impacts of climate change and urban development on water resources in the tualatin river basin,...

This article was downloaded by: [Texas A&M University Libraries]On: 14 November 2014, At: 16:32Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK

Annals of the Association of American GeographersPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/raag20

Impacts of Climate Change and Urban Development onWater Resources in the Tualatin River Basin, OregonSarah Praskievicz a & Heejun Chang ba Department of Geography , University of Oregonb Department of Geography , Portland State UniversityPublished online: 12 Feb 2011.

To cite this article: Sarah Praskievicz & Heejun Chang (2011) Impacts of Climate Change and Urban Development on WaterResources in the Tualatin River Basin, Oregon, Annals of the Association of American Geographers, 101:2, 249-271, DOI:10.1080/00045608.2010.544934

To link to this article: http://dx.doi.org/10.1080/00045608.2010.544934

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.

This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/raag20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/00045608.2010.544934

http://dx.doi.org/10.1080/00045608.2010.544934

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Impacts of Climate Change and Urban Developmenton Water Resources in the Tualatin River Basin,

OregonSarah Praskievicz∗ and Heejun Chang†

∗Department of Geography, University of Oregon†Department of Geography, Portland State University

We investigated the relative importance of future climate change and land use change in determining the quantityand quality of freshwater resources in northwestern Oregon’s Tualatin River Basin using the U.S. EnvironmentalProtection Agency’s Better Assessment Science Integrating Point and Nonpoint Sources (BASINS) modelingsystem. Models were calibrated and validated using historic flow and water quality data between 1990 and 2006.The goodness of fit for the calibrated models was high, with coefficients of determination ranging from 0.72 to0.93 in the calibration period. The calibrated models were run under a range of eight statistically downscaledclimate change, two regional land use change, and four combined scenarios. Results included average increasesin winter flows of 10 percent, decreases in summer flows of 37 percent, and increases in fifth-percentile flowsof up to 80 percent as a result of climate change in the Tualatin River Basin. For land use change, the resultsincluded an increase in annual flows of 21 percent for the development-oriented scenario and a decrease of16 percent for the conservation-oriented scenario. For combined scenarios of high climate change and highurban development, there is a projected increase in winter flows of up to 71 percent and decrease in summerflows of up to 48 percent. Climate change scenarios were more significant than urban development scenarios indetermining basin hydrological response. The results are relevant to regional planners interested in the long-termresponse of water resources to climate change and land use change at the basin scale. Key Words: climate change,hydrological modeling, urban development, water quality, water resources.

Investigamos la importancia relativa que tienen el cambio climatico futuro y el cambio del uso del suelo paradeterminar la cantidad y calidad de los recursos de agua dulce en la Cuenca del Rıo Tualatin, en Oregon,utilizando el sistema de modelaje BASINS de la Agencia de Proteccion Ambiental de EE.UU. Los modelos secalibraron y validaron utilizando datos de flujo historico y calidad del agua entre 1990 y 2006. La bondad de ajustepara los modelos calibrados fue alta, con coeficientes de determinacion de 0.72 a 0.93 en el perıodo de calibracion.Se corrieron los modelos calibrados en un ambito de ocho de cambio climatico degradado estadısticamente, dospara cambio regional de uso del suelo y cuatro escenarios combinados. Los resultados incluyeron incrementospromedios en flujos de invierno del 10 por ciento, decrecimientos en los flujos de verano de 37 por ciento, eincrementos en los flujos del quinto percentil de hasta el 80 por ciento, como un resultado del cambio climaticoen la Cuenca del Rıo Tualatin. Para el cambio del uso del suelo, los resultados incluyeron un incrementoen los flujos anuales del 21 por ciento para el escenario orientado hacia el desarrollo, y un decrecimientodel 16 por ciento para el escenario de orientacion conservadora. Para escenarios combinados de alta intensidad de

Annals of the Association of American Geographers, 101(2) 2011, pp. 249–271 C© 2011 by Association of American GeographersInitial submission, June 2009; revised submission, January 2010; final acceptance, March 2010

Published by Taylor & Francis, LLC.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014

250 Praskievicz and Chang

cambio climatico y alto desarrollo urbano, hay un incremento proyectado en los flujos de invierno de hasta el 71por ciento y decrecimiento en los flujos de verano de hasta el 48 por ciento. Los escenarios de cambio climaticofueron mas significativos que los escenarios de desarrollo urbano para determinar la respuesta hidrologica de lacuenca. Los resultados son relevantes para planificadores regionales interesados en la respuesta a largo plazo delos recursos hıdricos al cambio climatico y el cambio de uso del suelo a escala de cuencas. Palabras clave: cambioclimatico, modelado hidrologico, desarrollo urbano, calidad del agua, recursos hıdricos.

I n a world of changing climate, growing popula-tions, and increased human influence on the globalenvironment, hydrologic impact analysis has be-

come a thriving area of research. It is essential thatthe potential impacts of climate change be studied inconjunction with other trends of global change, no-tably urban development, given that these forces caninteract in complex ways (Clifford 2009). These twodriving forces of climate change and land use changeare likely to affect both water quantity and quality, atglobal, continental, regional, and basin scales, in geo-graphically disparate areas around the world (Changand Franczyk 2008). Because of the significance oftheir impacts on water resources, and the projectedchanges in climate and land use during the twenty-first century in many world regions, studying the sepa-rate and combined influences of these two variables isimportant for sustainable water resource management(Praskievicz and Chang 2009). Geography’s traditionalfocus on human–environment interactions and open-ness to interdisciplinary approaches (Turner 2002)make geographers uniquely well positioned to con-front global change issues in their research (Balling2000).

Anthropogenic climate change is expected to affectthe quantity and quality of global water resources and tonecessitate changes in the way these resources are man-aged (Oki and Kanae 2006; Kundzewicz et al. 2007).In the Pacific Northwest, where most precipitation fallsduring the winter, the most significant projected resultof climate change is a reduction in snowpack, whichis a major source of summer flows (Mote et al. 2003).Over the past fifty years, peak spring runoff in snowmelt-dominated and transient basins in the Western UnitedStates has been occurring earlier because of reducedsnowpack and warmer spring temperatures (Regondaet al. 2005; Barnett et al. 2008).

Climate change also has the potential to affect wa-ter quality. Higher water temperatures resulting fromincreases in air temperature promote the growth ofalgal blooms and decrease dissolved oxygen, loweringecological productivity. In areas where rainfall amountand intensity are expected to increase, more pollutants

might be flushed from land surfaces into water bodies,although this could be countered by an increase in dilu-tion (Murdoch, Baron, and Miller 2000; Chang, Evans,and Easterling 2001). Lower summer flows can increasethe concentration of pollutants because of reduced di-lution effects (Kundzewicz et al. 2007).

In addition to climate change, water quantity andquality are affected by land use changes. As relativelypermeable forest and agricultural land are converted tohighly impermeable urban land cover, less water infil-trates the soil to recharge aquifers. Additionally, thelower infiltration rates of urban land cover cause highersurface runoff, increasing flood risk (Dunne and Leopold1978). Urban development impairs water quality byadding both point sources of pollution such as waste-water treatment and industrial effluent, as more facili-ties are built to serve a growing population, and non-point source urban pollution (Atasoy, Palmquist, andPhaneuf 2006). The spatial patterns of urban develop-ment and conservation also have significant impacts onthe timing and magnitude of runoff and water pollution(Randhir and Hawes 2009).

Because of the dynamic interactions between climateand land cover, numerous integrated watershed model-ing studies have examined the relations among climatechange, land use change, and water resources (Mimikouet al. 2000; Chang 2003, 2004; Sharma 2003; Chen, Li,and Zhang 2005; Samaniego and Bardossy 2006; Wilbyet al. 2006; Davis Todd et al. 2007; Ducharne et al.2007; Choi 2008; Franczyk and Chang 2009). The hy-drological models used in these studies are capable ofsimulating responses to climatic and land cover inputsto project runoff and water quality outcomes. The wa-tershed is a natural unit with which to model regionalwater resources for planning purposes. Although previ-ous studies have examined combined impacts of climatechange and land use change on water resources, mostof these studies focused on one or two scenarios, andfew modeled impacts on both runoff and water qual-ity at the basin scale. Understanding the dynamics ofwater quantity and quality together under a range offuture scenarios is essential for sustainable, adaptive,integrated water resources management.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014

Impacts of Climate Change and Urban Development on Water Resources in Oregon 251

Figure 1. Location of the Tualatin River Basin within the state of Oregon. Note: CWS = Clean Water Services.

This research examines the relative importance offuture climate change and land use change in deter-mining the quantity and quality of freshwater resourcesusing the Tualatin River Basin (TRB) as a case study.It uses a geographic information systems (GIS)-basedhydrological model to investigate these relationshipsand project future impacts on water resources under arange of climatic and land use scenarios. The model-ing framework is the U.S. Environmental ProtectionAgency’s (EPA) Better Assessment Science Integrat-ing Point and Nonpoint Sources (BASINS). Althoughit is a case study, the integrated modeling and theresults of the study can be applied and compared toother regions that experience similar global changeissues.

