Central Arizona Project - Service Area Model

This is an article based on a poster presented at the GoldSim User Conference.

The Central Arizona Project (CAP) manages and delivers more than 1.5 million acre-feet of Colorado River water to the most arid and populated regions of Central and Southern Arizona. As shortages on the Colorado River grow more likely, there is an increasing need for greater clarity on the scale of water challenges faced by CAP and its customers, and how those challenges can be met in an affordable and sustainable manner. To help address future shortages and other various challenges and planning uncertainties, CAP’s Resource Planning & Analysis (RPA) department and has been engaged in a systematic and coordinated effort to expand its planning and analytical capacity through the development of simulation models, geographic information systems, and databases. These tools are integrated to form an in-house decision support tool for exploratory water resource management analysis.


The Central Arizona Project Service Area Model (CAP:SAM) was developed by in-house staff using GoldSim software to simulate water demands for all major water using entities in the CAP three county Service Area, and the water supplies they use to meet their demands. The model can be used to simulate a wide range of future conditions including variable rates and patterns of growth, shortage impacts, effluent reuse, aquifer recharge and recovery, and complex supply portfolio management decisions on an annual time step.

Figure 2 - Screen Capture of GoldSim Model

Model Architecture

CAP:SAM performs a large number of interrelated calculations using array and matrix sets, broadly organized into four conceptually simple steps performed for water users classified into major categories of municipal, agricultural, industrial, tribal, and institutional. CAP:SAM was developed using GoldSim software, which is a graphical, object-based modeling environment that allows complex mathematical relationships to be defined and calculations to be performed that would be difficult or impractical to implement using spreadsheets or traditional computer code.

CAP:SAM is designed to accommodate a large range of future scenarios related to water demand and supply reliability in order to identify and characterize risks and uncertainty faced by CAP. The CAP:SAM dashboard (part of which is shown below) is a user friendly interface for planners to build scenarios as well as to make adjustments to data elements in the model, such as water entitlement and preferences for use of supplies. The use of dashboards allows for a more transparent, accessible approach and helps facilitate interactions among stakeholders.

Rather than only using CAP:SAM to describe single sets of best-prediction scenarios, the model uses GoldSim’s Monte Carlo simulation capabilities to run hundreds to thousands of different sets of assumptions and simulate a large range of plausible futures. This approach can help CAP identify, evaluate, and choose strategies that perform well over a wide range of likely futures and better manage surprises. The graphic below illustrates the ranges of possible total shortage volumes to Arizona from the Colorado River Simulation System (CRSS) “Protect 1,000 ft.” run.

Visualization of CAP:SAM model runs can assist planners in analyzing results and facilitates conversations and understanding among stakeholders and decision makers. CAP:SAM runs are automatically exported into spreadsheets that create visualization of results that provide context by incorporating historical data. In the example below, future CAP water use is broken into categories of “destination” with single shortages projection included on the Colorado River.

Colorado River Shortage Probabilities Plot

Data Sources

CAP:SAM relies heavily on data managed in a formally structured relational database that includes information that originates from several departments within CAP, as well as data from AWBA, and ADWR. The CAP data includes historical and scheduled water deliveries, customers, rates, contract entitlements, water priorities, recharge activity, and basic information on operations and pumping plants. There are also numerous tables related to municipal water providers and agricultural water users, including water use by year and type, population and housing units, and data related to Designations of Assured Water Supply.

Historic and Future Water Supply Plot

Modeling Process

During a simulation, the GoldSim model calculates projected water demands based on user-defined scenario data and determines the supplies available under legal and physical constraints. Requests are made based on established water entitlements in order to fulfil the demands using these supplies. This process is further outlined as follows:

Project Demands

To simulate municipal water demand, CAP:SAM produces individual projections for 80 public and private water utilities accounting for more than 99% of the demand in the municipal sector, 23 Agricultural Irrigation Districts and other Grandfathered Irrigation Rights, 12 Tribes and Tribal Districts, and over 20 other user categories including the Central Arizona Groundwater Replenishment District (CAGRD), Arizona Water Banking Authority (AWBA), and industrial users such as mines and power plants. Municipal demand is modeled using current and projected housing units. Each housing unit is multiplied by a provider-specific factor of gallons per housing unit per day (GPHUD) that represent the demand from the housing unit itself as well as a fraction of the ancillary demands (e.g., new parks, commercial land uses, etc.). To help differentiate the effects of observed long-term declines in water use from future growth-related trends, the model separately considers existing and new municipal demand. For agricultural water use, the model projects water use on the basis of crop type and irrigated area.

