1

Improving the parameterization of land-surface

exchange in the GEOS-Chem Hg model

Shaojie Song and Noelle Selin

Massachusetts Institute of Technology

song33@mit.edu

The 7th International GEOS-Chem Meeting06 May 2015 @ Cambridge, MA

22

Atmosphere-surface exchange fluxes of Hg0 are still poorly constrained, limiting our ability to fully understand the mercury global biogeochemical cycle.

Global mercury cycle and the atmosphere-surface exchange of Hg0 (GEM)

adapted from Driscoll et al. (2013) and Pirrone et al. (2010)

33

Top-down constraints on global Hg0 emissions and parameters in the GEOS-Chem Hg model

Song et al., Atmos. Chem. Phys. Discuss., 2015, doi:10.5194/acpd-15-5269-2015

ground-based observations

model-observation comparison

ocean and soil emission fluxes

Bayesian inversion approach

4

Two-layer ocean model in GEOS-Chem

Henry’s law constantHg0 Diffusivity

Intermediate water Hg0 concentrations

HgII/HgP partitioning coefficient (KD)

Biological carbon pump

The parameter inversion optimizes two ocean parameters: dark oxidation rate of aqueous Hg0 (KOX2) and aqueous HgII/HgP partitioning coefficient (KD).

Reference simulationParameter inversion

Flux in Mmol yr-1

Song et al., Atmos. Chem. Phys. Discuss., 2015, doi:10.5194/acpd-15-5269-2015

55

For NOMADSS, We select background Hg0 data in the planetary boundary layer. We divide Hg0 observations into two categories: over-land and over-ocean.

Atmosphere-surface exchange of Hg0 in the Eastern US and Northwest Atlantic

6

GEOS-Chem nested-grid simulations over North America We conduct multiple GEOS-Chem simulations: reference simulation, parameter

inversion, and sensitivity simulations.

The sensitivity simulation #1 is based on the parameter inversion with elevated springtime soil emission in the mid-latitude region (by a factor of 4) and elevated summertime bromine columns in the tropics and subtropics (by a factor of 3).

An increase of springtime soil emission is consistent with results from the emission inversion. Agricultural activities and a larger land surface Hg pool due to deposition in the snow may contribute to such seasonal pattern.

Soil emissions from different simulations

7

Model-observation comparisonAMNet Hg0 concentration

1.33 ng m-3

1.30 ng m-3

Among different GEOS-Chem simulations, the sensitivity simulation #1 can best capture the seasonal pattern of AMNet Hg0 concentration.

In 2013 June/July, the sensitivity simulation #1 has a small negative bias (-0.03 ng m-3) when compared to the AMNet measurements.

The sensitivity simulation #1 gives the closest Hgaq

0 concentration among different simulations, but still 8-40% lower than the ship cruise measurements.

88

  Over-land Over-ocean Mann–Whitney U test

Measurement 1.44 ± 0.09 1.56 ± 0.11 p < 0.001

Reference simulation 1.40 ± 0.04 1.32 ± 0.03 p < 0.001

Parameter inversion 1.30 ± 0.03 1.28 ± 0.02 p < 0.001

Sensitivity simulation #1 1.29 ± 0.06 1.26 ± 0.05 p < 0.001

Sensitivity simulation #2 1.30 ± 0.04 1.29 ± 0.05 p = 0.992

N of samples 356 69  

Model-observation comparison Sensitivity simulation #2 increases ocean

emission of Hg0 from the Northwest Atlantic by 80% in summer.

The median Hg0 from the NOMADSS is 1.46±0.10 ng m-3 (N=443), ~ 10% higher than that from the sensitivity simulation #1 (1.29±0.04 ng m-3).

The overall uncertainty of the NOMADSS Hg0 measurements is 6~10%.

NOMADSS Hg0 concentration

99

Modeled Hg0 emission fluxes from the land and ocean during summer (JJA)

0.8

1.3

-0.9

1.5

-0.6

1.9

0.8

Acknowledgments: We thank NSF Atmospheric Chemistry Grant #1053648, and the NOMADSS team, Environment Canada, Ministry of the Environment (Japan), the SEARCH network, X. Feng & X. Fu (IGCAS, China), K. Crist (Ohio University), H. Amos, Y. Zhang, E. Corbitt (Harvard), C. Holmes (FSU), J. Kuss (IOW, Germany).

Thanks for your attention! Any questions?

Contact: Shaojie Song (song33@mit.edu)

Take-home message: GEOS-Chem simulations and observations suggest that the Eastern US terrestrial ecosystem is a net sink of Hg0 while the Northwestern Atlantic Ocean is a source.

Flux: ng m-2 h-1