RIVER RESEARCH AND APPLICATIONS

River. Res. Applic. 24: 941–959 (2008)
Published online 1 April 2008 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/rra.1112

ECONOMIC ANALYSIS OF WATER TEMPERATURE REDUCTION PRACTICES
IN A LARGE RIVER FLOODPLAIN: AN EXPLORATORY STUDY
OF THE WILLAMETTE RIVER, OREGONy
SAICHON SEEDANG,a*,z ALEXANDER G. FERNALD,bx
RICHARD M. ADAMS cô and DIXON H. LANDERS dk
a

b

d

Institute of Water Research, Michigan State University, Room 101A Manly Miles Building, 1405 S. Harrison Road, East Lansing,
MI 48823-5243, USA
Department of Animal and Ranges Sciences, New Mexico State University, Box 30003 MSC3-I, New Mexico State University, Las Cruces,
NM 88003, USA
c
Department of Agricultural and Resource Economics, Oregon State University, 200A Ballard Extension Hall, Corvallis,
OR 97331-3601, USA
U.S. Environmental Protection Agency, National Health and Environmental Effects Research Laboratory, Western Ecology Division, 200 SW
35th Street, Corvallis, OR 97333, USA

ABSTRACT
This paper examines ecosystem restoration practices that focus on water temperature reductions in the upper mainstem
Willamette River, Oregon, for the benefit of endangered salmonids and other native cold-water species. The analysis integrates
hydrologic, natural science and economic models to determine the cost-effectiveness of alternative water temperature reduction
strategies. A temperature model is used to simulate the effects of combinations of upstream riparian shading and flow
augmentations on downstream water temperatures. Costs associated with these strategies are estimated and consist of the
opportunity costs of lost agricultural production and recreation opportunities due to flow releases from an up-stream reservoir.
Temperature reductions from another strategy, hyporheic flow enhancement, are also examined. Restoration strategies
associated with enhanced hyporheic cooling consist of removal/reconnection of current obstacles to the creation of dynamic
river channel complexity. The observed red; uction of summer water temperatures associated with enhanced channel complexity
indicates that restoring hyporheic flow processes is more likely to achieve cost-effective temperature reductions and meet
the total maximum daily load (TMDL) target than conventional approaches that rely on increased riparian shading or/and
combinations of flow augmentation. Although the costs associated with the hyporheic flow enhancement approach are
substantial, the effects of such a long-term ecological improvement of the floodplain are expected to assist the recovery of
salmonid populations and provide ancillary benefits to society. Copyright # 2008 John Wiley & Sons, Ltd.
key words: water temperature models; hyporheic temperature; cost-effectiveness analysis; Willamette River; floodplain restoration;
economic analysis
Received 21 December 2007; Accepted 7 January 2008

INTRODUCTION
Over 70% of Oregon’s population lives in the Willamette River basin, and three-quarters of Oregon’s economic
output is derived from activities in the basin (Willamette Restoration Initiative, 2001). However, after 150 years of
human use and proximity to the river, there are signs of declines in the state of ecological health in the region. These
symptoms include changes in the diversity of river and floodplain landscapes, a recent degradation of river water
*Correspondence to: Saichon Seedang, Institute of Water Research, Michigan State University, Room 101A Manly Miles Building, 1405 S.
Harrison Road, East Lansing, MI 48823-5243, USA. E-mail: seedang@msu.edu
y
This manuscript has been subjected to the United States Environmental Protection Agency peer review process, and modified with permission
from the American Society of Civil Engineers (ASCE).
z
Environmental Scientist.
x
Associate Professor.
ô
Professor.
k
Senior Scientist.

Copyright # 2008 John Wiley & Sons, Ltd.

942

S. SEEDANG ET AL.

quality after several decades of improvement, a widespread increase of invasive plants and animal species and the
disappearance and decline of many native plants and animal species (Willamette Restoration Initiative, 2001).
Among these changes, an important public policy concern in the basin and in the Pacific Northwest region is the
decline of native salmonid fish species (National Research Council, 1996; National Marine Fisheries Service,
1999). Salmon have played an important economic, cultural and religious role in the Pacific Northwest for
centuries. These factors have motivated local, state and federal agencies as well as private organizations, to set
restoration goals for the Willamette River and other watersheds for the benefit of salmon and other species.
Numerous restoration plans have been proposed by public agencies and private organizations (e.g. the Pacific
Northwest Ecosystem Research Consortium (PNW-ERC), Oregon Biodiversity Partnership, Oregon Wetlands Joint
Venture, Willamette Valley Livability Forum and the state of Oregon). Several proposed habitat restoration
strategies include restoring main and side river channels, dam removal, replanting riparian forest and reducing
in-steam water temperature. However, few studies have attempted to measure the biological and economic tradeoffs
within the mix of activities and resources to be used in each management plan. The lack of comprehensive,
systematic assessments of proposed plans is due in part to the complexity of modelling important physical and
ecological processes, and in estimating the possible outcomes associated with manipulations to such systems. In
addition, analyses of the cost-effectiveness of proposed restoration plans have generally not been integrated into
current efforts. Another area of management interest that has not been addressed in current assessments is the
potential role of hyporheic flow enhancement1 in the suite of possible techniques to reduce in-stream
water temperatures. The research reported here provides an exploratory assessment of the cost-effectiveness of
alternative strategies, including the potential for hyporheic process enhancement, to reduce temperatures in a large
river system.
The overall objective of this research is to conduct an economic analysis of alternative restoration practices to
address excessively high water temperature in a large dynamic river floodplain system. Specifically, economic and
biophysical information will be used to assess the efficiency of various restoration plans associated with
compliance of in-stream temperature targets needed to restore salmonid populations. Information on restoration
costs of each strategy, the resulting ecological benefits (temperature reductions to enhance ‘aquatic life use’ as
required by Clean Water Act (Copeland, 1999)), and the physical and economic tradeoffs between proposed
strategies are needed to provide guidance when considering investments in ecosystem restoration and management
of a large river. The upper mainstem Willamette River is selected as the empirical focus for the study because prior
work indicates that this is an appropriate location for the proposals explored here (Gregory et al., 1998; Dykaar and
Wigington, 2000; Fernald et al., 2000). Water temperature is selected as the key parameter to determine the
effectiveness of the restoration approaches. This is motivated by current regulatory interest (i.e. Oregon Total
Maximum Daily Loads (TMDLs) development, Oregon Water Temperature Standard) and availability of water
temperature models to perform such a management analysis.

BACKGROUND
The mainstem Willamette River flows northward for 299 km through a large alluvial floodplain of the Willamette
Valley before entering the Columbia River near Portland (Altman et al., 1997). The upper river, the primary focus of
this research, has a higher gradient with a wide floodplain, and it is characterized as a complex system of
meandering and braided channels containing numerous side channels and islands (PNW-ERC, 2002). Mean annual
flow on the mainstem (measured at the Albany USGS Gauge 14174000) is approximately 400 m3 s
1 (Dykaar and
Wigington, 2000). Most of the flow in the Willamette River occurs from November to March as a result of winter
1
There is increasing recognition of hyporheic flow influencing water quality and other ecological functions, specifically in an alluvial river
landscape (Landers et al., 2000, 2002; Ward et al., 2002). Hyporheic flow is a relatively new area of research. As a result, its definition often
changes with each researcher’s focus, the physical characteristics of the study site and the methodology of the study. This research follows the
definition of hyporheic flow by Fernald et al. (2000, 2001 and 2006) as it relates to the Willamette River. Specifically, hyporheic flow is river
water flowing along a gradient and passing through various channel features (e.g. alcoves, bars, islands). During summer, a mixing of surface and
subsurface water flows beneath these features and results in temperature cooling in micro-habitats (e.g. alcoves, side channels) and the main river
channel. Greater channel complexity results in increased hyporheic flow and a substantial cooling effect which may considerably influence
overall river water temperature.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

