Intercomparison of the representations of the atmospheric chemistry of pre-industrial methane and ozone in earth system and other global chemistry-transport models

Derwent, Richard G.; Parrish, D. D.; Archibald, A. T.; Deushi, Makoto; Bauer, Susanne E.; Tsigaridis, Kostas; Shindell, D. T.; Horowitz, Larry W.; Khan, M. Anwar H.; Shallcross, Dudley E.

doi:10.1016/j.atmosenv.2021.118248

Cited by 8 publications

(4 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The aggressive-conservative performance gap converges when fusing >9 realisations, or >4 realisation-averaged models. It exposes the critical limitation of the conservative approach and that the innovative enhanced spacetime BNN will not perform satisfactorily when only a few models are used for fusion, because different models have used different chemistry mechanisms, or simplifications, or have other physical differences [94], so that limited numbers of models cannot capture the full variations of the realistic surface O 3 by BNN-based linearcombination. It also further justifies the necessity of the CMIP6 multi-model study from the perspective of raising the signal-noise ratio and enabling more credible surface O 3 datasets (the more models used in the fusion process the better the performance).…”

Section: Sensitivity Analysismentioning

confidence: 99%

Multi-stage Ensemble-learning-based Model Fusion for Surface Ozone Simulations: A Focus on CMIP6 Models

Sun

Archibald

2021

Preprint

View full text Add to dashboard Cite

Accurately simulating the geographical distribution and temporal variability of global surface ozone has long been one of the principal components of chemistry-climate modelling. However, the simulation outcomes have been reported to vary significantly as a result of the complex mixture of uncertain factors that control the tropospheric ozone budget. Settling the cross-model discrepancies to achieve higher accuracy predictions of surface ozone is thus a task of priority, and methods that overcome structural biases in models going beyond naïve averaging of model simulations are urgently required. Building on the Coupled Model Intercomparison Project Phase 6 (CMIP6), we have transplanted a conventional ensemble learning approach, and also constructed an innovative 2-stage enhanced space-time Bayesian neural network to fuse an ensemble of 57 simulations together with a prescribed ozone dataset, both of which have realised outstanding performances (R 2 > 0.95, RMSE < 2.12 ppbv). The conventional ensemble learning approach is computationally cheaper and results in higher overall performance, but at the expense of oceanic ozone being overestimated and the learning process being uninterpretable. The Bayesian approach performs better in spatial generalisation and enables perceivable interpretability, but induces heavier computational burdens. Both of these multi-stage machine learning-based approaches provide frameworks for improving the fidelity of composition-climate model outputs for uses in future impact studies.

show abstract

Section: Sensitivity Analysismentioning

confidence: 99%

Multi-stage Ensemble-learning-based Model Fusion for Surface Ozone Simulations: A Focus on CMIP6 Models

Sun

Archibald

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Figure 4 compares the observationalderived long-term changes from Fig. 3 with time series of annual mean surface ozone at Mace Head and Cape Grim simulated by six ESMs that participated in the sixth Coupled Model Intercomparison Project (CMIP6; Eyring et al, 2016). Figure S2 shows those same model simulations for the entire 1850-2014 simulation period with each of the six models identified.…”

Section: Discussionmentioning

confidence: 99%

Investigations on the anthropogenic reversal of the natural ozone gradient between northern and southern midlatitudes

Parrish¹,

Derwent²,

Turnock

et al. 2021

Atmos. Chem. Phys.

