Ensuring that all sensors continuously operate in proper working condition (e.g., ‘sensor uptime’) is an underlying goal for sensor-based networks. For an observatory of NEON’s magnitude, one that spans an entire continent, with sites in remote regions that are outfitted with environmental sensors, maintaining acceptable sensor uptimes poses a large challenge. To ensure NEON’s sensor uptime goals are met, a near real-time Sensor Health monitoring system was developed. This system identifies sensors that are dropping data, reporting data out of expected operational ranges, or are becoming noisy/stuck at a certain value. It alerts field staff of these issues in near-real time, allowing them to focus their attention on cleaning, removing/replacing, or re-calibrating problematic sensors during bi-weekly site visits.
In addition to actions that promote sensor uptime, a suite of quality assurance and quality control (QA/QC) procedures are continuously applied to incoming data. Nearly all sensor-based data collected within the Observatory stream from sensors in raw, uncalibrated form in electrical units such as volts and ohms. After being streamed to NEON headquarters in their raw form, calibration coefficients are applied to these data, converting them into meaningful scientific units, e.g., W m‑2. These individual, calibrated data are then put through a suite of nominal QA/QC tests. These tests include, but are not limited to, range (climatological or calibration data ranges), step/spike (large jumps in data), null (missing datum), and gap (consecutive missing data). Sensor-specific QAQC tests are applied to a handful of sensors in the network. Calibration checks are also applied to verify that valid (current) calibration coefficients are being applied.
This presentation will provide high level summaries of our ‘sensor health’ application and our QA/QC pipeline.