VOL.
?, NO.
5
WATER
RESOURCES
RESEARCH
OCTOBER
.1971
Clear-CutLoggingandSediment
Production
in the
Oregon Coast Range
GEORGE
W.
BROWN
AND
JAMES
T.
KRYGIER
School o/ Forestry, OregonState University, Corvallis, Oregon 97331
Abstract. The impact of road construction, two patterns of clear-cut logging, and controlled slash burning on the suspendedsediment yield and concentration from three small
watersheds in the Oregon Coast Range was studied for 11 years. Sediment production was
doubled after road constructionbut before logging in one watershed and was tripled after
burning and clear-cutting of another watershed. Felling and yarding did not produce statistically significant changes in sediment concentration. Variation in the relation between
sediment concentration and water discharge on small undisturbed streams was large. Conclusions about the significanceof all but very large changes in sediment concentration are
limited becauseof annual variation for a given watershed,variation between watersheds,
and variation with stage at a given point.
In the Pacific
Northwest
commercial
forests
cover much of the headwaterlandscape.Timber
harvest is the predominant land use in these
forests. In Oregon alone 500,000-700,000acres
are logged annually. Logging is preceded by
road construction.Clear-cuttingis the principal
logging technique, followed by slash burning.
The changein the appearanceof the landscape
is dramatic and abrupt. The crucial questionfor
water quality is: How does clear-cut logging
affect erosion and sedimentation
in headwater
streams?In Oregon this questionis crucial because salmon, steelhead, and trout use these
small streamsfor spawningand rearing.
The purposeof this paper is to describethe
effect of road building, clear-cut logging, and
slashburningon suspended
sedimentproduction
from three forested watershedsin the Oregon
Coast Range, where precipitation averages100
inchesannually and topographyis steep.
The effects of sediment on fish have been sum-
causemortality, delayed development,or poor
conditionin salmonand trout [Brannon, 1965;
Kramer, 1965; Koski, 1966; Shelton and Pollock, 19'66;Servizi et al., 1969]. Gravel size has
beenrelatedto the interchangeof dissolvedoxygen [OregonGame Commission,196.7;Ringler,
1970]. Cooper found that depositionwill occur
in spawninggravels at moderate concentrations
even though velocitiesare too high to permit
depositionon the surface.
Public concernfor pollution has led to the
establishmentof water quality standards.Oregon has now gone beyond interstate standards
(e.g., the Water Quality Act of 1965) and has
applied water quality standardsto subbasinsto
controlthe quality of upstreamwaters directly
[OregonDepartmento[ EnvironmentalQuality,
1969]. These standards have been set without a
full understandingof sedimentconcentrations
or
sediment production rates from mountain
streams under either natural conditions or con-
marized by Cordone and Kelley [196.1]. Excessive concentrations of suspended sediment
(20,000 ppm) can causegill injury or alteration
of behaviorpatterns.The mostimportant effects
of sediment at more typical concentrationsoc-
Forest hydrologistshave often related sediment productionto timber harvest operations
in headwaterareas. Road constructionpreceding logging is often the most serious cause of
cur because of alteration and destruction of bot-
erosion. In the volcanic formations of the Ore-
tom organismsand because of indirect influencesof sedimentdepositionon intragravel flow
and aeration. Mineral and organic sedimentsin
water or deposited in spawning gravels may
gon Cascades,
sedimentyieldsfrom three small,
steep watersheds tributary to the McKenzie
River seldom exceeded200 ppm before treatment [Fredriksen,1965, 1970]. Immediately
1189
ditionsinfluencedby loggingoperations.
1190
BROWN AND KRYGIER
after roads were constructed
across one water-
shed, • peak sediment concentration of 1780
ppm was observed,250 times that recordedin
• control watershed.
This initial
couldbe detectedin the concentrations,
possibly
becauseof inadequatesampling.Sheridanand
McNeil [19681 found small increasesin the
effect subsided
percentageof fine sedimentsdepositedin the
after 2 months, but concentrationsremained stream gravels after loggingin this same area.
two to three times the level predicted from the The probablesourceof this sedimentwas debris
control. These results did not include samples avalanches,which were common in clear-cut
from
landslide
events. In
19'61 and 1964 road
landslides produced average concentrations
about 34 times greater than that expectedfrom
the pretreatment relationship. Mean annual
sediment yield including bed load was 8000
t/mi • in a 9-year period, !09 times the lossfrom
an undisturbed control watershed.
