U.S. Department of Energy

Pacific Northwest National Laboratory

Agent-Based Energy-Water Model

Schematic of water demands on the San Juan River system. Competition over water includes Indian rights (Navajo Indian Irrigation Project), non-Indian irrigation along the river, City of Farmington plus several other smaller municipalities, and two large coal-fired power plants. Navajo reservoir is operated to deliver irrigation water and to ensure instream flows of 500 cfs along the full length of the river which terminates at Lake Powell in Utah. Several new water projects are under construction to meet mandated Indian water rights delivery.

River basin management is a challenging issue due to heterogeneous water users located in the watershed that often have different objectives for their water use purposes and different decision-making time frames. Realistic watershed management often uses a bottom-up decision making process driven largely by stakeholders’ concerns. An Agent-Based Modeling (ABM) framework is suitable for simulating such a bottom-up process to handle the integration and coordination of watershed activities (Yang et al. 2009; Yang et al. 2012; Berglund 2015). IM3 is working to dynamically connect an ABM with a water system model (Riverware [RW]) to simulate the allocation of water resources for various water users such as domestic and agricultural users while also considering the downstream flow requirement. The San Juan River basin in New Mexico is used as a case study area. An RW Model was developed for the entire San Juan River Basin. Three major tributaries (the Navajo, Animas, and La Plata Rivers) are considered in this model. The transboundary water diversion from the San Juan River Basin to Albuquerque, NM, which is part of the Rio Grande River Basin, is also incorporated in this RW model. Eleven major cities, nineteen major irrigation districts, and three major reservoirs are considered in the current RW modeling structure. The streamflow at the basin outlet is regulated by the Colorado River Compact and the San Juan River needs to provide a certain amount of water (500 cfs per day in the current modeling structure) to the downstream area. A watershed-scale ABM is developed and connected with the RW model to represent the objectives and behaviors of specific water users (and groups of users) in the San Juan basin. Agents, in this context, are entities that make decisions such as reservoir releases, crop choices, and the amount/timing of irrigation (e.g., an individual farmer or a local water management authority). Nineteen major irrigated districts are defined as agricultural agents in the current modeling structure and modeling the Navajo Reservoir is also an agent is under progress. The agricultural agents in the ABM adjust their irrigated area year-by-year based on the information of basin outflow and their own behavioral rules. Data Management Interface" inside RW facilitates this two-way coupling between RW and ABM allows information transfer at any predefined frequency. Example results of agents' irrigated area changes (acres in y-axis) throughout the entire simulation period (85 years in x-axis) are given in the following figure.

Point of Contact: Ethan Yang

>> Return to Modeling Capabilities


Berglund EZ. 2015. "Using Agent-Based Modeling for Water Resources Planning and Management." Journal of Water Resources Planning and Management. 141(11), DOI:10.1061/(ASCE)WR.1943-5452.0000544.

Yang Y-CE, X Cai, and DM Stipanović. 2009. “A Decentralized Optimization Algorithm for Multiagent System-Based Watershed Management: MAS-Based Watershed Management.” Water Resources Research. 45(8). DOI:10.1029/2008WR007634.

Yang Y-CE, J Zhao, and X Cai. 2012. “Decentralized Optimization Method for Water Allocation Management in the Yellow River Basin.” Journal of Water Resources Planning and Management. 138(4). 313–325. DOI:10.1061/(ASCE)WR.1943-5452.0000199.

| Pacific Northwest National Laboratory