943

rain and spring snowmelt. About 35% of the annual runoff is from snowmelt, either directly into the stream or
indirectly through the ground-water system (Laenen and Bencala, 2001). Summer and fall low-flow periods are
significantly affected by federally managed reservoir operations, which have historically been managed for flood
control and summer low-flow augmentation since the 1970s. As a result of low-flow augmentation, the amount of
water supplied to the main channel during the summer months is elevated over previous uncontrolled flows
(PNW-ERC, 2002). Unfortunately, there are no historical temperature data bases for the upper main channel of the
Willamette that are of acceptable quality to reconstruct historic, pre-dam temperature characteristics in the river
during summer low flow. However, the known reduction of complexity and connectivity that has resulted in about
an 80% loss of total channel complexity (PNW-ERC, 2002) has, undoubtedly, resulted in major temperature
alterations. The primary alteration is the loss in the quality and quantity of cold refugia that were derived from the
historic off-channel habitat cooled by hyporheic connectivity and shaded by extensive gallery forests. Moreover,
the loss of in-stream large wood has resulted in fewer cool scour holes associated with root wads. Currently,
summer water temperatures in the entire mainstem Willamette River exceed the state water temperature standard,
and the river is listed as ‘water temperature limited’ under the 2002-303(d) list. The primary purpose of the
temperature standard is to protect fish.2
Most salmonids in the basin, including Chinook salmon (Oncorhynchus tshawytscha) and steelhead trout
(Oncorhynchus mykiss), spawn in the tributaries of the Willamette River and use the mainstem river for rearing and
migration habitats (Altman et al., 1997). The decline of many native fishes in the Willamette River has been widely
recognized and reported. The primary causes are the decline of water quality, including a change in temperature
patterns, the decline of total channel length as a result of bank revetment and disconnection of main channel to side
channels and the loss of riparian shade, structure and diversity, especially in urban and agricultural areas (Benner
and Sedell, 1997; Willamette Restoration Initiative, 2001). In addition, Willamette anadromous salmonids face the
same ocean and migratory problems confronting other stocks of salmonids in the Northwest, such as ocean
conditions, harvest rates and other impediments to migration, and pollution. These external conditions are beyond
the scope of this study.
A number of studies have observed and modelled how stream water temperature is influenced by various factors
including human influences such as land use practices, destruction or removal of riparian vegetation and flow
diversion and dam operations (Beschta et al., 1987; Bartholow, 1991; Hostetler, 1991; Isaak and Hubert, 2001).
Knowledge about stream water temperature responses to these physical and biophysical changes can be linked to
develop and evaluate potential management practices. Hyporheic flow restoration is another potential approach in
addressing problems related to water temperature. In a large floodplain, water temperature reduction resulting from
hyporheic flow processes can be promoted by expanding the extent (i.e. connectivity and complexity) of river
channel characteristics (Boulton et al., 1998; Fernald et al., 2000). For example, the removal of constructed erosion
protection structures (i.e. rip-rap) along a river that tend to reduce channel complexity and diversity should be
considered a crucial restoration practice. Efforts to reconnect channels to the floodplain can also be promoted,
especially in the potentially active reaches that are currently not extensively used for societal and economic
purposes (Gregory et al., 1998). Reconnecting a river with its channel and floodplain may improve hyporheic flow
temperature and water quality, but also could result in retention and storage of water (which results in reduced
magnitude of peak flows), sustained base flow and increased habitat complexity (Coulton et al., 1996; Boulton
et al., 1997).
An important policy consideration when making decisions related to habitat restoration is determining which
strategies are most effective in meeting the restoration target at the lowest cost. There is a need to integrate
economic information and biophysical complexity of habitat ecosystems with restoration management practices.
Examples of efforts to link the complexity of the habitat ecosystem with economic information of management
decisions can be found in several studies. For example, Kuby et al. (2005) use basin level optimization models to
evaluate tradeoffs between ecological benefits and economics cost of dam removal in the Willamette basin. Paulsen
and Wernstedt (1995), Hickey and Diaz (1999) and Stevens et al. (2002) applied economic concepts to the design of
2
In the mainstem Willamette River, the 7-day average maximum water temperature criterion is 17.88C (648F), except in the lower portion of the
river (the first 50 miles from the river’s confluence with the Columbia River), where the standard is 208C (688F) (Oregon Department of
Environmental Quality, 2003).

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

944

S. SEEDANG ET AL.

environmental management policies involving complex physical and natural resource information and applied
optimization techniques to select management methods that attained the desired objective (e.g. a minimum stream
flow requirement) at the least cost. Johnson and Adams (1988) studied the economic benefits of flow levels on
steelhead populations in the John Day River, Oregon. They quantified the benefits of increased fish numbers to
anglers under various water allocations by combining a fish production model with an on-site economic valuation
analysis. Adams et al. (1993) later developed a bio-economic model for riverine management in the John Day
River. The model was used to develop a cost-benefit analysis comparing the efficiency of management actions in
terms of improved fish production and habitat.
Several studies have investigated investments in habitat conservation programmes in terms of their efficient
allocation of funds (e.g. Wu and Boggess, 1999; Wu et al., 2000; Wu and Skelton-Groth, 2002; Watanabe et al.,
2005). These studies found that an efficient allocation of funds must consider the cumulative effects and
environmental benefits, which depend on the threshold effects of conservation efforts of targeted restoration sites.
The conditions investigated by these studies involve traditional riparian habitat investments (e.g. reestablishing
riparian corridors) to reduce water temperature for anadromous fishes in streams and tributaries in small
watersheds. The research described here differs in that it investigates and compares investments in various
restoration practices to reduce water temperature in a large river floodplain. The results from this research, while
exploratory, provide information on selecting appropriate restoration investments suited to sites characterized as
large river floodplains.

PROCEDURES AND METHODOLOGY
Restoration practices
Three general categories of management practices that can be implemented in the study area to reduce water
temperature include: (1) increasing riparian shade by planting trees along the main channel; (2) flow augmentation
via releases of cooler water from upstream reservoirs and (3) restoring/recreating channel complexity to promote
hyporheic flow functions. Riparian planting is considered a key restoration strategy for any ecological approach
which aims to improve fish habitat and water quality in the Northwest (Kauffman et al., 1997). Active restoration
strategies, such as planting riparian buffer zones with native plant species along the upper mainstem Willamette
River, may recover some of ecological functions that have been lost over time primarily due to agricultural
conversion. In particular, shading of the steam surface has the potential to reduce stream temperatures by reducing
the incidence of solar radiation. The value of foregone agricultural production and the costs of establishing such
riparian vegetation buffer zones are the primary costs of this practice.
Stream temperatures can also be cooled by deep-water releases from reservoirs in the upper portion of the basin,
given that hypolimnetic water released from reservoirs is substantially cooler than incipient stream temperatures.
Costs of this practice include lost recreational opportunities in the reservoirs (due to accelerated drawdown), and
the lost value of the water for alternative uses.
Reconnection of the floodplain to the main channel to create more complex fish habitats and promote fluviohydro-geomorphic processes that contribute to hyporheic cooling is another potential restoration management
practice (Landers et al., 2002; Fernald et al., 2006). Over the past century, placement of rock boulders or revetments
(i.e. rip-rap) to protect farmland and property from erosion has created obstacles to the natural maintenance and
creation of fluvial hydro-geomorphic processes (e.g. erosion and sedimentation). Removal of some of this bank
protection would promote the ecological connection of river and floodplain. However, the quality of floodplain soil
is particularly well suited for high valued agricultural uses, such as orchards, nursery stock and other specialty
crops. As a result, any proposal to allow some of these farmlands to revert to more natural riparian state will require
compensation for lost production from such highly valued agricultural land.
The effectiveness of restoration practices
Two models are used to measure the effects of the selected management activities on water temperature
reduction in the study area. The first model is the CE-QUAL-W2 water temperature, which is used to measure the
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

