i r n e
(one-fold cross-r'alidation)(Craumlich 1987,
Michaelsen
et al. 1987.Michaelsen
1988.Loaiciga
et al. 1993)to determinepredictivepower.and
rcconstructionswere developedbackto 1687.the
point at which fiee-growth data are availablefbr
a total of 25 trees and all 7 sampling sites are
represented
il the dataset. Althoughthis is subjcctive. the variancepropefiiesofthe reconstmc-
tiors did not seemto be afTectedmuch by samplc
size moving fbrward in the timc seriesfiom this
point. Residualsfrom the regressionwere exanlined fbr departuretiom a nonnal disftibutionwith
the Shapiro Wilk statislic,andautocorelatiol'l sh'I]clurc !\ r\ le\leJ\\ ith theDulbin-Wlts,'n .trti;tie.
An identical procedurewas usedto developa
rcgression
modelwith,l PCsrelatingfrcc-ringchronologies(growthindiccs)k) annualchangesin lake
lcvcl since i908 (when lake level data are available). Absoluteiake levcls over lime were calculatedfrornthe reconstructed
changesin lakc level.
Resultsand Discussion
Annual tree growth is negativelycorrelatedwith
precipitation during winter prior to thc current
growingseason(Fig. 2). indicatingthat growth
of T. mertensiana
is enhlncedby a longersnowfiee period and reducedby deep snou,packsthat
Westside
Eastside
E
E
E
O N O J F M A M J J A S O N O J
A M J J A S O
Month
F M A M J J
A S
N O J F M A M J J A S
Month
Fisure 2. Coffelatiorrsof annual trcc gro$ th inde\ $ ith pfeclritation lbr the current (!car r) and pre!rcus (year /- /)hydrologicd
h e p t e m b c rl.9 l l l 9 9 l ) a r e s h o i ! nf b r 6 n l o u n t a i nh e m l o c k( 7 - r r 3 . r , r . , r / . r r r d r ms)i t e . i' n C r a t e r
y e . r ( O c t o b e r t h r o u gS
L.rke National Park. Although il i\ possiblelo calculrte \igniticance levelsfor thc corclations. alph.rlevel\ could be
infl.rted by potential ber$een month intcrcorchtion of the gfolvth-climare felationship. Conelati(nrsfi)r Wizard ls
l l l n dt r e e \ ( n o t s h o w n )a f e s i i n i l i r t o l l r o s cl o r t h c $ , j s l s i d e .l o $ - e l e ! . r t i o n\ i t e .
124
Peterson.
Silsbee,andRedmond
extendinto the summer. This climate-growth
relationship is similal to that ti)und for Z
nertensiana at other sites (Graunlich 19137.
Petersonand Petenon in review) and for other
subalpineconiler specicsin the PaciticNofthEttl andPete$on
andPeterson199.+,
west(Peterson
1995a,b,Pctersonl998, SrnithandLraoque1998)
andthe RoclryMountains(ColenuttandLuckman
199l. Villalbaet al. 199.1:althoughgrowthof these
specicshas an equal or greaterpositive correlation with summcrtemperature).Corrclationsfor
winter precipitationare greaterfol higher elevatiol'rsiteslocatcdnear the rirn of the CraterLake
caldera(Fig. 2). Thesesitesgencrallybegin to
accumulatesnowpackin latc October.havea peak
snowpackof >300 cm in April. and reach complete sn()wnrelt
by ear'lyJuly lRcdmond 1990).
by winterprecipitationdurGrowthis enhanced
ing the previousyear(positivcconelation)(Fig.
2) because
ofthe etlcctofhigh snowtallon longtelm soil moisturestorage(PetersontrndPeterson
Ettl aDdPeterson1995b).Augustprecipi199,1.
tationdudng the previousyeal also has a positive coffelation with grouth.
In addition.grorvthis positivelycorrelatedwith
curreDt-)/earspring and early summer tcnlpera
tures,indicatingthat growth is enhancedby warm
temperaturesfar orablc for photosynthesis andcarly
snowmelt(Petersonand Petersonin rcview).