At Castle Creek in California, where the primary influencewas roads,averagesedimentconcentrations
and loads from a 4-mi 2 watershed in-
creasedfivefold the first year, from 64 to 303
ppm (935-4600tfmi•). Concentrations
and yield
declined to twice the normal rate in the second
year [Rice and Wallis, 19,62; Anderson and
Wallis, 196.51.In the Idaho. granitic batholith,
roads associatedwith jammer logging (a high
density road system) in one seasonproduced
highly variable sediment yields from three
logged watersheds: 12,400, 8900, and 89 t/mi •
[Copeland, 19651. Neighboring drainageswithout roadsin this area of highly erodiblegranitic
soil produced no sediment; in watershedswith
roads high yields were attributed to inadequate
cross drains.
The effectsof the loggingoperation are often
difficultto separate.In many erosionstudiesthe
sedimentcontributedby road construction,skid
trails, and loggingare measuredtogether.One
such study at Fernow Experimental Forest in
West Virginia reportedan average.turbidity of
490 ppm (Jacksonturbidity units) duringtractor logging. One year la,ter the average turbidity droppedto 38 ppm; 2 years later it was
only i ppm. Another study at this forest illustrates the importance of planning logging
operations.On a well-plannedloggingoperation, the maximumturbidity was only 25 ppm.
An adjacentwatershedwas loggedwithout any
plan or direction,and maximum turbidities of
56,000 ppm were recorded [Reinhart and
Eschner, 1962].
Sediment was sampled before, during, and
after clear-cut logging in the Maybeso and
Harris River valleys in southeastern Alaska
areas.
Fredriksen [1970] reported that on a watershedclear-cutover a 3-year period with a skyline system,and thus without roads, concentrationswereonly modestlyaffectedduringlogging.
The mean concentration during storm periods
remainedbelow 10 ppm until slidestriggeredby
the record storms of 1965 brought about 800
tons of soil and rock material
into the channel.
Most of this material remained trapped by
loggingdebris.
Controlledslashburning is a commonpractice after clear-cuttingin the Pacific Northwest.
Little
information
exists about the effect of con-
trolled burning on sediment production from
forests.Burning after loggingwtih the skyline
system describedabove was also reported by
Fredriksen [1970]. Resulting sedimentconcentrations during two subsequentyears ranged
from !00 to 150 ppm and were 67 and 28 times
those recorded
on an undisturbed
watershed
during the same period. Fredriksen noted that
sedimenthad beentrapped in the loggingdebris
and was releasedonly after burning.
THE
STUDY
In 1958 OregonState University began a cooperative study of the effectsof loggingon the
water quality and fishery resourcesof three
small watershedsin the Alsea basin in Oregon.
These watersheds are located about 8 miles
south of Toledo and about 10 miles from the
Pacific Ocean (Figure 1). The watershedswere
forestedwith Douglas fir and alder. Mean elevationsare 740, 850, and 1000 feet. Mean slopes
are 35, 37, and 50%. The maritime climate
producesa mean annual precipitationof about
100 inches.Summersare dry, however,and most
of the rainfall
occurs between November
and
April. The soilsare derivedfrom the Tyee sandstone formation. Over 80% of the soilsare from
either the Slickrock or the Bohannon series. The
Slickrock soils are derived from sandstone col-
luvium and are fairly deep. The Bohannon
[Me,ehan et al., 19'691.No significantchange series,a shallow,stonysoil,is derivedfrom the
Sediment
Production
1191
sandstoneresiduum. Both seriesare moderately
DEER
CREEK
stable.
FLYNN
CREEK
The sediment yield characteristicsof the
watershedswere monitored for 7 years before
treatment (1958-1965). Suspended sediment
is measured at the mouth of each watershed and
at six small stream gagesin Deer Creek. The
gagesat the mouth of each watershedare operated by the U.S. Geological Survey and
integrate the effectsof land use on each watershed. The small gagesin Deer Creek are operated by the OregonState University School
of Forestry. These gageswere installedin 1963
to evaluate the effect of each cutting unit on
TO•EDO
the tributaries within the Deer Creek watershed.
NEEDLE
Logging roads were constructedinto Deer
Creek and Needle Branch between March
to evaluate
SCALE,
IMILES
and
August 1965. Flynn Creek, a 500-acre watershed, served as a control and remainedin its
natural conditionthroughoutthe study. Sediment sampleswere collectedduring the winter
of 1965-1966
TO
HARLAN
MILES
the effect of road
FISH
TRAP
ROAD
LABORATORY
STREAM
STREAM
LOGGING
GAGE
UNIT
WPORT
TOLEDO
STUDY
AREA
building. Loggingbegan in March 1966 and
ended in November of that year. The 175-acre
Needle Branch watershed was fully clear-cut.
The 750-acre Deer Creek watershed was 25%
clear-cut,with three smallunits (Figure 1). The
effects of these three units were measured at
four weirs. The 138-acre watershed above weir
WALOPORT
Fig. 1. Watersheds and sediment sampling stations in the Alsea watershed study.