945

response of water temperature to a change in riparian shading and tributary inflow into the upper mainstem
Willamette River. The second model is a hyporheic flow temperature model, used to determine water temperature
change associated with an increase in channel complexity.
CE-QUAL-W2 water temperature model. The CE-QUAL-W2 is a two-dimensional longitudinal/vertical
(laterally averaged), hydrodynamic water quality model. The model predicts water surface elevations, velocities
and temperatures as well as other water quality variables. A recent version of this model (Version 3.1)
incorporates the slope of river sections, and has the capacity to model the entire river basin with features such as
dams, rivers and lakes (basin simulation). The model incorporates the effects of both vegetative and topographic
shading in a stream channel. Vegetation characteristics represented in the CE-QUAL-W2 model are tree top
elevation, distance of tree from the river centreline and tree density. The algorithm of the CE-QUAL-W2 model
uses the position of the sun to determine which topographic inclination angles coincides with the direction of
incoming solar radiation. Eighteen directions around each model segment are defined in this model (Annear
et al., 2001). This Willamette temperature model has been calibrated to ensure appropriateness of all model
components and applicability by the research team at Portland State University (PSU) and the model has been
used for developing TMDL management programmes of the Oregon Department of Environmental Quality in the
entire mainstem and most of its major tributaries. The calibration model-data error statistics for the mainstem
Willamette River indicate that the model performs well in terms of temperature predictions. For example, the
maximum daily water temperature calibration error statistic at all 12 calibrated stations has an absolute mean
error (AME) of less than 18C for temperature predictions (Annear et al., 2003a). At the Corvallis observation
station used in this research (downstream of the upper mainstem river), the AME value is within 0.358C of the
observed data.3
The model reach of the upper mainstem Willamette River used for this study, approximately 92 km, is shown in
Figure 1. Downstream water temperature north of Corvallis was selected as the reference site to investigate the
temperature responses to upstream management scenarios. The CE-QUAL-W2 model was used to predict the
response of water temperature to various scenarios of riparian shading on the upper section of the mainstem
Willamette River and tributary flow increases from Cougar reservoir (a U.S. Army Corps of Engineers reservoir)
via the McKenzie River. Simulated water temperatures from each management scenario were exported to Excel
spreadsheets to calculate a 7-day moving average of daily maximum temperatures. These were compared with the
state water temperature standard throughout the summer from June through the end of September, the period of the
year when elevated water temperature is of greatest concern.
Hyporheic flow model. In addition to the CE-QUAL-W2 water temperature model that includes riparian
planting and tributary inflow management, another temperature model is needed to investigate the effects of
management practices that promote hyporheic flow functions. While the study of hyporheic flow effects on
temperature is a relatively new research area, there is recent research on hyporheic flows in the Willamette River
that is relevant to this study (Fernald et al., 2000, 2001, 2006). Fernald et al. examine the relationship of hyporheic
water temperature and dynamic channel features and show that hyporheic water temperature is closely associated
with hyporheic flow in that surface water cools as it flows through various channel features (e.g. side channels,
islands, sand bars, channel bed and banks), reemerges, and mixes with surface water downstream. The result of this
mixing decreases the ambient main channel surface water temperature. Current understanding indicates that the
extent of cooling depends on several factors, including hydraulic gradients, flow volumes, area and type of channel
features and vegetation types. We further explored these relationships and estimated hyporheic cooling, and
interpolated the results to policy analysis of restored channel complexity.
Hyporheic cooling was estimated along a 67 km reach of the Willamette River using a hydrometric estimate of
total hyporheic flow amount, an empirically based function for hyporheic cooling; and a mixing model to estimate
river cooling from hyporheic flow. Hyporheic cooling is most likely in summer when river water is relatively
warmer. Concurrent temperature measurements of hyporheic flow and main channel flow at a site near Corvallis,
3
The CE-QUAL-W2 model application to the mainstem Willamette River was calibrated by Professor Scott Wells and the Water Quality
Research Group at Portland State University. The results of model calibration using 2001 and 2002 data produce similar statistical results (Berger
et al., 2003, pages 143 and 151). For the detailed discussion of model calibration or validation/verification in the context of the water quality
model, CE-QUAL-W2, see pages 23 to 25 of Cole and Wells (2007). Additional publications and applications related to this model can be found
in Berger et al., 2002, Annear et al., 2003b, Wells et al., 2004, Rounds, 2007.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

946

S. SEEDANG ET AL.

Figure 1. Upper Willamette River model study. Note: The upper reach of the mainstem Willamette River is selected to investigate water
temperature responses under various management scenarios. Spatially, the length of the upper mainstem Willamette River model for this study is
approximately 92 km. The model starts from south of Eugene (near the confluence of the Middle Fork and Coast Fork Willamette) to north of
Corvallis. The model simulation used here was from 5 June 2001 to 30 September 2001. The Water Quality Research Group at Portland State
University provided assistance in separating this model from the main model for this reach of the Willamette River, as well as calibrating the
model for this research. This figure is available in colour online at www.interscience.wiley.com/journal/rra

Oregon, showed hyporheic water was cooler than main channel water from late May until early September, after
which hyporheic water was warmer than main channel water. An annual combined surface and subsurface heat
budget might show a net balance for hyporheic temperature effects. The focus in this study is on hyporheic cooling
that occurs in the critical period of June through August when river temperatures are high and most prone to
exceedance of regulatory standards for fish tolerance. Bed form interacting with flow velocity affects stream-bed
hyporheic exchange with rapid residence times in shallow depths into the stream bed (Cardenas and Wilson, 2007).
Channel morphology also affects surface–subsurface exchange, with high flux and short residence times within the
upper meter of the stream bed (Cardenas et al., 2004). Our method focuses on head gradient-driven hyporheic flow
and cooling that takes place over long hyporheic flow paths.
River features conducive to hyporheic flow, where hydraulic gradients conduct water into the river bed and
banks, were identified with 1998 aerial photos based on 1997–1999 field studies (Fernald et al., 2001, 2006). A GIS
coverage (Figure 2) was generated with three types of features: (1) islands in their entirety; (2) peninsulas separated
from the river bank by a line 308 off thalweg flow direction, representing the typical subsurface flow path from river
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

947

Figure 2. Example island, peninsula and point bar hyporheic flow features of the Willamette River study area, with lines denoting crosssectional width, WH, for each feature. This figure is available in colour online at www.interscience.wiley.com/journal/rra

to alcove head and (3) meander bends with less than 1000 m radius of curvature and within 80 m of the river bank,
where hyporheic residence times are expected to be short enough that subsurface water exhibits hyporheic
characteristics. Vegetated and bare gravel areas of each feature were distinguished in separate GIS coverages.
Hyporheic flow volume was calculated for each feature with Darcy’s law (Hubbert, 1969). Feature crosssectional width, WH (m), was digitized perpendicular to the hyporheic flow direction (Figure 2). Feature flow depth
multiplied by WH gave mean cross-sectional hyporheic flow area, AH (m2). Flow depth for peninsulas was set equal
to alcove depth derived from a function relating peninsula length to alcove depth based on bathymetry data from
16 representative alcoves (Landers et al., unpublished data). Peninsula flow depth ranged from 0.7 to 2.0 m. Based
on our field studies of hyporheic flow paths, we assigned representative flow depths of 2.0 m for meander bends
and 4.0 m for islands. Feature mean hyporheic flow path length, LH (m), was calculated as feature surface area
divided by WH. Hyporheic hydraulic gradient (DH/DL) was LH divided by the change in elevation over LH. The
change in elevation was calculated using distance travelled along the shoreline of each feature multiplied by

0.001, the representative river water slope. We used surface cover as a surrogate for saturated hydraulic
conductivity, Ksat (m s
1), based on the previous peninsula deposit study (Fernald et al., 2006) and dye tracer
studies (Fernald et al., 2001). Vegetated areas had lower Ksat (0.04 m s
1), and bare gravel areas had higher Ksat
(0.37 m s
1). Single feature total hyporheic flow, QH (m3 s
1), was equal to 
Ksat  AH  DH/DL.
River cooling was calculated with QH and individual feature hyporheic cooling, DTH (8C). A function of cooling
per metre of hyporheic flow path length was developed using LH and previous study hyporheic temperature
data (Fernald et al., 2006). Temperature data were from two periods, July and August 1999, and from six sites
representative of the range of Ksat from low (vegetated) to high (bare gravel) sites. For all features,
DTH ¼ 
30.08 þ 32.28  EXP (
0.001757  LH). To determine the hyporheic-cooling effect on river water
temperature, DTR (8C), we used a simple percentage dilution calculation with hyporheic flow, QH, and river flow, Q
(m3 s
1): DTH  (QH/Q) ¼ DTR. We tabulated DTR from each hyporheic flow feature and summed the change in
cooling from each feature to determine cumulative river cooling.
We identified 149 island, peninsula and meander bend features over 67 km of the Willamette River from river
km 278 to 211. Median feature area was 9680 m2, with all areas totalling 4.26 km2. Median QH of 0.061 m3 s
1
resulted from median DH/DL of 0.002, median AH of 120 m2 and median Ksat of 0.26 m s
1. Median feature
hyporheic cooling, DTH, of 
2.78C resulted in a river temperature change, DTR, of 
0.618C over the study reach.
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

