Hydrologic Sensitivity to Climate and Land Use Changes in the Santiam River Basin, Oregon1

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION

AMERICAN WATER RESOURCES ASSOCIATION

HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN, OREGON1

Cristina Mateus, Desiree D. Tullos, and Christopher G. Surﬂeet2

ABSTRACT: Future changes in water supply are likely to vary across catchments due to a river basin’s sensitivity to climate and land use changes. In the Santiam River Basin (SRB), Oregon, we examined the role elevation, intensity of water demands, and apparent intensity of groundwater interactions, as characteristics that inﬂu- ence sensitivity to climate and land use changes, on the future availability of water resources. In the context of water scarcity, we compared the relative impacts of changes in water supply resulting from climate and land use changes to the impacts of spatially distributed but steady water demand. Results highlight how seasonal runoff responses to climate and land use changes vary across subbasins with differences in hydrogeology, land use, and elevation. Across the entire SRB, water demand exerts the strongest inﬂuence on basin sensitivity to water scarcity, regardless of hydrogeology, with the highest demand located in the lower reaches dominated by agricultural and urban land uses. Results also indicate that our catchment with mixed rain-snow hydrology and with mixed surface-groundwater may be more sensitive to climate and land use changes, relative to the catchment with snowmelt-dominated runoff and substantial groundwater interactions. Results highlight the importance of evaluating basin sensitivity to change in planning for planning water resources storage and allocation across basins in variable hydrogeologic settings.

(KEY TERMS: climate change; land use change; GCMs; water scarcity; sensitivity; water resources planning; model uncertainty.)

Mateus, Cristina, Desiree D. Tullos, and Christopher G. Surﬂeet, 2014. Hydrologic Sensitivity to Climate and Land Use Changes in the Santiam River Basin, Oregon. Journal of the American Water Resources Association (JAWRA) 1-21. DOI: 10.1111/jawr.12256

INTRODUCTION Northwest (PNW) region is estimated to substantially reduce summer streamflow (Baker et al., 2004; Tague et al., 2008; Chang and Jung, 2010). Earlier Meeting multiple purposes of hydrosystems is spring runoff, shorter winter runoff periods, and likely to become increasingly difficult as demands for longer and drier summers (Tague and Grant, 2004, water, food, and energy increase in the future (Baker 2009; Chang and Jung, 2010) may lead to changes in et al., 2004), and as climate and land use changes the timing and quantity of water supply (Tague will impact water supply (Bureau of Reclamation, et al., 2008; Chang and Jung, 2010; Surfleet and 2011). For example, warming projected for the Pacific Tullos, 2013).

1Paper No. JAWRA-13-0164-P of the Journal of the American Water Resources Association (JAWRA). Received July 19, 2013; accepted August 5, 2014. © 2014 American Water Resources Association. Discussions are open until six months from print publication. 2PhD Candidate (Mateus), Water Resources Engineering, 116 Gilmore Hall, and Associate Professor (Tullos), Biological and Ecological Engineering, Oregon State University, Corvallis, Oregon 97331; and Associate Professor (Surﬂeet), Natural Resources Management and Environmental Science, California Polytechnic State University, San Luis Obispo, California 93407 (E-Mail/Mateus: mateuscm@ onid.orst.edu).

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 1 JAWRA MATEUS,TULLOS, AND SURFLEET

However, the degree of difficulty in meeting man- transpiration by trees can influence the amount of agement objectives is likely to vary across catchments water available for direct drainage to streams or depending on an individual basin’s sensitivity to aquifer recharge (Farley et al., 2005). Land use changes in climate and land use. A basin’s sensitivity change effects on water resources vary considerably is based on the degree to which a system is affected between different altitudes and vegetation covers by changes in climate and land use (Gallopin, 2006). (Muhamad et al., 2013). For example, increases in While the availability and management of water vegetation density can decrease snowpack accumula- resources are generally sensitive to climate (Johnson tion, increase transpiration, and increase the evapo- and Weaver, 2009; Brown et al., 2012) and land use ration of precipitation intercepted by vegetation changes (Vor€ osmarty€ et al., 2000), the sensitivity var- canopy, thus reducing runoff (VanShaar et al., 2002). ies with characteristics of the basin, such as eleva- While the links between basin characteristics and tion, hydrogeology, and existing land use. its sensitivity to land use and climate change are For example, elevation can play an important role becoming clearer (Lettenmaier et al., 1992; Guo et al., in a basin’s sensitivity to climate change, but in com- 2008), additional work is needed to identify the impli- plex ways. Basins at higher elevations with hydro- cations of basin sensitivity on the availability of logic regimes heavily influenced by snowmelt are water resources and water scarcity. In this study, we more vulnerable to temperature changes than precip- present analyses on the influence of climate and land itation changes (Chang and Jung, 2010). On the other use changes on the future availability of water hand, rainfall-dominated basins are more vulnerable resources across subbasins with different hydrogeo- to changes in precipitation than changes in tempera- logic and land use characteristics. The objectives of ture (Chang and Jung, 2010). However, seasonal this study were to investigate how subbasin charac- snow zones in the PNW can be particularly sensitive teristics, including elevation, intensity of water to changes in both temperature and precipitation demands, and apparent intensity of groundwater (Lettenmaier et al., 1992; Surfleet and Tullos, 2013). interactions, contribute to hydrologic sensitivity, to Hydrogeology also plays a central role in a basin’s climate and land use change response, and to water sensitivity to climate and land use changes. While scarcity. surface water systems are likely to respond more rapidly to changes in precipitation and temperature, groundwater-driven systems are likely to experience greater magnitudes of change (Tague and Grant, METHODS 2009; Safeeq et al., 2013). For groundwater systems, subsurface storage captures snowmelt runoff and releases it across longer time scales, thus maintain- Study Area ing more consistent, and sometimes higher, base-flow conditions than observed in surface water systems The Santiam River Basin (SRB) (Figure 1), drain- with rapid runoff responses to precipitation (Tague ing the Cascade mountain range, encompasses et al., 2008). However, the storage of precipitation in approximately 4,700 km2 of the eastern portion of the the subsurface of groundwater systems, particularly Willamette River Basin (WRB). Elevations range those dominated by snowmelt, may result in a larger from 50 m at the confluence of the Willamette River absolute change as the climate warms relative to sur- to 3,199 m at the summit of Mt. Jefferson. Land use face water systems (Tague and Grant, 2009; Safeeq is primarily forested (79%) in the headwaters, with et al., 2013). Declines in snowpack appear to have alluvial, agricultural (3%), urban (3%), and range limited effect on late summer low flows in surface areas (15%) in the lower reaches (Table 1). Approxi- water systems because catchments are already very mately 80% of precipitation falls between October dry in the summer. Larger effects on late summer and April and is characterized by rain below 350 m, base flow may occur in groundwater systems due to rain and snow between 350 and 1,100 m, and snow groundwater depletion, in combination with shifts in above 1,100 m (Surfleet and Tullos, 2012). Four dams precipitation from snow to rain that leads to earlier operated by the U.S. Army Corps of Engineers regu- peak runoff events (Safeeq et al., 2013). late runoff in the basin: Detroit and Big Cliff Dams In addition to a basin’s underlying elevation and in the North Santiam Basin (NSB); Foster and Green geology, land use can also play a pivotal role in a Peter Dams in the South Santiam Basin (SSB) (Fig- basin’s sensitivity to climate and land use changes. ure 1). The reservoirs are operated primarily for flood Factors that reduce infiltration and groundwater control, with secondary benefits of hydropower recharge, such as urbanization, can lower base-flow production, recreation, fisheries, and water quality levels. In addition, changes to vegetative cover that regulation, and irrigation water supply. Flood man- influence interception of precipitation and evapo- agement has modified the natural streamflow of the

JAWRA 2 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

FIGURE 1. Santiam River Basin (SRB), Oregon, Including Reservoirs, Elevations Range, Aquifer Characteristics, Vulnerability to Water Scarcity Location Points (small black dots), and Hydrologic Responses Location Points (large white circles). Left inset: SRB point of diversions (black dots), North Santiam Basin (NSB) in dark gray and South Santiam Basin (SSB) in light gray, and subbasins delimited by the white lines. Right inset: SRB location in Oregon.

Santiam River by decreasing the frequency and mag- North Santiam River. Urban areas are concentrated nitude of floods and elevating the magnitudes of in small communities situated along the lower main summer low flows (Risley et al., 2012). stem of the NSB, comprising 4.6% of the basin area. The NSB drains approximately 2,000 km2. High The SSB drains approximately 2,700 km2 of pre- Cascades geology dominates the upper regions of dominantly Western Cascades geology (Tague et al., NSB and is characterized by highly porous and 2008; Surfleet and Tullos, 2012). The Western Cas- permeable volcanic layers that contribute to high cades geology has steep with well-developed drainage groundwater recharge and low drainage densities networks. Shallow soils rapidly drain to 1-3 m deep (Tague and Grant, 2004) that sustain base flow confining layers, leading to shallow subsurface storm during the dry summer months (Tague et al., 2008; flow that generates rapid runoff responses, high peak Chang and Jung, 2010; Surfleet and Tullos, 2012). flows, high-flow variability, and little groundwater The NSB is primarily forested (Table 2), with agricul- storage (Tague and Grant, 2004). Compared to NSB, tural land use occurring downstream of the Little this basin has greater variation in land uses

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 3 JAWRA MATEUS,TULLOS, AND SURFLEET

TABLE 1. Santiam River Basin Characteristics.

