Using machine-learning to construct long-term, gap-free stratospheric species profile data sets based on satellite occultation measurements and TOMCAT 3-D model
- 1School of Earth and Environmental Sciences, University of Leeds, Leeds, United Kingdom of Great Britain – England, Scotland, Wales (s.s.dhomse@leeds.ac.uk)
- 2National Centre of Earth Observation, Leeds, University of Leeds, United Kingdom of Great Britain – England, Scotland, Wales
Monitoring the atmospheric concentrations of the reactive species in the stratosphere, including greenhouse gases (GHGs), is crucial in order to improve our understanding of their climate impact. Although progress has been made towards construction of long-term ozone profile data sets, limited long-term profile data are available for other species. Here, we merge TOMCAT chemical transport model (CTM) output and profile measurements from two solar occultation instruments, the HALogen Occultation Experiment (HALOE) and the Atmospheric Chemistry Experiment - Fourier Transform Spectrometer (ACE-FTS), to construct a long-term (1991-2021), gap-free stratospheric profile data set (hereafter TCOM). The Extreme Gradient Boosting (XGBoost) regression model is used to estimate the corrections needed to apply to the CTM profiles to bring them into line with the observations.
We have already released data sets for two important greenhouse gases: methane and nitrous oxide. For methane (TCOM-CH4), we use both HALOE and ACE satellite profile measurements (1992-2018) to train the XGBoost model and profiles from three later years (2019-2021) are used as an independent evaluation data set. As there are no nitrous oxide profile measurements for the earlier years, XGBoost-derived correction terms for TCOM-N2O profiles are derived using only ACE-FTS profiles for 2004-2018 time period with profiles from 2019-2021 again being used for the evaluation.
Overall, both TCOM-CH4 and TCOM-N2O profiles show excellent agreement with the available satellite measurement-based data sets. Biases in TCOM-CH4 and TCOM-N2O are less than 10% and 50% throughout the stratosphere, respectively. Daily zonal mean TCOM-CH4 and TCOM-N2O profile data sets on altitude (15--60~km) and pressure (300--0.1~hPa) are publicly available via https://doi.org/10.5281/zenodo.7293740 and https://doi.org/10.5281/zenodo.7293740, respectively.
Our presentation will discuss the construction, performance and availability of the TCOM data sets. We aim to release data sets for ozone (TCOM-O3), hydrogen chloride (TCOM-HCl), hydrogen fluoride (TCOM-HF), water vapour (TCOM-H2O), nitric acid (TCOM-HNO3), nitric oxide (TCOM-NO) and nitrogen dioxide (TCOM-NO2) shortly.
How to cite: Dhomse, S. and Chipperfield, M.: Using machine-learning to construct long-term, gap-free stratospheric species profile data sets based on satellite occultation measurements and TOMCAT 3-D model, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-4542, https://doi.org/10.5194/egusphere-egu23-4542, 2023.