948

S. SEEDANG ET AL.

We feel this is a conservative estimate of hyporheic cooling and represents the lower end of expected
hyporheic cooling of river water for a few reasons. We used conservative approaches in developing relationships,
as shown in the following three examples. First, alcove depth as an index of peninsula hyporheic flow depth may
underestimate hyporheic inflows that begin deeper in the river and arrive in the alcoves. Second, we used a
median hyporheic flow path length for alcoves, treating the peninsula features as triangles. However, in one
case we looked at in detail, the flow paths are longer than the median value and would result in more cooling than
we calculated. Third, our gradient estimates were conservative, being based on a calculation of shoreline distance
that in one case resulted in a hyporheic gradient less that the river gradient where we know there is actually a
steeper gradient through the feature than around the feature. Importantly, we only considered deposit features
visible from aerial photos. Field observations revealed at least 12 major riffle complexes associated with river
channel thalweg gradient breaks over the 30 upstream river km of the 67 river km we analysed. Photo analysis
suggested about the same number on the downstream 37 km. There is surely an important amount of hyporheic
flow through riffles and other features that create hydraulic gradients into the river bed. The cooling functions
of large river riffle complexes have not been fully described (as far as we know) and are an important area for
future research.
For scenario evaluation in the economic analysis, we employed results of previous studies to estimate possible
cooling effects of all hyporheic flow paths. Dye tracer studies suggested 70% of river flow passed through hyporheic
flow paths over the study reach (Fernald et al., 2001). Median hyporheic cooling of 
2.78C for this amount of total
hyporheic flow would result in river cooling of 
1.98C. Though we did not have direct field measurements of the
cooling associated with all hyporheic flow paths, this estimate may approximate the upper end of expected
hyporheic cooling.
An average cooling of 1.38C for these two estimation methods, [(0.61 þ 1.9)/2], is used as a baseline for
existing hyporheic cooling associated with an area of hyporheic features before channel restoration in the study
area. In predicting the cooling effects after channel restoration, we used channel surface area as an indicator
and assumed that when an area of channel surface increases, the channel complexity, the area of hyporheic
features and the degree of hyporheic cooling downstream will also increase proportionately. This assumption is a
very conservative estimation because the cooling effect from hyporheic flow after channel restoration could be
more significant due to a longer hyporheic flow path and larger areas of hyporheic interaction in river floodplain.
It should be noted that there are several factors that affect hyporheic flow and its cooling ability, including flow
magnitudes and their orientation to the features, channel and feature dimensions, and type of substrates and that
further study is needed.
The estimation of channel surface area before and after channel restoration (i.e. removing, breaching or
reengineering revetments and reconnecting old side channels) is obtained from the Pacific Northwest Ecosystem
Research Consortium (PNW-ERC, 2002).4 According to the PNW-ERC Conservation 2050 Scenario, it is
estimated that after restoration (year 2050), the total channel surface area of alcoves and side channels would
increase almost 160% relative to 1995 conditions. This percentage is used as a reference to interpolate the area of
hyporheic features which determine the potential area for hyporheic cooling.
Restoration costs
Estimates of the costs of each restoration action are needed for a comprehensive evaluation of the cost-effectiveness
of resources used in these management practices. These costs should include both the direct implementation costs
and the economic opportunity costs. Direct costs are associated with costs for design, construction, operation,
maintenance and monitoring of a restoration activity. Opportunity costs refer to a benefit forgone due to
implementation of a management practice (e.g. the agricultural benefits lost if farmland is taken out of production).
Major restoration costs associated with management practices include riparian planting, land purchase, bank
revetment removal and lost recreation opportunities in upriver reservoirs. Costs that involve multiple time periods (i.e.
4
The PNW-ERC developed three alternative future plans for the Willamette River Basin. These plans range from greater emphasis on
conservation to greater emphasis on economic development. The ‘Conservation Scenario 2050’ focuses on land and water allocation for
conservation and restoration of ecological functions, which includes reconnecting the river channel and removing and/or modifying some
revetments of meander bends of the mainstem Willamette River.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

949

maintenance and operation in support of the riparian shading approach) are adjusted to reflect present value dollar
amounts in order to make comparisons among management options of various time lengths.5
Riparian planting costs include site preparation, materials (trees), labour and operational costs. Planting costs are
associated with the first year of project implementation as well as the operational costs that occur over the life of the
project. Planting costs are based on a standard cost per unit for Oregon (estimated by the U.S. Department of
Agriculture, Natural Resource Conservation Service (NRCS) for the 2003 fiscal year). The average cost of planting
riparian trees per river kilometre (15 m width) for both sides of the river channels is estimated at $11 250 (or $3000
per acre). The total consists of two major costs, site preparation and planting (trees and labour costs). Other
site-specific costs (e.g. replacement due to first year tree mortality, fencing, tree shelter, irrigation water, stream
bank stabilization, etc.) are likely to vary by site and are not included in this analysis. Riparian planting costs
estimated from several other sites in Oregon (e.g. Grande Ronde River; Tualatin River and Beaver Creek) show
substantial variation because of the differing site characteristics and practices implemented (types of trees, methods
of planting and spacing). The estimated costs per river kilometer from these sites ranged from $7500 to $20 625
(Knoder, 1995; Bishaw et al., 2002; Watanabe, 2003). The costs used here fall within the general range observed at
these other sites.
It should be noted that riparian planting is not included in the hyporheic-cooling analysis of this study, given that
riparian planting may actually reduce channel migration rates (Micheli et al., 2004). In fact, as a result of dynamic
geomorphologic process, riparian vegetation development will occur naturally over time.
Another component of major costs associated with riparian shade and hyporheic flow enhancement practices is
the cost of land. Land prices from market transactions should reflect the discounted value of agricultural production
that could be derived from this land over time. However, obtaining market price information of farmland that is not
affected by other factors in the Willamette Valley, such as urbanization pressures, is difficult. In addition, the
sale price for riparian farmland exclusively is not available. Fortunately, the estimated real market value (RMV) is
available for all counties in the study area.6 Therefore, this research uses RMV estimates to represent opportunity
costs of taking agricultural land out of production. Estimated market values (i.e. RMV values) and land-use zoning
information are provided through Geographical Information System (GIS) or digital data formats obtained for the
three counties within the study area (Linn, Benton and Lane Counties). The average market value of farmland
within 100 m of the upper mainstem Willamette River is approximately $4600 per acre. Within the study area, the
average market value is highest in Benton County and lowest in Linn County. This average market value of
farmland, especially in Benton County, is comparable with the actual sale prices of several parcels of farmland in
2002 (personal communication with Toni Blessing, Appraiser, Benton County, 2003). The average land value is
used to represent the cost of land that would need to be acquired for restoration approaches (i.e. planting trees to
achieve shading or for reconnecting the river to its floodplain to achieve enhanced hyporheic cooling).
In addition to the cost of purchasing land, the hyporheic-cooling practice also include the average costs of
removing rock rip-rap placed along the banks to protect farmland from erosion. The USACE District Office in
Portland estimated an average cost of $10 per cubic yard ($13 per cubic metre) for revetment removal (personal
communication with Mike McAleer, the U.S. Army Corps of Engineers, Portland District, December 2006). The
cost estimate is based only on the removal of rock material. Other site-specific costs, such as the costs of bank slope
stabilization after removal of rip-rap, material disposal, transportation, access roads and permit acquisition are not
included.
To determine the locations and lengths of revetment to be removed, we utilized data from the ‘Conservation
Scenario 2050’ of the PNW-ERC that estimated the length of required revetment on the upper Willamette for
removal to be 2716 m, or about 5% of the upper river total bank length. This proposed restoration also estimated the
acreage of land needed for restoring floodplain natural processes in the next 50 years to be about 890 acres.7

5
All relevant costs are discounted to reflect the net present value with a 5.875% discounting rate, as used by the National Resource Conservation
Services (NRCS) and other federal agencies during the period of 2002–2003(Department of the Interior, 2002).
6
RMV is the estimated price for which property would sell in a transaction at the assessment date for the tax year. RMV is estimated by county
assessors based on several criteria, such as a comparison of sale prices from similar properties and/or physical inspection (ORS 308.205).
7
The excel spreadsheet, as well as other information related to this estimate, was obtained directly from Linda Ashkenas, Oregon State
University, Fisheries and Wildlife Department.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

950

S. SEEDANG ET AL.

Another important component of costs associated with restoration practices is the opportunity cost of
releasing upstream reservoir storage. One of these costs is the reduced flatwater recreation opportunities at
those reservoirs which are drawn down as a result of increases in streamflow. Although there are other uses for
the water stored in the McKenzie River reservoirs, the opportunity costs are judged to be low and are not
included in this analysis.8 The data used to estimate lost recreational benefits are primarily drawn from the
Willamette Basin Recreation Study Report.9 Other sources of data for estimating recreational values (e.g.
reservoir water level, volume, surface area and recreational visitation data) were obtained from the U.S. Army
Corps of Engineers, Portland District. The procedures employed to calculate recreation benefits consist of
three steps (see details in Seedang, 2004): (1) reservoir water levels under flow management scenarios were
calculated using regression analysis linking reservoir elevation, area capacity (acres) and flow release; (2) the
effect of reservoir water levels on recreation visitations was estimated using a coefficient on the water level
variable in the visitation model employed in the Willamette Basin Recreation Study Report and (3) the reservoir
recreation benefits were estimated by multiplying the change in visitations (estimated from step 2) with the
marginal visitation recreation benefit. Visitation recreation benefits (i.e. the willingness to pay or marginal
benefit of an individual recreationist making an additional trip to visit the reservoir), for Cougar reservoir were
estimated by the USACE in the Willamette Basin Recreation Study Report. The recreation benefits due to
changes in reservoir water level reflect the opportunity costs of recreation benefits forgone for the 2001 water
year from releasing additional reservoir water to downstream. The result showed that on average, the value of
reservoir recreation impact during summer (June–September) is $2.75 per acre-foot of water released
downstream (the highest impact is in September at $ 3.4 per acre-foot). This estimated value is low when
compared with other reservoir recreation studies (e.g. recreation values for releases from California Central
Valley reservoirs ranged from $6 to more than $600 per acre-foot (Ward et al., 1996)). This lower value for
Willamette Valley reservoirs may be due to a variety of reasons, such as the presence of alternative recreation
sites and opportunities within the upper basin (e.g. Blue River Lake, McKenzie River), or specific reservoir
characteristics such as shallow bank slopes, which result in less impact on recreation activities from reservoir
drawdown.

