Fault detection for salinity sensors in the Columbia estuary

Abstract

[1] Sensors deployed in the Columbia River estuary gather information on physical dynamics and changes in estuary habitat. Of these sensors, conductivity sensors are particularly susceptible to biofouling, which gradually degrades sensor response and corrupts critical data. Several weeks may pass before degradation is visibly detected. Since the onset time of biofouling is unknown, an indeterminate amount of measurement data is corrupted. To speed detection and minimize data loss, we develop automatic biofouling detectors based on machine learning approaches for these conductivity sensors. We demonstrate that our detectors identify biofouling at least as reliably as human experts. In addition, these detectors provide accurate estimates of biofouling onset time. Real-time detectors installed during the summer of 2001 produced no false alarms yet detected all episodes of sensor degradation before the field staff.

A Feature‐Based Procedure for Detecting Technical Outliers in Water‐Quality Data From In Situ Sensors

“…Early detection of these technical outliers will limit the use of corrupted data for subsequent analysis. For instance, it will limit the use of corrupted data in real-time forecasting and online applications such as online drinking water-quality monitoring and early warning systems (Storey et al, 2011), predicting algal bloom outbreaks leading to fish kill events and potential human health impacts, forecasting water level and currents, and so on (Archer et al, 2003;Glasgow et al, 2004;Hill & Minsker, 2006). However, because data arrive near continuously at high speed in large quantities, manual monitoring is highly unlikely to be able to capture all the errors.…”

“…Water-quality sensors are exposed to changing environments and extreme weather conditions and thus are prone to errors, including failure. Automated detection of outliers in water-quality data from in situ sensors has therefore captured the attention of many researchers both in the ecology and data science communities (Archer et al, 2003;Hill et al, 2009;Koch & McKenna, 2010;McKenna et al, 2007;Raciti et al, 2012). This problem of outlier detection in water-quality data from in situ sensors can be divided into two subtopics according to their focus: (1) identifying errors in the data due to issues unrelated to water events per se, such as technical aberrations, that make the data unreliable and untrustworthy and (2) identifying real events (e.g., rare but sudden spikes in turbidity associated with rare but sudden high-flow events).…”

A Survey on Data Quality for Dependable Monitoring in Wireless Sensor Networks

“…In consonance with the theme of the work herein, [ 71 ] presented an ANN-based approach to detect disaster events through an environmental sensor network. Additionally, [ 72 ] presents another ANN-based approach to detect biofouling events (thus, fault events) in an aquatic sensor network.…”

The Ocean Appliance: Complete Platform Provisioning for Low-Cost Data Sharing

“…The CTD reports a timeseries of point observations, while the ADP reports a timeseries of an array of 3-d vectors. The temperature sensor inside a CTD is robust, reliable, and requires modest maintenance, while the conductivity sensor can become severely bio-fouled in estuarine environments, requiring estimates of quality for proper interpretation [1].…”

“…We report on our experience using the appliance model to manage a cruise integrating real-time observations with the predictive capabilities of the CORIE forecasting system [1]. Finally, we describe an upcoming experiment involving three coordinated vessels linked by Ocean Appliances.…”