Commissionpersonnel.During stormssamples
2 was 30% clear-cut, the 100-acre watershed were taken at more frequent intervals to ascerabove weir 3 was 65% clear-cut, and the 39- tain sediment loads as stream levels changed.
acre watershed above weir 4 was 90% clear-cut. At the small weirs within Deer Creek, samples
The 572-acre watershed above weir 6, which were taken only during storms.
measuresthe combined effect of the upper
CI-IANGES
IN ANNUAL
SEDIMENT
LOAD
watersheds,was 25% clear-cut.
The slash on Needle ]Branch was burned in
Analyseswere run on two aspectsof the susOctober 1966. The upper units of Deer Creek pended sediment data: (1) annual sediment
remainedunburned; the lower unit was lightly load and (2) suspended
sedimentconcentration.
burned in October 1966. Sedimentsamplingon Changesin annual sedimentload for the three
Needle ]Branchthe following winter reflected watersheds were estimated by an averaging
the combinedeffect of road building, logging, techniquedesignedto reduce the variation in
and slash burning. Sediment measurementsat
sedimentassociated
with a changingstreamflow
the six stationson Deer Creek permitted evalu- regime.In streamflowstudiesconductedconation of the first two effects; U.S. Geological currently with these sedimentstudies,the timSurvey samplesincludedthe effect of burning ber harvest produced significantly increased
the lower unit.. Sediment sampling continued volumes of streamflow on both treated waterthrough the 1967-1968 and 1968-1969 storm sheds[Harper, 1969; Hsich, 1970].
seasons.Sedimentyields will be monitored for
A long-termflow-durationcurve was usedto
severalmore years.
estimateannual sedimentyields on the basisof
Routine suspendedsediment concentrations the flow regime during an average year. The
in parts per million were obtaineddaily at the followingprocedurewas usedin this averaging
U.S. GeologicalSurvey weirs by Oregon Game technique' (1) six years of data from the call-
1192
BROWN
TABLE
1.
AND
Total Annual Sediment Yield in Tons per Square Mile Computed from U.S. GeologicalSurvey
'•
Records
Flynn Creek
Water
Year
1959
1961
1962
1963
1964
1965
1966
1967
1968
1969
Deer Creek
Needle Branch
Normalized
Actual
Normalized
Actual
92
172
136
212
223
337
246
136
92
123
66
258
84
127
209
1237
300
137
59
139
114
193
178
285
231
308
577
251
101
162
82
286
97
160
199
1040
740
213
84
162
bration period were usedto estimatethe longterm
I•RYGIER
flow-duration
characteristics
of
each
Normalized
74
98
201
161
181
129
270
570
372
279
Actual
49
180
115
115
187
422
365
904
490
517
no reasonto suspectthat treatment would reducesedimentyield.
Annual sedimentyields were highly variable
during the pretreatment period. There was a
stream; (2) • relationbetweenmean daily sediment concentrationand mean daily dischargeat
each of the three weirs was developedfor each threefold difference between the minimum and
year of the study; and (3) the sedimentcon- maximumannualyieldson eachwatersheddurcentration-dischargerelationshipwas combined ing this period, even when normalized values
with the mean flow-duration curve for each weir
to obtain the mass of sediment carried in each
flow class.Summingthesevaluesprovided estimates of total annual sedimentyield from each
watershed. Thus this technique assumedthat
the flow each year was equal to the long-term
mean in volume and distribution.This assumption normalizedthe effect of abnormalyears by
reducing the variation in sediment yield associated with annual differencesin discharge.
The sediment yields thus attained provide an
indicationof the averageexpectancyof a change
associated with the treatments.
The normalized
were used.
Road building significantly increasedsediment yield in Deer Creek (the patch cut watershed) during the 1966.water year. One road
slide produceda sedimentyield of 349 tons
(40% of the nonnormalizedyield) for the 1966
water year.
The increasein annual sediment yield after
road buildingin Needle Branch was alsostatistically significantat the 95% level of probability.
Road drainageand erosionof sidecastmaterials
alongroadsseemthe most likely sourcesof this
increase.No large slide occurredon roads in
annual sedimentyield from each of the treated
watershedswas then compared to that of the
Needle Branch.
are shown in,Table 1. Included are both normal-
year. During the two postloggingwater years
(1968 and 1969') sediment yields returned to
prelogginglevels.
