Sensor networks, built on the backs of the latest digital communication technologies, are increasingly being deployed in urban sewer networks and at regional scales to monitor flooding and water quality of rivers (Habibi et al. 2017; Mullapudi et al. 2017; Jones et al. 2018; Yildirim and Demir 2019). Concurrently, control technologies are being deployed alongside sensors which allow for water resources operators to actively manipulate these systems in places and in a manner that was previously inconceivable (Kerkez et al. 2016). Together, sensing and control deployments mean massively more complex systems to address persistent water resources challenges such as flooding and water quality. Recent research studies demonstrate the potential for coordinated and increasingly automated management and control approaches to improve performance considering both water quality and quantity objectives (Muschalla et al. 2014; Sadler et al. 2019; Sharior et al. 2019; Mullapudi et al. 2020; Sun et al. 2020; Troutman et al. 2020).

However, decision-making methodology based solely on underlying physical and technical characteristics ignore the social and political structures which overlay and interact with them. Considering these structures as part of a singular, sociotechnical whole exposes normative questions of “right” and “wrong,” or more profound questions of societal ethics and values. Take for example, if some flooding is an inevitable outcome within a catchment with dynamic control capabilities, how ought a controller act to consider the societal implications of such flooding? Or, if a control scheme consistently recommends distributing harm in poor neighborhoods and benefits in richer neighborhoods, ought its directives be followed? To address this gap in knowledge, we previously developed a framework to construct data-driven ethical preference models in relation to water resources issues, such as flooding, using a combination of voting-based practices and machine learning methods (Ewing and Demir, 2020). Ethical preference models captured group “wisdom” and demonstrate that the framework is a candidate for use in decision support toolchains to support water resources decisions.

In this study, we propose to explore a novel methodology to incorporate concepts of ethics and justice into decision support toolchains for water quantity and water quality objectives by building on the preliminary work from our research lab (Ewing and Demir, 2020). To do so we will utilize a modeling framework, the Python programming module pystorms, which allows hydrologic and hydraulic simulation of water conveyance networks with dynamic control of storage assets (Mullapudi and Troutman 2020). Our extension will allow dynamic operation of storage assets while considering social and geographic data – such as census data of social vulnerability, land use type, and landowner type (i.e. public or private) – and voting-based ethical preference models. Through this experimental setup we will investigate how different dynamic control strategies affect the distribution of benefits and harms across communities and landscapes.

The findings from this research will provide greater insight into how to manage our next generation sociotechnical systems such that they preempt injustices, inequities, and inefficiencies. The findings will also inform how novel decision support toolchains for dynamic control may integrate into our water resources management in both catastrophic and quotidian management scenarios.