Modeling Approach

The Willamette River Basin (WRB) is a complex coupled system, one that includes both a natural system (the biophysical components) and a human system (the socio-economic components). The interactions, feedbacks, and evolving characteristics of these different components will control when and where water is abundant or scarce. Given the complexity of the system, and the detailed spatial and temporal scales at which these different components interact and influence each other, an explicit and quantitative representation of this system requires a computer model to incorporate the many processes and relationships in time and space between and among the natural and human system components. This allows us to predict how those processes are likely to change over time. As a result, we developed Willamette Envision. The model enables us to explore the interactions between land and water use, law and policy, and public management of land and water resources.

Willamette Envision includes modeling components that represent water supply, water allocation, and water demand (Fig. 2). We model water supply in the basin using a hydrologic model that translates daily values of meteorological conditions into a spatially distributed estimate of snow, soil moisture, and streamflow conditions. We model water allocation by incorporating the operating rules for the 13 federal Willamette Project reservoirs, and a model of water use constraints imposed by Oregon water law. We model water demand for four sectors: urban areas, agriculture, upland forests, and instream ecosystems (fish). We estimate water demand for urban areas as a function of factors such as water price and population, and simulate urban growth by relating land use changes to land characteristics and economic returns. We model water demand for agricultural and forested lands by estimating evaporative water loss from different crop and forest cover types. Connected modeling components determine farmer planting and irrigation decisions, and forest succession and disturbance by wildfire and harvest. We represent instream ecological water needs by modeling instream water rights and also federally required ESA-related minimum flows that are integrated into reservoir management rules.

Model Components

Willamette Envision incorporates:

• Hydrologic, ecological, and economic system sub-models called "plug-ins" – Willamette Envision incorporates a suite of sub-models to simulate processes that affect the distribution, movement, supply of, and demand for water in the WRB. Each sub-model was developed or adapted for the project by researchers with expertise in that field. Some sub-models are based on new empirical analyses, while others are process-based models originally developed for other purposes and adapted for this project. For example, the land use transitions model is an economic model developed for this project using historical parcel-level land value data, while the reservoir operations model was adapted from code in ResSim, a model developed by the US Army Corp of Engineers. The plug-in models within Willamette Envision represent both socioeconomic and biophysical components of the water system. Fig. 2 is a conceptual diagram showing those components of the regional human-natural system that have been developed into quantitatively modeled plug-ins, and linked together within Willamette Envision.

• Modeling frameworks that enable sub-models to run simultaneously and share data – The framework allows output from one modeling component to become input for others as they step through time. For example, hydrology is modeled within a modeling framework called FLOW that was developed for this project (Vaché et al., in review). FLOW models the movement of water through the stream network, while also allowing water to be added or removed at specific points where the model links to other plug-ins. This linkage allows economic processes such as the irrigation decision model, which determines the timing and amount of groundwater pumping and surface irrigation diversions, to be linked to the hydrologic model.

• A geographic information system (GIS) – Willamette Envision contains a spatial database that stores information about the WRB in map polygons called Integrated Decision Units (IDUs) and in a line network representing the river system. As the component models run, they store and retrieve data from this spatial database. At the end of a Willamette Envision model run, results are available as data, tables, and maps, which can be exported for further analysis.

• Alternative scenarios – model inputs and other model elements have been varied to create alternative scenarios with Willamette Envision. For example, we developed different scenarios that varied assumptions about fire suppression in the forest disturbance sub-model. That change then affected the distribution of forest types predicted by the forest state and transition sub-model, and consequently stream flows predicted by the hydrology sub-model. By comparing the reference case scenario with alternative scenarios, we can evaluate how sensitive the model is to particular assumptions.  We can also ask “What if?” questions that may help inform policy analysis. 

Technical Details

• Code – Willamette Envision consists of several hundred thousand lines of C++ code and runs as a 64-bit Windows application.

• Spatial representation – Data are stored in ESRI shapefiles representing the stream network and landscape polygons called Integrated Decision Units (IDUs). The polygons are meaningful spatial units, representing for example agricultural fields, forest stands, or developed areas.

  • The stream network and associated catchments were taken from the National Hydrography Dataset version 2 (NHD+V2) found at http://www.horizon-systems.com/NHDPlus/NHDPlusV2_home.php.

  • The IDU layer was developed by intersecting the catchment shapefile with a composite dataset representing landuse and landcover in the WRB. This catchment shapefile provides information needed by the hydrology model to connect the landscape to individual stream reaches. The land use/land cover dataset was developed by combining information from two sources: (1) a US Forest Service dataset (GNN) representing the forested portions of the basin and (2) as US Department of Agriculture datasets (NASS CDL) representing all other portions of the basin. The NASS data is derived from LANDSAT imagery and is designed to closely capture the crops grown as part of the agricultural system.