Natural or Major Groundwater Area Regulated Precipitation Inﬂuence Location Study Site (km2) Runoff Land Use Type Geology on Runoff

Santiam River SRB 4,719 Regulated 79% forest, 15% range, Rain and snow 60% Western Cascades, Moderate Basin 3% agriculture, 15% High Cascades, 2% urban, 10% Alluvium, 1% water 5% Basalt North Santiam NSB_Mouth 2,036 Regulated 77% forest, 16% range, Rain 52% Alluvium, High River at the 5% agriculture, 36% Western Cascades, Mouth 1% urban, 12% Basalt 1% water South Santiam SSB_Mouth 2,683 Regulated 80% forest, 15% range, Rain and snow 50% Western Cascades, Low River at the 2% agriculture, 38% Alluvium, Mouth 2% urban, 11% Basalt, 1% water 1% High Cascades North Santiam NSB_Boulder 1,172 Natural 97% forest, 2.8% water, Snow 90% High Cascades, High Basin below 0.2% urban 5% Western Cascades, Boulder Creek 5% Alluvium North Santiam NSB_Mehama 1,660 Regulated 91% forest, 2% range, Rain and snow 30% High Cascades, Moderate Basin at 4% agriculture, 60% Western Cascades, Mehama 1% urban, 10% Alluvium 1% water South Santiam SSB_Waterloo 1,192 Regulated 91% forest, Rain and snow 90% Western Cascades, Low Basin at 2% agriculture, 5% Alluvium, Waterloo 3% range, 2% urban, 3% Basalt, 2% water 2% High Cascades South Santiam SSB_Cascadia 456 Natural 99% forest, 1% water, Rain and snow 95% Western Cascades, Low Basin at 0.4% urban, 3% Basalt, Cascadia 0.2% range 2% High Cascades South Santiam SSB_Deer 319 Regulated 91% forest, 5% water, Rain and snow 95% Western Cascades, Low Basin at 3% range, 1% urban 3% Basalt, Deer Creek 2% High Cascades

(Table 2), with a greater proportion of the SSB basin evaluate the background sensitivity of the basins to occupied by urban and agricultural uses than in the changes in precipitation and temperature. We calcu- NSB. Nearly 98% of the urban development, rural lated seasonal runoff change and seasonal runoff var- residential, and agriculture occur below Green Peter iability to represent different aspects of hydrologic and Foster dams. responses to climate and land use changes. A sensi- In the lower sections of the SRB, just above the tivity analysis was conducted to investigate whether conﬂuence of the NSB and SSB (Figure 1), the hy- climate change or land use change is responsible for drogeology is characterized by substantial areas of the observed hydrologic responses. Finally, we evalu- alluvium and associated groundwater storage and ate basins’ sensitivity to water scarcity as the ratio of recharge from the Willamette Valley Aquifer (Lee demand to supply to identify areas most sensitive to and Risley, 2002). Approximately 90% of this valley water scarcity across the basin. Water demand by bottom area is in agriculture and range land use with land use is based on spatially variable but steady little forest and urban land uses. estimates from the Oregon Water Resources Depart- ment (OWRD, 2011).

Study Approach Estimates of Future Water Supply. We apply results from a coupled groundwater-surface water We apply (Figure 2) projections of future runoff model (GSFLOW) to investigate the effects of climate due to climate change (Surfleet and Tullos, 2012) and and land use changes on future water supply. land use change (Baker et al., 2004) to investigate GSFLOW represents a coupling of the United States future water supply in the SRB at nine locations Geological Survey’s (USGS), Modular Groundwater (Figure 1). Each location represents a collection of Flow Model (MODFLOW) simulating groundwater runoff and land use projections of subbasins that flow, and the USGS Precipitation-Runoff Modeling drain to that point. We calculated historical precipita- System simulating surface water flow (Risser et al., tion elasticity (Ɛp) and temperature sensitivity (S)to 2005; Markstrom et al., 2008). Integrated GSFLOW

JAWRA 4 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

TABLE 2. Land Use Projections in the North Santiam Basin (NSB) and South Santiam Basin (SSB) at Low-, Mid-, and High-Elevation Locations for Historic Time Period (2010), and Percent Changes in Land Use from Historic to Conservation 2050 and Development 2050 Scenarios. Low elevations: NSB_Mouth and SSB_Mouth; Mid elevation: NSB_Mehama and SSB_Waterloo; High elevation: NSB_Boulder and SSB_Cascadia.

NSB SSB

Historic Conservation Development Historic Conservation Development (km2) 2050 (%) 2050 (%) (km2) 2050 (%) 2050 (%)

Low elevation Agriculture 98 2 2 62 5 6 Forest 1,574 6 6 2,156 1 1 Range 323 11 11 404 7 9 Urban 27 1 1 45 0 1 Water 14 2 2 16 1 1 Mid elevation Agriculture 60 1 1230 0 Forest 1,511 0 0 1,086 0 0 Range 28 1 1 40 0 0 Urban 22 0 0 21 0 0 Water 39 0 0 24 0 0 High elevation Agriculture 0 0 0 0 0 0 Forest 1,132 0 0 450 0 0 Range 0 0 0 1 0 0 Urban 7 0 0 2 0 0 Water 33 0 0 2 0 0

FIGURE 2. Study Approach Flowchart. White boxes represent the inputs used in the model, the white circle represents future streamflow projections used for this analysis, and the gray boxes represent the metrics used to assess hydrologic sensitivity to projected changes in climate and land use. modeling for the SRB was parameterized for 13 sub- and the low-elevation, alluvial basins (SR_Mouth, basins that consisted of hydrologic response units SR_Morgan, NSB_Mouth, and SSB_Mouth). For the (HRUs), delineated by Chang and Jung (2010) based Upper North Fork subbasin, leakage of groundwater on the similarity in elevation, geology, soil type, from the High Cascades to the east side toward the slope, and aspect. For computational efficiency, MOD- Deschutes basin was represented by applying esti- FLOW was applied only in the subbasins where mates of groundwater loss from Lee and Risley substantial groundwater interactions occur. These (2002). Groundwater flow from the High Cascades include subbasins draining the High Cascades (Upper (Boulder Creek) was not transferred as surface water North Fork Santiam where Boulder Creek is located) flow to the mid-elevation basins of the NSB under

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 5 JAWRA MATEUS,TULLOS, AND SURFLEET the assumption that the groundwater flow is stored global climate model inputs, and model validation are in deep storage. Implications of this assumption are reported in Surfleet and Tullos (2012). considered in the Discussion section of this paper. Climate change was simulated within GSFLOW Groundwater flow was simulated in the low-elevation, (Surfleet and Tullos, 2012) using eight Global Cli- alluvial locations, calibrated by both streamflow and mate Model (GCM) projections and two greenhouse well observations. Thus, the lower basin groundwater gas (GHG) emission scenarios (A1B and B1) (IPCC, simulations included groundwater contribution from 2007) for the SRB. GCM results were statistically the alluvial aquifer but not from the High Cascades. downscaled using the bias correction and spatial To represent the uncertainty attributed to hydro- downscaling method (Wood et al., 2002). While the logic modeling parameters and to address equifinality downscaled GCM projections of changes in precipita- in GSFLOW simulations, Surfleet and Tullos (2012) tion and temperature were applied consistently over developed posterior distributions for 13 of the hydro- the spatial domain, the timing and delivery of runoff logic model parameters using a formal Bayesian were projected to vary spatially across the basin due parameter uncertainty approach, Differential Evolu- to the changes in precipitation type. For example, tion Adaptive Metropolis (DREAM) (Vrugt et al., the spatial distribution of precipitation varies 2009). DREAM applies a Markov Chain Monte Carlo between the High and Western Cascades as a result Sampling algorithm to a priori parameter distribu- of precipitation falling primarily as snow at the tions for the subbasins from Chang and Jung (2010). higher elevations of High Cascades and primarily The posterior distributions of parameter values for as rain at lower elevations of the Western Cascades behavioral solutions were determined in the DREAM (Tague and Grant, 2004). It was predicted that High assessment when the Gelman-Rubin convergence Cascades areas of the SRB will have a greater pro- diagnostic was less than 1.2 for each of the Markov portion of precipitation falling as rain in the future sampling chains, where an evolution of a chain is relative to the past, along with a shift to smaller but conducted for each parameter assessed. Due to com- more frequent rain-on-snow events (Chang and Jung, putational and time constraints, this uncertainty 2010; Surfleet and Tullos, 2012). The Western analysis and development of tuned parameter distri- Cascades, generally at lower elevations than High butions were performed for only three study subba- Cascades, was also predicted to have increased sins (SSB at Cascadia, NSB at Boulder Creek, and proportion of precipitation as rain but lower propor- SSB at Thomas Creek) representing the range of tions of rain-on-snow events due to less snow precipi- topographical and hydrogeologic settings of the SRB tation. (High Cascades, Western Cascades, and transitional Land use changes were simulated in GSFLOW by area of foothills to Willamette Valley floor, respec- adjusting model parameters representing percent tively). The parameter distributions from these three impervious areas, vegetation interception for rain basins were then transferred to the remaining basins and snow for summer and winter, vegetation cover based on similarity in the hydrogeology and topogra- density for summer and winter, and parameters con- phy. trolling snow interception. GSFLOW calculates the Model fit to observed historic (1960-2006) stream- hydrologic response of a subbasin based on a flow is high for both daily and monthly records in the weighted average of each parameter from HRUs three tuned parameter subbasins, with Nash Sutcliffe within subbasins. Each land use was assigned a Efficiencies (NSE) greater than 0.7 and 0.8, respec- parameter value indicative of that land use (Table 3). tively. However, model fit to observations for the To simulate land use changes, model parameters basins to which parameter distributions were trans- were adjusted based on the proportion of change in ferred varied between basins, ranging from a NSE land use area within HRUs. Changes in land use value of 0.73 for the Santiam River at Jefferson to were represented by adjusting both the vegetation NSE of 0.52 for the South Santiam Waterloo site (Sur- cover parameter as well as the amount of interception fleet and Tullos, 2012). The differences in model fit due to a change in vegetation. Interception of precipi- values are likely due to the differences between the tation by plants in GSFLOW is computed as a func- DREAM-tuned parameter subbasins and the trans- tion of plant-cover density and the storage available ferred basins with respect to the proportion of the on the plant cover type in each HRU. The intercep- basins draining the High Cascades geology. To repre- tion term was not adjusted within the uncertainty sent this uncertainty associated with modeled monthly assessment when parameter distributions were devel- streamflow, the lower (LCI, 2.5%), median (MCI, 50%), oped. Interception was only modified due to changing and upper confidence intervals (UCI, 97.5%) are vegetation cover with land change, following the reported for each of the metrics, as described below in model uncertainty assessment. further detail. Further details on model development Land use projections were based on Conservation and uncertainty assessment, parameter distributions, and Development scenarios obtained from the alter-