Annual sedimentyields in Needle Branch increasedmarkedly immediately after the watershed was logged and burned. The normalized
sedimentyield increasedfourfold over the pretreatment mean. The normalized yield on the
confro,1watershed dropped to, three-fourths of
The annual sediment yield observed during
controlby regression
analysis.A similaranalyti- the first year (1967) after logging in Deer
cal techniquehas been describedby Anderson Creek was significantlyhigher than that during
[1954].
the control period. This yield may include maAnnual sediment yields for each watershed terials depositedby the large slidethe previous
ized (weighted)yieldsand 'actual' yieldsprovided from the annual sediment hydrograph
analysesreportedby the U.S. GeologicalSurvey. Regressions
comparingnormalizedannual
sedimentyield on Flynn Creek (the control)
with that of each treated watershed are illus-
trated •n Figures2 and 3. 0nly the upper,95%
confidence limits are calculated because there is
Sediment
Production
1193
tion analysestherefore use only rising-stage
data. Rising-stagedata were further segregated
to include only storm dat• with discharges
I
6001966
greater than 5 csm (cubic feet per secondsquare
mile).
500-
Simultaneoussamplesof sedimentconcentration and dischargewere fitted to a• regression
400 --
equation of the form:
300
log S = aq- blog D
200
(1)
where S is the sedimentconcentration,D is the
dischargemeasuredwhen the sedimentsample
was obtained,and a and b are regressionconi
øo
i00
i
200
i
300
I
400
stants.
500
FLYNNCREEKSEDIMENT,T/MI 2
Fig. 2. Comparisonof no.rm•lized annual sediment yield from Flynn Creek (unlogged) and Deer
Creek (p•tch cut) for 7 yea.rsbefore treatment.
Comparative yields after road building (1966)
•nd logging (1967-1969) are shown in relation
to the 95% confidence limit on the pretre•tment
regression.
its pretreatment mean during this year. The
annual sediment yield declined during the following years as vegetationreturned, but yields
remained higher than those before loggingand
burning.
CHANGES
IN
SEDIMENT
varies with water-
shed treatment is of great significance,because
most of the new water quality standards for
sedimentare related to this relationship rather
than to annual yield.
Regressionscomparing instantaneous sediment
concentration
with
but that in dischargeas well. Thus the assump-
tion of orthogonality
in the test of individual
degreesof freedom may not apply, and the
standardtest for interactionbetweenregression
equationsmay not be appropriate.
The test statistic seleetedto circumvent this
CONCENTRATION
A secondanalysiscomparedthe instantaneous
concentrationsof sedimentduring storms with
the streamflow at which the sampleswere obtained. This analysiswas done both before and
after road building and logging.Understanding
how sediment concentration
Evaluating the differencesin sediment concentration resulting from different watershed
treatments proved a difficult task. Sample sizes
tend to be unequal. Variations in the sedimen•
concentration-discharge
equation can also occur
becauseof annual changesin runoff pattern.
Two additionaltypes of variation may be imposedby treatment: Clear-cuttingmay change
not only the variation in sedimentconcentration
the
streamflow
ob-
served when the sediment sample was taken
were preparedfor eachsamplingstation. Segregating the concentrationdata into two groups
for these analyses was necessary,becausethe
sediment-streamfiow relationship for rising
stageswas significantlydifferent from that for
falling stages at all sampling sites. A similar
procedurehas been used by Fredriksen [1970].
The correlationcoefficientr was generallymuch
higher for rising stage data. All the concentra-
600
--
--
i•1)67
•500
--
-
,.z,400 -
:•
-r
--
•6
300
-
ß
m200•
T00
o
•
1966
A
•
•oo
/
/
--
A
--
A
200
--
•o
•o
•
Fig. 3. Comparison of normalized annum sedimen• yield from Flynn Creek (u•logged) and
Needle Branch (clear-cut) for 7 years before lreatmen•. Comparative yields •f•er road building
(1•6) and logging and burning (1967-1969) are
shown in rel•ion
to the 95% confidence ]imi• on
•he pre•reatmen• regression.
1194
•Row•
A•D
from analysisof separatesamplesand annum
yield. In Deer Creekthe road slidein 1966pro-
difficulty was:
F2
SSB,,
-- (SSB•
q-
where SS• is the error sum of squaresobtained
by combiningor poolingthe sedimentconcentration-dischargedata for the (1) pretreatment
period and (2) each treatment year for each
watershed (,4 or B), S& is the error sum of
squaresfor the pretreatment period, and SS• is
the error sum of squaresfor the posttreatment
period. Watershed,4 representsthe control watershed (Flynn Creek) and watershedB repreents either of the treated watersheds (Deer
Creek or Needle Branch).
Mean
streamflow
and
sediment
concentra-
tions are shownin Table 2 for the U.S. Geological Survey weirs and in Table 3 for the small
weirs within Deer Creek for each year of the
study. 0nly 2 years of pretreatment data were
used in this analysis becausemore data were
not
available
from
the
small
KRYGm•
weirs
in
Deer
Creek. 'Notation of a significantincreaseis the
result of testing regressionsby equation 2
rather than by simpletests on the mean values.
