1 results match your criteria: "Shaanxi Atmospheric Observation Technical Support Center[Affiliation]"
Enhancing meteorological data reliability: An explainable deep learning method for anomaly detection.
J Environ Manage
January 2025
Shaanxi Atmospheric Observation Technical Support Center, Xi'an, 710000, China. Electronic address:
Article Synopsis
- Accurate meteorological observation data is crucial for human activities, but challenges like sensor malfunctions can lead to data inaccuracies.
- A new deep learning method using autoencoders, SHAP, and Bayesian optimization is proposed for detecting these data anomalies quickly and accurately.
- This method involves analyzing reconstruction errors in data, assessing the importance of different meteorological elements, setting appropriate anomaly thresholds, and fine-tuning model parameters to enhance detection accuracy, benefiting areas like agriculture and disaster prevention.