PhD Final Defense – Anav Vora

Modeling Human Interventions to Understand Streamflow Changes and Water Management

Implications in River Basins

Advisor: Professor Ximing Cai

Abstract

Hydrological models support improved understanding of hydrological processes, forecasts of extreme

events, water availability assessments, and evaluations of climate and land use change impacts. The

credibility of these outputs greatly depends on how accurately hydrological models represent human

water management. Despite substantial efforts on improving human representations, data limitations have

required simplifying assumptions, which introduces biases and obscures model-based inferences. This

dissertation improves the representation and evaluation of human water management in hydrological

models to understand human effects on streamflow and water availability at the river basin scale.

We start with streamflow regulation by reservoirs and examine if more realistic reservoir representation in

a watershed hydrologic model can generate more accurate streamflow simulations. Five reservoir

representations are coupled with a conceptual rainfall-runoff model to evaluate watershed outflow,

internal fluxes, and model parameters. The results show two representations derived from long-term

reservoir operation records improve not only the watershed outflow but also the internal flux (reservoir

inflow) across five watersheds. The identified reservoir representations are included in a national-scale

data inventory of human interventions in hydrological systems (Data_IH2) that we compile to conduct

basin-scale impact analysis of multiple human activities. Data_IH2 encompasses reservoir operations,

irrigation, drainage, stormwater management, inter-basin transfers, and water supplies. We also develop a

diagnostic framework to evaluate the utility of Data_IH2 in large-scale hydrological models without

calibration and parameterization biases. Applying Data_IH2 and the diagnostic framework to the

Mississippi River Basin shows that reservoir operations and irrigation jointly and substantially affect

streamflow. The diagnostic framework also identifies key data gaps and priorities for model improvement.

We use the Data_IH2-based Mississippi River Basin model to examine how upstream withdrawals,

consumption, and return flows affect downstream flow and water availability. By extending a human

water use accounting framework to explicitly consider routed return flows, we find that 7% of locations

receiving routed streamflow rely on upstream return flows to meet local withdrawals annually; the share

reaches its highest of 10.9% in August.

Overall, this dissertation addresses the human dimension of a hydrologic system based on most recent

data availability of various human interventions. Specifically, it provides procedures to evaluate systemlevel

representativeness of coupled models with human and natural processes, a national-scale data set

and diagnostic framework for modeling human interventions, and a water use accounting framework that

links upstream water management to downstream water availability. Moreover, the dissertation calls for

attention on data gaps and challenges in formulation and calibration of large-scale hydrologic models,

recognized as coupled nature-human systems.