The study has three primary objectives. The first is tocontribute to the body of knowledge about the separateand combined influences of climate change and urbandevelopment on water quantity and quality. The secondobjective is to test the application of the BASINS mod-eling framework in a mesoscale, low-elevation basin inthe Pacific Northwest. The third objective is to makerecommendations for the sustainable management ofregional water resources.

Study Area

Our study area is the 1,844-km2 TRB, located to thesouthwest of Portland, Oregon, and including portionsof the cities of Beaverton, Hillsboro, Lake Oswego, andTigard (Figure 1). The river originates in the CoastRange and flows for a length of 134 km before enter-ing the Willamette River near West Linn. River ele-vations range from 603 m at the river’s source to 18 mat its mouth, with very little elevation change in thelower reaches, and the basin’s elevation range is ap-proximately 600 m (Tualatin River Watershed Council1999). The predominant soil type in the basin is theCascade series, which is a clay loam with moderateto high erosive potential and high phosphorus levels(U.S. Geological Survey [USGS] 2008). The marinewest coast climate of the basin is characterized by mod-erate year-round temperatures (mean winter low of 0◦Cand high of 17◦C; mean summer low of 5◦C and high of28◦C, based on the period 1970–1999). Average annualprecipitation is approximately 965 mm at Hillsboro, ofwhich over 75 percent falls during the winter months ofNovember through April (1970–1999; Oregon ClimateService 2008). Snowfall is limited because of the basin’s

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


modest elevations. The annual distribution of flows gen-erally follows that of precipitation, with a winter peakand low flows of as little as 4 m3/s−1 during the drysummer.

The TRB was selected as the study area for severalreasons. First, previous research on climate change im-pacts in the Pacific Northwest suggests the potentialfor significant hydrological changes in low-elevationbasins, including the Tualatin, as a result of alteredprecipitation patterns and increased evapotranspirationfrom higher temperatures (Palmer et al. 2004; Changand Jung 2010). Second, the basin is located in rapidlyurbanizing Washington County, one of the fastest grow-ing regions of Oregon, making it an ideal area for study-ing the impacts of urban development. Finally, thebasin’s moderate size and rich availability of flow andwater quality data enable the successful application ofthe BASINS modeling system.

BASINS and WinHSPF ModelDescription

BASINS is an integrated environmental analysis sys-tem originally developed by the EPA in 1994 for usein evaluating compliance with total maximum dailyloads of pollutants and simulating hydrologic impacts ofmanagement decisions (Donigian, Bicknell, and Imhoff1995). BASINS is not in itself a model but includes anumber of submodels, the most important of which isthe Windows-based Hydrologic Simulation Program—Fortran (WinHSPF).

WinHSPF is generally classified as a lumped concep-tual hydrological model, because it represents physicalprocesses based on idealized system behavior andreports output for the entire watershed defined byuser-specified points (Watts 1997). It calculates a waterbalance for selected points based on inputs of pre-cipitation, with hydrological parameters for differentland cover classes (U.S. EPA 2001). WinHSPF usesseparate water balance equations to calculate runoff onpervious and impervious land surfaces.

The major type of time series data necessary for run-ning WinHSPF is hourly scale meteorological data.At a minimum, the program needs precipitation andevapotranspiration data, but other meteorological vari-ables, including maximum and minimum temperatures,wind speeds, and cloud cover, can also be included.These meteorological data can be accessed for manyweather stations in the United States through the

BASINS Data Download tool or imported from textfiles.

The WinHSPF module that simulates pervious landhydrology is called PWAT. Hydrological behavioris determined by several dozen parameters that areinitially estimated by WinHSPF based on the basin’sclimatic, topographic, and land cover characteristics,then adjusted manually or with the aid of an autocal-ibration program to optimize model performance (U.S.EPA 2000). Some of the major adjustable parametersinclude lower and upper zone soil moisture storage, soilinfiltration rate, length and slope of overland flow path,groundwater and interflow recession rates, evaporationcoefficients, groundwater zone partitioning, vegetationinterception, and Manning’s n roughness coefficient.The IWAT module, which is used for impervioussurfaces, uses many of the same climatic and topo-graphic parameters as PWAT but does not includesuch parameters as soil moisture storage, infiltration,groundwater, or vegetation interception. Instead,retention on an impervious surface is modeled based onan empirically derived estimate of the amount of waterthat must accumulate on such a surface before runoffoccurs.

In addition to simulating hydrology, WinHSPF cal-culates mass balances for selected water quality con-stituents, including sediment and nutrients. The outputof the sediment balance is the total sediment load trans-ported (U.S. EPA 2006). Major adjustable parametersthat determine sediment processes include coefficientsfor sediment washoff, soil matrix scour, solids washoff,and solids accumulation rate, which are initially esti-mated and then adjusted during the calibration process.Initial values are estimated from the input sediment loadsize fractions, determined by soil type, and shear stressin the flow plane, a function of topography (U.S. EPA2006). Another major water quality constituent mod-eled by WinHSPF is nutrient loading, including totalphosphorus, which is simulated in HSPF by a nutrientbudget. The output of the nutrient balance equationis the total load, which can be combined with runoffvolume to determine nutrient concentrations (Bicknellet al. 2001).

We chose to use BASINS and WinHSPF for thisresearch for three primary reasons. First, as one ofthe most commonly used public domain hydrologicalmodeling systems, BASINS has a large user communitywith abundant case study examples and technical sup-port (e.g., Hunter and Walton 2008; Ribarova, Ninov,and Cooper 2008; Chung and Lee 2009). Second,HSPF has an advantage over many other hydrological

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 1. Summary of data sets used for hydrological andwater quality modeling

Data sets Format Resolution Source

Digital elevationmodel

Raster 10 m USGS (2004)

Soil layer Shapefile 1:20,000 NRCS (2001)Watershed boundary Shapefile N/A USGS (2006)Land cover Raster 30 m U.S. EPA (2008)Stream flow gauge data .txt N/A CWS (2008)Water quality data .txt N/A CWS (2008)Forest Grove climate

data.wdm N/A OCS (2007)

Note: CWS data were obtained through personal communication withJan Miller, Clean Water Services Water Resources Program Manager, 21November 2008. USGS = U.S. Geological Survey; NRCS = National Re-source Conservation Service; U.S. EPA = U.S. Environmental ProtectionAgency; CWS = Clean Water Services; OCS = Oregon Climate Services.

models in that it simulates loadings of several waterquality constituents in addition to hydrology, thusallowing for comprehensive assessment of long-termimpacts on both water quality and quantity. Finally,unlike related models such as the Soil and WaterAssessment Tool (SWAT), which was developed pri-marily for agricultural watersheds, HSPF was intendedfor use in mesoscale, mixed land use basins, similar tothe TRB.

Data and Methods

Calibration and Validation Data

To calibrate and validate the hydrology and waterquality models, we obtained several categories of data.These include elevation, soils, the watershed bound-ary, land cover, stream flow, water quality, and climate.The characteristics of these data sets are summarized inTable 1.

Scenario Data

To simulate the effects of potential future conditionson water resources in the TRB, we selected several sce-narios representing a range of climate change and urbandevelopment in the basin. There were a total of eightclimate change and two land use change scenarios, eachof which we ran separately. In addition, we combinedthe highest change and lowest change climate scenarioswith each of the two land use scenarios, for a total offour combined scenarios.

Because different general circulation models(GCMs) can produce widely varying outcomes, wechose to use a total of seven GCMs to generate theclimate change scenarios to generate a range of possibleresults (Table 2). Three of these GCMs were drivenby the Intergovernmental Panel on Climate Change’sSpecial Report on Emission Scenarios A1B emissionscenario, three by the B1 emission scenario, and one,the Community Climate System Model (CCSM), byboth the A1B and B1 scenarios, for a total of eightGCM–emission scenario combinations. The two emis-sion scenarios assume different socioeconomic condi-tions with different use of fossil fuels in the future.The medium-emission A1B scenario projects rapid eco-nomic growth with increasing globalization that reducesregional differences in per capita income. The source ofenergy for this rapid growth is balanced, deriving fromboth fossil and nonfossil fuels. The low-emission B1scenario projects lower growth than A1 with a service-and information-oriented economy, which releases lessgreenhouse gases than the A1B scenario (Nakicenovicet al. 2000). We ran the eight climate scenarios for twotime periods: 2030 to 2059 (2040s) and 2060 to 2089(2070s).