Some interesting differencescan be drawn
duced an increasedannual yield that was significant at the 95% level of probability. The
relative significance
droppedin the analysisof
sediment
concentration.
The increase was not
significantat 95% but was significantat 90%.
The annual yield in the subsequentwater year
(1967) was still significantlyhigherthan that in
the pretreatmentperiod (at the 95% level of
probability).The comparison
of sedimentconcentrationsduring the 1967 water year with
thoseduringthe controlperiodrevealedno significant differences.The reasonfor this discrepancy is that there was a significantshift in the
sediment-discharge
relationshipof the control
watershedduringthis year.
The shift in the sediment concentration-dis-
charge relationshipof the control likely occurred
as a result of the flood of December
1964-January1965 and was producedby residual materials deposited during those major
events. Before the floods the maximum concen-
tration of suspendedsediment recorded on
Flynn Creek (the control) was 682 ppm, comparedto 969 ppm on NeedleBranch.During the
TABLE 2. Analysis of SimultaneousSampling of SuspendedSediment Concentration and Streamflow
at U.S. GeologicalSurvey Stations during Rising Stages and DischargesGreater than 5 csm
Sediment Concentration,
ppm
Water Year
Sample
Size
Range
Mean
Streamflow, cfs
Range
Mean
FlynnCreek (Control)
1964-1965
1966
1967
1968
1969
72
64
28
18
17
1-205
1-718
38-439
32-256
1-200
194
128
148
109
57
4.2-148
4.2-66
4.2-69
11-44
7-44
1964-1965
1966
1967
1968
1969
71
66
32
20
49
1964-1965
1966
1967
88
68
89
1-969
1-892
1-6300
116
179'
589
1.6-45
1.6-27
1.6-25
1968
1969
44
38
20-7670
70-738
640•
280•
1.9-32
3.1-24
32
22
32
26
20
Deer Creek (PatchCut)
1-1610
1-6960
53-670
35-345
6-381
267
337*
233
115
90
6-204
6-115
20-105
10-46
8-76
58
32
59
40
37
Needle Branch(Clear-Cut)
* Significant increaseat 90% level of probability.
• Significantincreaseat 95% level of probability.
13
9
11
10
12
Sediment Production
TABLE
3.
1195
Analysisof SimultaneousSamplingof SuspendedSediment Concentrationsand Streamflow
at Stations on Deer Creek during Rising Stagesand DischargesGreater than 5 csm
Sediment Concentration,
ppm
Water
Year
Sample
Size
Range
Mean
1964-1965
1966
1967
45
15
14
Deer Creek œ
1-716
2-178
3-242
1964-1965
1966
1967
43
15
14
Deer Creek 3
1-793
1-410
4-194
1964-1965
1966
1967
16
14
13
1-99
1-10
1-8
1964-1965
1966
1967
15
15
19
Deer Creek 6
3-462
1-720
1-366
Streamflow, cfs
Mean
Range
85
32
67
1.0-34
1.0-9
1.0-30
8.3
3.7
11.3
99
117
72
0.8-19
1.2-18
0.8-18
4.6
4.9
7.1
18
2
4
0.3-10
0.3-8
0.3-7
2.1
1.8
2.0
Deer Creek 4
152
176'
122
4.5-136
4.5-108
6-84
34.1
27.3
33.5
* Significantincreaseat 90% level of probability.
flood the maximum concentration on Flynn
Creek was 2050 ppm, comparedto 476 ppm on
Needle Branch. Most pools in Flynn Creek
were filled with sediment (R. C. Williams, unpublishedreport, 1965). Thus the controlwater-
shed respondeddifferently to the same storm
event from the other two watersheds,which
were treated the next year. Anderson [1968,
1970] hasnotedthe dissimilarresponses
of other
watershedsto the samelarge event. I-Ie has also
shown that materials deposited during these
events provide a sedimentreservoir for many
subsequent
years.Thus the classicalconceptof
a single'control'watershedfor sedimentstudies
of this type may not alwaysbe valid.
The samepattern appearson Needle Branch.
The differenceduring the first water year after
logging(19'67) is even more profound.A fivefold increase in mean sediment concentration
was not significantat the 90% level of probability becauseof the upward shift in the sediment-dischargeregressionof the control watershed during this same year. The increaseon
Needle Branch is significantat about 87%, but
the differencein statistical significanceis still
surprising.
The comparisonof changesin sedimentconcentration
for the small weirs in Deer Creek is
shownin Table 3. Road building producedsignificant changesonly at station 6. Station 3,
immediately below the road slide, showed no
increase
in
sediment
concentration
because
samples were not taken during this event.