Growth may be leduced by high temperatures
duringtheprevioussummer(negativeconelation)
b e c r u . e, ' l t h c i r n p a c to' l l , ' u * ' i l m u i . t u t eo n
carbohydratesbragc (Ettl and Petersen1995b).
The lirst 6 PCs derivedlion PCA. which ac
countfor 88% of thevariancein thc data(ofwhich
the first two PCs accountfbr 757c),were retained
in the stepwiseregressionprocedurefbr predictof precipi
ing annualprecipitation.Regression
tation on thesePCs resultsin a significant modcl
(rr=0.46,p<0.001).The correlationbetweenthe
valueofthe deletedobservationand
reconstuucted
is 0.60(p<0.001).Residuals
actualprecipitation
regression
show no signillcant
tiom the
(tiISt-orderautocorrelation=0.03,
autocol:relation
Durbin-Watsonstatistic=2.06)or depanurefrom
a normal distribution.
Using the same analytical approach,regrcssion of lakclevel changeon PCs 1, 3. 5, and 6
resultsin a signilicantrnodel(rr=0.40,p<0.001).
Crossvalidationindicatesthat the correlation
betweenthc reconstmctedvalue of the deleted
observationand the actualchangein lake level is
[ ! 7 r p < l t O 0 l r .R c ' i d u a l l' r , r n tr h er c g r c : s i o n
contain some serial colrelation (tlrst-order
tutocoffelation=0.14.Durbin-Watson statls
tic=1.70).andthe distributionofresidualsdeplllts
from normal in the shortnessof the tails (low
kurtosis,p<0.05).
Bccauseprecipitationis so stronglycorrelated
with lake-levelchangeat CraterLake, (Redmond
1990),our analysestbcus on the quantitativere
lationship of tfee growth, precipitation (most of
which falls as snow) and lake level. Measured
annualprecipitation(since1931)andreconstructed
precipitationat CraterLake (from 1687to 1930,
based on the model derived through PCA) are
shown in Fig. 3. The prccipitationmodel based
on tree growth representsinterannual variation
reasonablywell, although severalof the higher
Reconstructionsoften
valuesare underestimated.
unrlerestimlle
e\lreme \illue\ bectu-etree-rin!
chronologiesarebasedon a spatiallyvariableaggregateof treesused to developa timc sedesof
mcan, rathel than extreme, values (Fritts 1976,
F r i l t .u n dC u i u t l o q O l .H i g h n r e i i n i l i r l i t 'in\ i r l dicatedfor theearly andmid 1700s,the nid 1800s
and the early 1900s. Thesegeneraltrcndscorrobomlcpre\ i,'u. rec n.lruclion\of precipitatir'n
for the Pacific Northwest (Brubaker 1980,
Craurnlich1987.Fritts 1991).
Measurcdlake level (since190[i)and rcconstructedlake level at Cratcr Lake are sho\\,nln
Fig. .1. Reconstructedlake level representsthe
generalpattern of interannualvariation in measurementsreasonablywell, althoughthc magni
tudeofhigh andlow extremesis underestlmated.
Mean reconstructedlake level dudng the measurementperiod(1908-1991:-92.2 cm) is less
period(168?
thanthat of the pre measurement
lakelevel
1907: 175.7crn). Mcan reconstmcted
during the 20th century (-1J4.[icm) is much lower
than that prior k) 1850 (226.3cm). Lake level
startingin the late 1980shasbeenlower thanat
trny time during the past 300 yea$ except in the
early 1930sto rnid 19.10s.Pedodsof greatest
lake-levelincreasestrack periodsof gcnerally
precipitation(Brubaker1980.
higherrcconstructed
Frirrs1991,Graumlich1993)(Fig.3).
Reconstmctedlake level is generally much
higher during the Litle Ice Age than during the
20thcentury(Fig..1).The highestreconstructed
Ievelsduring thc late lTth centuryare9 m greater
Hydrological Variationat Crater Lake
125
240
220
I
I
I
I
I,
h
200
E
iliilililti
t) 1 8 0
z
o
tr
(
160
I
I
E
LIJ
(!
