Conference Presentation (Other)

FZJ-2019-04499

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png Prediction of Daily Maximum Ozone Threshold Exceedances by Artificial Intelligence Techniques in Germany

Gong, B. (Corresponding author)FZJ* ; Kleinert, F.FZJ* ; Schultz, M.FZJ*

2019

Please use a persistent id in citations: http://hdl.handle.net/2128/22729

Abstract: The objective of this research is to forecast local daily maximum ozone threshold exceedances by artificial intelligence techniques in Germany. We utilised synthetic minority over-sampling and under-sampling as preprocessing step to address the imbalanced data issues. We compared the traditional machine learning algorithms with state-of-the-art deep learning algorithms. The results demonstrate that such combination of preprocessing technologies and machine learning can effectively and accurately forecast ozone threshold exceedances. The performance of the current set-up deep learning is not better than the traditional machine learning techniques, and the prediction accuracy decreases significantly from 1 leading day to 2 days prediction.

---

Contributing Institute(s):

1. Jülich Supercomputing Center (JSC)

Research Program(s):

1. 512 - Data-Intensive Science and Federated Computing (POF3-512) (POF3-512)
2. IntelliAQ - Artificial Intelligence for Air Quality (787576) (787576)
3. PhD no Grant - Doktorand ohne besondere Förderung (PHD-NO-GRANT-20170405) (PHD-NO-GRANT-20170405)
4. Earth System Data Exploration (ESDE) (ESDE)

Appears in the scientific report 2019

---

Database coverage:

---