Evaluation of sulphur species and improved SO2 dry deposition parameterization in UKESM1
Catherine Hardacre1, Jane Mulcahy1, Richard Pope2, Colin Jones3, Steve Rumbold3, Can Li4,5
1 Met Office, Exeter, 2 School Earth and Environment, University of Leeds, 3 National Centre for Atmospheric Science (NCAS), 4 Earth System Science Interdisciplinary Center, University of Maryland, USA, 5 Atmospheric Chemistry and Dynamics Laboratory, NASA Goddard Space Flight Center, USA
Emissions of SO2 can be oxidised to form sulphate (SO42-) aerosol, which plays a key role in both acid deposition, atmospheric aerosol loading and cloud properties, thereby directly influencing the Earth’s radiative balance. An accurate representation of sulphur processes in models is therefore essential for constraining uncertainties associated with the impacts of aerosols on the Earth system and thus understanding the global climate. Although the global atmospheric loading of sulphur compounds has generally decreased since the 1980’s, there is substantial regional variation. For example, legislation has driven reductions in SO2 emissions and subsequently sulphate aerosol in Europe and North America, but emissions in Asia continue to increase (Aas et al., 2019). It is important that models can capture these trends and the resulting climate response so that confidence can be placed in simulations of future climate change.
In this study we evaluate simulated surface SO2 and sulphate concentrations simulated by UKESM1 against observations from ground based measurement networks in the USA and Europe for the period 1987–2014. We have also evaluated simulated total column SO2 (TCSO2) from UKESM1 against observations from a new retrieval product of TCSO2 from the Ozone Monitoring Instrument (OMI) on the Aura satellite (available at: https://aura.gesdisc.eosdis.nasa.gov/data/Aura_OMI_Level2/OMSO2.003). We use 4 members from the 19 member historical ensemble that was completed for UKESM1’s contribution to CMIP6 (Sellar et al., 2019). We also use a further 4 historical simulations in which we change the model’s SO2 dry deposition parameterization to increase SO2 removal from the atmosphere.
We used ground based observations of surface SO2 and sulphate concentrations, and SO2 dry deposition from the Clean Air Status and Trends Network (CASTNet; Finklestein et al., 2000; see http://epa.gov/castnet/javaweb/index.html) and the European Monitoring and Evaluation Program (EMEP; Torseth et al., 2012; see http://ebas.nilu.no/). CASTNet provides surface observations of mean seasonal SO2 and sulphate concentrations, and SO2 dry deposition which are available from 1987 to the present day at 97 sites. In this study we used observations from the CASTNet sites designated as “western reference” (USA-W) and “eastern reference” (USA-E). These sites have been reporting measurements since at least 1990 and are used for determining long term trends (EPA – National Air Quality and Emission Trends Report, 2000). They are located west and east of 100° longitude respectively and are generally subject to different pollution regimes due to the much larger number of SO2 sources in the eastern USA region. Surface SO2 and sulphate concentrations have been monitored at EMEP sites for the period 1972 to the present day. In this study we have used data from 144 SO2 sites and 99 sulphate sites, although not all sites have measurements over the full period. No SO2 dry deposition data were available from EMEP. The OMI TCSO2 product was available for the period 2005 – 2014 (Li et al., 2020). Further details on the OMI TCSO2 product, how it was processed to obtain TCSO2 concentrations and the assessment of the temporal resolution is given in Pope et al., (2020).
We find that UKESM1 captures the historical decrease in surface SO2 and sulphate concentrations in both the USA and Europe over the period 1987 – 2014 (see Figure 1). UKESM1 is also able to capture the spatial pattern of higher concentrations in USA-E, where there many more SO2 sources compared with USA-W. However, in USA-E and Europe, UKESM1 over predicts surface SO2 concentrations with simulated (observed) values of 20.36 µm-3 (6.31 µm-3) and 11.12 µm-3 (3.29 µm-3) respectively. In the cleaner USA-W region, the absolute bias is smaller, with simulated and observed values of 6.45 µm-3 and 0.53 µm-3 and for 1987–2014, but the normalised mean bias (NMB) is large (NMB = 11.6) compared with values of between 2.2–2.4 for USA-E and Europe over the same period. UKESM1 has small negative biases in surface sulphate concentrations in USA-E and Europe, where the simulated values of 3.17 µm-3 and 2.19 µm-3, respectively, are under-predicted relative to the observed values of 4.19 µm-3 and 2.82 µm-3. In contrast, UKESM1 over predicts surface sulphate concentrations in USA-W with simulated and observed values of 1.14 µm-3 and 0.74 µm-3 respectively.
Figure 1: Time series of observed and modelled surface SO2 concentration (top row), sulphate concentration (middle row) and SO2 dry deposition (bottom row) for USA-W (left column), USA-E (middle column) and Europe (right column).
