Pressure on freshwater resources has grown exponentially over the past century, with water use growing more rapidly than population. Water from streams, lakes, reservoirs, and aquifers is withdrawn in large quantities to meet the needs of agriculture and energy production. To meet food, energy, and water security goals in the future, managers will have to understand complex interactions in a system that involves both nature (for example, variability in weather and land conditions) and humans (for example, individual choices of farmers and decisions made by government agencies). As a small island nation highly dependent on irrigated agriculture and hydropower, Sri Lanka is an ideal location for research to address key questions about agricultural adaptation and provide insights for other places in the world. How do changing rainfall and temperature patterns, land use changes, and water allocation decisions affect water availability and rice yields? How are farmers responding to water scarcity? How do water allocation decisions reflect the priorities of decision makers in balancing water for irrigation and water for electricity generation?

The area in Sri Lanka known as the dry zone is the major region for production of food under irrigation. Analyses of rainfall and temperature data from 1880 to the present revealed that the critical planting months of March and September have been getting drier in the past few decades. Rice production was found to be correlated with calculated indices of drought leading to a hypothesis that climate forecasts could prove valuable to farmers and officials who decide on water allocations. Depending on how much water farmers expect to be released to them for irrigation, they can elect to plant rice which, with adequate water, provides a stable source of food and income or they can elect to plant an alternate crop, such as onions, which requires less water but must be sold in a market with highly variable prices. Farmers decisions reflect an assessment of expected climate conditions over an upcoming season and of their ability to deal with the risk of crop failure. Extensive surveys of farmers and interviews with key individuals conducted by the research team showed that decisions made by farmers are conditioned on their belief that they can adapt and on their perceptions of risk associated with different planting decisions. Informed by field research, a computer simulation model was constructed to evaluate the impact of provision of seasonal climate forecasts. Simulations show that a farmer using seasonal forecasts will have increased average agricultural income relative to farmers who do not make use of forecasts but at the expense of lower production of rice which is the staple food crop of the country. 

The decisions of individual farmers depend in part on higher level decisions about overall water diversions. In times when water is scarce, there is inadequate supply to fully satisfy both irrigation and hydropower demands. An analysis of ten years of data on river flows, paddy production, and hydropower output indicates that throughout dry spells water managers favor paddy production, even though economically rice is less valuable than power. When water is scarce, decisions are not driven by money; instead, social values influence Sri Lankans to support farmers. Future planning for electricity production must account for such preferences while incorporating an understanding of environmental and social implications of choices. A decision support analysis was conducted that included economic, environmental, technical and social objectives of various agencies. The results of the analysis demonstrate how different aspects are valued differently across agencies. Knowledge about the various valuations should facilitate the process of reaching consensus about the preferred mix of hydropower, thermoelectric power, and power from renewables in the future.

Throughout the project the researchers interacted with a Stakeholder Advisory Committee in Sri Lanka to ensure that results of the research were communicated effectively. Computer codes for drought calculation and decision support to inform farmers about using forecasts were developed and disseminated. Data and software programs were made available as publications appeared. A web-based data explorer tool was developed to make results from surveys easily accessible. The project supported training at multiple levels. Five post-doctoral scholars, six PhD students, and three MS students were supported at Vanderbilt University, the University of Colorado, and the University of North Florida.

Last Modified: 10/30/2017
Modified by: George M Hornberger