  • There are 164,892 polygons in the IDU layer, covering the nearly 30,000 square kilometer (11,500 square mile) extent of the WRB. Most IDUs cover an area of between 2 and 700 hectares (5 and 1730 acres).Climate data is gridded at a resolution of 2.5 arc-minutes (about 4 km).Refer to the stream layer and IDU metadata for more details about spatial geometry.

• Time step – Some processes are simulated at a daily timestep and others at an annual timestep.

  • The daily time step includes many processes such as:

    • stream flow
    • snow and canopy evaporation
    • infiltration and percolation
    • evapotranspiration
    • Irrigation decisions
    • Urban water use
    • exercise and enforcement of water rights
       
  • The annual time step includes many landscape change processes such as:
    • spatial distribution of population and income changes
    • expansion of urban growth boundaries
    • urban water price
    • crop choices
    • agricultural land value
    • forest wildfire
    • forest harvest
    • land use transition (for example from agricultural or forest uses to developed uses)
       
• Simulation period – Most WW2100 modeling scenarios simulate the period from January 1, 2010 through December 31, 2099. However, we also ran two scenarios with a simulation period from January 1, 1950 - December 31, 2009 to allow comparisons of modeled future conditions to a modeled past. These scenarios used simulated historical climate data as a model forcing and held population and land cover constant with conditions at the end of 2010. Refer to the scenarios page from more details.
• Run time – A single run of Willamette Envision takes about 48 wall clock hours. Up to six scenarios can be run at the same time without increasing elapsed time appreciably.

Methods

The Reference Case scenario addresses the first two of the four overarching objectives of the Willamette Water 2100 project: (1) to identify and quantify the linkages and feedbacks among human, hydrologic, and ecologic dimensions of the water system, and (2) to make projections about where and when human activities and climate change will impact future water scarcities. We then developed alternative scenarios to address the third and fourth objectives of the project: (3) to evaluate how biophysical and human system uncertainties affect these projections and (4) to evaluate how policy changes or other interventions might affect future water scarcities. This page provides details about the assumptions and model settings that are part of the Reference Case and alternative scenarios.

Reference Case Scenario

The Reference Case scenario modeled future conditions through the year 2099 and adopted mid-range assumptions about climate change, population and income growth; it also assumed that institutions such as water rights, the land-use planning system, reservoir operating rules, forest practices, and urban water pricing continue to operate in their present form, or with expectations for change that are most likely to occur (e.g., expansion of the urban growth boundaries). The lists below describe some of the specific data sources and model settings in the Reference Case scenario:

External Drivers

• Climate - Climate inputs from the regionally downscaled projections from the MIROC5 global climate model with the RCP 8.5 emissions scenario. Projections are in the middle of the range of possible changes predicted by a suite of global climate models that perform well for the Pacific Northwest. Annual mean temperature in the Willamette River Basin (WRB) increases ~4°C (~7.5°F) over the century. Climate input data was selected and downscaled through a multi-step process.

• Population - County population projections to 2050 from the Oregon Office of Economic Analysis (2011), linear extrapolation 2050-2100. County projections downscaled to cities and areas zoned for rural residential use in proportion to 2010 Census block population counts. Example population totals: WB as a whole in 2010 = 2.41M; 2050 = 3.73M; 2100 = 5.37M; within Portland UGB in 2010 = 1.43M ; 2050 = 2.20M; 2100 = 3.16M.

• Income - County level total household personal income forecasts to 2040 from Woods and Poole (2011); linear extrapolation 2040-2100. The annual projected growth rate in mean household income for WB counties 2012-2100 is 1.13%, a rate that is lower than the average historical growth rate for 1930-2013 of 1.63%. Mean household income in 2010 is $87,900 and reaches $242,000 (in 2005 dollars) in 2100. Income at the city level is assumed to be the same as at the county level. Note that the Woods and Poole measure of income is more inclusive of sources and types of income than the definition of “money income” used by the US Census Bureau and other government agencies; it includes all money income plus “exclusions to income” such as food stamps, agricultural payments-in-kind, and imputed rental value of owner-occupied housing. It also includes certain interest payments, other labor income, and a number of other measures that are not included as part of money income. The mean value for household income that we are using will typically be 30% or 40% higher than the median value.