The outputs of GCMs are far too coarse for hy-drologic impact analysis at the basin scale (Xu 1999).Accordingly, the GCM outputs were statistically down-scaled for the Pacific Northwest by the Climate ImpactsGroup (E. Salathe personal communication, 12 March2009). The statistical downscaling is based on the biascorrection and spatial downscaling method (Salathe,Mote, and Wiley 2007). First, the bias between theGCM-derived climate simulation and observed climatedata aggregated to 1/16◦ is corrected based on transferfunctions that are derived from the cumulative distri-bution functions for the observed and modeled timeseries climate data. Second, spatial downscaling wasmade for the bias-corrected modeled data by applyinga perturbation factor (the mean temperature differencebetween the bias-corrected modeled data and the 1/16◦

data) for temperature and a multiplicative scaling factor(the mean ratio of simulated and observed precipitationon the 1/16◦ grid) for precipitation. The output is amonthly transient time series of climate on a 1/16◦ gridfor the historic and future climate simulations. The dailytime series are created by resampling a historic monthwhose monthly mean spatial precipitation pattern mostclosely matches the calendar month to be disaggregated.The ratio of the downscaled monthly mean to the ana-log monthly mean finally yields a daily time series withthe appropriate monthly mean. Time series of maximum

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 2. Summary of characteristics of general circulation models used for climate change scenario modeling

Name Agency Atmospheric resolution Oceanic resolution Emission scenario(s)

Bjerknes Center for ClimateResearch

University of Bergen, Norway 2.8◦ 1.5◦ A1B

Community Climate SystemModel v.3

National Center forAtmospheric Research,United States

1.4◦ 1.125◦ A1B, B1

Coupled Global Climate Modelv.3

Canadian Centre for ClimateModelling and Analysis,Canada

2.5◦ 1.8◦ A1B

Parallel Climate Model National Center forAtmospheric Research,United States

2.8◦ 0.67◦ A1B

Centre National de RecherchesMeteorologiques v.3

Meteo France, France 2.8◦ 1.875◦× 2◦ B1

European Centre HamburgModel v.5

Max Planck Institute forMeteorology, Germany

2.8◦ 1.5◦ B1

Institut Pierre Simon Laplace v.4 Institut Pierre Simon Laplace,France

2.5◦× 3.75◦ 2◦ B1

and minimum temperature, precipitation, and windspeed were provided by the Climate Impacts Group,and the area-weighted average for the grid cells com-prising the TRB was computed. The resulting averagedtime series became the input to WinHSPF for climatechange scenario modeling. As shown in Figure 2, thedownscaled GCM data closely match with the observedhistorical climate data. Temperature differences are lessthan 0.5◦C at a monthly basis and 0.1◦C at an annualbasis. Monthly precipitation differences are less than10 percent for 9 months, but annual total precipitationdifference is less than 1.3 percent.

Figure 3 shows the change in average monthly tem-perature and total precipitation compared to the base-line 1970–1999 climate. There are increases in temper-ature in all months, higher in magnitude in the 2070s,particularly in the summer. Although there is somevariation among models in terms of the direction of

Figure 2. Comparison of observed and downscaled average pre-cipitation and temperature (1970–1999). Note: GCM = generalcirculation model.

change in precipitation, generally there are increases inwinter precipitation and decreases in summer precipita-tion, with a significant overall increase apparent in the2070s.

To simulate the impacts of urban development forthe TRB, we used two scenarios of possible land use forthe 2040s, developed by the Pacific Northwest Ecosys-tem Research Consortium (Baker et al. 2004). Thesetook the form of 30 m raster layers with Anderson landcover classifications (Anderson et al. 1976). Both sce-narios assume that the population of the WillametteRiver Basin, of which the Tualatin is a subbasin, willgrow to 3.9 million, but the type of growth that occursdiffers between the two scenarios. The developmentscenario assumes that market-oriented solutions willdominate land use in the region. It relaxes many exist-ing zoning regulations that protect rural areas from de-velopment and significantly expands the urban growthboundaries. In contrast, the conservation scenario as-sumes that the provision of ecological services will bethe priority driving land use in the future. This scenarioconcentrates most population growth within existingurban areas, while conserving and restoring natural veg-etation and wetlands. The differences between the twoscenarios are illustrated in Figure 4, which shows thatthe development scenario has a higher increase in itsurban area, whereas the conservation scenario includesa substantial increase in the water and wetland class asa result of restoration activities.

In addition to simulating the effects of climatechange and urban development separately, we also

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 3. Absolute change from baseline in average monthly temperature for the (A) 2040s and (B) 2070s and percentage changes frombaseline in total monthly precipitation for the (C) 2040s and (D) 2070s, according to eight climate change scenarios.

modeled the combined impacts of these two changes.We selected the climate change scenario with the high-est average change in basin runoff (IPSL B1, with anincrease in annual runoff of 16 percent) and the onewith the lowest average runoff change (CCSM3 B1,with a decrease in annual runoff of 6 percent). Wethen ran each of these two climate change scenarios incombination with the development and conservationland use scenarios, for a total of four combined scenar-

ios (high climate change/development land use change,high climate change/conservation land use change, lowclimate change/development land use change, and lowclimate change/conservation land use change).

Model Calibration and Validation

We ran BASINS/WinHSPF at a daily time step forthe period from 1990 to 2006 at the Weiss Bridge

Figure 4. Changes from baseline in percentage land use according to the two urban development scenarios.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 3. Initial and final values of calibrated hydrology parameters for the Tualatin River Basin

Parameter Description Recommended range Initial value Final value Unit Effect on runoff

LZSN Lower zone nominal soilmoisture storage

3.0–8.0 4.0–6.5 6.0–8.0 in. ↓

INFILT Index to infiltration capacity 0.01–0.25 0.16 0.19 in./hour ↓AGWRC Base groundwater recession 0.92–0.99 0.98 0.92 Ratio ↓INTFW Interflow inflow parameter 1.0–3.0 0.75 2 None ↓LSUR Length of overland flow 200–500 400 500 ft ↓DEEPFR Fraction of groundwater inflow

to deep recharge0.0–0.2 0.1 0 Ratio ↑

BASETP Fraction of remainingevapotranspiration frombaseflow

0.0–0.05 0.02 0.01 Ratio ↓

NSUR Manning’s n (roughness) foroverland flow

0.015–0.035 0.02 0.035 None ↓

IRC Interflow recession parameter 0.5–0.7 0.5 0.7 None ↓Note: Values of LZSN vary according to land use category.

station, near the mouth of the Tualatin River. The firstyear, 1990, serves as an initialization year and was notincluded in the evaluation of goodness of fit. We thendivided the remaining modeled period (1991–2006)into two halves. The first half (1991–1998) serves as thecalibration period, and the second half (1999–2006)is the validation period. We manually calibrated themodel for hydrology using an iterative process, guidedby U.S. EPA (2000). To evaluate the model’s goodnessof fit, we used the coefficient of determination, or R2,one of the most commonly used statistical measures formodel assessment (Weglarczyk 1998). The appropriate-ness of using the coefficient of determination for modelevaluation has been criticized, because this measure issensitive to outliers and to systematic bias in the model(Legates and McCabe 1999). Accordingly, we supple-mented the model evaluation by also calculating theNash–Sutcliffe model efficiency, E, and the annual de-viation of runoff volumes, or the percentage differencebetween modeled and observed annual flow volume,to evaluate the model’s water balance, as suggested byWatts (1997).

For the initial run in the calibration period (1991–1998), in which all parameters were kept at their defaultvalues, the R2 was 0.59 for the calibration period (1991–1998) and 0.58 for the validation period (1999–2006).Annual runoff was underestimated by an average of9.2 percent. To improve the goodness of fit over theinitial run, we iteratively adjusted model parameters.Table 3 contains the initial and final values of eachparameter adjusted during the calibration process. Thegeneral intent of the calibration was to increase the

overall surface flow and reduce the magnitude of peakflows to better match the observed data. The overallimpact of these changes is to make the TRB’s flows lessflashy and more dependent on subsurface flow than inthe initial run.

After adjusting the model parameters as outlined inTable 3, we evaluated the model and found R2 of 0.83and 0.76 for the calibration and validation periods, re-spectively, with an overall annual difference betweenobserved and modeled flows of 3.5 percent. The modelperformance improved significantly between the initialand final runs as a result of parameter adjustments madeduring calibration, although peak flows are somewhatoverestimated (Table 4). Given the relatively high finalfitness values, the model can be considered sufficientlycalibrated.

We used a similar procedure to calibrate the sedimentmodel as that used for hydrology. Table 5 shows theparameters adjusted during the calibration process. Themajor adjustments made were to the daily reductionin detached sediment, the exponent in the sedimentwashoff equation, and the fraction of solids removedper day from impervious surfaces.

After parameter adjustment, the final model fit wasan R2 of approximately 0.72 in the calibration periodand 0.55 in the validation period. Although the modelfit is good for the lower amounts of sediment, the peakloadings are somewhat overestimated. This deficiencyin the model is likely the result of the similar overesti-mation of peak flows in the hydrology model.

