Exploring Human Water Use and Water Allocation Algorithms in Hydrologic Models

Abstract:

Humans instigate water management and so representation of their actions and decisions are integral to hydrologic models. The challenge is that the representation of human decisions for water management is largely unconstrained and a potential source of substantial uncertainty in hydrologic models. Commonly used representations are based on hypotheses of the drivers that motivate water use. However, there are few studies that investigate the effects of such assumptions and modeler choices on the resulting prediction error. In fact, no framework exists for the systematic sampling of a set of alternative representational choices for human water use and infrastructure decisions. This study proposes to apply a combinatorial framework, such as the “method of multiple working hypotheses” to systematically explore the uncertainty due to alternative assumptions of decision making in hydrologic models. To do this, we sample the literature for representative hypotheses of water management decisions by water users and water managers. We implement these decision making models in an integrated model - HBV for hydrology and the PyWR model for infrastructure and water use decision making. Our case study is the Rufiji river basin in Central Tanzania. We systematically combine alternative choices within these models, record their predictions of water withdrawals, reservoir releases, and streamflow. We outline the range of predictions given different conceptualizations of human decision making in hydrologic models. Potential results include a minimum range of uncertainty given alternative model structures.