How useful are seasonal forecasts for farmers facing drought? A user-based modelling approach

Abstract

Seasonal forecasts of water availability have clear potential benefit for decisions in irrigated agriculture. This potential depends in part on how accurate the information provided is. The actual benefit, however, depends on how the information is used in the decisions, by whom, and the outcome of those decisions. In this paper we assess how useful seasonal forecasts are in supporting drought management decisions by farmers at the irrigation district level. We model the decisions irrigated farmers make on what and when to plant in the Ebro basin (Spain), and the interconnected decisions reservoir operators make on whether to apply curtailments to the water allocated to farmers. The modelled farmers are supplied from a reservoir with capacity for a single irrigation season and therefore their decisions are conditioned by the expected water availability through to the end of the season. Different farmer behaviours are considered as a function of their risk averseness and their technical capacity. The value of seasonal streamflow forecasts to inform these decisions is compared against that of current practice using extrapolated historical records, as well as against a reference forecast based on climatology. Results show that seasonal forecasts of water availability have skill, albeit limited. How salient information is to the decisions that farmers make, however, differs for each type of farmer as they take key decisions at different points in the season. As a consequence, seasonal forecast information is found to not serve the various farmer types considered equally. Our results illustrate how assessing the usefulness of information to servicing a decision can be approached from a combined technical and user-centric perspective.