Table 4 summarizes the sediment model evaluationstatistics for the calibration and validation periods. The

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 4. Hydrology and water quality model evaluation parameters for the calibration (1991–1998) and validation(1999–2006) periods in the Tualatin River Basin

Parameter Run Calibration R2 Calibration E Validation R2 Validation E

Flow Initial daily 0.59 0.72 0.58 0.36Final daily 0.83 0.81 0.76 0.49Final monthly 0.94 0.94 0.88 0.65

Sediment Daily 0.72 0.24 0.55 0.45Monthly 0.78 0.27 0.59 0.49

Orthophosphate Daily 0.93 0.82 0.93 0.83Monthly 0.99 0.94 0.99 0.95

R2 values fall below the target of 0.8, and the Nash–Sutcliffe model efficiency values are also lower thandesired, although they are positive, meaning that themodel does explain some of the variance in sedimentloads. The relatively poorer performance of the sedi-ment model can be explained by the fact that it includesall of the uncertainty of the hydrology model, plus ad-ditional uncertainty from the sediment transport equa-tions, thus leading to cascading errors. Additionally, theobserved sediment data are not as reliable as the ob-served hydrology data, so it is inherently more difficultto match the modeled to the observed time series. Thesediment model’s performance is, however, adequate foraverage conditions; it is mainly in the higher loadingsthat the overestimations are significant. Accordingly,the peak loading results should be treated with caution,but the average loadings can be considered reasonablyaccurate.

As with the hydrology and sediment models, we usedan iterative procedure to adjust parameters in the nutri-ent loading model. The nutrient constituent modeledwas total orthophosphate (PO4). Orthophosphate loadwas simulated by making it a function of sediment load,because it readily adsorbs to sediments, with no furtherchanges made to the calibrated hydrology and sedimentparameters.

The orthophosphate load, modeled as a function ofsediment load, achieved an R2 of approximately 0.93in both the calibration and validation periods. The cal-

ibration and validation time series show that the modelsomewhat underestimates peak loads. The model eval-uation statistics, however, indicate that the fit is verygood (Table 4), probably because orthophosphate ishighly sensitive to the flushing effects of increased flow.There are grounds for a high level of confidence in thescenario modeling results for orthophosphate.

Results

Impacts of Climate Change on Hydrology

Figure 5 shows the changes in mean monthly flowfrom the baseline period (1970–1999) resulting fromthe eight climate change scenarios for the 2040s and2070s. Although there is some variation among thescenarios, the general pattern is increases in winter flowand decreases in summer flow, with somewhat greaterchanges in magnitude by the 2070s (Table 6).

In addition to modeling changes in the mean hydrol-ogy, we examined potential changes in extreme hydro-logical events resulting from climate change. Figure 6shows the modeled flow duration curves for the base-line, 2040s, and 2070s for the high-change (IPSL4 B1)and low-change (CCSM3 B1) climate scenarios. Thelowest 5 percent of flows, those exceeded 95 percentor more of the time, are of lower magnitude under theclimate change scenarios than the baseline. This meansthat, under climate change, lower low flows should be

Table 5. Initial and final values of calibrated sediment parameters

Parameter Description Recommended range Initial value Final value Unit Effect on load

KRER Coefficient in the soil detachment equation 0.15–0.45 0.325 0.15 None ↓JRER Exponent in the soil detachment equation 1.5–2.5 2 2.5 None ↑AFFIX Daily reduction in detached sediment 0.03–0.1 0.03 0.1 Ratio ↓JSER Exponent in the sediment washoff equation 1.5–2.5 2 2.5 None ↑REMSDP Fraction of solids removed per day 0.03–0.2 0.03 0.2 Ratio ↑

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 5. Absolute changes inmonthly flow resulting from the eightclimate change scenarios for the (A)2040s and (B) 2070s.

expected. It is not only the low flows that are likelyto be affected by increased hydrologic variability underclimate change. The flow duration curves for only thehighest 5 percent of flows, those exceeded less than 5percent of the time, show that these flows are higherin magnitude under climate change than the baselinefor both the high-change and low-change climate sce-narios. The magnitude of these increases in high flowranges from 69 to 80 percent for the high-change sce-nario and 25 to 32 percent for the low-change scenario(Table 7). This analysis indicates that the TRB is likelyto experience increased hydrologic variability as a resultof climate change, with lower low flows and higher highflows.

In addition to mean and extreme runoff, we also mod-eled changes in several other hydrological parametersresulting from the climate change scenarios. Figure 7shows changes in potential evapotranspiration (PET)as it relates to precipitation and changes in the parti-tioning of runoff into surface flow and groundwater flow

for the high- and low-change climate scenarios. Thesummer soil moisture deficit, when PET exceeds precip-itation, is projected to grow more severe as a result ofclimate change, because of decreasing precipitation andincreasing PET from higher temperatures. Currently,the dry summer conditions are ameliorated by a rela-tively steady input of groundwater, but as groundwaterflows become more seasonally variable, these summergroundwater flows will decline. Consequently, ground-water might not reliably supplement summer flows un-der a future of climate change.

Impacts of Urban Development on Hydrology

Figure 8 shows the results of the development andconservation land use scenarios on monthly flow in theTRB. The development scenario results in increasedflow in each month, whereas the conservation scenariodecreases each month’s flow. This is probably becausethe development scenario includes a large increase in

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 6. Percentage changes in monthly flow resultingfrom the averaged 2040 and 2070 climate change scenariosand the development and conservation land use scenarios

2040 2070Climate Climate Development Conservation

Month (%) (%) (%) (%)

January 3.97 10.74 25 −10February 4.49 9.33 26 −10March 7.13 1.72 27 −9April 5.82 5.50 27 −9May −4.59 –6.72 25 −11June −15.28 −24.12 24 −13July −26.34 −39.91 27 −13August −35.00 −30.39 10 −33September −22.63 −22.76 11 −28October −11.37 −17.60 9 −28November −3.54 8.65 18 −18December 6.92 12.68 25 −11

urban land, with its associated impervious surfaces,whereas the conservation scenario includes increasedwetland area because of restoration activities, resultingin increased storage of precipitation. The average mag-nitude of these changes is approximately a 21 percentincrease for the development scenario and a 16 percentdecrease for the conservation scenario.

As with climate change, we created flow durationcurves for the baseline, development, and conservation

land use scenarios, to determine whether urban devel-opment will affect hydrologic variability in the basin.Flows are higher for the development scenario than thebaseline and lower for the conservation scenario (Figure9A). Fifth-percentile flows are 23 percent higher thanbaseline for the development scenario and 14 percentlower for the conservation scenario. This means thatflooding is more likely to occur under the developmentscenario than present conditions and, under the con-servation scenario it is somewhat less likely to occur.

Figure 10 shows the partitioning of runoff into sur-face and groundwater flows under the baseline, develop-ment, and conservation land use scenarios. Groundwa-ter flows are higher than baseline for the conservationscenario and lower than baseline for the developmentscenario. This is in line with the overall pattern of hy-drological changes in the basin, with increased wetlandarea providing a source of flows in the conservation sce-nario, whereas the higher impervious surface area of thedevelopment scenario limits groundwater recharge.

Combined Impacts of Climate Change and UrbanDevelopment on Hydrology

In addition to modeling the separate impacts ofclimate change and urban development on basin hy-drology, we modeled their combined impacts, as both

Figure 6. Daily flow duration curves for baseline and future periods resulting from the (A) high-change and (B) low-change climate scenarioand for Q5 flows only for the (C) high-change and (D) low-change climate scenario.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 7. Percentage changes in 5 percent highest flows forthe 2040s and 2070s resulting from the high- and

low-change climate scenarios

Period High change (%) Low change (%)

2040s 80 252070s 69 32

types of changes are likely to occur in the TRB overthe next several decades. Figure 11 shows the re-sults of these combined scenarios for the 2040s and2070s. The general pattern is the same as for theclimate change scenarios, with increases in winterflow and mostly decreases in summer flow, and thereare greater differences between the climate changescenarios than between the land use scenarios. Thisindicates that climate change is likely to more sig-nificantly impact hydrology than land use change inthe TRB over the study period. For the 2040s, the in-creases in winter flow range from 11 percent (low cli-mate change/conservation land use change [LCCL] inFebruary) to 69 percent (high climate change/develop-ment land use change [HCDL] in January). For the2070s, this increases to a minimum winter increase of12 percent (low climate change/development land use

Figure 8. Absolute changes in monthly flow resulting from the de-velopment and conservation land use scenarios for the TualatinRiver Basin.

change [LCDL] and LCCL in February) and a max-imum increase of 78 percent (HCDL and high cli-mate change/conservation land use change [HCCL] inJanuary).

We generated flow duration curves to examinedifferences in extreme flows among the four com-bined climate change and urban development scenar-ios compared to the baseline for the 2040s and 2070s(Figure 9B–C). Under most scenarios, fifth-percentileflows more than double (Table 8).