JAWRA 6 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

TABLE 3. Parameter Ranges for Varying Land Use and Vegetation Types and References.

Road, Mature Urban Lots Urban Med Urban No Structure, Cover Type Doug Fir Hardwood Grass Trees Trees Trees Shrubs Corn etc.

Percentage of 0 0 0 10-50 10-50 10-50 0 0 100 impervious areas (%) Cover density 10-20 8-15 0.1 7-11 4-8 2-5 10-20 5-10 0 (winter) (%) Cover density 24-35 10-20 0.2 30-36 10-18 1-10 10-20 20-35 0 (summer) (%) Snow interception 2-6 1-3 0 1-2 0.5-1 0 1-2 0 0 (max) (in) Summer rain 2-3 2-4 0.1-0.3 0.5-1.5 0.2-1.0 0-0.2 0.1-0.3 0.1-0.3 0 interception (max) (in) Winter rain 10-15 1-2 0.1 0.5-1.5 0.2-1.1 1-0.2 0.1-0.3 0.1-0.3 0 interception (max) (in) Source (i) (ii) (ii) (iii) (iii) (iii) (ii) (ii)

Source: (i) Rothacher, 1963; (ii) Crockford and Richardson, 2000; (iii) Xiao et al., 1998. nate futures analysis developed by the Pacific North- emphasizing ecosystem protection and restoration. west Ecosystem Research Consortium (PNW-ERC, Projections indicated increases in forested and range 2011) for historic (2010) and future (2050) time peri- areas with reductions in urban and agriculture areas ods (Table 2). The PNW-ERC alternate futures analy- (Table 2). For consistency with GSFLOW projections ses were reported as projected changes relative to using current land use, we grouped the 101 land uses historic conditions. Future land use scenarios reflect of PNW-ERC (2011) into five classes: forest, range, assumptions about land and water use, as well as the urban, water, and agriculture. Across the entire impact of these land use scenarios on ecological (ter- basin, these policy scenarios primarily lead to differ- restrial wildlife, water availability, ecological condi- ences (Table 2) in the area occupied by range, agri- tion of streams) and socioeconomic endpoints (Baker culture, and water. Differences between the NSB and et al., 2004). Their scenario approach was based on a SSB exist, with the SSB experiencing substantially set of plausible assumptions from stakeholders and smaller changes for nearly all land use categories. ecological consequences regarding future trends of potential management actions, reflecting potential Estimating Historical Sensitivities to Precipi- future trends in urbanization, rural residential devel- tation and Temperature. To evaluate streamflow opment, agriculture, forestry, and water use (Baker sensitivity of the basins to changes in precipitation et al., 2004). Thus, Conservation and Development and temperature, we conducted a precipitation elas- land use scenarios were not predictions of future ticity (Ɛp) and temperature sensitivity (S) analysis on change, but rather visualizations of the likely out- historical discharge (USGS NWIS, 2010), daily pre- comes of the stakeholder assumptions. Under both cipitation, and air temperature data (NOAA COOP, scenarios, population for the entire WRB was 2010; NRCS SNOTEL, 2010). We compared Ɛp and S expected to nearly double by 2050, resulting in for Boulder in the NSB and Cascadia in the SSB changes in land use. Land use change projections since both USGS gauge stations are located above the (e.g., from forest to agriculture or urban areas) reservoirs. Ɛp was calculated following the methods of reflected the shift from past human uses to new uses, Sankarasubramanian et al. (2001), which applies a rather than the expansion of the impacted area of nonparametric estimator to evaluate the simulated natural ecosystems. The Development scenario impacts of climate change on annual streamflow. We assumed market-oriented approaches to land use and were unaware of any definitive method for estimating water use, and reflected loosening of current land use temperature elasticity in basins due to the many restrictions. This policy was represented in the Devel- challenges associated with delayed timing of runoff opment scenario as reductions in agricultural areas response to changes in temperatures and the interac- and increases in urban and rural residential uses tions between temperature and other system vari- (Table 2). The Conservation scenario assumed restric- ables that influence runoff (e.g., precipitation, tive development policies would be implemented, evapotranspiration, etc.). For these reasons, tempera-

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 7 JAWRA MATEUS,TULLOS, AND SURFLEET ture elasticity is generally modeled (e.g., Dooge, 1992; ratio of the standard deviation (r) for either Decem- Vano et al., 2012). Instead, we applied a simplified ber, to represent the winter season, or August, to approach using historical data at the monthly time represent the summer season, to its mean (l) (Vogel scale to develop a general understanding of the rela- et al., 2007). tive sensitivity between the two study basins. We cal- To evaluate the degree to which variation in culated S for August to represent the summer December and August runoff is influenced by changes months and December to represent the winter in land use or climate, we performed a sensitivity months separately as the percent change in mean analysis taking into account three forcing variables: monthly flow to the percent change in mean monthly (1) percent change in land use over time; (2) percent temperature. In reporting the results, we acknowl- change in precipitation for A1B and B1 scenarios; edge that their interpretation is greatly limited by and (3) percent change in air temperature for A1B assuming that any interactions across processes and and B1 scenarios. We performed a linear regression time are not fully represented in this analysis. His- sensitivity analysis to evaluate the strength of the 2 torical observations of mean annual precipitation relationship (R ) and the effect (slope, b1) between across the 1974-2006 water years (October-Septem- hydrologic responses and forcing variables. ber) were 1,620 mm at Cascadia, while the mean annual runoff depth was 1,583 mm. Over the same Estimating Sensitivity to Water Scarcity time period, Boulder’s mean annual precipitation was within and across Basins. Consistent with the 2,099 mm, while the mean annual streamflow was water stress index (Water Systems Analysis Group, 1,593 mm. Therefore, about 98% of precipitation was UNH, 2009), we evaluated scarcity as the ratio of translated into streamflow for the Cascadia site, demand to supply (Equation 1), with changes in whereas only 76% of precipitation was translated into water supply derived from GSFLOW simulations and streamflow for the Boulder site. the distribution of water demand estimates derived from water rights in the OWRD (2011), described in Characterizing Hydrologic Responses within further detail below. Under this approach, a region and across Basins. We analyzed seasonal runoff was considered to experience water scarcity if change and seasonal runoff variability at six locations demand was greater than supply, with a water scar- in the SRB (Table 1, Figure 1) to represent different city ratio less than one. We assessed water scarcity aspects of hydrologic responses to climate and land at eight locations in the SRB (Figure 1, Table 1), use changes. We selected six basins for analysis based on the availability of demand estimates from based on availability of projected streamflow and dif- the OWRD and on the differences in geology and ele- ferences in basin geology and water sources: the vation: at the mouth and at Morgan Creek in the North Santiam River at the mouth, Mehama and Santiam River; at the mouth, Mehama and Boulder Boulder Creek and in the South Santiam River at the Creek in the North Santiam River; and at the mouth, mouth, Waterloo and Cascadia. Analysis was con- Waterloo and Deer Creek in the South Santiam ducted at each of the six locations for three scenarios River. We note that one location (Deer Creek) in the representing combinations of land use and climate SSB is different from the calculations of hydrologic changes: (1) Climate Change only (CC); (2) Develop- responses due to the differences in the availability of ment + Climate Change (Dev-CC); and (3) Conserva- water demands. However, because Deer Creek and tion + Climate Change (Cons-CC). For the analysis of Cascadia are of similar drainage areas, elevations, hydrologic responses to these scenarios, we represent land use, and geology (Table 1), we believed that the future time as the average of the water years these two sites represent similar hydrogeologic set- 2030-2059, and the simulated historical time period tings. as the average of the water years from 1960 to 2010. To analyze the seasonal runoff change, we estimated the change in monthly runoff (mm) for the Demand CU þ IS Water Scarcity ¼ ¼ ½cmð1Þ 2.5 (LCI), 50 (MCI), and 97.5 (UCI) percentile values Supply Q þ ST for the future time period compared to simulated historical at each location and for scenarios (1), (2), where Q is the mean monthly streamflow obtained and (3). Monthly streamflow was represented as the from GSFLOW simulations for the 2.5, 50, and 97.5 modeled average daily streamflow for each month percentiles; ST defines the existing storage in reser- multiplied by the number of days in that month. To voirs; CU are the out-of-stream consumptive uses; analyze the changes in seasonal variability of the and IS represents existing instream water rights. streamflow and uncertainty in future seasonal runoff Thus, the water scarcity ratio indicates pressures under different scenarios, we used a coefficient of from irrigation, industrial, commercial, municipal, variation (CV) of the seasonal streamflow. CV is the and domestic sectors as well as environmental flows.

