Agent-based Model: We will use agent-based modeling (ABM) to link physical and social systems

for a broader view of how uptake, assimilation and exchange of information from scientists to water

managers, smallholder farmers and other stakeholders may occur, and to identify patterns and barriers to

adoption of forecast information. In the ABM framework, we define farmers and water managers

throughout the target communities as agents having divergent behaviors characterized by physical

situations (e.g., water stress or sufficiency), financial conditions, technology (e.g., irrigation systems), risk

tolerance, and habits (e.g., response to forecast, tolerance to soil dryness.) This iterative and adaptable

method allows for understanding the potential long-term effects of providing forecast information,

specifically how the use or non-use of the information by stakeholders can modulate agriculture

productivity, hydropower output, etc. It also enables us to generate new, testable hypotheses that can be

implemented to improve adoption of the forecasting tool advice in an iterative fashion over the course of

the project. Given its easy user interface and intuitive nature, we will use the ABM as a collaborative

outreach and educational tool with groups of stakeholders and students to introduce the probabilistic

nature of forecast information. ABM has been shown to be an effective tool for studying coupled

human/engineered/environmental complex systems [124], including applications in Ethiopia to investigate

conflict between herder groups over the utilization of common grazing land and water resources [125].

The ABM will be informed largely through the ethnographic studies (years 3-4), specifically how and

from whom agents (stakeholders) receive information, the likelihood of acceptance and transmission to

other agents, and factors influencing these processes. Agent classes will include scientists, national and

Kebele government officials, non-governmental organization (NGO) employees, water managers, local

community leadership and smallholder farmers. Factors influencing the open or closed nature of the community will be explicitly modeled. 

ethnographic outcomes, we will use a collaborative process with stakeholders to verify its accuracy. The

ABM will help us identify barriers to information uptake and exchange, rank each factor’s importance, see

how forecast-based advice can influence outcomes and enable us to generate testable hypotheses.

During the 4th and 5th year of the project, these hypotheses will be tested to refine the model and improve

uptake of the advice. This iterative approach will enable us to verify and validate the model. The fully

validated model will be used to test the ability of the forecasting tools to increase the resiliency of

Ethiopian communities to water stresses.