Figure 7. Precipitation and potential evapotranspiration for the baseline and (A) high-change and (B) low-change climate scenario andsurface and groundwater flows for the (C) high-change and (D) low-change climate scenario. Note: PET = potential evapotranspiration.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 9. Flow duration curves forthe (A) land use scenarios andcombined climate change and landuse scenarios for the (B) 2040sand (C) 2070s. Note: HCDL =high climate change/development landuse change; HCCL = high cli-mate/conservation land use change;LCDL = low climate/developmentland use change; LCCL = low cli-mate/conservation land use change.

Figure 12 shows the partitioning of total runoff intosurface and groundwater flows under the baseline andfour combined climate change and land use change sce-narios for the 2040s and 2070s. For all four combined

scenarios, groundwater flows are lower than the base-line, as a result of the increased seasonality of precipita-tion under climate change and the increased impervioussurface area from urban development. Again, the two

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 10. Surface and groundwater flows for the baseline, devel-opment, and conservation land use scenarios.

combined scenarios with the same climate change sce-nario are more similar to one another than the two withthe same land use scenario.

Sediment Loading

The modeled impacts of climate change on sus-pended sediment loading closely track the hydrologicalchanges. Figure 13A and B show the monthly changes

Figure 11. Absolute changes in monthly flow resulting from thecombined climate change and urban development scenarios forthe (A) 2040s and (B) 2070s. Note: HCDL = high climate/development land use change; HCCL = high climate/conservationland use change; LCDL = low climate/development land usechange; LCCL = low climate/conservation land use change.

in sediment loading for the 2040s and 2070s resultingfrom the climate change scenarios. As with runoff, thescenarios produce different results, but the general pat-tern is increasing winter and decreasing summer load-ings under climate change.

As with the climate change scenarios, the responseof suspended sediment loading to the land use scenar-ios is similar to the hydrological response (Figure 14A).Similar to basin flow, there are increases in sedimentloading under the development scenario and decreasesunder the conservation scenario, indicating that sed-iment load is essentially a proxy for surface flow. Asindicated in Table 9, the relative magnitudes of thesechanges average an 18 percent increase for the devel-opment scenario and an 18 percent decrease for theconservation scenario.

Figure 15A and B shows the modeled changes insediment loading resulting from the combined climatechange and urban development scenarios for the 2040sand 2070s. Sediment loading increases in all months,except for decreases in summer loading under most sce-narios. The high climate change scenarios generallyhave greater increases in sediment loading than eitherof the low climate change scenarios, indicating thatclimate change is more likely than urban developmentto impact future sediment dynamics in the basin. Ta-ble 10 indicates that the magnitude of the increases issignificant, with a doubling or more of sediment loadingfor most scenarios, although the summer decreases aremore modest.

Orthophosphate Loading

Changes in orthophosphate loading closely trackchanges in flow and sediment loading, with increasesin winter and decreases in summer (Figure 13C andD). The relative magnitude of the changes reaches a

Table 8. Percentage changes in 5 percent highest flowsresulting from the four combined climate change and urban

development scenarios

Scenario 2040 (%) 2070 (%)

HCDL 181 146HCCL 178 143LCDL 121 123LCCL 87 120

Note: HCDL = high climate/development land use change; HCCL = highclimate/conservation land use change; LCDL = low climate/developmentland use change; LCCL = low climate/conservation land use change.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 12. Surface and groundwaterflows for the baseline and four com-bined climate change and urban de-velopment scenarios for the (A) 2040sand (B) 2070s. Note: HCDL = highclimate/development land use change;HCCL = high climate/conservationland use change; LCDL = low cli-mate/development land use change;LCCL = low climate/conservationland use change.

winter increase of 26 percent and a summer decreaseof 9 percent (Table 11). The modeled changes in or-thophosphate resulting from the land use scenarios aresimilar to the projected changes in flow and sedimentloading (Figure 14B). The average annual increase forthe development scenario is approximately 7 percent,and the average decrease under the conservation sce-nario is 32 percent. The changes in orthophosphateload resulting from the combined scenarios generallyfollow those resulting from climate change. There aredecreases in the summer and increases in the othermonths (Figure 15C and D). In the 2040s, the maximumwinter increase in orthophosphate loading is approxi-mately 50 percent and the maximum summer decreaseis 39 percent. In the 2070s, this range increases to amaximum winter increase of 56 percent and a maxi-mum summer decrease of 43 percent.

Discussion

As in all hydrologic modeling studies, there is con-siderable uncertainty associated with the results of thisresearch. For this study, at least three major sources ofuncertainty can be identified. The first source of uncer-tainty is the choice of emission scenarios and GCMs.The second is associated with transferring large-scaleclimatology to regional-scale climatology appropriatefor hydrologic and water resource impact assessment,namely, downscaling processes. The third is related tothe parameters and structures of hydrologic models usedfor impact assessment.

Climate change scenarios are developed based ondifferent assumptions of future emission scenarios thatreflect varying socioeconomic conditions. Because cli-mate change scenarios are typically generated by GCMs

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 13. Absolute changes in monthly suspended sediment load resulting from the eight climate change scenarios for the (A) 2040s and(B) 2070s and absolute changes in monthly orthophosphate load resulting from the eight climate change scenarios for the (C) 2040s and (D)2070s.

that have different structures and assumptions, they ex-hibit a wide range of direction and magnitude of futureclimate change. These scenarios are used as inputs foreither statistical or dynamic downscaling techniques,which introduce additional uncertainty, so that impactscan be modeled at the regional scale. Studies indicatethat the greatest source of uncertainty in the climateimpact modeling chain is the GCM (Wilby et al. 2006;Graham et al. 2007). Because they all model atmo-spheric conditions and feedbacks differently, GCMsvary widely in their projections, particularly for pre-cipitation. The choice of emission scenario is less im-portant for the near term, because most scenarios showvery similar levels of emissions through the 2050s andit takes time for the atmosphere to respond (Wilby andHarris 2006).

A final important source of uncertainty stems fromthe choice of a hydrological model in climate impactassessment. Different hydrologic models vary in theirparameters and assumptions and are suited to simulaterunoff at certain spatial and temporal scales. However,studies have shown that results of climate impact stud-ies are less sensitive to the hydrological model than

the climate change scenario (Graham et al. 2007; Kayet al. 2009). In other words, different hydrological mod-els tend to produce similar outcomes, given the sameclimatic inputs, but the same hydrological model rununder different GCM simulations might give widelydiffering results. Based on these studies, analyzing un-certainty in climate change impact studies, the greatestsource of uncertainty in this research is likely the GCM.We have attempted to address this uncertainty by usingseveral different GCMs and presenting their outputs asa range of possible results rather than a prediction offuture conditions.

The modeling results from this study found thatclimate change is likely to significantly affect water re-sources in the TRB during the twenty-first century, withpotential average increases in winter flow of 10 percentand summer decreases of 37 percent by the 2070s. Theseresults are similar to the findings of Franczyk and Chang(2009), who used the SWAT, driven by a downscaledECHAM5 A1B scenario and a range of syntheticscenarios, including an increase in mean monthly tem-perature of 2◦C or 4◦C and increases in mean winter pre-cipitation and decreases in mean summer precipitation

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Figure 14. Absolute changes in (A)monthly suspended sediment load and(B) monthly orthophosphate load re-sulting from the development and con-servation land use scenarios.

of 10 or 20 percent, to model impacts of climate changeon runoff in the Rock Creek Basin, a subbasin of theTRB, for the 2040s. The findings in this study are alsoin line with those of Mote et al. (2003), who used four

Table 9. Percentage changes in seasonal suspendedsediment and orthophosphate loads resulting from the

development and conservation land use scenarios

Parameter Season Development (%) Conservation (%)

Sediment Winter 9 −18Spring 11 −17Summer 53 −8Fall 1 −28Average 18 −18

Orthophosphate Winter 1 −31Spring 3 −30Summer 14 −33Fall 12 −34Average 7 −32

GCMs, with temperature increases ranging from 1.5◦Cto 3.2◦C and precipitation changes ranging from –2 to+22 percent, to drive the Variable Infiltration Capacitymodel to predict changes in the flows of the ColumbiaRiver, and found winter increases of up to 22 percentand summer decreases of 6 percent for the entireColumbia Basin. A study with a more similarly sizedrain-fed basin in a marine West Coast climate, the Up-per Campbell River in British Columbia, whose flowswere modeled using the University of British ColumbiaWatershed Model under the CGCM A1 climate sce-nario, with a temperature increase of 3.5◦C to 4.1◦C andan annual precipitation increase of about 13 percent,resulted in a 71 percent increase in winter flowsand a 59 percent decrease in summer flows (Loukas,Vasiliades, and Dalezios 2002). Although muchattention is often given to impacts of climate changein snowmelt-dominated basins, the results of all thesestudies indicate that changes can be significant in rain-

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 10. Percentage changes in seasonal suspended sediment load resulting from the combined climate change and urbandevelopment scenarios for the 2040s

HCDL 2040 HCCL 2040 LCDL 2040 LCCL 2040 HCDL 2070 HCCL 2070 LCDL 2070 LCCL 2070Parameter Season (%) (%) (%) (%) (%) (%) (%) (%)

Sediment Winter 425 451 270 261 410 436 269 259Spring 129 146 164 156 41 53 53 47Summer 54 49 −61 −67 −52 −59 −77 −81Fall 569 595 460 430 769 812 537 503

Orthophosphate Winter 50 40 19 10 56 46 20 11Spring 88 76 57 47 56 45 55 45Summer 9 −8 −23 −39 −9 −27 −26 −43Fall 82 55 95 65 79 51 74 47

Note: HCDL = high climate/development land use change; HCCL = high climate/conservation land use change; LCDL = low climate/development land usechange; LCCL = low climate/conservation land use change.

fed basins as well, because of changing precipitationpatterns and higher evapotranspiration.