JAWRA 8 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

The approach used to calculate existing storage season were different than those needed for reproduc- (ST), consumptive uses (CU), and instream uses (IS) tion during the Summer Chinook spawning season is based on the OWRD methodology (Cooper, 2002). between September and October. For this analysis, we considered water demand to be constant in the future based on the assumption that, under Oregon’s water rights law, surface water is no longer available for new uses and groundwater sup- RESULTS plies may also be limited (OWRD, 2012). Therefore, future changes in water use will result in changes in the locations where the water is extracted, rather Historical Sensitivities of Upper Basins to than the amount of water being extracted. In addi- Precipitation and Temperature tion, by representing water demand as the entire allocation of water rights in the basin, we overesti- The sensitivity of two sites in the upper basins to mated the actual amount of water diverted from the changes in precipitation and temperature indicated stream, making our evaluation of the potential for that the SSB hydrogeology may have greater sensi- water scarcity higher than the actual scarcity experi- tivity to climate change than the NSB. Results of the enced in the basins. However, this assumption was elasticity analysis indicated that increases in precipi- not in conflict with the goal of this analysis, which tation could increase streamflow by a factor of 1.47 at was to evaluate the spatial and temporal patterns Cascadia and 1.0 at Boulder. Values of the CV for and geographic relationships with water scarcity, the precipitation elasticities were lower at Cascadia rather than the absolute magnitude of change in each than at Boulder, with elasticity CV values of 37 and basin. Therefore, our results should be interpreted 46%, respectively. The results of our simplified analy- not as absolute values but as the relative potential sis of monthly discharge and temperature ratios indi- for scarcity across basins. cated that the upper SSB also exhibited greater The water that is retained in the four reservoirs is sensitivity to temperature than the NSB in the his- allocated as existing storage. ST was calculated as torical record. Increases in temperature were related the volumetric change in reservoir storage from one to a decrease in streamflow by a factor of 0.5 at Cas- month to the next, as reported at the end of each cadia and to no change in streamflow at Boulder month for each reservoir based on historical storage. during the winter months. The impact of increasing ST was then converted to a discharge (e.g., cm) by temperature on summer month runoff appeared to be dividing by the time between the two months. Con- a decrease at both Cascadia and Boulder, by factors sumptive uses account for irrigation, municipal, of 5.5 and 1.3, respectively. Both the temperature industrial, commercial, and domestic demands. The and precipitation elasticity analyses indicated that CU values were based on the part of the diversion the groundwater basin was historically less affected consumed via evaporation and transpiration, assum- by changes in precipitation and temperature com- ing that evaporation and transpiration rates will pared to the surface water basin. We speculated that remain constant in the future, irrespective of land the greater historical sensitivity to precipitation and use change. Water not consumed, for example, with temperature in the unregulated SSB was associated municipal use, was assumed to return to the stream with rapid runoff response due to the lack of ground- and was not subtracted from the total supply. CU water storage. was calculated by multiplying the maximum diversion rate allowed for the water right by a CU coefficient. For example, consumptive uses for municipal Hydrologic Responses within and across Basins use was the actual diversion multiplied by a consumptive use coefficient for summer (0.45) and winter Projected Change in Seasonal Runoff (0.15). In the case of consumptive uses for irrigation, Depth. Whereas the depths of December runoff the number of irrigated acres was multiplied by crop were projected to increase for all subbasins except type, irrigation method. Therefore, return flows were SSB at Waterloo (Figure 3a), August low flows were calculated by subtracting the water being withdrawal projected to decline in the future for all subbasins from the CU. IS uses were not considered consump- (Figure 3b). These projections were consistent across tive because they represent water rights and scenic all land use and climate scenarios. However, while waterway flows held by the OWRD for beneficial uses subbasins exhibited similar trends in monthly runoff, such as recreation or wildlife protection. IS uses of the magnitude of projected changes varied by land water depended on the stream, the intended use, and use, GHG emission scenario, and location within the varied by season. For example, minimum flows that basin. To illustrate the differences between the hy- are provided for adult fish habitat during the summer drogeologic settings of the NSB and SSB, we com-

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 9 JAWRA MATEUS,TULLOS, AND SURFLEET

FIGURE 3. Monthly Change in Runoff Depth (mm) for (a) December and (b) August for the North Santiam Basin at: the Mouth of the North Santiam River (NSB_Mouth), Boulder Creek (NSB_Boulder), and Mehama (NSB_Mehama); and for the South Santiam Basin at: the Mouth of the South Santiam River (SSB_Mouth), Waterloo (SSB_Waterloo), and Cascadia (SSB_Cascadia) under Conservation Scenario (x axis) and Development Scenario (y axis). Black figures represent A1B scenario and gray figures represent B1 scenario; solid figures represent the NSB and figures with a pattern represent the SSB. The black lines represent the upper and lower confidence interval for each scenario. Note that the scales for December and August plots are different, but the grid line spacing is consistent. pared pairs of sites across the two basins of similar At the mid-elevation basins, SSB_Waterloo and elevation and drainage areas. Differences that were NSB_Mehama (Figure 1), located 21- and 34-km larger than the uncertainties are discussed below. upstream in the respective basins, projected winter Considering the entire NSB and SSB, we found and summer runoff indicate that elevation and geol- indications of differences in the two basins’ sensitivity ogy play an important role in differentiating runoff to climate and land use changes. For example, while responses at these locations. At SSB_Waterloo, a winter runoff (Figure 3a) was projected to increase at rainfall dominated subbasin draining the Western the mouth of both the SSB and NSB, the increase in Cascades geology, substantial decreases in runoff winter runoff was higher for the SSB_Mouth than depths are projected for both December (Figure 3a) the NSB_Mouth. The higher runoff in the SSB and August (Figure 3b). In contrast, the NSB_Meha- occurred for the median confidence intervals under ma subbasin, located in a transient zone with mixed all GHG emission and land use scenarios, although rain and snow due to its mid-range elevations, is pro- the differences were not larger than the uncertainties jected to undergo increases in winter runoff and when considering the upper and lower confidence small decreases in summer runoff relatively to Water- intervals. This result suggested that, taken as the loo. Larger decreases in summer runoff are projected entire basin, the SSB may be more sensitive than the for SSB_Waterloo compared to NSB_Mehama, sug- NSB to both land use and climate changes that influ- gesting higher sensitivity to changes in climate and ence winter runoff over the next 40 years. Reductions land use for SSB_Waterloo during the summer in summer runoff depths were also larger for the months. Changes in both winter and summer runoff SSB_Mouth than the NSB_Mouth (Figure 3b), with depth at these mid-elevation locations are all signifi- all differences larger than the modeled uncertainties. cant when uncertainties are considered. Thus, the larger reductions in summer runoff indi- The most upstream locations in the basin, NSB_ cated that the SSB broadly is also more sensitive Boulder and SSB_Cascadia, were projected to than the NSB to land use and climate changes that undergo consistently low changes in winter and sum- influence summer runoff. mer runoff regardless basin hydrogeology. For both

JAWRA 10 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON locations, small increases in December runoff (Fig- At the low-elevation locations in the basin, a ure 3a) were projected for the future, with higher greater decrease in streamflow variability was pro- increases for the NSB than the SSB. However, jected during August for SSB_Mouth than increases in winter runoff were only larger than NSB_Mouth (Figures 5a-5f). August streamflow vari- uncertainties for the SSB_Cascadia location and only ability for A1B CC, CC-Cons, and CC-Dev was pro- for the B1 Conservation scenario. Summer runoff jected to decrease by 0.06, 0.1, and 0.09, respectively, (Figure 3b) was projected to decrease in the future at SSB_Mouth, whereas the same scenarios produced for both locations, although changes were smaller relatively small (0.01) changes for the NSB_Mouth. than the uncertainties under all scenarios. These For the B1 scenario, August streamflow variability at results suggested that seasonal runoff responses to the SSB_Mouth was projected to decrease by 0.05 for changes in climate and land use were similar at the the CC scenario and 0.08 for CC-Cons and CC-Dev two high-elevation locations. scenarios, whereas changes were again small (0.02) at the NSB_Mouth. All changes in August streamflow Projected Change in Streamflow Variabil- variability are all significant when model uncertainty ity. Differences in seasonal streamflow variability was considered. are projected across GHG emission scenario, land use While December streamflow variability was pro- scenario, and location within the basins. August jected to undergo small decreases at the mouth of streamflow variability (Figure 5) was generally both basins under all scenarios (Figure 4), slightly higher in the SSB than the NSB, indicating greater greater decreases in variability were projected for the variability in summer water supply for the SSB com- NSB_Mouth (Figures 4b-4f) than the SSB_Mouth, pared to the NSB in the future. No consistent pat- with the exception of A1B scenario under CC land terns in the projections for December streamflow use scenario (Figure 4a). December variability for variability (Figure 4) was observed for the both A1B at the SSB_Mouth was projected to decrease by basins. These results suggested that, in terms of vari- 0.02, 0.01, and 0.02 for CC, CC-Cons, and CC-Dev, ability, summer streamflow may distinguish the two respectively, whereas these same scenarios projected basins better than winter streamflow. a decrease of 0.02, 0.02, and 0.03 for the NSB_Mouth,

