Ehsan Kahrizia, Jeffery S. Horsburgha
aDepartment of Civil and Environmental Engineering, Utah Water Research Laboratory,
Utah State University
Pipeline | Abstract | Features | Getting Started
Pipeline Overview
Abstract
Advances in environmental sensors and data infrastructure have rapidly increased the availability of aquatic datasets, but effective quality control (QC) remains challenging. Traditional manual QC processes are subjective, time-consuming, and insufficient for real-time data monitoring, highlighting a high demand for effective automated QC methods. While statistical and machine learning approaches show enhancements, their effectiveness depends heavily on dataset characteristics, anomaly diversity, and categorization approaches, which limit their generalizability for different cases. This study presents standardized benchmark testing datasets aiming to test and evaluate existing automated QC algorithms. Using the Logan River Observatory in-situ sensor data, a set of standardized benchmark testing datasets was developed to create a comprehensive resource for evaluating existing automated QC performance in aquatic sensor data.
Features
- Detects anomalies using physically rule–based functions combined with statistical laws
- Labels and categorizes anomalies (e.g., spikes, drifts, gaps)
- Includes visualization and reporting tools
- Labeled datasets and ready for use
Getting Started
git clone https://github.com/YourUsername/AutomatedAnomalyDetectionLabeling.git
cd AutomatedAnomalyDetectionLabeling
pip install -r requirements.txt
python AnomalyDetection\ for\ \'T&SpCond\'_Ver.2.7_Feb.23.py