Sensing and Cyberinfrastructure for Smarter Water Management: The Promise and Challenge of Ubiquity

The pace of environmental data collection is increasing. As sensors become cheaper, larger numbers of land-based and remote sensors are being deployed in networks dedicated to collecting high-quality environmental data for both operational and scientific purposes—a trend accelerated by advances in sensing technology and data-intensive science (National Research Council 2012). Beyond these growing dedicated environmental measurement networks, data collected by the small, low-cost sensors that are embedded in everyday consumer products, such as mobile phones, have recently begun to be considered for environmental monitoring (National Research Council 2011). In combination, these dedicated scientific-quality sensors and embedded pervasive sensors present an opportunity for environmental sensors to exist nearly everywhere—in other words, these sensors present an opportunity for ubiquitous environmental sensing.

As illustrated in Fig. 1, water data can possess three characteristic V’s that typify big data: volume, velocity, and variety (Madden 2012). Thus, the explosion of new data production expected from the adoption of ubiquitous sensing will usher in a new era of big water data that has the potential to transform the practice of water resources planning and management.

First, these data will enable scientific inquiry into the behaviors of very large-scale integrated water resource systems, allowing a number of existing knowledge gaps to be closed:

Through investigating questions such as these, we will gain a better understanding of interdependencies between management decisions and water resource system responses. Presently, however, spatiotemporal variability in environmental and social processes challenges the application of current technologies to some of the world’s largest cities—a challenge that could be met by both increasing our understanding of managed complex behaviors of the water resource systems and by increasing the available real-time data describing the current state of the system.

Second, this vast amount of data will also facilitate the development of adaptive water management systems and deployment of an intelligent water resource infrastructure. Many cities (e.g., Singapore; Panama City, Florida; and Caceres, Spain) are currently operating or launching intelligent drinking water operations that use real-time sensor networks to detect pipe leaks and optimize drinking water operations. However, considerably fewer cities are implementing intelligent sewer systems such as Quebec City’s GO-RTC intelligent sewer system (Colas et al. 2005), which uses real-time environmental sensor data streams to increase efficiency and reduce the risk of sewer overflows. The relative rarity of intelligent sewer systems is due in part to the current challenges of accurately quantifying rainfall at spatiotemporal scales relevant to urban drainage. This limitation, combined with the challenge of integrating data from multiple sources, serves as a barrier for more integrated adaptive urban water management systems that holistically consider drinking water production, drinking water distribution, sewage and stormwater collection, and water treatment. Advances in ubiquitous water sensing will lower barriers to such integrated systems by providing access to real-time high-resolution data sets on which to base management decisions.

Adoption barriers to ubiquitous water sensor networks are grounded in both technology and policy. The technological challenges resemble those of other big data applications (National Research Council 2013): (1) sensor hardware must be developed to quantify many of the most important physical parameters driving water resource models; (2) large volumes of data must be aggregated, processed, and analyzed; (3) new methods for referencing data streams from an increasing number of potentially mobile sensors over large geographical areas must be created; and (4) new approaches for integrating and fusing various data types available from heterogeneous sensors (e.g., video, audio, text, photos, etc.) must be pursued.

To overcome these challenges and meet the potential for big water data to inform SWM, higher-level data and model services are needed that integrate multiscale data from multiple sources into best estimates and uncertainty bounds of historical, current, and future (model predicted) environmental states that can be delivered instantly from the Cloud, in a format specified by data standards. Early prototypes have shown the promise of such approaches (e.g., Hill et al. 2011), but a significant effort would be needed to transform prototypes into an operational water data infrastructure that is sufficiently comprehensive to be truly useful. While these technological challenges are nontrivial, nonetheless the major challenges to SWM will be policy oriented.

First, leveraging embedded pervasive sensor data as part of a ubiquitous sensing network presents the opportunity for location data or other sensitive information about individual citizens to be collected, stored, and analyzed—an apparent loss of citizen privacy. Although technological solutions to the anonymization of confidential data are being explored (e.g., Kifer and Gherke 2006), the significance of the perception of citizen privacy was recently highlighted by the public response to the revelation that the U.S. National Security Agency has been collecting and mining data originating from mobile phones operating in the United States (“In secret” 2013). Like many types of data collected by sensors embedded in the built environment, these cell phone data have the potential to locate the behaviors of individuals in space and time. Thus, ubiquitous sensing of urban systems has the potential to be viewed as an Orwellian technology leading to decreased social freedom rather than increased security from natural disasters. Changing this perception will require not only convincing people of the benefit of ubiquitous sensing technology to water resources, but also addressing public concerns over what data are collected and how these data will be used and secured from unauthorized use.

Second, because data collected by ubiquitous environmental sensors will be used to support management of systems to protect human health and well-being, these systems will need to be robust to sensor errors (a technical challenge), but also the assignment of fault and liability needs to be clarified for autonomous systems. In water resources management, the consequence of poor decision making can be catastrophic: a region could exhaust its drinking water supply or a city could suffer a damaging flood. Adaptive water systems could be even more vulnerable to erroneous information if they depend solely on data without effective error checking. Intelligent, integrated infrastructure systems may also be more vulnerable to cascading failures compared with conventional isolated systems due to their elevated interdependencies. Thus, it is also important to explore liability structure in the context of intelligent systems, a dialogue that is already beginning as regulators draft provisions for driverless vehicles (Canadian Broadcast Corporation 2013).

Third, integration of data from a large number of producers complicates the issue of data ownership rights. Large discrepancies exist between federal and private-sector policies on data ownership and openness and this can impede essential public-private partnerships for ubiquitous sensing. Permission from data owners is often required for data sharing and data owners may expect to receive a portion of monetary or nonmonetary benefits that are derived from their data. In the case of negative benefits, data ownership is often used to apportion liability. Thus, without clear, transparent policies governing data ownership, such as those forged around health-care data (Evans 2011), the data integration upon which ubiquitous environmental sensing is based may never be realized.

As future sensing, computation, and communication hardware are adopted for broad use, the immense potential to address critical issues in water resources management should be realized and that will enable more resilient population centers. However, using these technologies to support the management of our most precious natural resource requires not only that the technology be robust, but also that we accept the level to which this technology has become embedded in our everyday lives.