(!
ll
1l
140
'120
100
- - Measured
Reconstructed
80
1700
1750
1800
1850
1900
1950
1990
YEAR
Figure 3. Measured(dashedline) andreconstructed(solid linc) prccipilarion1or thc CracrLakc wca$er nadon. basedon hydrologjcal year (Oclober through Septeinber). The reconsrructionwas cxlcndcd 10 1681. lhc point at $hich tree growth
daraare availablefor a total of 25 rreesand all 7 sanrplingsilcs arc rcprcscnlcdin thc da|a set.
than the lowest measuredlevels of the 20th cen
tury. The magnitudeof this lake level changeis
suppofiedby the fact that only youngertrees(pd
muily T.meftensionuandA.magniJica<150 yeas
old) arcfoundalongthelowerwall ofthe caldera
andshorelineof Wizard Island.comparcdto much
older treesat higher elevations. The lowest lake
levels (measuredand reconstructed)occur during the early 1930sto mid 1940s,following a period
of low precipitation in westem North Amedca
(Graumlich 1993). including the Pacific Northwest(Graumlich1987),with additional1o* levels startingin the late 1980s. The highestlake
levels(rcconstructed)
occurin thenid 1700s.with
evidencefor additional high levelsin the late 16fi)s.
Even the lowerpeak in lake level in the nid 1800s
is higher than any measuredin the 20th century.
Ifthe model underpredictsextlemevaluesassug
gestedby datafor the20th century,thenthehigher
actuallake-levelchangesduringthe Little IceAge
may havebeenevenhigherthanindicatedby re126
Peterson,
Silsbee,andRedmond
(Fig. 4): loweractuallakelcvcl
conshuctedvalues
chan-tesmay havebeenlower than reconstructed
aswell.
Becausethe reconstructedparameteris lakelevel ./?.rr?ge
rather than absolirlelake level (the
sum of interannualchanges,as shown in Fig. 4),
it is difticult to directly link reconstructedpre
cipitation (Fig. 3) and absolutc lake levei visually in the figures of reconstructedvalues. An
other factor that obscuresdirectionality oftrcnds
ir thrt hoth stanLlrdization
proce
h regression
dure) and regressionnrodelingof the growth-index values produce time series with lower
magnjtudeof variancethan in the original data.
Therefore,a smallbut persistent
anomalyin precipitation which would affect lake level
c h a n g e e o u l db e o b \ c u r ei n l h er c c , ' n : t r u e t i ( ' n .
Even ifprecipitationwere relativelyconstantover
time.but temperature
waslowet as it surelywas
during the Litde Ice Age, then lake lcvel could
have been higher prior to the 20th century as a
u, 40
o
z
$ o
-t!
600
o -40
q
400
o
d
o<\
-co
-go
200
LUF
> o
-40
-80
0
40
80
(cm)
LAKELEVELCHANGE
MEASUREO
\-\n.
J'rf
... o
\lI
)<c
<9)
r 6
q)
./t . ^,\il
-200
-400
-600
- - Measured
Reconstructed
'1700
1750
1800
'1850
'1900
'1950
1990
YEAR
Figure ,1. Mersured (dashedlinc) and reconsnucted(solid line) lakc lcvcl ibr Crater Lake. Lakelevei variationsxre adjusled to
l0 Scplcmber. Water stageis relerenced1()an elevationof I ,882-I ,l m abore nean sealcvcl and lho\ln as a departure
from ihi s elevatid (Redmond 1990). The reconstructionwas exreDdedto I 61i7.lhe poi nt at which tree-growlhd ata are
available ibr a lotal of 25 trees and all 7 sampling sires arc rcpreseniedjn the data set. The scatterylot shows thc
relationshipbctween predictedand me sured(since 1908) lake level change(r:=0.57).
resultoflower evaperativerates.Lower temperature. cr)ulLlal.o ha\e promoledice lirrm:rlionon
the lake, a very rare occunencein the 20th cenu i n t e re v a p , r r n t i o n .