These types of hydrological changes are particularlysignificant in regions like the Pacific Northwest thathave pronounced seasonal variability of precipitationand therefore of water availability. The results of thisstudy suggest that the TRB will experience lower low

flows and higher high flows in the future as a result ofclimate change. These findings are in agreement withthose for other midlatitude basins (Milly et al. 2002;Arnell 2003).

Changes in hydrological variability might be moresignificant for water resource management than changesin mean hydrology. Although annual basin runoff might

Figure 15. Absolute changes in monthly suspended sediment load resulting from the combined climate change and urban developmentscenarios for the (A) 2040s and (B) 2070s and absolute changes in monthly orthophosphate load resulting from the combined climate changeand urban development scenarios for the (A) 2040s and (B) 2070s. Note: HCDL = high climate/development land use change; HCCL =high climate/conservation land use change; LCDL = low climate/development land use change; LCCL = low climate/conservation land usechange.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Table 11. Percentage changes in seasonal orthophosphateload, averaged for all eight climate change scenarios, for the

2040s and 2070s

Season 2040 (%) 2070 (%)

Winter 16 26Spring −2 4Summer −8 −9Fall 43 62

not change much (for the 2070s composite climatechange scenario compared to the baseline, t = –2.00,which is not significant), amplified seasonality of flowscan prove problematic in areas like the TRB. The basinhas already experienced problems with damaging flood-ing (Chang et al. 2010), notably during the springof 1996, and regularly suffers from summer low flowsthat degrade the river’s ecological and aesthetic valuethrough high water temperatures and poor water qual-ity (Boeder and Chang 2008). The results of this studyindicate that such problems are likely to worsen as aresult of climate change and land use change. Increasedhydrologic variability poses a challenge for water re-source managers, who have traditionally relied on theconcept of stationarity, the assumption that the prob-ability distribution of climatic and hydrological vari-ables does not change over time (Milly et al. 2008).In light of climate change, there have been calls for anew paradigm in water resource management to accom-modate the future of potentially increased variability.The institutional conservatism of many water resourcemanagement agencies and the long life span of muchwater infrastructure, however, represent significant ob-stacles to incorporating climate change impacts intowater resource management in the near term.

The modeled changes in land use in this study re-sulted in an average increase in basin runoff of 21 per-cent under the development scenario and an averagedecrease of 16 percent under the conservation scenario.The results from the development scenario are similarto those of Tang et al. (2005), who modeled a mesoscaleMichigan basin using the Long-Term Hydrologic Im-pact Assessment model and found an increase in runoffof about 25 percent under a scenario of urban sprawlin which basin urban area increases from less than5 percent to 11 to 20 percent, depending on the sub-basin. In the TRB, the urban area increases from nearly17 percent to over 26 percent in the development sce-nario. In the conservation scenario, meanwhile, urbanarea increases to only 19 percent, whereas the water and

wetlands category increases from less than 1 percent tonearly 2 percent because of wetland restoration activ-ities. These differences explain why the developmentscenario results in an increase in basin runoff and theconservation scenario causes a decrease and illustratethe importance of land use policies in managing basinhydrology.

The modeling results from the combined climatechange and urban development scenarios suggest thatthese two changes will jointly increase winter flow by upto 71 percent and decrease summer flow by up to 48 per-cent by the 2070s. These changes are similar in pattern,but greater in magnitude, compared to the combinedclimate change and land use change modeling results ofFranczyk and Chang (2009) for the Rock Creek Basin,in which there was a maximum of a 24 percent in-crease in runoff. The larger changes found for the TRBare likely the result of this study’s greater range of cli-mate change scenarios and longer time period, as well asdifferences in spatial scale and hydrologic models usedfor impact assessment.

In comparison to the urban development scenariosalone, the combined scenarios generally predict largerchanges in runoff. Additionally, the combined highclimate change scenarios tend to be more similar toone another than the combined high land use changescenarios. These differences suggest that the climatechange scenarios have more significant impacts thanthe land use change scenarios on the hydrology of theTRB over the study period. Barlage et al. (2002), Chang(2003), and Franczyk and Chang (2009) also found thatbasin hydrology is more sensitive to climate change thanland use change. The combined scenarios also generallypredict higher increases in winter runoff and higher de-creases in summer runoff than the climate change sce-narios alone. The results of this study, then, indicatethat urban development will exacerbate the problemsof increased seasonal variability in flows caused by cli-mate change. This is because the greater impervioussurface area associated with urban development meansa higher proportion of rainfall becomes surface runoffrather than groundwater recharge, thus increasing theoverall flashiness of the basin. The reduced infiltrationmight ultimately deplete aquifers, which are an impor-tant source of cool water during the summer dry period.Although some of this groundwater recharge can bemade up by increased winter precipitation, the higherintensity storms projected under climate change sce-narios can overwhelm the infiltration capacity of thesoil, leading to increased surface runoff and flooding.Urban development might therefore contribute to the

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


increased seasonality of hydrology associated with cli-mate change.

The modeling results of this study indicate thatsuspended sediment and orthophosphate loads arehighly dependent on flow in the TRB, and changesin the loading of these water quality constituentsresulting from climate change and urban develop-ment are likely to closely track hydrological changes.Because increased flow can impact water qualitythrough nonlinear flushing or dilution effects, however,increased load does not necessarily imply increased con-centrations of pollutants, nor does reduced load entaillower concentrations. The strong relation between flowand sediment and orthophosphate load is in line withwhat one would expect, because sediment load is con-trolled largely by erosion and scour from surface runoff,and orthophosphate tends to adsorb to soil particles andso is highly correlated with sediment.

Under the climate change scenarios, winter or-thophosphate load increases by an average of 26 per-cent and summer load decreases by an average of9 percent by the 2070s, changes that are consistentwith the pattern of hydrological change and also withthe results of Arheimer et al. (2005), who modeled a50 percent increase in total phosphorus transport ina Swedish basin under climate change. Bouraoui et al.(2004) modeled even larger increases in winter nutrienttransport (up to 85 percent) as a result of climate changein a Finnish basin. Chang, Evans, and Easterling (2001)also modeled large increases in phosphorus loading un-der scenarios of climate change in a Pennsylvania basin,with loads more than doubling in some subbasins. Theclimate change–related increase in phosphorus loadingin this research is smaller than for these other studies,partly because the changes in annual flow are not aslarge and also possibly because of the TRB’s naturallyphosphorus-enriched soils, which mean the basin’s cur-rent phosphorus loading might be nearer to saturation.

As with hydrology, the dominant pattern of changein water quality parameters under climate change andurban development is increased seasonality. It is likelythat the combination of higher intensity precipitationand higher impervious surface area will flush significantamounts of pollutants into streams during winter storms,and the lower summer flows will reduce the capacity fordilution. Existing water quality problems in the basinmight therefore worsen during the twenty-first century,although the impacts can be somewhat ameliorated byconservation-oriented land use planning.

Examining the separate and combined potential im-pacts of climate change on basin hydrology leads to a

consideration of the dynamics of climate change mitiga-tion and adaptation. Mitigation refers to steps taken toprevent climate change from occurring, such as reduc-ing greenhouse gas emissions or increasing carbon stor-age. Adaptation, in the context of climate change, is theprocess of deliberately taking actions, such as increasingreservoir storage or instituting water conservation poli-cies, to reduce the harm caused by anticipated climatechange. There is often a tension between these twotypes of action, with mitigation being seen as proactiveand adaptation as primarily reactive. Although mitiga-tion of greenhouse gas concentrations is clearly nec-essary to reduce the severity of climate change in thefuture, adaptation is also necessary, particularly at thelocal scale, because the inertia of the climate systemmeans that some climate change is unavoidable as a re-sult of past emissions (Nelson, Adger, and Brown 2007).One way to reconcile mitigation and adaptation is torecognize their differing scales of implementation. Be-cause the atmosphere is a global system, greenhouse gasemissions from any source in the world contribute to theproblem, so coordinated global action is necessary forsuccessful mitigation. The impacts of climate change,however, are highly place specific, and therefore so arethe actions needed for adaptation. Climate change islikely to affect water resources during the twenty-firstcentury, regardless of any international mitigation ef-forts, but local and regional policymakers have a fargreater degree of control over how water resource man-agement in the basin adapts to climate change impacts.The modeling results of this study, which show sub-stantial differences in hydrological response betweendevelopment-oriented and conservation-oriented ur-ban growth, suggest that one potentially powerful wayto adapt to climate change impacts is to plan for com-pact development with preservation and restoration ofnatural vegetation and wetlands. An urban develop-ment pattern similar to that in the conservation landuse scenario used in this study could partially amelio-rate some of the winter flooding and summer low-flowconditions projected to result from climate change inthe basin.