Logging did not produce significantincreases
in sediment concentrationat any of the sampling stations in Deer Creek.
A frequencydistributionof mean daily sediment concentrationduring low flow periods is
shownin Table 4. Mean daily flow of lessthan
5 csm occurredduring 60-70% of the year on
these coastal watersheds.
Even
with the severe
treatment given Needle Branch during the 1967
water year, mean daily concentrationswere less
than 10 ppm during about 97% of these lowflow days. These data substantiatethe fact that
in mountain watersheds the majority of the
sediment load is carried during a few large
storms. The best indication
of treatment effect
is shown in the increased maximum concentration at low flow on Needle Branch from the
1966-1969 water years. The pattern is similar to
that of sedimentyield shownin Table 1.
DISCUSSION
The results of this intensive study of sediment yield and land use clearly illustrate the
1196
BROWN
TABLE
4.
AND
KRYGIER
Frequency
Distribution
ofMeanDailySuspended
Sediment
Concentrations
during
Days
with Mean
Flow of Less than 5 csm
Concentration Class, ppm
Water
Year
0-5
5-10
10-20
1959-1965
1966
1967
1968
1969
91.9
84.4
97.4
98.1
92.8
5.4
11.0
2.2
1.1
6.8
2.1
4.2
0.4
0.4
0.4
1959-1965
1966
1967
1968
1969
89.3
84.7
77.1
79.3
89.4
7.7
11.5
16.7
17.9
4.6
2.5
3.4
1.2
0.8
4.5
1959-1965
1966
1967
94.4
89.8
96.6
1968
1969
20-30
Maximum
Concentration,
ppm
Days
per
Year
0.3
26
21
13
26
12
231
263
237
267
251
0.2
28
28
52
53
37
237
261
240
257
216
15
74
220
413
230
235
256
238
250
230
• 30
Flynn Creek
0.3
0.4
0.4
Deer Creek
0.3
0.4
3.8
0.4
0.9
1.6
0.9
4.7
9.0
0.8
Needle Branch
0.8
0.1
0.4
0.4
1.4
0.4
0.4
0.8
93.2
2.4
2.0
2.4
93.4
3.5
1.3
1.4
0.4
Values are percentageof days in each classwhen flow was lessthan 5 csm.
effect of several forest managementpractices
on water quality. The influence of roads on
sedimentyield hasagainbeendemonstrated
by
this study. Our results substantiatethe conclusionsdrawn by Fredriksen [1970]. The road
system in Deer Creek, for example, was carefully located,constructed,and used.The roads
were locatednear the ridges.They enteredthe
watershed from the back of the ridges; thus
road mileage within the watershed was minimized. The roads were well graveled and were
not used during the winter months. Even with
theseprecautions,one slide occurredand its effect was quite significant.A large volume of
material still remainstrapped behind a logjam
in the upper part of the watershedand provides a potential sourcefor additional sediment
yield at somelater date.
About
1.5. miles
of road
were
constructed
within the Needle Branch watershed. This con-
struction, together with the landings for logs,
exposed mineral soil over about 7% of the
watershed.This exposedsoil was undoubtedly
the sourceof sedimentin the 1966 water year.
High-lead logging alone did not produce
amountsof sedimentsignificantlydifferent from
thosein the calibrationperiod. This result also
coincideswith those observedelsewhere[Fredriksen, 1970; Lull and Satterlund, 1963; Meebernei al., 1969; Packer, 19'67]. The maximum
sediment
concentration
observed at weir 4 in
Deer Creek was lessthan 20 ppm. The watershed was 90% clear-cut but unburned. This
concentrationcan be compared with a maximum sedimentconcentrationof over 7000 ppm
observedafter loggingand burning in Needle
Branch.
Slashburning after clear-cuttingis a common
managementpractice in the Pacific Northwest.
Arguments both for and against burning are
numerous. A review of this controversy is
clearly beyondthe scopeof the presentpaper.
The sedimentdata collectedduring the study,
however,indicatethe effectof burning on water
quality.
The slashfire in NeedleBranchwasextremely
hot; mineral soil was exposedthroughoutmost
of the watershed. High sediment yields can be
expectedwhen mineral soil is subjectedto more
than 100 inchesof rain during a 6-month period.