l u r y .I h e r e b )r e d u c i n g
Variationin atmosphericcirculationcanresult
in multidecadal anomaliesof sufficient magnitude to producesignificant changesin the water
level of lakes and other bodies of water, as has
beenshownfbr California's Siera Nevada(Stine
1990,1994). The midlatitude storm track of the
northernhemi.fhere.u hich hasr strongimprel
on general precipitation patterns in tar westem
Nonh America(Graunrlich1993.Stine1994),may
have tracked consistcntlyover southernOregon
during the Little lce Age. Theseconditions are
suppofied by previous climatic reconstructions
which indicatethe prevalenceoflow pressureand
high precipitationin this region during the latter
half of the 17th century and much of the lSth
century(Fritts1991).
During the 20th century,periodsoflower lake
level coincidewith the positive(dry, warm) phase
of the PacificDecadalOscillation(PDO) (Mantua
et al. 1997),and periods of higher lake level coincide with the negative(wet, cool) phaseof the
PDO (Fig.,+). In addition,mountainhemlock
chronologiesfiom treeline at Cfater Lake and
throughout most of the Pacific Northwest ffe
generally in phasewith the PDO Index (PDOI)
(PetersonandPetersonin review),andtreesattain
Hydrological Variation at Crater Lake
121
more growthduringperiodswith lesssnowpack
(Smithand Laroque1998). Conversely,
lorl''-el(especiallyin southernOrevationchronologies
cgon),fi)r which high summertemperature
can
limit growth.aregenerallyout of phasewith the
PDOI (Petersonand Petersonin rcview). These
phenomena suggest that there are strong teleconncctionsamongatmospheric
circulationpatterns,tree growth. and hydrology in southern
O r c g o n .B e c r u r eC r J l e rL J k ei \ : r r r e \ f , ' n \ i \ eI , ,
variation in precipitation, the PDO may have a
s i g n i f i c a rirnt l l u e n e e , ' inn t e r r l e , . r r rl r lr i r r t i , r inn
lake levels.
Thc climatically-sensitive
hydrological
system
of Crater Lake. coupled with the dendrochronol o g i c r lr e c o r di.s c a p r h l eo l t r a c k i n gv a r i r t i o ni n
regional precipitation. Even small changesin
annualprecipitation,incIudinga 1ewextrerneyears.
can have a substantialcumulativeimpact on lake
Ievel. For example. the 9 m difference in lake
level cited above can be accountedfor by a deviationof only 3 cm/yr in precipitation(<27nof
mean annual precipitation during the measurementperiod).Most assessments
ofthe impactof
climaticvariabilityduringthe nextcenturyhave
focused on how potential tefiperetLtreftctexses
would aflect natuml resources(e.g.. Watson et
a l . l q q 6 J .H o w e \ e r e
. \ e n \ r n i r l l c h a n g ei n' 7 , r e cipitation can have major efl'ectson hydrological systems(Arnell et al. 1996).
Lake-level variation and rates of lake level
changeare stfaightforwardhydrological effects
LiteratureCited
A r n e l l . N . , B . B . L t e s ,H . L a n g , J . J . N l a g n u s o n .x n d P
Mulholhnd. 1996. Hydrolog) rnd fresh$rter ecology. In R. T. Wal.,on. M. C. Zinyoucra. and R. H.
. l i m a t eC h a n g e1 9 9 5 :l m p a c t s . A d a p t a
\ , l o s s( e d s . )C
tions and Mitigation of Climate Chrnge: Scientific
T c c h i i c a l A r a l ) s e s . C a l n b i d g e U n i v e l s i t ) -P f e s s .
Cambridgc. Unitcd Kingdon. Pp. 325 363.
Blasing. T. J. and D. Duvick. 198,1.Reconslructionof pre
cipitatifir hillory in ),iorlh Amcrican corn bch using
r e e r i n g s .N a t u l e2 0 , 1 : 1 , 11- l, 1 5 .
Brifia. K. R.. P D. Jones. and F. H. Schweingruber 1992.