Economic Policy and Public Management Assumptions

• Forest management - The level of wildfire suppression is held at historical rates throughout the simulation. But because of changing climate and forest conditions, the area of forest burned per year rises over the simulation from 0.2% per year in 2010 to 0.6% per year in 2100. Harvest by clear cut is maintained at historical rates (8,000 acres per year on public lands + 29,000 acres per year on private lands). There is no harvest of protected areas. Harvest only occurs on stands older than 40 years on private lands, and between 40-80 years on public lands.

• Urban development - All future development occurs within Urban Growth Boundaries (UGBs); UGBs expand when 80% developed (72% for Eugene-Springfield). No development of parks, wetlands, or other protected areas. Land zoned for Exclusive Farm Use is added to UGBs when all other contiguous areas are exhausted. Growth of the Portland Metro UGB confined to urban reserves through 2060.

• Reservoirs operations - Operations of the 13 federal reservoirs that are part of the Willamette Project modeled according to rule curves implemented as of 2011 and recommendations from the Biological Opinion as of 2009 except the selective withdrawal structure at Cougar Dam. Reservoirs begin refilling Feb 1, with a target to fill reservoirs by May. Smaller, non-federal reservoirs are not modeled. Pulse flows in sub-basins are not implemented.

• Urban water demand and pricing - Water demand estimated for UGBs as a whole based on residential and commercial income shares, water price, population, income, housing density, and season. Bull Run (water source for Portland Metro) is modeled according to the municipal water right, which limits diversions to 636 cfs, and in the model is allowed to contribute not more than two-thirds of the Portland Metro municipal demand. Baseline information on water rates and price structure obtained from Water Management and Conservation Plans for Portland (2010), Salem (2009), Corvallis (2005), and Eugene (2012), and through personal communication for Springfield (2012). Example baseline residential prices: Portland UGB= $2.44 per ccf; Eugene-Springfield UGB= $1.25 per ccf; rural residential areas= $0.3 per ccf (reflects pumping cost). Reference Case scenario assumes an increase in water price 2011-2015 (6% per year) and 2016-2025 (1.5% per year; in real, inflation-adjusted dollars); actual and anticipated cost increases to cover infrastructure backlogs, seismic upgrades, etc., then prices held constant in inflation-adjusted terms for a given city population size; resulting average per capita municipal use = 95 gal per day in 2100.

• Crop choice and irrigation - Crop types are set annually based on land characteristics, water rights and irrigation decision, crop prices, and climate. Crop types are limited to hay, grass seed, corn, pasture, wheat, clover, fallow, and other crops. Lands in orchards, vineyards and tree farms remain in those land cover types and do not change throughout the simulation and are largely not irrigated. On lands with water rights, the decision to irrigate is made annually based on land characteristics, climate (including June precipitation), and energy price. The price of wheat ($64 per ton), grass seed ($5 per bushel), and energy are held constant (in real terms). Irrigation diversions are limited by water right and constrained to legal limits including a max irrigation rate 1/80th cfs per acre and duty 2.5 acre-feet per acre. About two thirds of acres with water rights are irrigated in an average year; the result is that about 280,000 acres are irrigated at the simulation start. Crop and irrigation models are interdependent. Some crops are always irrigated and some crops are never irrigated.

• Water rights - Modeling includes surface and groundwater rights for irrigation, municipal, and instream uses as documented in the Oregon Water Resources Department Water Rights Information System as of July 2012. Priority is determined by seniority among all rights within the Willamette Basin. For municipal and irrigation rights to surface water, when requested water cannot be allocated due to insufficient water in the stream, junior water rights in upstream reaches may be regulated off for the rest of the season. Tallies are kept by modeling polygon of the frequency of water shortages lasting at least seven consecutive days. The Reference Case scenario assumes that groundwater wells tap an unlimited source. No new water rights or contracts for stored water from the Willamette Project are added over the length of the simulation. The scenario includes instream water rights implemented as of 2010. When more than one instream water right applies at the same time to the same reach, the water rights model applies both water requirements.

Alternative Scenarios

We evaluated 21 alternative scenarios in this project. Fig. 1 depicts the Reference and the 21 alternative scenarios grouped by scenario element and purpose. The Reference Case (shaded grey) modeled future conditions (2010-2099) and set the baseline assumptions for each scenario element. Eighteen scenarios (shaded orange, red, blue, and purple) then varied a single element or assumption, and held all others the same as in the Reference Case scenario. For example, the LateRefill scenario explored the effect of making a specific change in federal reservoir operating rules while all other biophysical and human systems operated as in the Reference Case scenario. Table 1 details specific assumptions of the Reference Case scenario and identifies how those assumptions were changed in the single variable alternative scenarios.