The cause of increasedsediment yield after
loggingin Deer Creek is somewhatobscureand
may be the result of interacting factors. Upstream, the sediment contribution of the two
Sediment
Production
1197
clear-cuttings,as indicated by changesin the chargerelationshipis stagedependent.A much
sediment concentration-discharge
regressions, better correlation between sediment concentrawas not significant.Downstream,two possible tion and dischargewas observedduring rising
sourcesof sedimentexist: (1) materialsthat ac- stages.Thus one concludesthat a great deal of
cumulated in the stream channel as a result of
experience,together with an intensive, rigidly
the road slidethe previousyear (the mean con- standardizedsamplingschemebasedon flow recentration of sediment at station 6 in Deer
gimes,is requiredbeforea judgmentwith a preCreek, though not statistically significant,re- cisionof 10% can be made.
mained high, some residual sourcethus being
Our ability to makeaccuratejudgmentsabout
indicated) and (2) the clear-cutting farthest changesin the sedimentconcentration-discharge
downstream,(the fact that the clear-cuttingbe- relationship in this study would have been
tween station 6 and the U.S. GeologicalSurvey greatly improved by replicating the control.
station was lightly burned after loggingcould The assumption that neighboring watersheds
have contributedto increasedsedimentyield).
respondin similar fashion to similar events,
The data indicatethat sedimentyieldsshould regardlessof magnitude,is certainly questionapproach pretreatment levels 5-6. years after
able. Thus it would seem that any sediment
completeclear-cuttingand burning.Two impor- monitoring systemwould require more than one
tant concepts must be understood about this
controlfor comparison.
recovery: (1) These results pertain to a case
A further constraintin sedimentsamplingor
study conductedin an area in which vegetation monitoringis imposedby the influenceof a few
grows rapidly and returns to unoccupied sites
very quickly. Exposure time for mineral soil is
thus minimized. (2) Erosion is significantfor
both terrestrial and aquatic habitats. Although
the supply of sedimentfrom the slopesmay deeline rapidly, the presenceof this material in
the stream gravels may persist. Such an accumulationof fine materialsin spawninggravel
can significantlyreduce the emergenceof salmonid fry [Hall and Lantz, 1969].
What inferencesabout sediment sampling or
monitoringcan be drawn from the data? This
questionis crucial, not only from the point of
view of studying sediment transport processes,
but from that of water quality as well. The
Oregon water quality standards [Oregon Department o.• Environmental Quality, 1969], for
example,specify that no activities will be permitted that cause 'any measurableincreasesin
natural stream turbidities when natural turbidi-
ties are less than 30 Jackson Turbidity Units
(JTU) or more than a 10 percent cumulative
increase in natural
stream turbidities
stream turbidities are more than 30 JTU
large storms on annual sedimentyields. If these
eventsare not sampledadequatelyor are missed,
conclusions
about the treatment effect are likely
to be erroneous.This problem is compounded
by the treatment itself, which imposesa greater
variation on the sediment-dischargerelationship, particularly at high flows. Thus monitoring a specificstream to detect a 10% changein
sediment concentrationwill require more than
just a few random samples.
We have shown that clear-cut logging may
produce little or no change in sediment concentrations in small streams. The greatest
changeswere associatedwith the road building
operation that preceded logging and the controlled slash burning afterward. We have also
shown that unlessthese changesare large, it
may be very difficult to separate man-caused
changesin sediment concentrationfrom those
imposedby natural variation, particularly if
very large runoff events occur within the measurement period.
when
...'
Acknowledgments. Study cooperators include
the Georgia-Pacific Corporation, the U.S. GeoThe important questionis how to obtain the logical Survey, the U.S. Forest Service, the Federal
best standard of comparison.What, in other Water Quality Administration, Oregon State Uniwords,is a 'natural' sedimentconcentrationfor versity, and the Oregon Game Commission. Funds
for this portion of the study were provided by
small streams? Our data indicate that rather
large annual variations in the sediment-dis- Federal Water Quality Administration grant WP423 and Oregon State University. The authors are
charge relationship can occur on undisturbed particularly indebted to Dr. Henry Anderson and
watersheds.Variations between watershedsmay Dr. Scott Overton for their assistance in data
also be large. Variation in the sediment-dis- reduction and analysis, Wayne Hug for 5 years of
119,8
BROWN AND 1{RYGIER
intensive sediment sampling, and the U.S. Geological Survey for making their recordsavailable.
REFERENCES
Anderson,It. W., Suspendedsediment dischargeas
related to streamflow, topography, soil, and land
use,Eos Trans.AGU, $5(2), 268-281, 1954.
Anderson,It. W., Major flood effect on subsequent
suspended sediment discharge (abstract), t•os
Trans. AGU, 49(1), 175, 1968.
Anderson, H. W., Relative contributions of sediment from source areas, and transport processes,
in Proceedings of Symposium on Forest Land
Usesand Stream Environment, pp. 34-6,2,Oreg.,
State Univ., Corvallis, 1970.
Anderson, It. W., and J. R. Wallis, Some interpretations of sediment sourcesand causes,Pacific Coast basins in Oregon and California,
Proceedings of the Federal Interagency Sedimentation Conference,1963,Misc. Publ. 970, pp.