Trcc ring densityrcconstructiollol summer tempera
turepafternsacro\s\\'estern),iorthr\merica since 1600.
J o u r n . rol f C l i m r t e 5 : 7 3 5 - 7 5 ' 1 .
Brinkman.$'. A. R. 1981.Whtersuppliesto the GfeatLakes
reconstructedfioln tree-rings. Journal of Climalolo g y a n dA p p l i e dN l e r e o f o l o g2) 6 : 5 3 0 - 5 3 8 .
128
Peterson.
Silsbee.and Rednond
ofprecipitation trends,and will reflect naturalor
human-inducedvariability in climate that may
occur during the next century. Large,deeplakes
are sensitiveindicatorsof variation in precipitation ifthe lakescomprisea sufficientlylargepropofiion of a watershed. Given the correlation
betwecn lakc-level change and tree growth determinedin this study,it may alsobe possibleto
use lake-levelchangeas a meansof estimating
interannualand interdecadaltree growth. This
would be a sirnple,economicalapproachfor estimating tenporal variability in forestproductivity. which is an impofiant indicatorofecosystem
health. Cratcr Lake and other large lakes that
dominate watershedsurface area may be valu
able tbr monitoring tuture changesin hydrology
and productivity that result from variability in
climatic patternsin the Pacific Nofihwest.
Acknowledgements
We appreciate review comments by Linda
Brubaker.RobertCollier, Lisa Graumlich.Emily
Heyerdahl,Gary Larson,Andrea Lloyd, Hugo
Loaiciga.StevenMcKay,DonaldMcKenzie,David
Meko, David W. Peterson,Linda Wilkinson,
Connie Woodhouseand Clinton Wright. Statistical review was provided by Loveday Conquest
and Michacl Arbaugh. Paige Eagle and Beth
Rochefbrt assistedwith tigures. Researchwas
supportedby the Global ChangeResearchProgram of the U.S. Geological Sun'ey, Biological
Resources
Division.
Ilnnk)ran.W.,{. R. 1992.Local tree dngs andteleconnections:
wa)s to improve the feconstructionof I-ake Supcrjor
. | l a r e rs u p p h e ! . J o u r n a l o l C r c a r L , r k e s R e s e a r c h
l8il57 361.
Brinlman. \\'. ,{. R. 1989. ComparisoDof i\ro indic.rtofsof
climatic change:tree growth and lake Supefior $'ater
m p p l i e s .Q u a t e f n a r yR e s e a r c h
30:36-52.
B r u b a k e rL. . B . 1 9 E 0 S
. p a t i a l p a t t c r nosf t r c c g r o $ l h r r o n a l i . s i n t h e P r c i l i c N o f t h w e \ t .E c o l o g y6 1 i 7 9 8 - 8 0 7 .
C o l e n u t t .N , l .E . . r n d B . H . L u c k m a n . 1 9 9 1 .D e n d r o c h r o n o
logical investigationof k/i-r h.r//i:{ Lurch Valley.
Albcrta.CanadienJoumalol ForestRes*u:ch2l:12211231.
C o l l i c r . R . , J . D y m o n d ,J . M c M a n u s ,a n d J . L u p r o n . 1 9 9 0 .
Chemic. rnd physical propenies of the rvater col
umn at Cmtef Lake. Oregon. In l-1.T. Drakc. G. L.
Larson.J. Dynond. andR. Collier (eds.).CruterLake:
Alr EcosystemStudy. American Associadon ior thc
Advancementof Science.San Francisco.Pp. 69 79.
Cook,E. R. .LndG. C. J.rcob).I 981.Pot'mac Rivcr streanrllow
since 1730 !s rcconstruclcdb! lree rings. Journal of
C l i m r o l o g ) a n d A p p l i c d N 4 e t e o r o l o2g2] : I 6 5 9 - I 6 7 1 .
Cook. E. R. and K. Peters. 1981. The snnxrthing spline: a
ne!r'approachto standafdizingfbresl iDterior1rccring
rviddr sefie\ for dendrcclim.rtic studies. Trcc Ring
B u l l e t i n, l l : , 1 5 - 5 3 .
