Preprints
https://doi.org/10.5194/gmd-2022-2
https://doi.org/10.5194/gmd-2022-2
Submitted as: model description paper
19 Jan 2022
Submitted as: model description paper | 19 Jan 2022
Status: a revised version of this preprint was accepted for the journal GMD and is expected to appear here in due course.

Global, high-resolution mapping of tropospheric ozone – explainable machine learning and impact of uncertainties

Clara Betancourt1, Timo T. Stomberg2, Ann-Kathrin Edrich3, Ankit Patnala1, Martin G. Schultz1, Ribana Roscher2,4, Julia Kowalski5, and Scarlet Stadtler1 Clara Betancourt et al.
  • 1Jülich Supercomputing Centre, Jülich Research Centre, Wilhelm-Johnen-Straße, 52425 Jülich, Germany
  • 2Institute of Geodesy and Geoinformation, University of Bonn, Nußallee 17, 53115 Bonn, Germany
  • 3Aachen Institute for Advanced Study in Computational Engineering Science (AICES), RWTH Aachen University, Schinkelstrasse 2a, 52056 Aachen, Germany
  • 4Data Science in Earth Observation, Technical University of Munich, Lise-Meitner-Str. 9, 85521 Ottobrunn, Germany
  • 5Methods for Model-based Development in Computational Engineering, RWTH Aachen University, Eilfschornsteinstr. 18, 52062 Aachen, Germany

Abstract. Tropospheric ozone is a toxic greenhouse gas with a highly variable spatial distribution which is challenging to map on a global scale. Here we present a data-driven ozone mapping workflow generating a transparent and reliable product. We map the global distribution of tropospheric ozone from sparse, irregularly placed measurement stations to a high-resolution regular grid using machine learning methods. The produced map contains the average tropospheric ozone concentration of the years 2010–2014 with a resolution of 0.1° × 0.1°. The machine learning model is trained on AQ-Bench, a precompiled benchmark dataset consisting of multi-year ground-based ozone measurements combined with an abundance of high-resolution geospatial data.

Going beyond standard mapping methods, this work focuses on two key aspects to increase the integrity of the produced map. Using explainable machine learning methods we ensure that the trained machine learning model is consistent with commonly accepted knowledge about tropospheric ozone. To assess the impact of data and model uncertainties on our ozone map, we show that the machine learning model is robust against typical fluctuations in ozone values and geospatial data. By inspecting the feature space, we ensure that the model is only applied in regions where it is reliable.

We provide a rationale for the tools we use to conduct a thorough global analysis. The methods presented here can thus be easily transferred to other mapping applications to ensure the transparency and reliability of the maps produced.

Clara Betancourt et al.

Status: closed

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CEC1: 'Comment on gmd-2022-2', Juan Antonio Añel, 23 Feb 2022
    • AC1: 'Reply on CEC1', Clara Betancourt, 03 Mar 2022
  • RC1: 'Comment on gmd-2022-2', Anonymous Referee #1, 25 Feb 2022
    • AC2: 'Reply on RC1', Clara Betancourt, 14 Apr 2022
  • RC2: 'Comment on gmd-2022-2', Anonymous Referee #2, 13 Mar 2022
    • AC3: 'Reply on RC2', Clara Betancourt, 14 Apr 2022

Status: closed

Comment types: AC – author | RC – referee | CC – community | EC – editor | CEC – chief editor | : Report abuse
  • CEC1: 'Comment on gmd-2022-2', Juan Antonio Añel, 23 Feb 2022
    • AC1: 'Reply on CEC1', Clara Betancourt, 03 Mar 2022
  • RC1: 'Comment on gmd-2022-2', Anonymous Referee #1, 25 Feb 2022
    • AC2: 'Reply on RC1', Clara Betancourt, 14 Apr 2022
  • RC2: 'Comment on gmd-2022-2', Anonymous Referee #2, 13 Mar 2022
    • AC3: 'Reply on RC2', Clara Betancourt, 14 Apr 2022

Clara Betancourt et al.

Clara Betancourt et al.

Viewed

Total article views: 945 (including HTML, PDF, and XML)
HTML PDF XML Total BibTeX EndNote
734 187 24 945 12 18
  • HTML: 734
  • PDF: 187
  • XML: 24
  • Total: 945
  • BibTeX: 12
  • EndNote: 18
Views and downloads (calculated since 19 Jan 2022)
Cumulative views and downloads (calculated since 19 Jan 2022)

Viewed (geographical distribution)

Total article views: 837 (including HTML, PDF, and XML) Thereof 837 with geography defined and 0 with unknown origin.
Country # Views %
  • 1
1
 
 
 
 
Latest update: 20 May 2022
Download
Short summary
Ozone is a toxic greenhouse gas with high spatial variability. We present a machine learning-based ozone mapping workflow generating a transparent and reliable product. Going beyond standard mapping methods, this work combines explainable machine learning with uncertainty assessment to increase the integrity of the produced map.