Selection of cost-effective strategies
The final step in the evaluation procedure is to combine the information and methods into a decision-making
framework. Specifically, once the responses of water temperature and the costs associated with each proposed
restoration practice is determined, various restoration options can be compared. A cost-effectiveness analysis is
then used to determine the least cost option that achieves a given level of temperature reduction or ecosystem
function target. A cost-effectiveness frontier for these options can also be constructed with these targets.
The cost-effectiveness frontier represents the least cost envelope of points for restoration options that achieve
various levels of temperature reduction. In cost-effectiveness analysis, the preferred alternative is the one
that produces the socially desirable temperature reduction at minimum cost. Simulation results of water
temperature reduction scenarios from riparian shading, flow augmentation and their combinations (total of
25 scenarios) from the CE-QUAL-W2 water temperature model are used for evaluating and constructing the
cost-effectiveness frontier.

8
We used only lost recreational opportunities as the opportunity costs of increased stream flows because the other uses of these storage reservoirs
(flood control, hydropower and irrigation) are unlikely to be affected by the releases for increased streamflows during August and September.
Specifically, there is very limited agricultural use (approximately 1600 ac-ft) of these stored waters. Total hydropower generated will not change
as a result of changes in release patterns, only the KWH rate may vary. However, shifts from one month to the next in this generation are not likely
to result in a significant change in revenues generated. Finally, flood control is provided by drawing down the storage reservoirs in the fall to
create storage for winter and spring flows. Releases of additional stream flows in August would not change the storage capacity of the reservoirs
available in the winter. Should these assumptions change, the costs of the flow augmentation strategy may be underestimated.
9
The Willamette Recreation Study Report is part of the U.S. Army Corps of Engineers Portland District’s (USACE) Willamette Basin Reservoir
Feasibility Study Project. In this report, the operations of the 13 dams of USACE are evaluated to consider the effects of changes in water
demand.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

951

RESULTS AND DISCUSSION
Base case scenario
The base case model simulates a situation where no restoration effort takes place. The downstream profile shows
that summer water temperatures (on the basis of a 7-day moving average of daily maximum temperatures)
during the simulated period (5 June–30 September 2001) violated the state water temperature standard more than
75% of the time, with most of these violations occurring during the months of July and August. Moreover, for 25%
of the summer period, the water temperature exceeded 208C (Figure 3). It is expected that such persistent high water
temperatures during these periods will cause stress to several cold-water fish species, especially salmonids rearing
in this area (Altman et al., 1997). The longitudinal profile of simulated water temperatures for the base case at
several locations is also examined. Water temperatures at all locations are higher than the Oregon state standard
(17.88C). The general profile of water temperature is an increase in temperature as it flows downstream. However,
in some locations, such as the middle section of the river, there is a major cooling effect from the tributary and
possible hyporheic processes. Several dye-tracer studies (i.e. Fernald et al., 2001; Laenen and Bencala, 2001)
performed on this part of the river showed significant existence of hyporheic flow.

Maximum temperature reduction from riparian shading and flow augmentation
The simulation results from the maximum riparian shade scenario show limited water temperature reductions
(average of less than 18C) at the downstream observation reach. In addition, water temperature on most days in July
and August exceeds the TMDL standard (Figure 4). Several factors limit the effectiveness of riparian shade in
reducing water temperature in the Willamette. One factor is the inability of tree height to provide adequate shading
for the river surface area exposed to solar energy. Given the broad nature of the mainstem river channel, and the
north–south orientation, the ability of riparian trees to provide shadows across the entire channel is limited. Also, as
the volume of the main stem river flow is relatively large, it is difficult for cumulative riparian shade to have a
substantial effect on water temperatures over a long length of the river. Cumulative cooling effect from shade may
have a significant effect on water temperature at the local scale or at the site immediately downstream due to a
shorter travel-time of flow before getting exposed to the daily heat. Rounds (2007) used the same model to evaluate

Figure 3. Comparison of simulated water temperatures for the base case scenario with water temperature standard 17.88C (State of Oregon
Water Temperature Standard). Note: The State of Oregon has designated the upper mainstem Willamette River as an area for salmonid rearing
and migration. No activity is permitted that will raise the water temperature of the receiving water body if the receiving water body already
exceeds 17.88C on the basis of a 7-day moving average of daily maximum temperature. The ODEQ uses this information in developing the
TMDL for water temperature for the upper mainstem Willamette River. This figure is available in colour online at www.interscience.wiley.com/
journal/rra
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

952

S. SEEDANG ET AL.

Figure 4. Comparison of water temperature reduction scenarios during the summer months with the base case. Note: Water temperatures
reported here are the 7-day moving average maximum daily temperatures. Base case scenario is the simulated existing condition of water
temperature during the summer of 2001; maximum riparian shade scenario is based on full riparian shade for the entire 92 km with maximum tree
height; and maximum riparian shade with flow increments (þ50%) is the combination of the maximum riparian shade and the additional flow
(þ50%) increments from the McKenzie River upstream tributary of the mainstem Willamette River. This figure is available in colour online at
www.interscience.wiley.com/journal/rra

shading effects on selected-5 mile reaches of the mainstem Willamette and found significant local effect on water
temperature, while there is little cooling effect of temperature over the entire length of the river.
The analysis of the effects of increased water volume in the upper mainstem Willamette River indicates that
augmented flow from the McKenzie River also has limited impact on temperatures. We found cooling effects of
flow augmentations (þ25 and þ50%) to be around 0.25–0.488C. Rounds (2007) also found overall cooling effects
in the mainstem Willamette River were in the similar ranges of our study and the effects of cooling were further
diminished downstream. In our study, even when we combined flow augmentations with maximum riparian shade
efforts, downstream water temperature is not reduced sufficiently to meet the state water temperature standard for
the months of July and August. Specifically, the average summer water temperature can only be reduced by
approximately 18C or about 5% from the base case scenario. However, even a reduction of this magnitude is still
biologically important to salmonid rearing and migration habitats (Hostetler, 1991; Altman et al., 1997).
Cost-effectiveness results from riparian shading and flow augmentation scenarios
A total of 25 management scenarios (including existing condition or no investment) were simulated to
investigate the effectiveness of their impacts on water temperatures. The scenarios are sets of combinations of
riparian shading on the upper mainstem Willamette River (which is further divided into three segments; upper,
middle and lower sections, to investigate the differences of temperature responses). The scenarios also consider
flow augmentation practices (which also considered the temperature responses between releasing flow during the
hottest months (July and August) compared to the entire summer (July to September)). Costs (in present values)
corresponding to the per cent of temperature reduction of the 25 management scenarios are plotted against their
effective temperature reduction (from the base case) to construct a cost-effectiveness frontier. Results show that
only 10 of the 25 scenarios are cost-effective solutions for a given budget and a desired level of temperature
reduction (Figure 5). The other 15 scenarios result in less cooling for the same or higher expenditures. The results
show that to achieve small temperature reductions, increasing reservoir releases in July and August alone are
cost-effective. However, as the target level temperature reduction increases, riparian shade combined with
increased flow is required. Within the combinations of riparian shade and flow increases, the cost-effectiveness of
each point begins with planting riparian for shading on the upper sections of the mainstem river and moves towards
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

