Projecting hydrologic change under land use and climate scenarios in an agricultural watershed using agent-based modeling

Watershed systems are changing due to human activities within the landscape and shifting precipitation patterns, and quantifying the coupled effects of these two factors is necessary for anticipating future hydrologic response. In this study, we use a model that combines an agent-based model (ABM) with a semi-distributed hydrologic model to assess how projected changes in precipitation and temperature affect streamflow when simultaneously considering how those variables impact the land use decisions that also influence watershed response. We use a flood-prone watershed which is characteristic of many agriculturally-dominated watersheds in the central Midwest US. In the ABM, farmer agents make decisions that affect land use based on factors related to profits, past land use, neighbor influence, and internal behavior. A city agent aims to reduce urban flooding by paying farmer agents a subsidy for allocating land to conservation practices that reduce runoff. We run the model for the 2018-2097 period using the Representative Concentration Pathway (RCP) 4.5 and RCP 8.5 climate scenarios and different decision-making options. The model reveals that under future precipitation, which becomes increasingly intense leading to more in-field flooding, farmers increase conservation land by up to 60%. This land conversion results in a 4-7% decrease in mean 95th percentile discharge relative to scenarios where conservation land is held constant at the historical mean. If farmers are allowed to modify their internal behavior and preferences, the mean 95th percentile discharge decreases further (up to 16%). Using the assumption that land owners are willing to adapt their personal preferences to mitigate the negative effects of climate change on their land and external incentives exist to do so, upstream runoff mitigation practices could reduce downstream impacts from more frequent intense precipitation. However, farmer agents converted, on average, <10% of their land to conservation even when this variable was unrestricted. By the end of the century, precipitation has the dominant influence on discharge, given the significant changes in projected precipitation and the limited land conversion.