Development of Inferential Sensors for Real-time Quality Control of Water-level Data for the Everglades Depth Estimation Network

Abstract

The Everglades Depth Estimation Network (EDEN) is an integrated network of real-time water-level gaging stations, ground-elevation models, and watersurface models designed to provide scientists, engineers, and water-resource managers with current (2000-present) water-depth information for the entire freshwater portion of the greater Everglades. The generation of EDEN waterlevel surfaces is derived from real-time data. Real-time data are automatically checked for outliers using minimum, maximum, and rate-of-change thresholds for each station. Smaller errors in the real-time data, such as gradual drift of malfunctioning pressure transducers, are more difficult to immediately identify with visual inspection of time-series plots and may only be identified during on-site inspections of the gages. Correcting smaller errors in the data often is time consuming and water-level data may not be finalized for several months. To provide water-level surfaces on a daily basis, EDEN needed an automated process to identify errors in water-level data and to provide estimates for missing or erroneous waterlevel data. A technology often used for industrial applications is “inferential sensor.” Rather than installing a redundant sensor to measure a process, such as an additional waterlevel gage, an inferential sensor, or virtual sensor, is developed that estimates the processes measured by the physical sensor. The advantage of an inferential sensor is that it provides a redundant signal to the sensor in the field but without exposure to environmental threats. In the event that a gage does malfunction, the inferential sensor provides an estimate for the period of missing data. The inferential sensor also can be used in the quality assurance and quality control of the data. Inferential sensors for gages in the EDEN network are currently (2010) under development. The inferential sensors will be automated so that the real-time EDEN data will continuously be compared to the inferential sensor signal and digital reports of the status of the real-time data will be sent periodically to the appropriate support personnel. The development and application of inferential sensors is easily transferable to other real-time hydrologic monitoring networks. INTRODUCTION The Everglades Depth Estimation Network (EDEN) is an integrated network of approximately 250 real-time water-level gaging stations, ground-elevation models, and water-surface models designed to provide scientists, engineers, and water-resource managers with current (2000-present) water-depth information for the entire freshwater portion of the greater Everglades (Telis, 2006). The U.S. Geological Survey Greater Everglades Priority Ecosystems Science program provides support for EDEN with the goal of providing quality-assured hydrologic data for the Comprehensive Everglades Restoration Plan (CERP) (U.S. Army Corps of Engineers, 1999). Presented on a 400-square-meter grid spacing, the EDEN offers a consistent and documented data set that can be used by scientists and managers to: (1) guide large-scale field operations, (2) integrate hydrologic and ecological responses, and (3) support biological and ecological assessments that measure ecosystem responses to the CERP. These data establish a large data set of baseline conditions prior to the implementation of the CERP that offers investigators a single repository for historic hourly water-level data. While EDEN data are of great importance to many scientific and resource management activities, some of the massive amounts of data being collected by EDEN are inaccurate for reasons such as sensor malfunction, data communication errors, and other types of hardware issues. Detecting these issues can be time consuming and problematic, especially when they are not obvious by inspection, such as detecting drift. It can be time consuming to correct these types of problems because of the remoteness of the monitoring sites and the expense of having qualified technical personnel travel to the gages. In order for these data to be used for important assessments they need to be validated and sometimes corrected, further adding to the expense and time required to disseminate the data. A technology often used for industrial applications is the inferential sensor. Rather than installing a redundant sensor to measure a process, such as an additional water-level gaging station, an Figure 1. Location of the inferential sensor within the data stream of EDEN. inferential sensor, or virtual sensor, is developed that estimates the processes measured by the physical sensor. The inferential sensor typically is an empirical or mechanistic model using inputs from one or more proximal gages. The advantage of using an inferential sensor is that it provides a redundant signal to the sensor in the field but without exposure to the environmental threats (floods or hurricanes, for example). In the event that a physical sensor does malfunction, the inferential sensor provides an estimate for the period of missing or erroneous data. The inferential sensor also can be used in the quality assurance and quality control of the data. The virtual signal can be compared to the real-time data and if the difference between the two signals exceeds a certain tolerance, corrective action can be taken. Inferential sensors for gages in the EDEN network are currently under development. The inferential sensors (fig. 1) will be automated so that the real-time EDEN data will continuously be compared to the inferential sensor signal and digital reports of the status of the real-time data will be sent periodically to the appropriate personnel.

5 Figures and Tables

Cite this paper

@inproceedings{Daamen2011DevelopmentOI, title={Development of Inferential Sensors for Real-time Quality Control of Water-level Data for the Everglades Depth Estimation Network}, author={Ruby C. Daamen and E. A. Roehl and Paul A. Conrads}, year={2011} }