Figure 2 shows that UKESM1 over-predicts TCSO2 over much of the globe, but particularly over the source regions of India, China, the USA and Europe, and volcanic sources. In these areas UKESM1 over-predicts TCSO2 by 0.6–1.0 Dobson Units (DU). Over the background regions, including much of the ocean, the model over predicts TCSO2 by up to approximately 0.3 DU. UKESM1’s over-prediction of TCSO2 is generally greater during the winter season. The model’s over-prediction of TCSO2 over land and source regions agrees with the comparison with the EMEP and CASTNet observations (see Figure 1a-c). Further, as seen in the ground-based observations, the satellite data shows an East-West divide in the USA with greater TCSO2 in the eastern USA compared with the western USA.
Figure 2: Median total column sulphur dioxide (TCSO2; Dobson units, DU) for 2005–2014 for UKESM1 in December-January-February (DJF),(a); UKESM1 in June-July-August (JJA), (b); OMI in DJF, (c); OMI in JJA, (d); UKESM1 – OMI in DJF, (e), and UKESM1 – OMI in JJA (f). Model fields have been spatio-temporally co-located with the OMI overpasses and the satellite air mass fractions updated to account for the model vertical structure. OMI data has been filtered for volcanic events, with associated model samples excluded as well.
Figure 1 and Figure 3 show the impact of the changes to the SO2 dry deposition parameterization on UKESM1’s simulation of surface SO2 concentration, surface sulphate concentration, SO2 dry deposition and TCSO2. Figures 1a, 1b and 1c clearly show that the changes to the SO2 dry deposition parameterization reduce surface SO2 concentration in all three regions and thus reduce UKESM1’s positive bias. Figure 3 shows that TCSO2 is reduced in UKESM1-SO2 relative to UKESM1 and that the comparison with OMI is therefore moderately better. In Figures 3a and b, the absolute difference in UKESM1-SO2 – OMI TCSO2 was 0.3–0.5 DU over source regions and 0.1–0.3 DU over background regions. This represents a decrease in the global TCSO2 for model – OMI bias of 20–30%. Over the outflow regions (e.g. off the USA eastern seaboard), TSCO2 has reduced by 30–50%. Over the source regions, this varies by 30–50% over South East Asia, 20–30% over Europe and 10–30% over the USA.
However, in increasing SO2 dry deposition flux the existing positive bias observed at the CASTNet sites has been exacerbated, with bias in the mean annual flux increasing from 7.3×10-5 kg m-2 y-1 to 1.0×10-4 kg m-2 y-1 in USA-W and from 1.9×10-4 kg m-2 y-1 to 4.5×10-4 kg m-2 y-1 in USA-E (see Figures 1g and 1h). Negative model bias in surface sulphate concentrations is also exacerbated by the changes to the SO2 dry deposition parameterization in the polluted regions of USA-E and Europe, but UKESM1’s over prediction of surface sulphate concentration in USA-W is reduced.
Figure 3: Median TCSO2 (DU) differences (model – OMI) for 2005 to 2014 for UKESM1 – OMI in DJF, (a); UKESM1 – OMI in JJA, (b); UKESM1-SO2 – OMI in DJF, (c) and UKESM1-SO2– OMI in JJA (d).
The evaluation of UKESM1 SO2 against ground based and satellite observations suggests that the model over predicts atmospheric concentrations of this species. This may result from too little removal of SO2 from the atmosphere, or too high emissions. The two main removal pathways for SO2 are oxidation to sulphate and deposition (both wet and dry) to the Earth’s surface. In investigating changes to the parameterization of SO2 dry deposition in UKESM1 we find that we can reduce the simulated atmospheric SO2 concentrations through increased dry deposition. This reduces the positive model bias in surface SO2 concentrations in the USA and Europe, and in TCSO2 across most of the globe. We have also demonstrated that this leads to a decrease in the simulated atmospheric loading of sulphate aerosol and a moderate reduction in the UKESM1 mid-century cold bias. However, the comparison between UKESM1 and the ground-based observations of sulphate and SO2 dry deposition show that UKESM1’s bias in surface sulphate concentrations and SO2 dry deposition flux increases when the changes to the parameterization of SO2 dry deposition are included in the model, suggesting that uncertainty also exists in other parts of the UKESM1’s representation of the sulphur cycle. This evaluation of UKESM1’s surface sulphate concentrations agrees with Mulcahy et al., (2020) who suggest that the conversion of SO2 to SO42- in Europe and USA-E is oxidant limited, rather than SO2 limited. The role of oxidation will be addressed in future research.
UKESM1 uses SO2 emissions from CMIP6 (Coupled Model Intercomparison Project Phase 6, Erying et al., 2016). In comparing these with the HTAP-OMI (Liu et al., 2018; https://avdc.gsfc.nasa.gov/pub/data/project/OMI_HTAP_emis/) and EDGAR (Crippa et al., 2018; https://edgar.jrc.ec.europa.eu/overview.php?v=432_AP) data sets, Pope et al., (2020) showed that the total global SO2 emissions in CMIP6 (115 Tg y-1) are moderately larger than the HTAP-OMI emissions (100 Tg y-1). Notably, while the HTAP-OMI and EDGAR emissions higher in a few regions where there are large point sources, the CMIP6 emissions are larger in most areas. This may therefore contribute to the model’s positive bias in surface SO2 concentration and TCSO2, particularly over remote regions.