Conclusions

Understanding future changes in hydrology causedby combined forces of climate change and urban de-velopment has significant implications for sustainablewater resources management. Using the TRB as a casestudy and a GIS-based hydrologic simulation model, we

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


projected the potential impacts on basin hydrology andsediment and nutrient loading resulting from a series ofseparate and combined scenarios of climate change andurban development. The model evaluation statistics in-dicate high goodness of fit for the calibrated hydrol-ogy, suspended sediment, and orthophosphate modelscompared to observed data. Projected impacts of cli-mate change include higher winter flows, lower sum-mer flows, increased hydrologic variability, increases inwinter sediment and nutrient loading, and decreasesin summer sediment and nutrient loading. The devel-opment land use scenario results in increased runoff,sediment, and orthophosphate loads, whereas theconservation scenario produces decreased runoff anddecreased sediment and orthophosphate loads. Thecombined climate change and urban development sce-narios generally produce hydrological and water qualityresults that track the results from climate change alone,suggesting that the water resource impacts from climatechange are more significant than those from land usechange in the TRB over the study period, althoughthis needs to be confirmed with a sensitivity analysis.The development and conservation scenarios do differin their hydrological and water quality outcomes, how-ever, thus representing a potential opportunity for localadaptation to climate change by pursuit of sustainableurban development.

The main contribution of this research is as a casestudy application of the WinHSPF model in a mesoscaleurban basin in the Pacific Northwest. It is one of thefew studies to model changes in both water quantity andquality resulting from a range of both climate changeand urban development scenarios. Most existing re-search focuses on only one type of change or one type ofimpact, no doubt because of the difficulty involved inselecting multiple scenarios, calibrating and validatingseveral different models, and evaluating and compar-ing the results. It is important, however, to assess thecombined influence of climate change and urban devel-opment in many river basins like the Tualatin, becauseboth changes are likely to occur during the twenty-firstcentury and because the two types of change could am-plify or ameliorate one another’s effects. Understandingthe relative importance of climate change and urbandevelopment in determining future conditions is im-portant for water resources management, because localpolicymakers have more control over land use policythan global climate policy. Furthermore, much of theexisting research has focused on only the hydrologicalchanges resulting from climate change and urban devel-opment, but water quality is also likely to be affected by

these changes, as this study demonstrates. By modelingbasin response to the potential range of future condi-tions, this study provides a comprehensive view of thetypes of challenges likely to be faced by water resourcemanagers over the twenty-first century and provides anexample that can be applied in other basins facing sim-ilar global change issues.

As concurrent climate change and urban develop-ment progress, water resources will be affected in boththeir quantity and quality. Despite the significant uncer-tainty involved, hydrological modeling studies such asthis one are useful for projecting the likely direction andmagnitude of these changes primarily driven by anthro-pogenic environmental changes so that water resourcemanagers are prepared to adapt. This work is necessaryto ensure that the economic, aesthetic, recreational,and ecological values of river basins will continue to beprovided in the future.

Acknowledgments

This research was partially supported by a grant(code# 1-9-3) from the Sustainable Water ResourcesResearch Center of the 21st Century Frontier ResearchProgram in Korea, a sustainability grant from the JamesF. and Marion L. Miller Foundation, and a Faculty En-hancement Grant at Portland State University. Weappreciate Eric Salathe at the University of Washing-ton for providing downscaled climate change simula-tion data, Il-Won Jung at Portland State University forextracting downscaled data for hydrologic model sim-ulation, and Jan Miller of Clean Water Services forproviding flow and water quality data. The preliminaryfindings of this research were presented at the annualmeeting of the Association of American Geographersin Las Vegas in 2009. We greatly appreciate commentsfrom an anonymous reviewer and editor Richard As-pinall, who helped improved the quality of the article.

References

Anderson, J. R., E. E. Hardy, J. T. Roach, and R. E. Wit-mer. 1976. A land use and land cover classification systemfor use with remote sensor data. Washington, DC: U.S.Government Printing Office.

Arheimer, B., J. Andreasson, S. Fogelberg, H. Johnsson, C. B.Pers, and K. Perrson. 2005. Climate change impact onwater quality: Model results from southern Sweden. Am-bio 34:559–66.

Arnell, N. W. 2003. Relative effects of multi-decadal climaticvariability and changes in the mean and variability ofclimate due to global warming: Future streamflows inBritain. Journal of Hydrology 270:195–213.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Atasoy, M., R. B. Palmquist, and D. J. Phaneuf. 2006. Esti-mating the effects of urban residential development onwater quality using microdata. Journal of EnvironmentalManagement 79:399–408.

Baker, J. P., D. W. Hulse, S. V. Gregory, D. White, J. VanSickle, P. A. Berger, D. Dole, and N. H. Schumaker.2004. Alternative futures for the Willamette River Basin,Oregon. Ecological Applications 14:313–24.

Balling, R. C. 2000. The geographer’s niche in the green-house millennium. Annals of the Association of AmericanGeographers 90:114–22.

Barlage, M. J., P. L. Richards, P. J. Sousounis, and A. J.Brenner. 2002. Impacts of climate change and land usechange on runoff from a Great Lakes watershed. Journalof Great Lakes Research 28:568–82.

Barnett, T. P., D. W. Pierce, H. G. Hidalgo, C. Bonfils, B. D.Santer, T. Das, G. Bala, et al. 2008. Human-inducedchanges in the hydrology of the Western United States.Science 319:1080–83.

Bicknell, B. R., J. C. Imhoff, J. L. Kittle, Jr., T. H. Jobes,and A. S. Donigian, Jr. 2001. Hydrological Simula-tion Program—Fortran: HSPF version 12 user’s manual.Mountain View, CA: Aqua Terra Consultants.

Boeder, M., and H. Chang. 2008. Multi-scale analysis of oxy-gen demand trends in an urbanizing Oregon watershed,USA. Journal of Environmental Management 87:567–81.

Bouraoui, F., B. Grizzetti, K. Granlund, S. Rekolainen, andG. Bidoglio. 2004. Impact of climate change on the watercycle and nutrient losses in a Finnish catchment. ClimaticChange 66:109–26.

Chang, H. 2003. Basin hydrologic response to changesin climate and land use: The Conestoga River Basin,Pennsylvania. Physical Geography 24:222–47.

———. 2004. Water quality impacts of climate and land usechanges in southeastern Pennsylvania. The ProfessionalGeographer 56:240–57.

Chang, H., B. M. Evans, and D. R. Easterling. 2001. Theeffects of climate change on stream flow and nutrientloading. Journal of the American Water Resources Associa-tion 37:973–86.

Chang, H., and F. Franczyk. 2008 Climate change, land usechange, and floods: Toward an integrated assessment Ge-ography Compass 2:1549–79.

Chang, H., and I-W. Jung. 2010. Spatial and temporalchanges in runoff caused by climate change in a com-plex large river basin in Oregon. Journal of Hydrology388:186–207.

Chang, H., M. Lafrenz, I-W. Jung, M. Figliozzi, D. Platman,and C. Pederson. 2010. Potential impacts of climatechange on flood-induced travel disruption: A case studyof Portland in Oregon, USA. Annals of the Association ofAmerican Geographers 100:938–52.

Chen, J., X. Li, and M. Zhang. 2005. Simulating the impactsof climate variation and land-cover changes on basinhydrology: A case study of the Suomo Basin. Science inChina Series D: Earth Sciences 48:1501–09.

Choi, W. 2008. Catchment-scale hydrological response toclimate-land-use combined scenarios: A case study forthe Kishwaukee River Basin, Illinois. Physical Geography29:79–99.

Chung, E., and K. S. Lee. 2009. Prioritization of water man-agement for sustainability using hydrologic simulation

model and multicriteria decision making techniques.Journal of Environmental Management 90:1502–11.

Clifford, N. J. 2009. Globalization: A physical geographyperspective. Progress in Physical Geography 33:5–16.

Davis Todd, C. E., A. M. Goss, D. Tripathy, and J. M.Harbor. 2007. The effects of landscape transformationin a changing climate on local water resources. PhysicalGeography 28:21–36.

Donigian, A. S., Jr., B. R. Bicknell, and J. C. Imhoff. 1995.Hydrological Simulation Program—Fortran. In Com-puter models of watershed hydrology, ed. V. P. Singh, 395–442. Littleton, CO: Water Resources Publications.