22-30, U.S. Dep. of Agr., Washington, D.C.,
1965.
Brannon, E. L., The influence of physical factors
on the development and weight of sockeye
salmon embryos and alevins, Progr. Rep. 12,
26 pp., Int. Pac. Salmon Fish Comm., Seattle,
Wash., 1965.
Copeland, O. L., Land use and ecological factors
in relation to sediment yields, Proceedings of
the Federal Interagency .Sedimentation Conference, 1963, Misc. Publ. 970, pp. 72-84, U.S.
Dep. of Agr., Washington, D.C., 1965.
Cordone, A. J., and D. E. Kelley., The influence
of inorganic sediment on the aquatic life of
streams,GaliJ.Fish Game, 47, 189-228,1961.
Fredriksen, 1• L., Sedimentation after logging
road construction in a small western Oregon
watershed, Proceedings of the Federal Interagency Sedimentation Conference, 1963, Misc.
Publ. 970, pp. 56-59, U.S. Dep. of Agr., Washington, D.C., 1965.
Fredriksen, R. L., Erosion and sedimentation following road construction and timber harvest on
unstable soils in three small western Oregon
watersheds,Pac. Northwest Forest Range Exp.
$ta. Res. Pap. PNW-104, 15 pp., U.S. Dep. of
Agr. Forest Serv., Portland, Oreg., 1970.
Hall, J. D., and R. L. Lantz, Effects of logging
on the habitat
of coho salmon
and
cutthroat
trout in coastal streams, in .4 Symposium on
Salmon and Trout in Streams, edited by T. G.
Northcote, pp. 353-375, University of British
Columbia, Vancouver, 1969.
Harper, W. C., Changesin storm hydrographsdue
to clearcut logging in coastal watersheds, M.S.
thesis, Oreg. State Univ., Corvallis, 1969.
ttsieh, F. S., Storm runoff response from roadbuilding and logging on small watershedsin the
Oregon coast range, M.S. thesis, Oreg. State
Univ., Corvallis, 1970.
Koski, K. V., The survival of coho salmon (Oncorhynchus lcisutch) from egg deposition to
emergencein three Oregon coastal streams, M.S.
thesis, Oreg. State Univ., Corvallis, 1966.
Kramer, R. It., Effects of suspendedwood fiber
on brown and rainbow trout eggs and alevins,
Trans..4met. Fish $oc., 94, 252-258, 1965.
Lull, It. W., and D. R. Satterlund, What's new in
municipal watershed management, Proc. Soc.
.4mer. Forest., pp. 171-175, Washington, D.C.,
1963.
Meehan, W. R., W. A. Farr, D. M. Bishop, and
J. It. Pattic, Effect of clearcutting on salmon
habitat of two southeast Alaska streams, Pac.
Northwest Forest Range Exp. $ta. Res. Pap.
PNW-82, 45 pp., U.S. Dep. of Agr. Forest
Serv., Portland, Oreg., 1969.
Oregon Department of Environmental Quality,
Water quality standards for the Umpqua River
basin, Oregon, 34 pp., Portland, 1969.
Oregon Game Commission, A study of the effect
of logging on aquatic resources,1960-66, Progr.
Memo., Fish. 3, 28 pp., Res. Div., Portland, 1967.
Packer, P. E., Forest treatment effects on water
quality, in International Symposium oi Forest
Hydrology, edited by William E. Sopper and
Howard W. Lull, pp. 687-689, Pergamon, New
York, 1967.
Reinhart, K. G., and A. R. Eschner, Effect on
streamflow of four different forest practices in
Allegheny Mountains, J. Geophys. Res., 67(6),
2433-2445, 1962.
Rice, R. M., and J. R. Wallis, How a logging
operation can affect streamflow, Forest Ind., 89
( 11), 38-40, 1962.
Ringler, N., Effects of logging on the spawningbed
environment in two Oregon coastalstreams,M.S.
thesis, 96 pp., Oreg. State Univ., Corvallis, 1970.
Servizi, J. A., R. W. Gordon, and D. W. Martens,
Marine disposal of sediments from Bellingham
Harbor as related to sockeye and pink salmon
fisheries,Progr. Rep. 23, 38 pp., Int. Pac. Salmon
Fish Comm., Seattle, Wash., 1969.
Shelton, J. M., and R. D. Pollock, Siltation and
eggsurvival in incubation channels,Trans. Amer.
Fish Soc., 95, 183-187, 1'966.
Sheridan, W. L., and W. J. McNeil, Some effects of
logging on two salmon streams in Alaska, J.
Forest., 66(2), 128-134, 1968.
(Manuscript received January 25, 1971;
revised May 4, 1971.)