CC CONS-CC DEV-CC c) e) a) 0.6 0.25 0.25 0.5 0.23 0.23 0.4 0.20 0.20 0.3 0.18 0.18

A1B 0.2 0.15 0.15 0.1 0.13 0.13 0.0 0.10 0.10

d) f) b) 0.6 0.25 0.25 0.5 0.23 0.23 0.4 0.20 0.20 B1 0.3 0.18 0.18 December Coefficient of Variation of Variation Coefficient December 0.2 0.15 0.15 0.1 0.13 0.13 0.0 0.10 0.10

FIGURE 4. Coefﬁcient of Variation for December for: A1B (top) and B1 (bottom) Climate Scenarios, and Climate Change (CC); Conservation + Climate Change (Cons-CC); and Development + Climate Change (Dev-CC) Land Use Scenarios. The squares represent the historic time period, the diamonds represent the future time period, and the vertical black lines represent the uncertainty associated with each location. The scale for CC scenario differs from the scale for Cons-CC and Dev-CC.

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 11 JAWRA MATEUS,TULLOS, AND SURFLEET

FIGURE 5. Coefficient of Variation for August for: A1B (top) and B1 (bottom) Climate Scenarios, and Climate Change (CC); Conservation + Climate Change (Cons-CC); and Development + Climate Change (Dev-CC) Land Use Scenarios. The squares represent the historic time period, the diamonds represent the future time period, and the vertical black lines represent the uncertainty associated with each location. The scale for CC scenario differs from the scale for Cons-CC and Dev-CC. respectively. For B1 scenario, the SSB_Mouth is pro- basins’ response to land use and climate changes. All jected to undergo a 0.02, 0.01, and 0.01 smaller of these changes in seasonal streamflow variability reduction in December runoff for CC, CC-Cons, and were outside the uncertainty bands, indicating that CC-Dev, respectively, compared to the NSB_Mouth. December and August streamflow variability may Changes in winter streamflow variability were small decrease at both sites in response to both climate and but significant under all scenarios when uncertainties land use changes, but that the degree of decrease were taken into account with the exception of A1B depends on the land use and climate scenarios. Development scenario (Figure 4e). For the farthest upstream locations in the basin, At mid-elevation locations, December and August in most scenarios, August and December streamflow streamflow variabilities were both generally projected variabilities were projected to increase at SSB_Cas- to decrease in the future, with the exception of some cadia and to decrease at NSB_Boulder. Historically, mixed trends in August variability. While December August variability was lower at NSB_Boulder and streamflow variability was projected to decrease for was projected to further decrease, whereas the both SSB_Waterloo and NSB_Mehama under all sce- higher historical values of variability at SSB_Casca- narios (Figure 4), there were no clear trends in dia were projected to further increase in the future December runoff to distinguish the two basins’ hydro- (Figure 5), creating a greater distinction between the logic response to land use or climate change scenar- streamflow variability at the two sites. All the ios. In most cases, SSB_Waterloo had higher changes in August streamflow variability were signif- variability for December streamflow compared to icant when model uncertainty is considered. For NSB_Mehama, with the exception of A1B-CC sce- December streamflow variability (Figure 4), values nario (Figure 4a). Historic August streamflow vari- were much more similar historically between the two ability was also higher at SSB_Waterloo compared to basins than for the summer flow variability. In the NSB_Mehama (Figure 5), but the change to future future, only CC and Dev-CC scenario differences variability is mixed at both locations. Similar to were outside the uncertainty bands, with small December runoff, there were no clear trends in increases in December streamflow variability pro- August runoff variability to distinguish the two jected for both sites.

JAWRA 12 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

Relative Influence of Projected Land Use and based on R2 values, the effect is opposite with a posi- Climate Changes on Seasonal Runoff. To evalu- tive relationship between temperature and December ate whether land use or climate was driving higher runoff and a negative relationship between tempera- winter and lower summer runoff at the basin scale, ture and August runoff. In both December and we evaluated the strength of relationships between August, the magnitude of the effect of temperature runoff depths and changes in land use, precipitation, changes on runoff is greater under the B1 scenario and temperature. Changes in winter and summer than under A1B. Therefore, increases in December runoff exhibited a weak relationship with percent runoff and decreases August runoff due to changes in change in land use, based on the low R2 (Table 4). temperature are significant regardless of land use The larger magnitude of the b1 coefficient for land changes. use change under A1B indicated that land use change The strength of the relationship between precipita- has a stronger influence on runoff under the more tion change and runoff change was weakest for the aggressive land use scenario, Development, than Con- scenarios with land use change relative to the sce- servation (Table 4). Comparing sensitivities to land nario with climate change only. Based on the very use change between winter and summer runoff, we low R2 values, there was a stronger negative relation- found that the slope, b1, for December runoff was ship between precipitation change and December positive and steeper than the negative slopes for runoff under the climate change scenario compared August runoff, indicating a stronger influence of land to scenarios that included land use changes. While use change on December runoff than on August run- the relationship between precipitation and December off. Furthermore, the very low R2 values for August runoff for B1 existed under the land use scenarios, runoff indicated that land use has very little effect there was no relationship between precipitation and on summer streamflows in the basin. Thus, results December runoff for A1B. Therefore, the relationship suggest that land use appeared to have limited effect between precipitation and runoff change appears to on summer and winter runoff. be significant only for winter runoff when land use In considering the effect of temperature on runoff changes are not considered. depths, we find that runoff is most strongly related to temperature changes when no land use change is Water Scarcity within and across considered. When adding the effect of the two land Basins. Defining water scarcity as the condition use scenarios to the climate change scenarios, the when demand exceeds supply, projections indicated relationship between temperature and runoff depths that summer sensitivity to water scarcity was gener- is far weaker (Table 4). However, while the strength ally higher in the NSB, but varied with intensity of of the relationship is similar between December and water demands and elevation. Water scarcity was August for the scenario with no land use change, found to be more prevalent under current hydrologic

2 TABLE 4. Sensitivity Analysis: Represented by the Strength of the Relationship (R ) and the Size of the Effect (slope, b1) between Hydrologic Responses and Forcing Variables for Climate Change Projections Only (CC), Climate Change + Conservation Land Use Projections (CC-Cons), and Climate Change + Development Land Use Projections (CC-Dev). Values with the strongest relationships are highlighted in gray.

December August

2 2 Y=b0 + b1x b1 b0 R b1 b0 R

% Change in land A1B CC-Cons 0.70 4.94 0.13 0.30 15.03 0.02 use vs. % change CC-Dev 0.90 0.85 0.38 0.28 16.01 0.02 in runoff B1 CC-Cons 0.01 11.27 0.00 0.11 17.79 0.00 CC-Dev 0.26 4.64 0.04 0.09 19.09 0.00 % Change in A1B CC 3.09 24.95 0.39 3.79 67.04 0.31 temperature vs. CC-Cons 1.59 27.09 0.01 1.63 53.07 0.04 % change in runoff CC-Dev 0.59 11.54 0.00 1.79 57.34 0.04 B1 CC 3.32 32.64 0.67 4.17 53.25 0.22 CC-Cons 1.54 5.70 0.01 3.80 79.58 0.08 CC-Dev 3.66 34.22 0.11 4.07 85.56 0.09 % Change in A1B CC 3.58 27.14 0.27 0.86 18.72 0.00 precipitation vs. CC-Cons 3.48 26.32 0.03 5.71 4.43 0.01 % change in runoff CC-Dev 0.83 9.00 0.00 4.87 7.15 0.01 B1 CC 2.86 17.98 0.27 1.86 9.75 0.00 CC-Cons 3.25 44.08 0.15 9.71 2.87 0.03 CC-Dev 5.53 33.06 0.13 9.43 1.08 0.03

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 13 JAWRA MATEUS,TULLOS, AND SURFLEET