Self Cite

View full text Add to dashboard Cite

Abstract. Our quantitative understanding of natural tropospheric ozone concentrations is limited by the paucity of reliable measurements before the 1980s. We utilize the existing measurements to compare the long-term ozone changes that occurred within the marine boundary layer at northern and southern midlatitudes. Since 1950 ozone concentrations have increased by a factor of 2.1 ± 0.2 in the Northern Hemisphere (NH) and are presently larger than in the Southern Hemisphere (SH), where only a much smaller increase has occurred. These changes are attributed to increased ozone production driven by anthropogenic emissions of photochemical ozone precursors that increased with industrial development. The greater ozone concentrations and increases in the NH are consistent with the predominant location of anthropogenic emission sources in that hemisphere. The available measurements indicate that this interhemispheric gradient was much smaller and was likely reversed in the pre-industrial troposphere with higher concentrations in the SH. Six Earth system model (ESM) simulations indicate similar total NH increases (1.9 with a standard deviation of 0.3), but they occurred more slowly over a longer time period, and the ESMs do not find higher pre-industrial ozone in the SH. Several uncertainties in the ESMs may cause these model–measurement disagreements: the assumed natural nitrogen oxide emissions may be too large, the relatively greater fraction of ozone injected by stratosphere–troposphere exchange to the NH may be overestimated, ozone surface deposition to ocean and land surfaces may not be accurately simulated, and model treatment of emissions of biogenic hydrocarbons and their photochemistry may not be adequate.

show abstract

“…The setup provides insights into potential Earth twins but is restricted to atmospheres with Earthlike compositions. Compared to the development of atmospheric chemistry modules for Earth-climate models (e.g., Kinnison et al 2007;Derwent et al 2021) and the progress of simulating aerosols (Lee et al 2016;Lines et al 2018;Steinrueck et al 2021), a robust chemical scheme with photochemistry capacity is still lacking and urgently needed in exoplanet science.…”

Section: Introductionmentioning

confidence: 99%

A mini-chemical scheme with net reactions for 3D general circulation models

Tsai

Lee

Pierrehumbert

2022

A&A

View full text Add to dashboard Cite

Context. Growing evidence has indicated that the global composition distribution plays an indisputable role in interpreting observational data. Three-dimensional general circulation models (GCMs) with a reliable treatment of chemistry and clouds are particularly crucial in preparing for upcoming observations. In attempts to achieve 3D chemistry-climate modeling, the challenge mainly lies in the expensive computing power required for treating a large number of chemical species and reactions. Aims. Motivated by the need for a robust and computationally efficient chemical scheme, we devise a mini-chemical network with a minimal number of species and reactions for H2-dominated atmospheres. Methods. We apply a novel technique to simplify the chemical network from a full kinetics model, VULCAN, by replacing a large number of intermediate reactions with net reactions. The number of chemical species is cut down from 67 to 12, with the major species of thermal and observational importance retained, including H2O, CH4, CO, CO2, C2H2, NH3, and HCN. The size of the total reactions is also greatly reduced, from ~800 to 20. We validated the mini-chemical scheme by verifying the temporal evolution and benchmarking the predicted compositions in four exoplanet atmospheres (GJ 1214b, GJ 436b, HD 189733b, and HD 209458b) against the full kinetics of VULCAN. Results. The mini-network reproduces the chemical timescales and composition distributions of the full kinetics well within an order of magnitude for the major species in the pressure range of 1 bar–0.1 mbar across various metallicities and carbon-to-oxygen (C/O) ratios. Conclusions. We have developed and validated a mini-chemical scheme using net reactions to significantly simplify a large chemical network. The small scale of the mini-chemical scheme permits simple use and fast computation, which is optimal for implementation in a 3D GCM or a retrieval framework. We focus on the thermochemical kinetics of net reactions in this paper and address photochemistry in a follow-up paper.

show abstract

Intercomparison of the representations of the atmospheric chemistry of pre-industrial methane and ozone in earth system and other global chemistry-transport models

Abstract: Eight chemical mechanisms are carefully examined for pre-industrial times.• Inter-mechanism differences stem from representation of C 1 -C 3 chemistry.• Intra-mechanism ranges are wider than inter-mechanism ranges.

Cited by 8 publications

References 42 publications

Multi-stage Ensemble-learning-based Model Fusion for Surface Ozone Simulations: A Focus on CMIP6 Models

Multi-stage Ensemble-learning-based Model Fusion for Surface Ozone Simulations: A Focus on CMIP6 Models

Investigations on the anthropogenic reversal of the natural ozone gradient between northern and southern midlatitudes

A mini-chemical scheme with net reactions for 3D general circulation models

Contact Info

Product

Resources

About