Overall, we find that UKESM1 is able to capture the temporal and spatial patterns in surface SO2 concentration, surface sulphate concentration, SO2 dry deposition and TCSO2. However, compared to observations we find that the model is biased, depending on the parameter, region and species. We reduce model bias in surface SO2 concentrations and TCSO2 by improving the dry deposition parameterization, but model biases are also likely to result from uncertainty in oxidation and the prescribed SO2 emissions. We find that the evaluation of UKESM1 against satellite data is robust, and although temporally limited to 2005–2014, can provide a valuable insight into the model’s ability to represent atmospheric SO2 global scale, particularly over the ocean and other remote regions.
References
- Aas, W., Mortier, A., Van Bowersox, V., Cherian, R., Faluvegi, G., Fagerli, H., Hand, J., Klimont, Z., Galy-Lacaux, C., Lehmann, C.M.B., Myhre, C. L., Myhre, G., Olivié, D., Sato, K., Quaas, J., Rao, P. S. P., Schulz, M., Shindell, D., Skeie, R. B., Stein, A., Takemura, T., Tsyro, S., Vet, R., and Xu, Z, “Global and regional trends of atmospheric sulfur”. Nature Scientific Reports. https://doi.org/10.1038/s41598-018-37304-0
- Crippa, M., Guizzardi, D., Muntean, M., Schaaf, E., Dentener, F., van Aardenne, J. A., Monni, S., Doering, U., Olivier, J. G. J., Pagliari, V., and Janssens-Maenhout, G.: Gridded emissions of air pollutants for the period 1970–2012 within EDGAR v4.3.2, Earth Syst. Sci. Data, 10, 1987–2013, 2018, https://doi.org/10.5194/essd-10-1987-2018.
- S. Environmental Protection Agency (EPA), Clean Air Status and Trends Network (CASTNet) Annual Report 2004, November 2005 (Available at: https://www3.epa.gov/castnet/docs/annual_report_2005.pdf, accessed 01/12/2020).
- Eyring, V., Bony, S., Meehl, G. A., Senior, C. A., Stevens, B., Stouffer, R. J., and Taylor, K. E.: Overview of the Coupled Model Intercomparison Project Phase 6 (CMIP6) experimental design and organization, Geosci. Model Dev., 9, 1937–1958, https://doi.org/10.5194/gmd-9-1937-2016, 2016.
- Finkelstein, P. L., Ellestad, T. G., Clarke, J. F., Meyers, T. P., Schwede, D. B., Hebert, E. O., and Neal, J. A., “Ozone and sulfur dioxide dry deposition to forests: Observations and model evaluation.” Geophys. Res.-Atmos. 105, 15365–15377, \doi{10.1029/2000JD900185}, 2000.
- Li, C., Krotkov, N. A., Leonard, P. J. T., Carn, S., Joiner, J., Spurr, R. J. D., and Vasilkov, A. “Version 2 Ozone Monitoring Instrument SO2 product (OMSO2 V2): new anthropogenic SO2 vertical column density dataset”, Atmos. Meas. Tech., 13, 6175–6191, 2020, https://doi.org/10.5194/amt-13-6175-2020
- Liu F, et al., “A new global anthropogenic SO2 emission inventory for the last decade: a mosaic of satellite-derived and bottom-up emissions”, Atmospheric Chemistry and Physics, 18, 16571–16586, 2018, https://doi.org/10.5194/acp-18-16571-2018
- Pope, R. J., Hardacre, C., Li, C., and Chipperfield, M. P., “United Kingdom Earth System Model – Ozone Monitoring Instrument Total Column Sulphur Dioxide Comparisons”, Hadley Centre Technical Note, Met Office Hadley Centre, in prep.
- Sellar, Alistair A., Colin G. Jones, Jane P. Mulcahy, Yongming Tang, Andrew Yool, Andy Wiltshire, Fiona M. O’connor et al. “UKESM1: Description and Evaluation of the UK Earth System Model.” Journal of Advances in Modeling Earth Systems 11, 12, 2019, 4513-4558.
- Tørseth, K., Aas, W., Breivik, K., Fjæraa, A. M., Fiebig, M., Hjellbrekke, A. G., Myhre, C. L., Solberg, S. and Yttri, K.E., “Introduction to the European Monitoring and Evaluation Programme (EMEP) and observed atmospheric composition change during 1972–2009”, Chem. Phys., 12, 5447–5481, 2012, doi:10.5194/acp-12-5447-2012
- Turnock, S. T., Mann, G. W., Woodhouse, M. T., Dalvi, M., O’Connor, F. M., Carslaw, K. S. and Spracklen, D. V., “The impact of changes in cloud water pH on aerosol radiative forcing.”, Geophysical Research Letters, 46, 2019, https://doi.org/10.1029/2019GL082067