FIGURE 6. Subbasins Sensitivity to Water Scarcity (demand > supply) in the Santiam River Basin under Current 2010, Conservation 2050, and Development 2050 Land Use Scenarios. (1) Santiam River at the Mouth; (2) Santiam River at Morgan; (3) North Santiam River at the mouth; (4) North Santiam River at Mehama; (5) South Santiam River at the mouth; (6) South Santiam River at Waterloo; (7) South Santiam River at Deer Creek. Vulnerability to water scarcity varies across subbasins and is localized in lower regions of the basin. conditions for the lower elevation locations of the demands decrease and water supply begins to have NSB (Figures 6, 7c-d, and 8a-b) than the SSB (Fig- more influence. ures 6, 7a-b, 8c-d). The higher scarcity of the entire In the valley bottom (SR_Morgan and SR_Mouth), NSB, relative to the SSB, was projected to further water demands exerted a strong influence on esti- increase in the future (Figure 6) as supply drops. mated sensitivity to water scarcity (Figure 7). Below However, basin sensitivity to water scarcity appeared the confluence of NSB and SSB, demand was greater to be primarily driven by intensity of water demands than supply from July to September for both Conser- at each location. In fact, for both NSB and SSB, sub- vation and Development scenarios at SR_Morgan basins’ sensitivities to water scarcity decreased mov- (Figures 7c-7d). However, moving 9 km downstream ing upstream (Figure 6), due primarily to decreasing to the mouth of the Santiam River, demand was water demand. However, groundwater contribution to greater than supply only for the low confidence (Fig- summer low flows also appeared to influence sensitiv- ures 7a-7b). Both locations have similar hydrogeolo- ity to water scarcity farther upstream in the NSB gy, with groundwater recharge from the alluvial and SSB basins (Figures 9a-b and 10b), where water areas of the Willamette River aquifer. However, the

JAWRA 14 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

2010 A1B 2010 B1 2050 A1B 2050 B1 Demand Conservation Development a) b) 200 200

150 150

100 100

50 50

0 0 Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct c) d) 200 200

150 150

100 100

SR_Morgan SR_Mouth SR_Morgan 50 50

0 0 Streamflow (cms) Streamflow (cms) Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct

FIGURE 7. Summer Streamflow Supply and Demand Comparisons for Low-Elevation Sites: Santiam River Mainstem at the Mouth (SR_Mouth) and Santiam River at Morgan Creek (SR_Morgan) under Conservation (left) and Development (right) Land Use Scenario. Supply is presented as the median streamflow with gray lines and the error bars representing the lower confidence interval and upper confidence interval. B1 scenario is represented by the dashed lines and the solid lines represent A1B scenario. Water demand is represented by the solid thick black line. water demands at SR_Morgan were 111 cm higher and future time period under Conservation for both than SR_Mouth from June to October, associated A1B and B1 scenarios (Figure 6). Thus, despite the with irrigation diversions and municipal water diver- slightly higher supply generated by the upstream sions for the City of Jefferson. These diversions led to hydrogeology in the NSB relative to the SSB, sub- scarcity at SR_Morgan during summer months under stantially larger water demands in the NSB_Mouth both the current and future scenarios. In contrast, at resulted in greater sensitivity to water scarcity for SR_Mouth, only the LCI values under the future sce- current and future climate and land use conditions. nario for July-September fell below the demand line At the mid-elevation locations, NSB_Mehama and (Figures 7a-7b), indicating that this reach generally SSB_Waterloo, sensitivity to late summer water scar- is unlikely to experience water scarcity in the future. city appeared to be driven less by water demands and Changes in land use, which do not modify water more by water supply, although the current threat of demands in our analysis, had minimal impact on water scarcity was low for both sites (Figure 9). supply and thus on potential scarcity for these two While demand was higher at NSB_Mehama, late locations. summer supply was also higher at this site, relative Moving upstream to the NSB_Mouth and to SSB_Waterloo. Thus, despite lower demands, SSB_Mouth sites, differences in water supply were SSB_Waterloo appeared to have slightly greater small, but differences in water demands greatly dis- potential for late summer water scarcity (Figure 6) tinguished the basins’ sensitivity to water scarcity both currently and into the future. Water scarcity at (Figure 8). Despite slightly greater supply from the SSB_Waterloo was projected only for the LCI during upper NSB, demand at the NSB_Mouth was greater August and September for the current (Figure 9c) than supply (Figures 8a-8b) under both current and and future (Figure 9d) scenarios. Only small future time periods. Demands at the NSB_Mouth increases in water scarcity were projected for were up to 100 cm higher than the SSB_Mouth. SSB_Waterloo that were not outside the uncertainty Thus, SSB_Mouth was projected to experience water bounds of the analysis. Water scarcity at NSB_Meha- scarcity only for the LCI (Figures 8c-8d) under cur- ma was projected only for the LCI during August rent time period for Development and A1B scenario, under Conservation and Development scenario for

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 15 JAWRA MATEUS,TULLOS, AND SURFLEET

2010 A1B 2010 B1 2050 A1B 2050 B1 Demand Conservation Development a) b) 120 120 100 100 80 80 60 60 40 40 20 20 0 0 Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct c) d) 120 120 100 100 80 80 60 60

SSB_Mouth 40 NSB_Mouth 40 20 20 0 0 Streamflow (cms) Streamflow (cms) Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct

FIGURE 8. Summer Streamflow Supply and Demand Comparisons for Low-Mid-Elevation Sites: North Santiam River (NSB) at the Mouth (NSB_Mouth) and South Santiam River (SSB) at the Mouth (SSB_Mouth) under Conservation (left) and Development (right) Land Use Sce- nario. Supply is presented as the median streamflow with gray lines and the error bars representing the lower confidence interval and upper confidence interval. B1 scenario is represented by the dashed lines and the solid lines represent A1B scenario. Water demand is represented by the solid thick black line. the current time periods (Figures 9a-9b). Despite pro- DISCUSSION jected reductions in water supply, the late summer differences between supply and demand were projected to remain low, with similar levels of water While we found evidence of basin-wide patterns in scarcity at the two sites due to contrasting levels of hydrologic responses and sensitivity to climate and supply and demand. land use changes, it appears that subbasin character- We found little evidence of water scarcity at the istics, including elevation, intensity of water most upstream locations in both subbasins demands, and groundwater interactions, influence (NSB_Boulder and SSB_Deer) due to the low water the direction and degree of hydrologic response, as demands, regardless of the differences in hydrogeolo- well as sensitivity to water scarcity, in several key gy (Figure 6). The NSB_Boulder subbasin, located ways. 59 km upstream from the mouth of the North Sant- iam River, exhibited no evidence of water scarcity for all time periods and scenarios (Figures 10a-10b). Hydrologic Responses and Sensitivity to Climate and Although demand is projected to be higher at Land Use Changes across Hydrogeologic Settings NSB_Boulder than at SSB_Deer, summer supply was also higher in the NSB due to groundwater contribu- Patterns of hydrologic response to climate and land tions sustaining base flow. As a result, it appears use changes vary between subbasins in SRB by unlikely that the NSB_Boulder will experience water hydrogeology and elevation. Beginning with runoff scarcity now or in the projected future. The depth changes, we projected that future water supply SSB_Deer, located 46 km above the mouth of the will decrease during the dry months and increase Santiam River, was found to also experience little during the wet months in most subbasins (Figure 3), scarcity, with supply dropping below demand only for consistent with other studies in the PNW (Tague and the LCI under both land use scenarios in August and Grant, 2004; Tague et al., 2008; Chang and Jung, September for the current time period (Figures 10c 2010; Hamlet et al., 2010). However, the direction and 10d). and magnitude of changes in summer and winter

JAWRA 16 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

2010 A1B 2010 B1 2050 A1B 2050 B1 Demand Conservation Development a) b) 100 100

80 80

60 60

40 40

20 20

0 0 Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct c) d) 100 100 Streamflow (cms) Streamflow (cms) 80 80

60 60

SSB_Waterloo NSB_Mehama NSB_Mehama SSB_Waterloo 40 40

20 20

0 0 Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct

FIGURE 9. Summer Streamflow Supply and Demand Comparisons for Mid-Elevation Sites: North Santiam River (NSB) at Mehama (NSB_Mehama) and South Santiam River (SSB) at Waterloo (SSB_Waterloo) under Conservation (left) and Development (right) Land Use Scenario. Supply is presented as the median streamflow with gray lines and the error bars representing the lower confidence interval and upper confidence interval. B1 scenario is represented by the dashed lines and the solid lines represent A1B scenario. Water demand is represented by the solid thick black line. streamflow varied between and along the two basins. ter-driven systems have slow summer recession and The largest increase in December runoff was pro- are likely to experience greater magnitude of change jected for the SSB_Mouth and the greatest decrease in low flows, relative to surface water systems that in August runoff was projected for the SSB_Mouth. already have very little water in late summer, due to At most sites, including the low- and high-elevation groundwater depletion. We note that the configuration locations, the differences in December runoff change of the groundwater model may have impacted our esti- between the NSB and SSB were not significant when mates of August streamflow variability for the mid- uncertainty was considered. and low-elevation sites. As noted in the Methods, the More important to water scarcity, we found signifi- upper and lower NSB were modeled separately with cant differences between the basins for late summer no transfer of groundwater from the high-elevation flow at low- and mid-elevation locations, where we pro- subbasin (Boulder Creek) to the mid-elevation subba- jected larger reductions in summer runoff at SSB rela- sins such that only surface runoff was transferred to tive to NSB. These results were supported by the the mid-elevation subbasins from Boulder Creek. elasticity analysis for the upper basin sites, which Thus, any groundwater generated from recharge was indicated that the SSB has historically been more sen- assumed to be stored deep in the aquifer and did not sitive to changes in temperature and precipitation. contribute to summer base flow at the lower elevations These results may also reflect the impact of elevation in the system. As such, the model configuration likely (Nolin and Daly, 2006; Safeeq et al., 2013), whereby underestimated the impact of groundwater discharge late summer flow may be more sensitive to changes in contributing to base flow on the NSB. However, given climate in systems located along the rain-snow transi- the effects of lower historical precipitation elasticity tion, of which the SSB occupies a greater proportion, and lower temperature sensitivity for the NSB relative than higher elevation with snow precipitation, of to the SSB, as well as the effect of elevation, which which the NSB occupies a higher proportion. However, influences the type of precipitation, it is not clear how these results contradict findings in other Cascades important the model configuration is on the results. basins (Tague and Grant, 2004, 2009; Safeeq et al., While the degree of change in seasonal streamflow 2013), where it has been demonstrated that groundwa- variability varies at each location and scenario, gen-

