Adaptively Monitoring Hydrologic Systems Using Web-enabled Sensor Nodes

Abstract:

Hydrologic measurement campaigns are often limited by battery resources or, as in the case of automated water quality samplers, the number of actual samples that can be taken. As such, it becomes imperative to strategically capture only events of interest, such as the “first flush” of a hydrograph. We address the need to capture abrupt signal changes in hydrologic systems by leveraging a network of low-cost, low-power sensor nodes that continuously monitor their environment and dynamically update their sampling frequency in response to real-time, on-board flow models and live precipitation forecasts obtained from WeatherUndergound. In our case, this enables the use of an automated sampler only when a sample is deemed necessary, permitting samples to be spaced out throughout entire storm events while preserving battery resources and extending the deployment period of each automated sampler. We discuss the deployment and evaluation of a web-enabled “intelligent” sensor network in an urban watershed in Michigan. Equipped with a suite of in-situ aquatic sensors, each node continuously monitors and transmits hydrologic measurements (e.g., temperature, conductivity, and gage height) in real-time. Results indicate that the system effectively and autonomously captures signals of interest while simultaneously reducing the use of sampling resources. We also investigate how these real-time sensor nodes provide insight into hydrologic systems during storm events.