D r a k c .E . T . . C . I - . l - a r s o nJ. . D , v n m n da, n dR . C o l l i e r( e d s . ) .
1990. Crarcr l-akei An Ecoslsiem Study. Anericrn
Associadolii or !hc Adrancenrentol Science.SanFrunE t t l . C . . 1 .a n d D . L . P e t e f s o n 1
. 995aE
. xtrene climalcand
\,xriation ln tfee growth: individualinic rcsponseln
subrlpine fir (Arnr l./.!tu..r,p.r). Clobal Change B1o l o g r -1 : 1 3 1 - 2 . 1 0 .
Flttl. C. J. xnd D. L. Peterson. 1995b. Growfi rcsponscof
subalf,inefir (Arlcr ldrl'rcd\rd) tocliln c in (hc Olym
pic N{ountains,\!rshinston. USA. Global Changc
t s i o l o g , vl : 2 l l - 2 3 0 .
F o l l a n d .C . K . . T . R . K l l l l . u n d K . Y u . V i n n i k o ! . 1 9 9 0 .O b
sen edclimate\'.lliutionsandchange.ln J. T. Houghlon.
G . J . J e n k i n s .a n d J . J . E p h r a u s ( e d s . ) .C l i m a t e
Charge: Thc IPCC Asscssnrenl.Cambridge Lhiversitr-Press.Canbidgc. UiilcdKingdom. Pp. 195 138.
Fritts. H. C. 1916.Tree Rings and Cllinatc. Acadcmic Press.
Nex York.
F r i t t s .H . C . l 9 9 l . R e c o n s t r u c l i nLga r - e cS c a l cC l i m a l i cP a t
terns iioln Trcc Ring Da!a: A Diagnostic Anall\is.
L h i l c r \ i t ! o l A r i / o n a P r e s s .l u c s o n .
Frirl\. H. Cl.a d J. Guior. 1990. Methodsof calibf ation. veri
l l c a t i o n .a n d r e r o r s t r u c t n . I n E . R . C o o k a r d L . A .
K d r i u k s r j ! ( e d s . ) ,N { e t h o d \o f D e n d f o c h r o n o b g r - :
A p p l i c r d d s i r r t h c E n v i r o n m e n t aSl c i e n c e sK. l u w e l
Acadcnic Publishcrs.I)ordrechi, lhe Netherlands.
F r i r t s ,H . C l . a n d J . M . 1 - o u g h I. 9 8 5 . A n e s t i m a t eo l a \ , e r r g e
annualtcrnpcfaturc\'ariationr tbr Nofth Americr. I 602
r o 1 9 6 1 .C l i m a t i cC h a n g e1 : 1 O 3 - ' 7 ) 1 .
C r a u m l i c h .L . J . 1 9 8 7 .P f e c i p i t a t i o !D. r r i a t i o ni n t h c P a c i l i c
N o f i h . | l e s t( 1 6 7 5 - 1 9 1 5 )i r s r e c o n s t r u c ( eldi o m l r c c
rings. Annal\ of the American Association of Gcog
r a p h e r s7 7 : 1 9 - 1 9 .
ofoee grorth to clinatic
G r a u n l i c h .L . J . 1 9 9 3 .I t e s p o n s e
!aiarion in the nlixed conifef and deciduousibrests
ofthe upper Grert Lakes fegion. Crnadian Journalol'
F o f e s tR e s e a f c h2 3 : 1 3 3 1 . 1 3 .
G r a u m l i c h .L . J . r n d L . B . B r u b a k c r .1 9 8 6 .R c c o n n r u c d o n
o f . u n u u l l e n p c l u r c ( 1 5 9 0 1 9 7 9 1f o r L o n g n i f e ,
\\'.rshingtdr. deri\ed iion rcc nngs. Quarcmar,vRe
s e a f c hl 5 : 2 2 3 - 2 3 , 1 .
Hol me\. R. L. I 983. Computer'rssisledqualil,"-contrcl rn lrce
r i n g d a r i n g a n d m e a \ u r e n e n t .T r e e R i n g B u l l c r i .