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 17 JAWRA MATEUS,TULLOS, AND SURFLEET

2010 A1B 2010 B1 2050 A1B 2050 B1 Demand Conservation Development a) b)

100 100

80 80

60 60

40 40

20 20

0 0 Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct c) d)

100 100

80 80

60 60 SSB_Deer NSB_Boulder 40 40

20 20

Streamflow (cms) Streamflow (cms) 0 0 Jun Jul Aug Sep Oct Jun Jul Aug Sep Oct

FIGURE 10. Summer Streamflow Supply and Demand Comparisons for High-Elevation Sites: North Santiam River (NSB) at Boulder Creek (NSB_Boulder) and South Santiam River (SSB) at Deer Creek (SSB_Deer) under Conservation (left) and Development (right) Land Use Scenario. Supply is presented as the median streamflow with gray lines and the error bars representing the lower confidence interval and upper confidence interval. B1 scenario is represented by the dashed lines and the solid lines represent A1B scenario. Water demand is represented by the solid thick black line. erally, the projected summer and winter variability managers that both influences and is influenced by a in streamflow for the two basins indicated that the range of anthropogenic and environmental factors. low variability in NSB streamflow will become even This study’s results demonstrated how, regardless of less variable, while the higher variability in SSB hydrogeology, demand for the water has a strong streamflow will increase. The historical and future influence on the potential for water scarcity. Areas differences between these basins are likely associated most sensitive to water scarcity in the SRB were with hydrogeology. Lower groundwater contribution found in the lower reaches of the basin (Figures 7c-d, and rapid runoff responses at the SSB influenced by 8a-b, 9b), where municipal and agricultural water flashier Western Cascades geology may explain the demands were high. While water supply generated by higher variability in this basin compared to the the upstream hydrogeology of the NSB (Figures 8a- highly porous High Cascades geology that drains 8b) was higher compared to the SSB (Figures 8c-8d), most of the NSB, providing less variable flow higher water demands in the NSB resulted in greater throughout the year. From a management perspec- sensitivity for current and future climate and land tive, this result was interpreted that summer water use conditions. Thus, an elevation gradient in sensi- supply will be less predictable in the future in the tivity to water scarcity was evident in our results, SSB than in the NSB. with areas sensitive to water scarcity decreasing in the upstream direction (Figure 6) due to decreasing demands. In the NSB, the upstream reduction in Influence of Demand on Sensitivity to Water Scarcity water scarcity was due to both reductions of water demand and sustained base flow during the dry Water scarcity, defined as the value of allocating months sourced by the High Cascades hydrogeology. units of water to different users in a certain area While geologic distinctions and elevation differences (Jaeger et al., 2013), is a major concern for water in snow accumulation and melt are the two mecha-

JAWRA 18 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON nisms controlling summer streamflow (Tague and logic responses varied with subbasins and the charac- Grant, 2004; Tague et al., 2008), our results demon- teristics that contribute to a basin’s hydrologic strated the need for a third mechanisms of water sensitivity to climate and land use, including eleva- demand to be included in the analysis of water tion, apparent intensity of groundwater interactions, resources to translate the impact of water scarcity and intensity of water demands. In addition, we eval- into impacts on people. uated the relative importance of changing water sup- The assumption that water demand will remain ply and the spatial distribution of water demands on constant in the future, based on currently allocated water scarcity, assuming that demand remained con- rights, likely resulted in conservative estimates of stant due to policy constraints for each subbasin. Our water scarcity. By allocating the full water right to results highlighted how: (1) Water demand exerts a each location, we estimated the potential for water strong influence on basin’s sensitivity to water scar- scarcity as a worst-case scenario, in which all water city regardless of basin characteristics, whereby sen- rights were exercised. While some have argued that sitive areas in our study basins tended to be located future changes in water use will mostly result in in the lower reaches where agricultural and munici- changes in the locations where water is extracted, pal demands were highest; (2) Changes in land use, rather than the amount of water being extracted which did not modify water demands in this analysis, (OWRD, 2008), others (Baker et al., 2004) indicate had minimal impact on supply, thus on potential that water demands in WRB may increase in the scarcity, but did reduce the signal of climate change; future, a scenario that will lead to further scarcity (3) Late summer low flow in low-elevation areas with than was estimated herein. mixed rain-snow catchments and a mixed groundwater-surface water system can be more sensitive to changes in climate and land use relative to high-ele- Relative Influence of Land Use and Climate Changes vation areas with snowmelt runoff patterns and substantial groundwater interactions; (4) Though the Results from the sensitivity analysis on land use degree of change varies with future scenario and ele- and climate change impacts suggested that the vation, summer runoff variability was projected to strength of the relationship and the effect between distinguish surface water and groundwater-driven percent change in runoff and percent change in land basins more than winter runoff variability, with sum- use, temperature, and precipitation vary depending mer runoff variability increasing in the surface water on the season and scenario. The strongest relation- system and decreasing in the groundwater system. ship between runoff and climatic forcing variables of Results emphasize the need for water managers to temperature and precipitation existed under the cli- consider water demands, basin hydrogeology, eleva- mate-only scenarios. In almost all cases, adding land tion, projected land use and climate changes, and use change to climate projections via the Develop- hydrologic model uncertainties when allocating and ment and Conservation scenarios reduces the distributing water to different users within the basin. strength of the relationships between temperature For example, lower groundwater contribution and and precipitation climatic forces and runoff to insig- rapid runoff response at the SSB may generate a nificant values of R2. Thus, land use change dimin- flashier and more variable hydrograph with larger ished the portion of the runoff change explained by winter peak flows and smaller summer base flows climate change alone. In addition, runoff sensitivities now and in the future. However, the degree of sensi- to land use change indicated that land use has a tivity even within the SSB varies with elevation, stronger influence on December runoff than August intensity of water demands, and degree of land use runoff. Finally, while temperature and precipitation change. While our results apply to a basin with changes did not explain the majority of the variation unique hydrogeologic characteristics, improving the in runoff, temperature changes appeared to have understanding of the sensitivity to and increased stronger impact on runoff compared to precipitation uncertainty caused by climate and land use changes, changes. and thus its effect on water resources, will increase the capacity for planning the storage and allocations of water resources across basins with variable hydrogeologic settings. CONCLUSIONS

ACKNOWLEDGMENTS Future water scarcity is likely to vary across river basins relative to their sensitivity to climate and land This material is based on the work supported by the National use changes. We investigated how patterns in hydro- Science Foundation under grant no. 0846360. Any opinions, ﬁnd-

