The protocol responds to the cross-border character of air pollution and the need to quantify the influence of distant emission sources. The original 1999 Gothenburg Protocol established national commitments to reduce sulphur dioxide, nitrogen oxides, volatile organic compounds and ammonia. It was amended in 2012 to include particulate matter and black carbon, extending its scope to human-health effects and climate co-benefits. A 2022 review concluded that existing air-quality legislation in the UNECE region was insufficient to meet the convention’s long-term clean-air objectives. In December 2023, the convention’s Executive Body began a revision process and asked that a possible role for methane, an ozone precursor, be considered. Earlier assessments by the Task Force on Hemispheric Transport of Air Pollution showed that intercontinental transport strongly influences surface ozone and demonstrated the value of large model ensembles for quantifying that influence and its uncertainty.

HTAP3-OPNS is organized around four linked questions. The first asks how much air pollution and its impacts in different world regions arise from sources within the same region compared with sources elsewhere. The second examines how suitable current models are for quantifying those contributions, while the third seeks to explain differences among models. The fourth asks how the contributions may change under realistic future emissions and potential future climate change. Compared with previous HTAP exercises, the new design places greater emphasis on ozone damage to vegetation, methane’s role in ozone formation, wildfire transport and total atmospheric deposition. It also calls for comparison of source perturbation, chemical tagging and adjoint sensitivity methods, and adds free-running chemistry-climate simulations to the source-receptor framework used to construct an ensemble emulator.

Three complementary sets of simulations form the core experimental plan. Perturbation experiments use global chemical transport models, and chemistry-climate models operated with prescribed dynamics, to derive source-receptor relationships for 2040. Transient future experiments use chemistry-climate models from 2010 to 2050 to assess changing emissions under SSP2-4.5 climate forcing and to distinguish effects associated with climate change from those caused by emissions. Transient historical simulations cover 2003 to 2020, using specified meteorology and updated global inventories to test model skill, establish a baseline for wildfire studies and generate comprehensive deposition fields for the World Meteorological Organization’s Measurement-Model Fusion for Global Total Atmospheric Deposition exercise. Regional modelling groups are also invited to downscale global simulations for Europe and other domains.

Future anthropogenic inputs come from version 2.1 of the GAINS LRTAP scenarios, produced specifically to inform the protocol revision. The Current Legislation pathway, or CLE, incorporates existing and planned air-pollution controls. The Maximum Technical Feasible Reduction pathway, or MTFR, assumes full implementation of measures shown to be technically feasible, including methane controls, regardless of cost-effectiveness. Both pathways use economic activity broadly consistent with SSP2-4.5. A hybrid HILO pathway combines MTFR emissions for conventional air pollutants with CLE methane emissions, representing high ambition for air-pollution control but minimal additional methane control. Methane concentrations differ between CLE and HILO even though their prescribed methane emissions are the same because their different nitrogen oxide and non-methane volatile organic compound emissions affect ozone chemistry, the hydroxyl radical and methane lifetime.

The perturbation simulations will use a 2040 CLE base case together with common 2015 meteorology, land cover, biomass-burning emissions and natural-emission settings. Most sensitivity runs lower relevant emissions or boundary concentrations by 20% to reveal how source regions and sectors influence ozone, particulate matter and deposition elsewhere. Priority-one runs include global changes in methane, nitrogen oxides, volatile organic compounds, carbon monoxide and combined anthropogenic emissions. They also include 20% reductions in land-based anthropogenic emissions from North America, the EMEP region, East Asia, South Asia and the Southern Mediterranean, plus a global reduction in international shipping emissions. Secondary runs isolate nitrogen oxides and non-methane volatile organic compounds within major source regions, while later-priority experiments subdivide regions further and examine aviation and biomass burning. Modelling groups may also apply tagging or adjoint methods to equivalent questions, allowing comparison of alternative attribution techniques.

A central limitation is that all perturbation runs use one meteorological year. The designers selected 2015 because it corresponds to the base year then being considered for revised emission-reduction commitments, but it was also the warmest year on record at the time and featured a strong El Niño event. Such climate variability influences atmospheric circulation, chemical production of ozone and the transport of pollution. Consequently, the resulting source-receptor coefficients should be interpreted as representative of 2015 meteorological conditions rather than invariant properties of emissions. Repeating the full perturbation matrix across multiple years was judged impractical because of computational demands. Earlier HTAP assessments nevertheless indicated that a single meteorological year can provide useful information about the relative importance of emission sources and transport pathways. Model evaluation, comparisons among participating systems and consistency of model responses to perturbations will help identify where this simplification matters.