953

Figure 5. Selected scenarios on the cost-effectiveness frontier. Total cost per scenario reported here is in present value terms, and maximum
temperature reduction in each scenario is calculated as the percentage difference from the base case. Note: e0, base case (no management) with
existing riparian shade; e23, existing riparian shade plus tributary flow increments (þ25%) in July and August; d20, existing riparian shade plus
tributary flow increments (þ50%) in July and August; d18, full riparian shade only upstream plus tributary flow increments (þ25%) in July and
August; c14, full riparian shade only upstream plus tributary flow increments (þ50%) in July and August; b10, full riparian shade only upstream
plus tributary flow increments (þ50%) in summer (June, July, August, September); b8, full riparian shade only downstream plus tributary flow
increments (þ50%) in summer (June, July, August, September); a5, full riparian shade upstream and downstream plus tributary flow increments
(þ50%) in July and August; a2, full riparian shade upstream and downstream plus tributary flow increments (þ50%) in summer (June, July,
August, September); a1, full riparian shade on all sections of the river (upstream, middle stream and downstream sections) plus tributary
flow increments (þ50%) in summer (June, July, August, September). This figure is available in colour online at www.interscience.wiley.
com/journal/rra

additional planting on the lower sections as greater cooling is required. Importantly, note that the scenarios of
riparian planting for shade alone are more costly than flow augmentation. The total cost of selected scenarios as
mapped on the cost-effectiveness frontier range from no cost for no management effort, to a total cost of almost
$3 million when combining incremental flow augmentation (þ50%) and maximum riparian shade for the entire
92 km associated with a 5% reduction of water temperature (18C temperature reduction) from the base case.
Note that since this analysis is not premised on maximization of total benefits (or benefit-cost analysis), there is
no single optimal or ‘best’ solution in terms of investments in temperature reduction practices, and we are only
searching for least costs ways to obtain a temperature reduction. The solutions shown are based on their costeffectiveness, and can therefore be useful in selecting least cost alternatives.
Hyporheic-cooling effects and costs
The results from the first set of management options indicate that neither riparian planting for shading, nor flow
augmentation, or any combination of both, can reduce water temperatures in the study area to a level that meets the
current Oregon standard. This suggests that improving water quality and fish habitats in large rivers with active
floodplains like the Willamette will need a broader suite of management approaches, including enhancement of
hyporheic flow functions. This section examines temperature reductions and costs as a result of manipulating
hyporheic flows through increasing channel complexity.
Table I summarizes the results of estimated hyporheic cooling associated with existing channel features
and predicted temperature reductions following 50 years of channel restoration. We estimated that there would be a
maximum net reduction of 1.998C from 1995 conditions due to the increase in channel complexity as a result of
restoration activities (through the year 2050).
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

954

S. SEEDANG ET AL.

Table I. Predicted hyporheic cooling associated with channel complexity
Description
Total channel surface area (km2)
Hyporheic feature area (km2)
Cooling effect (8C)

Before restoration

After restoration

Net change

13.6
4.3y
1.3z

35.2
11.1
3.3

21.6
6.8
2.0x


Based on GIS data of the Willamette River in 1995 and a 50-year estimation by the PNW-ERC. Total channel surface area includes main
channel, sloughs and side channels.
y
Based on GIS coverage of active channel from 1996 to 1998 by Fernald et al. (2006). The hyporheic features include mid-channel bars, point
bars and islands.
z
Based on an average cooling from two estimation methods (see details in Procedures And Methodology section).
x
Assuming that when an area of channel surface increases, the area of hyporheic features, and the degree of hyporheic cooling downstream, will
increase by the same proportion.

Estimated costs of restoring channel complexity are based on the PNW-ERC future scenarios (Conservation
2050 Scenario).10 In this research, we focus only on the promotion of hyporheic cooling arising from increased
channel complexity. As a result, the major costs of this approach are land purchases to allow for channel migration,
and the removal of rip-rap. We estimate that the total costs for a net reduction of 1.998C of this approach is
$4.3 million and the average cost of water temperature reduction by this approach is approximately $2.15 million
per 18C reduction. The majority of the costs for this approach is from the cost for purchasing land (about 95% of
total cost).
Several important points regarding these results and associated costs of channel restoration should be noted.
First, this hyporheic approach does not include the costs of any other restoration activity, such as riparian planting.
However, even if riparian buffers are not planted, natural reforestation of the riparian landscape will occur over
time.11 Second, it is uncertain that implementation of a riparian forest buffer is even necessary or desirable after
removal of some rock revetments from the side channels. Since the nature of channel complexity restoration
requires the dynamic processes of erosion and sedimentation, combined with lateral movement of the main river
channel to recreate the hyporheic channel features, replanting riparian forest in some areas (unstable reaches) may
be subject to removal due to future erosion (Dykaar and Wigington, 2000; Landers et al., 2000). This in itself is not
a negative occurrence since trees deposited in the river create channel complexity by contributing ‘large wood’ to
the river system. Moreover, the sediment eroded into the river has the potential of contributing to hyporheic cooling
by being deposited downstream. Therefore, including all costs of forest buffer reestablishment with implementation
of this approach likely overstates the costs of this plan. In addition, since the majority of channel features associated
with hyporheic cooling are in the area near the main channel, cost per degree of hyporheic cooling estimated
here may be over estimated. Third, a restoration approach that focuses on increased hyporheic features as well as
riparian establishment (i.e. cottonwood colonization), would not be successful without considering restoration
of some historical flow regimes which have been altered since the construction of upstream dams. Thus, flow
management, such as increases in the frequency and magnitude of flows (e.g. high flow or bank full discharge) is
desirable and would increase the rate of restoration of channel complexity and connectivity (Dykaar and
Wigington, 2000; Landers et al., 2000).
Restoration of some aspect of historical flow regimes for the Willamette River is mechanically a relatively
straightforward and inexpensive activity due to the various dams and control structures located on the tributaries
that feed into the upper river. The operation of these dams is under the jurisdiction of the U. S. Army Corps of
Engineers and they are currently (January 2007) considering operating the dams to re-establish ‘environmental
10
As suggested by the PNW-ERC team, the restoration processes involved are located near both sides of the mainstem, side channels are planting
with riparian buffers, some bank revetments are removed and some old side channels are reconnected to the main channel.
11
Based on tree ages we have collected, mature riparian forests (10 m high) can be established via natural seeding in 15 to 35 years or an
average of 25 years (Steve Cline, unpublished manuscript). There is the issue of a ‘reset’ period where a newly established gravel bar may
experience a period of dynamic equilibrium (frequent disturbance) that prevents establishment of what will become mature riparian forests. This
has been included in to the above estimate.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

Copyright # 2008 John Wiley & Sons, Ltd.

Reestablishment of maximum
riparian forest on both sides of
the main channel (a buffer
width of 15 m)
An increase of flow from a
reservoir on the McKenzie River
during summer months (June
to September)
Removal of some bank
revetments and restoration of
historical channels to
increase the interaction
between surface and
subsurface water and
promote channel complexity

Riparian shadingy

Costs for removal of bank
revetment and costs of
purchasing lands for channel
migration, including potential loss
of land near the river to erosion

Reservoir recreation benefit
loss from drawdown during
summer months

Costs of purchasing lands
and costs of planting

Cost component

1.99 (9.5)

0.48 (2.5)

0.63 (3.3)

Maximum temperature
reduction from the
base case, 8C (%)

4.28

0.64

2.68

Total cost
(millions)

2.15

1.33

4.25

Average cost per degree
of reduction (millions)


All costs are measured as net present value. The majority of costs for the riparian planting approach are assumed to occur within the first year of the project. The recreation costs for the flow
augmentation are estimated as an annual cost. All costs for the hyporheic-cooling approach are assumed to occur within the first year.
y
This approach assumes shading will provide maximum temperature reduction at a tree height of 35 m, beginning after 10 years of growth and planting activities are within the first year of project
to ensure that shading is maximized within 50 years.
z
Flow augmentation involves increments to mainstem flows through upper reservoir releases during the summer months using two levels of flow increments (þ25% and þ50%).
x
This approach is assumed to provide maximum effectiveness in temperature reduction within 50 years. It is also assumed that the restoration activities occur within the first year, so that it will
take 50 years of sufficient channel complexity to realize the full temperature reduction.