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 19 JAWRA MATEUS,TULLOS, AND SURFLEET ings, and conclusions or recommendations expressed in this mate- Johnson, T.E. and C.P. Weaver, 2009. A Framework for Assessing rial are those of the author(s) and do not necessarily reflect the Climate Change Impacts on Water and Watershed Systems. views of the National Science Foundation. The research was sup- Environmental Management 43(1):118-134. ported, in part, by the National Science Foundation through Tera- Lee, K. and J. Risley, 2002. Estimates of Ground-Water Recharge, Grid resources provided by Purdue University under grant number Base Flow and Stream Reach Gains and Losses in the Willia- TG-ECS100006. Funding was also provided by Secretaria Nacional mette River Basin, Oregon. U.S. Geological Survey Water de Educacion Superior, Ciencia, Tecnologıa e Innovacion SENES- Resources Investigations Report 01-4215, 53 pp. CYT, ECUADOR. We also gratefully acknowledge two anonymous Lettenmaier, D.P., K.L. Brettmann, L.W. Vail, S.B. Yabusaki, and reviewers for their thoughtful and detailed comment on an earlier M.J. Scott, 1992. Sensitivity of Pacific Northwest Water version of this manuscript. Resources to Global Warming. Northwest Environmental Jour- nal 8(2):265-283. Markstrom, S.L., R.G. Niswonger, S. Regan, D.E. Prudic, and P.M. Barlow, 2008. GSFLOW—Coupled Ground-Water and Surface- LITERATURE CITED Water Flow Model Based on the Integration of the Precipitation- Runoff Modeling System (PRMS) and the Modular Ground-Water Baker, J.P., D.W. Hulse, S.V. Gregory, D. White, J. Van Sickle, Flow Model (MODFLOW-2005). 6, Vol. 1; USGS. P.A. Berger, and D. Dole, 2004. Alternative Futures for the Wil- Muhamad, Q., K. Hubacek, M. Termansen, and L. Fleskens, 2013. lamette River Basin, Oregon. Ecological Applications 14(2):313- Modelling Land Use Change across Elevation Gradients in Dis- 324. trict Swat, Pakistan. Regional Environmental Change 13(3):567- Brown, C., Y. Ghile, M. Laverty, and K. Li, 2012. Decision Scaling: 581. Linking Bottom-Up Vulnerability Analysis with Climate Projec- NOAA COOP (National Oceanic and Atmospheric Administration tions in the Water Sector. Water Resources Research 48: Cooperative Observer Program), 2010. Daily Precipitation and W09573, doi: 10.1029/2011WR011212. Maximum & Minimum Temperature Data. http://www.nws. Bureau of Reclamation, 2011. Reclamation: Managing Water in the noaa.gov/om/coop, accessed July 2010. West. Basin Report No. SECURE Water Act, Section 9503(c). Nolin, A.W. and C. Daly, 2006. Mapping “At Risk” Snow in the US Department of the Interior Bureau of Reclamation. http://www. Pacific Northwest. Journal of Hydrometeorology 7(5):1164-1171. usbr.gov/climate/SECURE/factsheets/columbia.html, accessed June NRCS SNOTEL (Natural Resources Conversation Service Snow 2012. Telemetry System), 2010. Map of Oregon SNOTEL Sites. http:// Chang, H. and I. Jung, 2010. Spatial and Temporal Changes in www.or.nrcs.usda.gov/snow/maps/oregon_sitemap.htm, accessed Runoff Caused by Climate Change in a Complex Large River July 2010. Basin in Oregon. Journal of Hydrology 388(3-4):186-207. OWRD (Oregon Water Resources Department), 2008. Statewide Cooper, R., 2002. Determining Surface Water Availability in Ore- Water Needs Assessment: Oregon Water Supply and Conserva- gon. Open File Report SW 02-002. State of Oregon, Water tion Initiative. Oregon Water Resource Department, HDR Engi- Resource Department, Salem, Oregon. neering, Inc., http://www.oregon.gov/owrd/law/docs/owsci/owrd_ Crockford, R.H. and D.P. Richardson, 2000. Partitioning of Rainfall demand_assessment_report_final_september_2008.pdf, accessed into Through Fall, Streamflow and Interception: Effect of Forest January 2010. Type, Ground Cover and Climate. Hydrologic Processes 14:2903- OWRD (Oregon Water Resources Department), 2011. Water Avail- 2920. ability Report System: Santiam Basin in the Willamette River Dooge, J.C.I., 1992. Sensitivity of Runoff to Climate Change: A Basin. Oregon Water Resources Department. http://apps.wrd. Hortonian Approach. Bulletin of the American Meteorological state.or.us/apps/wars/wars_display_wa_tables/search_for_WAB. Society 73:2013-2024. aspx, accessed June 2011. Farley, K.A., E.G. Jobbagy, and R.B. Jackson, 2005. Effects of OWRD (Oregon Water Resources Department), 2012. Water Afforestation on Water Yield: A Global Synthesis with Implica- Resources Department: Water Protections and Restrictions. tions for Policy. Global Change Biology 11:1565-1576. http://www.oregon.gov/owrd/pages/pubs/aquabook_protections. Gallopin, G., 2006. Linkages between Vulnerability, Resiliency, and aspx, accessed March 2012. Adaptive Capacity. Global Environmental Change 16(3):293- PNW-ERC (Pacific Northwest Ecosystem Research Consortium), 303. 2011. Data Access Page. http://www.fsl.orst.edu/pnwerc/wrb/ Guo, H., Q. Hu, and T. Jiang, 2008. Annual and Seasonal Stream- access.html, accessed January 2011. flow Responses to Climate and Land-Cover Changes in the Poy- Risley, J., J. Wallick, J. Mangano, and K. Jones, 2012. An Environ- ang Lake Basin, China. Journal of Hydrology 355(1):106-122. mental Streamflow Assessment for the Santiam River Basin, Hamlet, A.F., P. Carrasco, J. Deems, M. Elsner, T. Kamstra, C. Oregon. U.S. Geological Survey Open-File Report No. 2012- Lee, and S.-Y. Lee, 2010. Final Report for the Columbia Basin 1133, 60 pp. Climate Change Scenarios Project. Climate Impacts Group. Risser, D.W., R.W. Conger, J.E. Ulrich, and M.P. Asmussen, 2005. http://www.hydro.washington.edu/2860/, accessed January 2013. Estimates of Ground-Water Recharge Based on Streamflow- IPCC (Intergovernmental Panel on Climate Change), 2007. Climate Hydrograph Methods: Pennsylvania. U.S. Geological Survey Change 2007: The Scientific Basis. IPCC Contribution of Work- Open-File Report 2005-1333, 30 pp. ing Group I to the Fourth Assessment Report of the Intergov- Rothacher, J., 1963. Net Precipitation under a Douglas-fir Forest. ernmental Panel on Climate Change. Cambridge University Forest Science 9(4):423-429. Press, Cambridge, United Kingdom. http://www.ipcc.ch/publica Safeeq, M., G. Grant, S. Lewis, and C. Tague, 2013. Coupling tions_and_data/publications_and_data_reports.shtml, accessed Snowpack and Groundwater Dynamics to Interpret Historical September 2011. Streamflow Trends in the Western United States. Hydrological Jaeger, W.K., A.J. Plantinga, H. Chang, K. Dello, G. Grant, D. Hul- Processes 27:655-688, doi: 10.1002/hyp.9628. se, J.J. McDonnell, S. Lancaster, H. Moradkhani, A.T. Morzillo, Sankarasubramanian, A., R.M. Vogel, and J.F. Limbrunner, 2001. P. Mote, A. Nolin, M. Santelmann, and J. Wu, 2013. Toward a Climate Elasticity of Streamflow in the United States. Water Formal Definition of Water Scarcity in Natural-Human Sys- Resources Research 37:1771-1781. tems. Water Resources Research 49:1-12, doi: 10.1002/wrcr. Surfleet, C. and D. Tullos, 2012. Uncertainty Assessment of Hydro- 20249. logic Response to Climate Change for the Santiam River,

JAWRA 20 JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION HYDROLOGIC SENSITIVITY TO CLIMATE AND LAND USE CHANGES IN THE SANTIAM RIVER BASIN,OREGON

Oregon. Hydrological Processes. http://dx.doi.org/10.1002/hyp. 9485, 16 pp. Surfleet, C.G. and D. Tullos, 2013. Variability in Effect of Climate Change on Rain-on-Snow Peak Flow Events in a Temperate Climate. Journal of Hydrology 479(2013):24-34. Tague, C. and G. Grant, 2004. A Geologic Framework for Interpret- ing the Low-Flow Regimes of Cascade Streams, Willamette River Basin, Oregon. Water Resources Research 40:W04303. Tague, C. and G. Grant, 2009. Groundwater Dynamics Mediate Low-Flow Response to Global Warming in Snow-Dominated Alpine Regions. Water Resources Research 45:W07421, doi: 10. 1029/2008WR007179, 12 pp. Tague, C., G. Grant, M. Farrell, J. Choate, and A. Jefferson, 2008. Deep Groundwater Mediates Streamflow Response to Climate Warming in the Oregon Cascades. Climatic Change 86(1-2):189- 210. USGS NWIS (U.S. Geological Survey National Water Information System), 2010. USGS Surface-Water Data for the Nation. http:// waterdata.usgs.gov/nwis/sw, accessed July 1, 2010. Vano, J.A., T. Das, and D.P. Lettenmaier, 2012. Hydrologic Sensi- tivities of Colorado River Runoff to Changes in Precipitation and Temperature. Journal of Hydrometeorology 13:932-949, doi: 10.1175/JHM-D-11-069.1. VanShaar, J., I. Haddeland, and D. Lettenmaier, 2002. Effects of Land-Cover Changes on the Hydrological Response of Interior Columbia River Basin Forested Catchments. Hydrologic Pro- cesses 16:2499-2520. Vogel, R.M., J. Sieber, S.A. Archfield, M.P. Smith, C.D. Apse, and A. Huber-Lee, 2007. Relations among Storage, Yield and In- stream Flow. Water Resources Research 43(5):W05403. Vor€ osmarty,€ C.J., P. Green, J. Salisbury, and R.B. Lammers, 2000. Global Water Resources: Vulnerability from Climate Change and Population Growth. Science 289(5477):284. Vrugt, J., C. Braak, C. Diks, D. Higdon, B. Robinson, and J. Hy- man, 2009. Accelerating Markov Chain Monte Carlo Simulation by Differential Evolution with Self-Adaptive Randomized Sub- space Sampling. International Journal of Nonlinear Sciences and Numerical Simulation 10(3):273-290. Water Systems Analysis Group, UNH, 2009. Relative Water Stress Index. http://webworld.unesco.org/water/wwap/wwdr/indicators/ pdf/A3_Relative_water_stress_index.pdf, accessed March 2012. Wood, A., E. Maurer, A. Kumar, and D. Lettenmaier, 2002. Long- Range Experimental Hydrologic Forecasting for the Eastern United States. Journal of Geophysical Research-Atmospheres 107:4429-4443. Xiao, Q., E. Gregory, J.R. Simpson, and S. Ustin, 1998. Rainfall Interception by Sacramento’s Urban Forest. Journal of Arbori- culture 24(4):235-244.

JOURNAL OF THE AMERICAN WATER RESOURCES ASSOCIATION 21 JAWRA