Future chemistry-climate simulations will run under the CLE, MTFR and HILO emission scenarios with SSP2-4.5 forcing and, where resources permit, repeat them with climate fixed at 2015 conditions. This paired design is intended to separate a climate-related response to pollution from changes in emissions. Under the selected pathway, the climate signal by mid-century may remain smaller than interannual variability in some regions and for some metrics, making its emergence above that variability an explicit analytical objective. The source study associates the 2040–2049 period with approximately 0.8 degrees Celsius of additional global warming relative to 2015 under SSP2-4.5. In fixed-climate runs, atmospheric composition is decoupled from radiation and aerosol-cloud interactions are disabled, allowing emissions-driven chemical changes to proceed without rapidly altering the prescribed climate.

The requested data extend well beyond simple concentration maps. Teams must provide hourly surface ozone, while hourly particulate matter and nitrogen dioxide are strongly encouraged; chemistry-climate groups report hourly fields for 2010–2019 and 2040–2049. Detailed ozone-deposition and meteorological variables will support vegetation-impact metrics, including crop and forest dose indices. Monthly dry and wet deposition fields cover ozone, organic matter, black carbon, sea salt, dust, sulphur compounds, ammonia, ammonium and oxidized nitrogen. Additional diagnostics for ozone production and loss, halogen chemistry, hydroxyl-radical production and lifetime, together with machine-readable model chemical mechanisms, may help explain disagreement among models. Groups conducting the historical simulations must submit total deposition for the WMO effort, while all participants must document model resolution, chemistry, emissions, land cover, deposition treatment and spin-up procedures.

Planned analysis will test model output against surface observations, including harmonised ozone data from the Tropospheric Ozone Assessment Report network. Perturbation-derived relationships will be used to construct an ensemble emulator for ozone, particulate matter, nitrogen and sulphur, enabling rapid assessment of alternative mitigation scenarios while retaining information about nonlinearity and model uncertainty. Hourly ozone will support maximum daily eight-hour average metrics, peak-season measures and estimates of health outcomes, while exposure and deposition fields will help assign impacts to particular sources. Chemistry-climate simulations will provide an independent assessment of future air quality and a check on the emulator, and additional chemical diagnostics may clarify differences in ozone production, halogen loss and hydroxyl-radical lifetimes. Related HTAP3-Fires and MCHgMAP exercises will coordinate common fire-emission data and model configurations where possible. The protocol itself provides no new concentration results or evidence that any mitigation option will succeed; its findings will emerge when the modelling groups complete the common experiments and publish their analyses.

Subject of Research: Multi-model assessment of long-range ozone, particulate matter and nitrogen and sulphur deposition

Article Title: HTAP3-OPNS: ozone, PM, nitrogen and sulphur deposition – multi-model experiments to support the revision of the CLRTAP Gothenburg Protocol

Article References: Butler, T., Ansari, T., Belis, C. A., Bergas-Masso, E., van Caspel, W., Fagerli, H., Flemming, J., Garcia Vivanco, M., Griffiths, P., Hamilton, D. S., Hernandez Trujillo, C., Höglund-Isaksson, L., Huijnen, V., Kaiser, J. W., Koren, G., Klimont, Z., Lindl, F., Lupascu, A., Mertens, M., … Keating, T. (2026). HTAP3-OPNS: ozone, PM, nitrogen and sulphur deposition – multi-model experiments to support the revision of the CLRTAP Gothenburg Protocol. Geoscientific Model Development, 19(20), 10023-10049. https://doi.org/10.5194/gmd-19-10023-2026

Image Credits: AI Generated

DOI: 10.5194/gmd-19-10023-2026

Keywords: HTAP3-OPNS, ground-level ozone, particulate matter, chemical transport models, chemistry-climate models, transboundary air pollution, methane emissions, Gothenburg Protocol, nitrogen deposition, sulphur deposition, wildfire emissions, ensemble emulator