.11:69-78.
Holmes,R. L. 198,t.Dendrochonoiogy progranrlibriry user's
nanu.rl. Lrborrtor) ol Trcc Ring Research.fucson.
L a n d $ c h r .J . N l . a n d N . C ' . \ 4 a t a h \ . 1 9 8 6 .O n t h e n a t u r eo f
persistenccin dcndrcc|ronologic recordswith implic r t i o n sl b r h y d r o l o g ) .J o u r n a o
l f H!drologt 86:219111.
Larson.C. L. 1990.Siatusoflhe ten year limnological study
of Cfatef Lake Crater Lake National Park. In E. T.
Drake.G. L. Larson,J. Dymond. and R. Collier(edi.).
Craler Lakci An Ecosyslcm Study. Amcflcan Asso
ci.rlion fof the Advancement of Science. San Franc i \ c o .P p . 7 1 8 .
I - o a i c i g a .H . A . , L . H a s t o n , a n d J . M i c h a e l $ n . 1 9 9 3 .
Dendrch) drology .rndlong lerm hydrologic phenom
c n a .R c l i c w s o f G c o p v s i c s3 l r l 5 1 l 7 l .
L o a i c i g a .H . A . , J . M i c h a e l s e nL, . H a s t o n ,S . F a r ! e r .a n dR .
B . l - c i p n i k .1 9 9 2 D
. r o u g h t si n r i v e r b a s i n os f t h e w e s t
e r n U D i l e d S l a r e s .G c o p h l s i c a l R c s c a r c hL c t t c r !
l9:2051-205.1.
\fmtua. N. J.. S. R. Harc. Y. Zbang.J. M. wallacc. and R. C.
F r a n c i s .1 9 9 7 A
. P a c i f i ci n t e r d e c a d a
c l i m a t eo s c i l l a tion with inpacts on salnon production. Bulletin of
thcAncrican N'lclcorologicalSociet]'78i1069 1079.
Nleko. D. Nl. and C. W Stockton. 1985.Tree-ringinferences
on hislorical changcsin the level of Crea! Salt I-ake.
In P A. Kay and H. F. Diaz (eds.),Problens of and
Prospects
for Predicting GreatSaltLake Le\ els.Center
fbf Public Ailidrs and Adnininration. U i\ersil) of
uuh. SBh Lake Cir!. Pp. 63 69.
\'Ieko. D. N{.. C. W. Siockton, and W R. Boggess-1980.A
tree-rirg reconshrclion of drcughr in soulhcm Cali
forlia. whter Reiourcestsulledn I6:59,1-600.
MichaelseD.J. 1988. Cross \alidalion in slali\tical clinalc
forecasrmodels.Joufnal of Cl imrtoloey rnd Applied
Nlcrcodogr. 26t | 589 I 60t-).
Nlichaelsen.J.. L. Hasron. nd F. $'. D.r\is. 1987.400 vears
of ccnlral Calilbmia prccipitalion rariabilit_vrccon
structed from tree ring\. $'ater Re\ources Bulletin
l3:809-818
Nathcnson.M. 1992.waterbalance lbr Crarerl-ake.Oregon.
U . S . G e o l . S u n . O p e n - F i l eR e p . 9 2 - 5 0 5 .U . S . G e o logical Sur!e,v.N{enlo Park. Calif.
Petefson,D. L. 1998.Clinrle, limiling lactors and cr!iron
nrentalchangein high-altit de fofestsof$,esternNofth
Americ.r. In N,l.Benislon and J. L. lnnes (eds.).The
Impacts of Clinrate variabiliiy in Forests.Sprrnger,
H e i d e l b e r gP. p . 1 9 1 - 2 0 8 .
Peterson,D. L.. M. J. Arbaugh. and L. J. Robinson. i991.
Effecisofozone andclimateon ponderosapine (l',lra
prrrd€lor., gro$lh in thc ColoradoRock)'.lllountains.
C a n r d i r nJ o u r n a o
l f F o r e s tR e s e a r c h
2 3 : 1 7 5 01 7 5 9 .