Hyporheic coolingx

Flow augmentationz

Description

Approach

Table II. Descriptions, cost components, assumption notes and cost estimations for each restoration approach

WATER TEMPERATURE REDUCTION PRACTICES

955

River. Res. Applic. 24: 941–959 (2008)

DOI: 10.1002/rra

956

S. SEEDANG ET AL.

flows’ that contribute towards restoration of historic ecosystem services, such as providing more habitat and cooler
temperatures for threatened and endangered native fish species. There would be little additional cost associated
with improving environmental flows in this fashion. For fish habitat to be significantly improved, it is not necessary
to cool the entire river system. By increasing the overall temperature heterogeneity in the upper Willamette River
through restoring habitat and hyporheic cooling opportunities, fish would be able to find cool temperature refugia
more easily and use these refuges during periods of maximal thermal stress.
Finally, the most difficult aspect of river restoration is not the science or economics—both of which appear to be
favourable, but the societal aspect of getting a diverse group of landowners to appreciate the societal importance of
creating this ‘Green Infrastructure’ and its documented ecological and societal benefits. Moreover, these
landowners (mostly family owned farms) can supply products potentially as valuable, if not more so, to society as
the actual crops they are currently producing. These new products would be ecosystem services such as aquatic
habitat, terrestrial habitat, water quality and flood protection. Currently, efforts to establish market incentives, such
as Oregon’s Willamette Partnership, are developing an innovative ‘ecosystem market-based’ approach designed to
improve water quality in the Willamette River basin. This programme encourages landowners to implement
ecosystem restoration and conservation practices such as planting trees or restoring floodplains to reduce
stream temperatures and they would be able to trade (sell) their ‘credits’ to other regulated discharge activities
(e.g. wastewater and storm water discharges from municipalities and industries).
Cost estimation for each approach and its potential maximum temperature reduction is summarized in Table II
along with their descriptions and assumptions. The riparian shading approach has the highest cost per degree of
cooling because it has only a small impact on water temperature, and riparian establishment costs occur at the
beginning of the project. An increase in tributary flows is the least costly approach; however, it has limited potential
for temperature reduction (0.488C), given the small amount of water available in the tributary for such a purpose
during the low-flow summer months. The average cost of a hyporheic-cooling approach is based on an investment
plan that needs to be finished within the first year of the plan, in order to realize the full temperature reduction
within 50 years.

CONCLUSIONS AND FURTHER RESEARCH
Economic and ecological information is needed to assess the cost-effectiveness of various practices when
developing restoration policies to improve water quality and fish habitat. In addition, restoration approaches need to
be adjusted to the specific environmental settings of the area of interest when designing a suite of appropriate
restoration practices for salmonids. Water temperature is one of the most important habitat qualities that influence
cold-water fishes like salmonids. Flow augmentation and riparian shading are considered practical ecological
restoration practices designed to reduce water temperature. These conventional practices have worked well in
reducing temperatures in many small rivers and tributaries; however, they appear to be incapable of lowering water
temperatures in a large floodplain river such as the Willamette, where efforts of flow augmentation have proven
ineffective and the river itself is too large to provide effective shading. An alternate restoration practice that
focuses on floodplain restoration to promote hyporheic cooling may provide a more effective reduction of water
temperatures in the upper Willamette River, which has experienced loss of heterogeneity and cool microsites.
Promoting heterogeneity and hyporheic cooling may be the only feasible approach to provide locations of cool
water in this reach of the river. The benefit of restoring channel complexity and/or connectivity of river and flood
plain is not only for cooling, but it also provides habitat heterogeneity for fish and other wildlife. Though the costs
of this approach are difficult to estimate and may appear high, there are other unintended economic benefits,
including reduction of flood damage (by reducing the amount of agriculture in the lowest floodplains and by
allowing storage and conveyance on floodplains), reduced maintenance costs (not maintaining rip-rap and levees),
improved aesthetics, increased aquatic and terrestrial habitat diversity and recreational opportunities.
This research provides an initial starting point in terms of evaluating alternative management approaches for
investment in water temperature reduction in a large river floodplain. However, these results need to be conditioned
by several features of the analysis. First, the hyporheic-cooling alternative may prove a viable alternative for
restoration management, especially when a resource manager wishes to achieve a relatively large temperature
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

957

reduction target. However, given the many interpolations involved in the estimation of hyporheic cooling reported
here, use of these results should be limited to exploring the potential for alternate restoration policies. Before
specific policies are implemented, additional quantitative analyses of the relationships between hyporheic cooling,
channel complexity and other variables related to temperature reduction are needed.
Second, as with the derivation of ecological outcomes, the economic information developed here is subject to
uncertainty arising from the data and assumptions. For example, some site-specific costs are not included in the
restoration activities and only one opportunity cost (recreation) is assigned to the flow augmentation strategy. While
these may cause an underestimate in costs per unit of cooling, this potential bias may be offset by the potentially
greater degree of cooling from hyporheic flow increases than was used in this study.
Third, water temperature reduction is not the only benefit of ecological restoration. Other benefits exist,
especially those associated with restoration of riparian forest and river floodplain habitat (i.e. source of food supply
for wildlife species, habitat diversity, increased habitat area, bank stabilization, etc.). An accounting of these
associated benefits, in combination with temperature reduction targets, needs to be part of any exercise in selecting
an appropriate restoration approach. Accounting for these other TMDL and environmental targets will reduce the
net costs per unit of temperature reduction by explicitly valuing associated non-target benefits. Given current
interest in tradable permit systems for TMDL’s in Oregon and elsewhere, questions such as how one denominates
these ecosystem services (temperature reduction and other targets) in units that can be traded, needs attention.
Finally, while cost-effectiveness may be helpful in determining the best policy when facing specific budget
constraints, it does not provide either the optimal biological or economic solution (i.e. situations where the net
benefits to society of restoration are maximized). To move towards maximization of net benefits, development of a
fish population model associated with water temperature and other environmental factors is needed. In addition,
quantification of the value of fish benefits into monetary values must be done if the goal is to complete a benefit-cost
analysis.

ACKNOWLEDGEMENTS

The authors are thankful to Dr Scott Wells, Dr Robert Annear and Mr Mikel McKillip at Portland State University
for their assistance with the CE-QUAL-W2 water temperature model application to the mainstem Willamette
River; Linda Ashkenas and the Pacific Northwest Research Consortium for helping provide the complete GIS data
for the Willamette Basin; Mike McAleer and the engineering team from the U.S. Army Corps of Engineers,
Portland District for providing data on revetment removal costs and the Willamette reservoir study; Dr JunJie Wu,
Department of Agricultural and Resource Economics, and Dr Stephen Lancaster, Department of Geosciences, at
Oregon State University for their support and providing guidance for this research. This research is partially funded
through financial support from the Department of Agricultural and Resource Economics at Oregon State
University.

REFERENCES
Adams RM, Berrens R, Cerda A, Li H, Klingeman P. 1993. Developing a bioeconomic model for riverine management: a case of the John Day
River, Oregon. River 4: 213–226.
Altman B, Henson CM, Waite IR. 1997. Summary of information on aquatic biota and their habitats in the Willamette Basin, Oregon, through
1995. U.S. Geological Survey Water-Resources Investigations, Report 97-4023, Oregon.
Annear RL, Berger CJ, Wells SA. 2001. CE-QUA-W2 version 3.1: Shading algorithm. Technical Report (EWR-05-01), Department of Civil
Engineering, Portland State University, OR.
Annear RL, McKillip M, Khan SJ, Berger CJ, Wells S. 2003a. Willamette Basin temperature TMDL model: boundary conditions and model
setup, Technical report (EWR-03-03). Department of Civil and Environmental Engineering, Portland State University, Portland, OR.
Annear RL, Wells SA, Berger CJ, McKillip ML, Khan SJ. 2003b. Willamette River system temperature waste load allocation model.
Proceedings of Getting it Done: The Role of TMDL Implementation, Watershed Restoration Conference 2003, Stevenson, WA, October 29–30,
2003.
Bartholow JM. 1991. A modeling assessment of the thermal regime for an Urban sport fishery. Environmental Management 15: 833–845.
Benner PA, Sedell JR. 1997. Upper Willamette River landscape: a historic perspective. In River Quality: Dynamics and Restoration, Laenen A,
Dunnette D (eds). Lewis Publishers: New York; 23–47.
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

