Table S1. Explanatory variables for Hg interannual variability and trendsa
Variable
Unit
Notes
Surface (2-m) air temperature over sea ice (TA,i)
K
-Surface skin temperature over open ocean (TS,o)
K
b
Sea ice threshold occurrence (Ni)
dimensionless
c
Sea ice fraction (fi)
dimensionless
d
Planetary boundary layer depth - ocean (ho)
m
e
Planetary boundary layer depth - ice (hi)
m
e
Net shortwave radiation at surface (FS)
W m-2
-Surface wind speed (v)
m s-1
f
Ozone column (Ω)
DU
g
Freshwater discharge (Q)
m3 s-1
h
-1
Net primary productivity (NPP)
Tg C a
i
Arctic Oscillation Index (AO)
dimensionless
j
a
Variables used in interpreting Hg variability in the Arctic through correlation analyses. Data
are from the MERRA meteorological analysis (used to drive GEOS-Chem) unless otherwise
indicated. Assimilated MERRA data originally at 0.5°x0.667° horizontal and hourly temporal
resolution are used here as two-month means for April-May and June-July. For correlation with
observed Hg0, MERRA variables are sampled in the 4°x5° grid box containing the measurement
site. For correlation with model results, MERRA variables are averaged over the Arctic Ocean
domain shown in Figure 1.
b
Skin temperature is used as proxy for sea surface temperature. For this analysis, only grid
squares that are partially non-ice-covered at the original hourly resolution are included in the
spatial and temporal averaging.
c
As described in the auxiliary material, bromine radical generation in GEOS-Chem (leading to
AMDEs) occurs for grid squares at 4°x5° resolution with at least 10% sea ice cover in at least
50% of native resolution (0.5°x0.667°) grid squares and less than 100% sea ice cover in any
native resolution grid square. The threshold occurrence variable Ni is the total number of hours in
a given month that this sea ice threshold criterion is met.
d
Sea ice fractions in MERRA are not an assimilated product but instead are prescribed from the
Reynolds et al. [2002] weekly product, interpolated to the MERRA time-step [Rienecker et al.,
2011].
e
The depth of the planetary boundary layer (PBL) differs significantly over ice-free and icecovered ocean boxes. Here we distinguish between these two populations with two PBL
variables. hi is the mean PBL depth over ocean boxes with at least 10% sea ice cover and ho the
mean PBL depth over all remaining ocean boxes.
f
Calculated from the horizontal wind vectors (u, v) at 10-m altitude.
g
Ozone columns are monthly mean observed ozone data from TOMS/SBUV gridded to 4°x5°
horizontal resolution.
h
Discharge data are from the Arctic-Rapid Integrated Monitoring System
(http://rims.unh.edu/index.shtml). Freshwater discharge represents the total monthly discharge
from the three largest circumpolar rivers (Yenisei, Lena, Ob) which define the IAV of the
riverine Hg flux in GEOS-Chem, as described in the auxiliary material. GEOS-Chem assumes no
riverine Hg inputs in November-April.
i
Interannual variability in net primary productivity (NPP) is from the NPP-sea ice extent
relationship of Arrigo and van Djiken [2011], applied to spatial distributions derived from
combining the POP-ICE sea ice-ecosystem model with MODIS satellite data, as described in the
auxiliary material.
j
Monthly mean Arctic Oscillation Index from the National Oceanic and Atmospheric
Administration (NOAA) / National Weather Service (NWS) National Centers for Environmental
Prediction (NCEP) Climate Prediction Center (CPC), available at
http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao.shtml.