Determine Supplies

The CAP:SAM model tracks the total legal and physical supply availability for 16 Supply Types in each projection year. Supply categories include Effluent, Surface Water, including the Salt River Project, and CAP water. The supply of Long Term Storage Credits (LTSCs), and Groundwater Allowances are tracked and debits and credits occur through time. To model the CAP supply from the Colorado River, CAP:SAM allows the user to input an annual diversion supply for Arizona, demands from on river users, total system losses, and the net storage to CAP’s storage reservoir. CAP:SAM also utilizes externally generated Colorado River supply scenarios from the Colorado River Simulation System (CRSS). CAP:SAM contains multiple scenarios, with each scenario containing 105 future realizations of Colorado River supplies.

Request Supplies

Requests represent the maximum preferred use of a particular supply that is legally available to an entity without consideration of the demands of that entity. The basic approach is to load in an array of the provider’s maximum entitlement for each supply type. Entitlement volumes can then be individually adjusted either by percentage, or by setting a limit to represent specific preferences or operational limitations. Requests for CAP supplies are disaggregated into treated water and groundwater delivered through annual storage and recovery (ASR), as well of preferences for earning underground Long Term Storage Credits (LTSC) with the remaining CAP entitlement. The request portion also includes ranked preference for 36 Storage Facilities that include deliveries to irrigation districts as in-lieu groundwater storage, or to spreading basins. Entitlements can also be modified through time based on transfers, leases, exchanges, reallocations, and priority conversions.

Fulfill Demands

In the final model step information from each of the other steps is brought together and reconciled. The model steps through each supply type in a defined sequence, incrementally satisfying the demand of each entity based on their request and their volume of unsatisfied demand. For each entity, demand fulfillment with a particular supply is based on the lessor of either the demand not yet met with other supplies, or the request (if any) for that particular supply. If the total supply available is not sufficient to meet the sum of the requests, a proportional reduction is applied. At the end of each fulfillment step, the demands from the start of the step are recalculated by subtracting the amount that was fulfilled. Those recalculated (unfulfilled) demands are then passed to the next fulfillment step.

Spatial Analysis

CAP:SAM is able to simulate spatial information using sets of matrixes defining relationships between array labels. The model’s major provider categories include 80 Municipal Providers, 23 tribes and tribal districts, and 12 Agricultural Irrigation Districts. Municipal water provider areas were defined by CAP staff on the basis of a provider’s current service territory, as well as Municipal Planning Areas, Certificates of Convenience and Necessity, and incorporated boundaries. Tribal and Irrigation District layers are provided by the Arizona Department of Water Resources.

Spatial Housing Unit Distribution

CAP:SAM models new municipal water demand as a function of projected housing units — measured as Gallons Per Housing Units per Day (GPHUD) — which provide a stronger relationship with demand than does population growth. The housing unit projections originate from Maricopa Association of Governments (MAG), Central Arizona Governments (CAG), and Pima Association of Governments (PAG). Growth projections were supplied at a level of geographic detail — Transportation Analysis Zone (“TAZ”) — that was sufficient to generate provider-specific projections by overlaying a GIS layer of each water provider’s projected service area. Knowing the projected location of future development is critical for determining which water provider will serve that development, along with the water use characteristics and water supply portfolio of that provider.

Alternative Growth Projections

Alternative housing unit projections in CAP:SAM are provided by the Sun Corridor Socioeconomic Allocation Model (SCSAM) developed by Applied Economics. SCSAM allocates population, housing and employment growth for 46 Study Areas within the three-county CAP Service Area. The model combines established land use analysis methodology with the use of linear regression techniques to relate historical growth patterns with the variables that may influence growth.

A wide range of growth scenarios were developed for CAP:SAM with adjustments to eliminate several of the speculative and largely unsubstantiated development projects; by controlling the importance of travel time as indicator of the location of development near the urban core; and through carrying capacity adjustments based on an assessment of potential for redevelopment.

Agricultural Data

Agricultural crop acreage estimates are developed using the Cropland Data Layer (CDL), a crop-specific land cover GIS data layer from the US Department of Agriculture’s National Agricultural Statistical Service. The CDL is derived annually from satellite image observations at 30-meter (0.22 acres per pixel) resolution and extensive agricultural ground surveys. Past data layers are available going back to 2008. Combined with crop consumptive use estimates and historical agricultural water use data from the Arizona Department of Water Resources, CAP:SAM is able to develop historical and projected by crop type. Spatial Housing Unit scenarios are also used to project urbanization of agricultural land.

Future of CAP:SAM

CAP:SAM capabilities and results will continue to be explored in a larger project, which is referred to as the "Service Area Analysis". The goals of this project are to: 1) Broaden and deepen understanding of CAP water supply and demand issues; 2) Explore the impacts of big picture phenomena on the Service Area; 3) Examine and test potential solutions to water challenges; 4) Provide a tool to help inform water management decisions.


Download Summary Posters of CAP Water Demand and Supply:


Making Better Decisions In An Uncertain World