958

S. SEEDANG ET AL.

Berger CJ, Annear RL, Wells SA. 2002. Willamette River and Columbia River waste load allocation model. Proceedings of the 2nd Federal
Interagency Hydrologic Modeling Conference, Las Vegas, NV, July 28–August 1, 2002.
Berger CJ, McKillip M, Khan SJ, Annear R, Wells S. 2003. Willamette Basin temperature TMDL model: model calibration. Technical Report
(EWR-04-03), Department of Civil and Environmental Engineering, Portland State University, Portland, OR.
Beschta RL, Bilby RE, Brown GW, Holtby LB, Hofstra TD. 1987. Stream temperature and aquatic habitat: fisheries and forestry interactions. In
Streamside Management: Forestry and Fishery Interactions, Solo EO, Cundy TW (eds). College of Forest Resources and Institute of Forest
Resources, University of Washington, Seattle: Washington; Contribution No. 57; 191–232.
Bishaw B, Emmingham W, Rogers W. 2002. Riparian forest buffers on agricultural lands in the Oregon Coast Range: Beaver Creek riparian
project as a case study. Forest Research Laboratory, College of Forestry, Oregon State University, Corvallis, Oregon.
Boulton AJ, Scarsbrook MR, Quinn JM, Burrell GP. 1997. Land-use effects on the hyporheic ecology of five small streams near Hamilton, New
Zealand. New Zealand Journal of Marine and Freshwater Research 31: 609–622.
Boulton AJ, Findlay S, Marmonier P, Stanley EH, Valett HM. 1998. The functional significance of the hyporheic zone in streams and rivers.
Annual Review of Ecology and Systematics 29: 59–81.
Cardenas MB, Wilson JL, Zlotnik VA. 2004. Impact of heterogeneity, bed forms, and stream curvature on subchannel hyporheic exchange. Water
Resources Research 40: DOI: 10.1029/2004WR003008
Cardenas MB, Wilson JL. 2007. Effects of current–bed form induced fluid flow on the thermal regime of sediments. Water Resources Research
43: DOI: 10.1029/2006WR005343
Cole T, Wells SA. 2007. CE-QUAL-W2: a two-dimensional, laterally averaged, hydrodynamic and water quality model, version 3.5, Instruction
Report EL-2007. USA Engineering and Research Development Center, Waterways Experiment Station, Vicksburg, MS.
Copeland Claudia. 1999. Clean Water Act: A Summary of the Law. Congressional Research Services, January 20, 1999.
Coulton KG, Goodwin P, Perala C, Scott MG. 1996. An evaluation of flood management benefits through floodplain restoration on the
Willamette River, Oregon, U.S.A. Philip Williams and Associates, San Francisco, California.
Department of the Interior, Bureau of Reclamation. 2002. Change in discount rate for water resources planning. Federal Register 67:
76756–76757.
Dykaar BB, Wigington PJ Jr. 2000. Floodplain formation and cottonwood colonization patterns on the Willamette River, Oregon, USA.
Environmental Management 25: 87–104.
Fernald AG, Landers DH, Wigington PJ Jr. 2000. Water quality effects of hyporheic processing. In Proceedings of the International Conference
on Riparian Ecology and Management in Multi-Land Use Watersheds, Wigington PJ Jr, Beschta RL (eds). American Water Resources
Association: Middleburg, VA; 167–172.
Fernald AG, Wigington PJ Jr, Landers DW. 2001. Transient storage and hyporheic flow along the Willamette River, Oregon: field measurements
and model estimates. Water Resource Research 37: 1681–1694.
Fernald AG, Landers DW, Wigington PJ Jr. 2006. Water quality changes in hyporheic flow paths between a large gravel bed river and off-channel
alcoves in Oregon, USA. River Research and Applications 22: 1111–1124. DOI: 10.1002/rra.961
Gregory SV, Hulse DW, Landers DH, Whitelaw E. 1998. Integrations of biophysical and socioeconomic patterns in riparian restoration of large
rivers. In Hydrology in a Changing Environment, Wheater H, Kirby C (eds). John Wiley and Sons: Exeter, U.K; 231–247.
Hickey JT, Diaz GE. 1999. From flow to fish to dollars: an integrated approach to water allocation. Journal of the American Water Resources
Association 35: 1053–1067.
Hostetler SW. 1991. Analysis and modeling of long-term stream temperatures on the Steamboat Creek basin, Oregon: implications for land use
and fish habitat. Water Resources Bulletin 27: 637–647.
Hubbert MK. 1969. The Theory of Ground-Water Motion and Related Papers. Hafner Publishing Co., Inc.: New York; 287–310.
Isaak DJ, Hubert WA. 2001. A hypothesis about factors that affect maximum summer stream temperatures across montane landscapes. Journal of
the American Water Resources Association 37: 351–366.
Johnson S, Adams RM. 1988. Benefits of stream flow: the case of the John Day River steelhead fishery. Water Resource Research 24: 1839–1846.
Kauffman JB, Beschta RL, Otting N, Lytjen D. 1997. An ecological perspective of riparian and stream restoration in the Western United States.
Fisheries 22: 12–24.
Knoder E. 1995. Benefits and costs of riparian habitat improvement in the Tualatin River basin. Report Number 10, Oregon Water Resources
Research Institute.
Kuby MJ, Fagan WF, ReVelle CS, Graf WL. 2005. A multiobjective optimization model for dam removal: an example trading off salmon passage
with hydropower and water storage in the Willamette basin. Advances in Water Resources 28: 845–855.
Landers DH, Haggerty PK, Clline S, Carson W, Faure F. 2000. The role of regionalization in large river restoration. Verhandlungen
Internationale Vereinigung für theoretische und angewandte Limnologie 27: 344–351.
Landers DH, Fernald AG, Andrus C. 2002. Off-channel habitats. In Willamette River Basin Planning Atlas: Trajectories of Environmental and
Ecological Change, Hulse D, Gregory S, Baker J (eds). The Pacific Northwest Research Consortium, Oregon State University Press:
Corvallis, Oregon, USA; 24–25.
Laenen A, Bencala K. 2001. Transient storage assessments of dye-tracer injections in rivers of the Willamette Basin, Oregon. Journal of the
American Water Resources Association 37: 367–377.
Micheli ER, Kirchner JW, Larsen EW. 2004. Quantifying the effect of riparian forest versus agricultural vegetation on river meander migration
rates, Central Sacramento River, CA. River Research and Applications 20: 537–548.
National Research Council. 1996. Upstream: Salmon and Society in the Pacific Northwest. Report on the Committee on Protection and
Management of Pacific Northwest Anadromous Salmonids for the National Research Council of the National Academy of Sciences. National
Academy Press, Washington, DC.
Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra

WATER TEMPERATURE REDUCTION PRACTICES

959

National Marine Fisheries Service. 1999. Endangered Species Act proposed 4(d) rule for Pacific Salmon, NMFS, Northwest Region. December,
1999.
Oregon Department of Environmental Quality. 2003. Oregon’s existing water quality criteria for stream temperature. Oregon Department of
Environmental Quality, Salem, OR.
Pacific Northwest Ecosystem Research Consortium (PNW-ERC). 2002. Willamette River Basin Atlas: Trajectories of Environmental and
Ecological Change (2nd edn), Hulse D, Gregory S, Baker J (eds). Oregon State University Press: Corvallis, OR.
Paulsen CM, Wernstedt K. 1995. Cost-effectiveness analysis for complex managed hydro systems: an application to the Columbia River Basin.
Journal of Environmental Economics and Management 28: 388–400.
Rounds SA. 2007. Temperature effects of point sources, riparian shading, and dam operations on the Willamette River, Oregon. Scientific
Investigations Report 2007-5185. U.S. Geological Survey.
Seedang S. 2004. Economic analysis of restoration practices to improve water quality and fish habitat of a large river floodplain: a case study of
the Willamette River, Oregon. Dissertation, Oregon State University, Corvallis, OR.
Stevens JB, Adams RM, Barkley D, Kiest LW, Landry CJ, Newton LD, Obermiller FW, Perry GM, Seely H, Turner BP, Li HW, Walks DJ. 2002.
Benefits, costs, and local impacts of market-based streamflow enhancements: the Deschutes River, Oregon. River 7: 89–108.
Ward FA, Roach BA, Henderson JE. 1996. The economic value of water in recreation: evidence from the California drought. Water Resource
Research 32: 1075–1081.
Ward JV, Tockner K, Arscott DB, Claret C. 2002. Riverine landscape diversity. Freshwater Biology 47: 517–539.
Watanabe M. 2003. A spatially explicit for allocating conservation efforts in watershed: The Grande Ronde River basin, Oregon. Ph.D.
Dissertation, Oregon State University, Corvallis, Oregon.
Watanabe M, Adams RM, Wu JJ, Bolte P, Cox MM, Johnson SL, Liss WJ, Boggess WG, Ebersole JL. 2005. Toward efficient riparian restoration:
integration economic, physical, and biological models. Journal of Environmental Management 75: 93–104.
Wells SA, Berger CJ, Annear RL, McKillip ML, Khan SJ. 2004. Willamette River basin temperature modeling study. Proceedings Watershed
2004, Dearborn, MI, July 11–14, 2004.
Willamette Restoration Initiative. 2001. Restoring a river of life: the Willamette restoration strategy overview. Recommendations for the
Willamette Basin supplement to the Oregon Plan for Salmon and Watersheds. Salem, Oregon. Available online at http://www.oregonwri.
org/wri_report.pdf [Accessed September 2003].
Wu JJ, Boggess W. 1999. The optimal allocation of conservation funds. Journal of Environmental Economics and Management 38: 302–321.
Wu JJ, Adams RM, Boggess W. 2000. Cumulative effects and optimal targeting of conservation efforts: steelhead trout habitat enhancement in
Oregon. American Journal of Agricultural Economics 82: 400–413.
Wu JJ, Skelton-Groth K. 2002. Targeting conservation efforts in the presence of threshold effects and ecosystem linkages. Journal of Ecological
Economics 42: 313–331.

Copyright # 2008 John Wiley & Sons, Ltd.

River. Res. Applic. 24: 941–959 (2008)
DOI: 10.1002/rra