Ocean Biogeochemistry Control on the Marine Emissions of Brominated Very Short-Lived Ozone-Depleting Substances: A Machine-Learning Approach

22 pags., 11 figs., 2 tabs. -- Open Access funded by Creative Commons Atribution Licence 4.0Halogenated very short lived substances (VSLS) affect the ozone budget in the atmosphere. Brominated VSLS are naturally emitted from the ocean, and current oceanic emission inventories vary dramatically. We present a new global oceanic emission inventory of Br‐VSLS (bromoform and dibromomethane), considering the physical forcing in the ocean and the atmosphere, as well as the ocean biogeochemistry control. A data‐oriented machine‐learning emulator was developed to couple the air‐sea exchange with the ocean biogeochemistry. The predicted surface seawater concentrations and the surface atmospheric mixing ratios of Br‐VSLS are evaluated with long‐term, global‐scale observations; and the predicted vertical distributions of Br‐VSLS are compared to the global airborne observations in both boreal summer and winter. The global marine emissions of bromoform and dibromomethane are estimated to be 385 and 54 Gg Br per year, respectively. The new oceanic emission inventory of Br‐VSLS is more skillful than the widely used top‐down approaches for representing the seasonal/spatial variations and the annual means of atmospheric concentrations. The new approach improves the model predictability for the coupled Earth system model and can be used as a basis for investigating the past and future ocean emissions and feedbacks under climate change. This model framework can be used to calculate the bidirectional oceanic fluxes for other compounds of interest.S.‐Y. W. is supported by the NCAR Advanced Study Program (ASP) Postdoctoral Fellowship. The CESM project is supported primarily by the National Science Foundation (NSF). This material is based upon work supported by NCAR, which is a major facility sponsored by the NSF under Cooperative Agreement 1852977. The computing and data storage resources, including the Cheyenne supercomputer (doi:10.5065/D6RX99HX), were provided by the Computational and Information Systems Laboratory (CISL) at NCAR. The Atmospheric Tomography Mission (ATom) is funded by the Earth Science Project Office at NASA (NNX15AJ23G). The NOAA ground‐based measurements were supported in part by NOAA's Atmospheric Chemistry, Carbon Cycle and Climate Program of its Climate Program Office. NASA's ATom mission was funded as a NASA Earth Science Venture ‐ suborbital mission. We thank NASA ESPO, the NASA DC‐8 crew, and the ATom Science Team for their exceptional professionalism in support of this mission. Franziska Ziska (GEOMAR, Germany) and colleagues are acknowledged for compiling the HalOcAt data set. Tomás Sherwen (University of York, UK) and Britton Stephens (NCAR, USA) are acknowledged for helpful discussions. CAM‐chem is a component of the NCAR CESM which is publicly available on the project website (http://www.cesm. ucar.edu/). The HalOcAt data set used for the training of the machine‐learning emulator is available in Ziska et al. (doi: https://doi.org/10.5194/acp‐13‐8915‐ 2013). The NCAR CESM Large Ensemble data set is available from the project website (http://www.cesm.ucar. edu/projects/community‐projects/ LENS/). The NCAR TOGA, UCI WAS, and NOAA PFP measurements during the NASA ATom campaign is available from Wofsy et al. (2018).Peer reviewe