The color coding in Fig. 1 further identifies the alternative scenarios and their purposes. Tables 1 and 2 list specific assumptions and model settings for each of these scenarios. Orange shading indicates the two scenarios that modeled a 60-year time span from the past (1950-2009). These “simulated historical” scenarios were used to compare the simulated future to the model’s own simulated past. They each used simulated historical climate data that was generated by the same global climate models used to develop future climate conditions for the other scenarios.

Blue shading indicates the future climate scenarios. These scenarios include different assumptions about future climate, but left all other assumptions the same as in the Reference Case scenario. The purpose of the future climate scenarios was to compare the range of possible outcomes, given uncertainty about future climate conditions. Along with the mid-range assumptions of the Reference Case, the HighClim and LowClim scenarios represent the range of possible future climates projected by a suite of global climate models determined to perform well for the Pacific Northwest (Rupp et. al., 2013).

Red shading in Fig. 1 identifies the human dimensions scenarios. These each vary one assumption about the human system, and most explore the effect of specific policy interventions such as changes in water prices, water law, or reservoir management rules. These scenarios explore the direction and magnitudes of change in relation to specific management interventions, so as to be able to differentiate between associated outcome sets, their costs and consequences. One advantage of running these policy scenarios with a single change or intervention relative to the Reference Case scenario is that it allowed us to attribute changes in outcomes to individual changes in policy.

Purple shading identifies the “counterfactual” scenarios. The term “counterfactual” means intentionally modeling a situation that will not occur – i.e., counter to the facts. These scenarios help us measure the impact of changes in the model by comparing the Reference Case scenario with a scenario that omits one source of change. For example, we have included one scenario of this kind where the climate does not change (but the population grows), and also one scenario where there is no change in population or personal income (but where climate change does occur). Comparing these counterfactual scenarios to the Reference Case scenario provides insights into the magnitude of the effects on future water scarcity due to one of these elements versus the other. In one other case we have modeled a scenario with no agricultural production (all fallow). This is clearly counter to the facts; comparison of this scenario to the Reference Case or other alternative scenarios provides evidence of the impact of agriculture on water use and stream flows.

The last three columns shown in Fig. 1 (red and green shading) show scenarios that vary multiple elements relative to the Reference Case scenario. Their purpose is to explore the effect of multiple, simultaneous, and plausible changes to the modeled social-ecological system. We developed two of these scenarios in collaboration with our Technical Advisory Group (TAG), a group with diverse expertise in Willamette Basin land and water use and management. The TAG developed two overarching themes for their scenarios, and with input from the modeling team, selected model settings consistent with each theme (Table 2). The Extreme scenario combined High Climate Change with high population growth and other model settings designed to maximize resource use and water demand for cities and agriculture. It also included a mechanism to take individual federal reservoirs “offline” occasionally for maintenance. The Managed scenario is similar to the Reference Case scenario in that it adopted the mid-range assumptions about climate change and population growth. However it also included several management choices that the TAG felt reflected recent trends in resource management. These included differential rates of fire suppression on private and public forest lands, a per-capita constraint on future municipal water use, and the establishment of new contracts for future use of water stored in federal reservoirs. Assumptions in the Worst Case scenario were selected to maximize water scarcity. Scenario elements include mid-range climate warming and high population growth, combined with increased fire suppression and high utilization of irrigation and instream flow rights.

Table 1. Detailed assumptions for the 18 single element alternative scenarios that were designed to isolate the influence of individual model settings or policy choices. Each varies one scenario element and holds all others the same as in the Reference Case scenario.

Table 2. Assumptions for the three Multiple Variable Alternative Scenarios that each differ from Reference Case scenario in multiple ways. The assumptions for the Extreme and Managed scenarios were developed in collaboration with the WW2100 Technical Advisory Group (TAG), a group with diverse expertise in Willamette Basin land and water use and management.

¹ See reports including Hutchison, James M., Kenneth E. Thompson, and John D. Fortune Jr. The fish and wildlife resources of the upper Willamette basin, Oregon, and their water requirements. Basin Investigations Section, Oregon State Game Commission, 1966; Hutchison, James M., and Warren W. Aney. The fish and wildlife resources of the Lower Willamette Basin, Oregon, and their water use requirements. Oregon State Game Commission, 1964; The fish and wildlife resources of the middle Willamette basin, Oregon, and their water use requirements. Report to the Oregon State Game Commission, Basin Investigations Section, 1963.