Ducharne, A., C. Baubion, N. Beaudoin, M. Benoit, G.Billen, N. Brisson, J. Garnier, et al. 2007. Long-termperspective of the Seine River system: Confronting cli-matic and direct anthropogenic changes. Science of theTotal Environment 375:292–311.

Dunne, T., and L. Leopold. 1978. Water in environmentalplanning. New York: Freeman.

Franczyk, J., and H. Chang. 2009. The effects of climatechange and urbanization on the runoff of the Rock Creekbasin in the Portland metropolitan area, Oregon, USA.Hydrological Processes 23:805–15.

Graham, L. P., S. Hagemann, S. Jaun, and M. Beniston.2007. On interpreting hydrological change from regionalclimate models. Climatic Change 81:97–122.

Hunter, H. M., and R. S. Walton. 2008. Land-use effectson fluxes of suspended sediment, nitrogen and phospho-rus from a river catchment of the Great Barrier Reef,Australia. Journal of Hydrology 356:131–46.

Kay, A. L., H. N. Davies, V. A. Bell, and R. G. Jones. 2009.Comparison of uncertainty sources for climate changeimpacts: Flood frequency in England. Climatic Change92:41–63.

Kundzewicz, Z. W., L. J. Mata, N. W. Arnell, P. Doll, P.Kabat, B. Jimenez, K. A. Miller, T. Oki, Z. SÍn, andI. A. Shiklomanov. 2007. Freshwater resources and theirmanagement. In Climate change 2007: Impacts, adaptationand vulnerability. Contribution of Working Group II to theFourth Assessment Report of the Intergovernmental Panelon Climate Change, ed. M. L. Parry, O. F. Canziani, J.P. Palutikof, P. J. van der Linden, and C. E. Hanson,173–210. Cambridge, UK: Cambridge University Press.

Legates, D. R., and G. J. McCabe, Jr. 1999. Evaluating theuse of “goodness-of-fit” measures in hydrologic and hy-droclimatic model validation. Water Resources Research35:233–41.

Loukas, A., L. Vasiliades, and N. R. Dalezios. 2002. Climaticimpacts on the runoff generation processes in BritishColumbia, Canada. Hydrology and Earth System Sciences6:211–27.

Milly, P. C. D., J. Betancourt, M. Falkenmark, R. M. Hirsch,Z. W. Kundzewicz, D. P. Lettenmeier, and R. J. Stouffer.2008. Stationarity is dead: Whither water management?Science 319:573–74.

Milly, P. C. D., R. T. Wetherald, K. A. Dunne, and T. L.Delworth. 2002. Increasing risk of great floods in a chang-ing climate. Nature 415:514–17.

Mimikou, M. A., E. Baltas, E. Varanou, and K. Pantazis.2000. Regional impacts of climate change on water re-sources quantity and quality indicators. Journal of Hy-drology 234:95–109.

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014


Mote, P. W., E. A. Parson, A. F. Hamlet, W. S. Keeton,D. Lettenmaier, N. Mantua, E. L. Miles, et al. 2003.Preparing for climatic change: The water, salmon, andforests of the Pacific Northwest. Climatic Change 61:45–88.

Murdoch, P. S., J. S. Baron, and T. L. Miller. 2000. Poten-tial effects of climate change on surface-water quality inNorth America. Journal of the American Water ResourcesAssociation 36:347–66.

Nakicenovic, N., J. Alcamo, G. Davis, B. DeVries, J.Fenhann, S. Gaffin, K. Gregory, et al. 2000. Special reporton emissions scenarios: A special report of Working GroupIII of the Intergovernmental Panel on Climate Change.Cambridge, UK: Cambridge University Press.

National Resource Conservation Service. 2001. STATSGOSoils Dataset, Oregon. http://www.brc.tamus.edu/swat/doc databases-soils.html (last accessed 7 May 2008).

Nelson, D. R., W. N. Adger, and K. Brown. 2007. Adaptationto environmental change: Contributions of a resilienceframework. Annual Review of Environment and Resources32:395–419.

Oki, T., and S. Kanae. 2006. Global hydrological cycles andworld water resources. Science 313:1068–72.

Oregon Climate Service. 2008. Zone 2—Climate dataarchives. http://ocs.orst.edu/index.html (last accessed 7May 2008).

Palmer, R. N., E. Clancy, N. T. VanRheenen, and M. W.Wiley. 2004. The impacts of climate change on the TualatinRiver Basin water supply: An investigation into projectedhydrologic and management impacts. Hillsboro, OR: CleanWater Services.

Praskievicz, S., and H. Chang. 2009. A review of hydrologicmodeling of basin-scale climate change and urban devel-opment impacts. Progress in Physical Geography 33:650–71.

Randhir, T. O., and A. G. Hawes. 2009. Watershed land useand aquatic ecosystem response: Ecohydrologic approachto conservation policy. Journal of Hydrology 364:182–99.

Regonda, S. K., B. Rajagopalan, M. Clark, and J. Pitlick.2005. Seasonal cycle shifts in hydroclimatology over thewestern United States. Journal of Climate 18:372–84.

Ribarova, I., P. Ninov, and D. Cooper. 2008. Modeling nu-trient pollution during a first flood event using HSPFsoftware: Iskar River case study, Bulgaria. Ecological Mod-elling 211:241–46.

Salathe, E. P., Jr., P. W. Mote, and M. W. Wiley. 2007.Review of scenario selection and downscaling methodsfor the assessment of climate change impacts on hydrol-ogy in the United States Pacific Northwest. InternationalJournal of Climatology 27:1611–21.

Samaniego, L., and A. Bardossy. 2006. Simulation of theimpacts of land use/cover and climatic changes on therunoff characteristics at the mesoscale. Ecological Mod-elling 196:45–61.

Sharma, U. C. 2003. Impact of population growth and climatechange on the quantity and quality of water resources in

the northeast of India. IASH-AISH Publication 281:349–57.

Tang, Z., B. A. Engel, B. C. Pijanowski, and K. J. Lim. 2005.Forecasting land use change and its environmental im-pact at a watershed scale. Journal of Environmental Man-agement 76:35–45.

Tualatin River Watershed Council. 1999. Tualatin River wa-tershed action plan. Hillsboro, OR: Tualatin River Wa-tershed Council.

Turner, B. L. 2002. Contested identities: Human–environment geography and disciplinary implicationsin a restructuring academy. Annals of the Association ofAmerican Geographers 92:52–74.

U.S. Environmental Protection Agency (U.S. EPA). 2000.BASINS technical note 6: Estimating hydrology and hy-draulic parameters for HSPF. Washington, DC: U.S.EPA.

———. 2001. Better assessment science integrating point andnonpoint sources: BASINS version 3.0 user’s manual.Washington, DC: U.S. EPA.

———. 2006. BASINS technical note 8: Sediment parameterand calibration guidance for HSPF. Washington, DC: U.S.EPA.

———. 2008. Multi-Resolution Land Characteristics Con-sortium (MRLC). http://www.epa.gov/mrlc/ (last ac-cessed 5 December 2008)

U.S. Geological Survey (USGS). 2004. National ElevationDataset (NED) 1/3 arc second. http://seamless.usgs.gov/Website/Seamless/products/3arc.asp (last accessed 8May 2008).

———. 2006. 8-digit hydrologic unit code boundaries.http://water.usgs.gov/GIS/huc.html (last accessed 8 May2008).

———. 2008. Tualatin River Basin water quality assess-ment. http://or.water.usgs.gov/tualatin/pn356.html (lastaccessed 5 December 2008).

Watts, G. 1997. Hydrological modelling in practice. InContemporary hydrology: Towards holistic environmen-tal science, ed. R. Wiley, 151–93. Chichester, UK:Wiley.

Weglarczyk, S. 1998. The interdependence and applicabil-ity of some statistical quality measures for hydrologicalmodels. Journal of Hydrology 206:98–103.

Wilby, R. L., and I. Harris. 2006. A framework for assess-ing uncertainties in climate change impacts: Low-flowscenarios for the River Thames, UK. Water ResourcesResearch 42:W02419.

Wilby, R. L., P. G. Whitehead, A. J. Wade, D. Butterfield,R. J. Davis, and G. Watts. 2006. Integrated modelling ofclimate change impacts on water resources and qualityin a lowland catchment: River Kennet, UK. Journal ofHydrology 330:204–20.

Xu, C. 1999. From GCMs to river flow: A reviewof downscaling methods and hydrologic modellingapproaches. Progress in Physical Geography 23:229–49.

Correspondence: Department of Geography, University of Oregon, Eugene, OR 97403, e-mail: [email protected] (Praskievicz); Departmentof Geography, Portland State University, Portland, OR 97207, e-mail: [email protected] (Chang).

Dow

nloa

ded

by [

Tex

as A

&M

Uni

vers

ity L

ibra

ries

] at

16:

32 1

4 N

ovem

ber

2014

impacts of climate change and urban development on water resources in the tualatin river basin,...

Documents