Pcrcrson.D. w. and D. L. Peterson.1994.ElTectsofclimate
on radial growih of mbalpine conilers in the Nonb
C.rscudeMountains. CanadianJoumal of Forcsl Rc
s e a r c h2 , 1 : 1 9 2 11 9 3 2 .
Petefson.D. W .rnd D. L. Peterson.Growth responsesol
mounlainlenrlock (Illgri n",t"rliirL,
to annualand
decadalclimatic vadability in the Pacilic Nofihwest.
Manuscripl rn revie\|'.
Redmond.K. T. I 990. Cralcr I -akcclimatc andlakc l$'cl \ ari
a b i l i r y . I nE . T . D r a k e ,G . L . L a r s o n .J . D ) n l o n d .a n d
R. Collier (eds.).Crater Lake: An EcosystemStudy.
Anrcrican Association for the Advancementof Sci
e n c e .S r n F f r n c i \ c o .P p . l 2 l - 1 . 1 1 .
Hydrological Variation at Crater Lake
129
R o b e r t s o nE. . O . a n dL . A . J o s z a .1 9 8 8 .C l i n a d c r e c o n s t r u c
tion frdn tree rings ar BaDff. Canadian Joumal of
Forest Research18:888 900.
Shen, H. W aDdG. Q. Tabois. 1995. Drough! analysisu,ith
reser!oirs using tree ring reconstructcdllows. Jour
n a t o f H l d r a u l i c E r g i n e e r i n gl 2 l r , { 1 3 , l l l .
Snrlth, D. J. and C. P Laroque. 1998- Mountain heniock
gro\r'th dynamjcs on Vancouvef Island. Northwest
Scicncc 72:67 70.
Stahlc. D. W and M. K. Cleareland. 1993. Reconstruction
and analysis of spring rainfall over the southeastern
tJ.S. ibr the pasr 1000 years-Bulletin of the Aneric a n \ , l e t e o r o l o g i c aSlo c i e t y7 l : 1 9 : 1 71- 9 6 1 .
Slahle.D. w.. tr{. K. Cleaveland,and J. G. Hehf- I98E. Norh
Carolim climac changcsrcconstructedtrom treerlng!:
A . D . 3 7 2 t o 1 9 8 5 .S c i e n c e2 , 1 0 : 1 5 1 71 5 1 9 .
Stine, S. 1990. Late Holocene fluctuations of Mono Lake.
eastemCalifornia. Palaeogeography,
Palueoclimalol
o g y .a n d P a l a e o e c o b g7y 8 : 3 3 3 ' 3 8 1 .
Received.hnuan3l, 1999
Acceptedfor publication May 21, 1999
130
Peterson.
Silsbcc,and Redmond
Slinc. S. 199:1.Er.lrcmc and persistentdroughr in California
andPatagoniaduring mediaevaltime. Nature369: 5,165,19.
Stockton,C.W 1990.Clnnaiic. hydrologic and water supply
iDferencesfrom tree rings. Civil Engineering Prac
t i c e .S p r i n g1 9 9 0 : 3 75 2 .
Villalba. R., T. T. Veblen, and J. Ogden. 199:1.Clinalic in
fluenceson the gfowth ofsubalpiDetreesin the Colorado Front Range.Ecolog) 75: I 450- I :16l.
Witson, R.t. M. C. Zin]'owera. and R. H. Moss, R.H. (eds.).
1996.Chnate Charge 1995:Inp cls,Adaptadonsand
Mitigarion of Climale Changei Scientific Tcchnical
Anallscs. Cambndgc University Press.Cambridge.
Unitcd Kingdom.
Woodhousc.C.A. l993.Trec grcwlh response!o l-NSO evenls
in lhe central Col.JradoFlonl Rangc.Ph!sical Ccog
raphy 1,1:,117
435.
'low
Young. K. C. 1994.Reconslructingstrear
trl1lcscnesrn
centralArizlrna using monlhl) prccipitatronand lrcc
ring rccords.Journalof Clilnate 7:361 17.1.