Natural emissions

Natural NOx emissions

Natural emissions of NO$_{x}$ by microbes in soils are included in the above EDGAR database as category AGL. [Olivier et al., 1999] states this is the current GEIA inventory from [Yienger and Levy, 1995]. In 1990 it contained 5.5 Tg(N)/yr (to be checked with paper of [Yienger and Levy, 1995], The [Yienger and Levy, 1995] emission imventory can be downloaded from the GEIA website at
http://eosweb.larc.nasa.gov/HORDERBIN/HTML_Start.cgi (select biomass burning as Project and Parameter). http://www.geiacenter.org/emits/noxsoil.html. GEIA header states that 3.08 Tg(N)/yr) is contained in the file soilNOX90mn1.1a.

Natural CO emissions

For CO there are also natural emissions from the ocean and vegetation, read in from cobio.d from GEIA [Houweling et al., 1998] valid for 1990. Preprocessing is done with script non_annual. In subroutine sources_sinks of TM3 the total natural CO emissions from soils and oceans are set to biocoyr=115 Tg CO (49 Tg C), assuming that the emission distribution bioco from file cobio.d is normalised to 1.

As described by [Houweling et al., 1998], the ocean CO emissions are distributed in latitude bands of 15 degrees, based on the work of [Bates et al., 1995]. The annual total CO emission from the oceans is a compromise between [Bates et al., 1995] and [Erickson, 1989].

CO fluxes from Bates

[Bates et al., 1995] gives an estimate of 13 Tg(CO) (5.5 Tg(C), 0.46 Tmol) with an uncertainty of a factor of 2. They measured CO partial pressures in the ocean surface layer and in the atmosphere above it from a ship. These data are available at
http://saga.pmel.noaa.gov

The CO flux estimate is based upon application of the flux relationships given by [Liss and Merlivat, 1986] and [Wanninkhof, 1992]: F=K$_{l}$L${\triangle}$p$_{CO}$ where K$_{l}$ is the gas transfer velocity (in m/s), L is the gas solubility at the surface seawater temperature (concentration/Pa), and ${\triangle}$p${_CO}$ is the CO partial pressure difference between air and surface sea water. The gas solubility L ([Wiesenburg and Jr., 1979]) was calculated as a function of the SST from the 40-year COADS climatological data set. The estimate depends strongly on the transfer velocity which in turn depends upon the used wind field (locally measured or averaged winds, e.g. from a reanalysis or COADS) [Wanninkhof, 1992]. For climatological winds (and SST, salinity): K${_l}$= 0.39 u$^{2}$ (Sc/660)$^{-0.5}$ For local winds (and SST, salinity) a value of about 0.31 instead of 0.39 would need to be used [Wanninkhof, 1992]. Sc is the Schmidt number, the ratio of kinematic viscosity of seawater and the molecular diffusivity of CO.

The gas diffusivity in fresh water as a function of SST and salinity (COADS) were taken from [Wise and Houghton, 1968] and reduced by 6following [Jahne et al., 1987]. Sc is 580 at 20 degrees C and a salinity of 35 ppt.

Ocean CO fluxes from Erickson

CO emissions from vegetation

The CO emissions from vegetation have been scaled to the global distribution of net primary production (NPP) as derived from the climate assessment model IMAGE ([Minnen et al., 1996], [Kreileman, 1996]). The total amount is based on the work of [Bauer et al., 1979] ...

SO$_{2}$ from volcanoes

For SO$_{2}$ there are also (natural) volcanic emissions taken from [Andres and Kasgnoc, 1998], read in from Gvolcs.d. This emission data set is extensively documented on the GEIA website, see http://www.geiacenter.org/emits/volcano.html. The emission data are processed with script non_annual using fortran program compvolc. We only use the emissions from 43 continuously degassing volcanoes. The program requires an input file with the heights of the TM layers heights_TM3.d that is dependent on the chosen number of TM levels. The program to make this file is still missing.
It includes both emissions of SO$_{2}$ and H$_{2}$S, but they are emitted only as SO$_{2}$ in sources_sinks of TM3.

The total emissions from continuously erupting volcanoes is multiplied by 1.39 to account for all non-SO$_{2}$ sulfur emissions (e.g. H$_{2}$S) 50 height of the volcano top and 50 top will be well above the orography of ECMWF and TM. So is this correct? For sporadical eruptions total emissions of a factor 1.85/4.8 times the total is emitted at 6 km geometrical height. In TM5 this factor is not included anymore.

This uses:

l6km=lvlpress(50000.0, 98400.0)

Gondwe et al state that the total emission from volcanoes is 13.4 Tg/yr. How much is there in the input file?

Natural DMS and H2S emissions

Oceans

DMS concentrations (nmol/l) in surface water ([Kettle et al., 1999] or [Kettle and Andreae, 2000]) valid for 1985) are read in from DMSconc.d which is calculated by interp_dms from the original file dms_array.dat.

The Kettle DMS concentrations can be downloaded from

http://saga.pmel.noaa.gov/dms/.

The measurements were made between 11 March 1972 and 30 August 1998 between 77 S and 90 N and -180 to 180 E. 15675 measurements are included. The monthly mean average surface water concentration distribution used in TM3, dms_array.dat, were provided by J. Kettle (pers. comm., 1998), and are given on a 1x1 degree grid. Also available are gridded flux estimates, but these are not used.

In sources_sinks of TM3 the ocean surface DMs concentrations are translated into fluxes in subroutine getDMS. The ocean-air fluxes are calculated following the parameterisation [Liss and Merlivat, 1986], as described by [Jeuken, 2000]. See also the desciption given for the CO fluxes from the ocean.

The flux F = $k_{w} {\triangle}C$ , where ${\triangle}C$ is the concentration difference.

For 10 m wind speed, v, the Liss and Merlivat windspeed dependency of transfer velocity is used, including the proposed correction depending on the Schmidt number:

• Smooth surface regime, $0 \leq v \leq 3.6$ m/s : C=0.17v; k$_{w}$=C (Sc/600)$^{-2/3}$
• Rough surface regime, $3.6 m/s < v \leq 13$ m/s : C=2.85v-9.65 ; k$_w$=C (Sc/600.)$^{-0.5}$
• Breaking wave (bubble) regime, $v > 13$ m/s : C=5.9v-49.3 k$_{w}$=C (Sc/600.)$^{-0.5}$

Here K$_{w}$ is in cm/h and v in m/s.

The Schmidt number is calculated as:

Sc=3652.047271-246.99T+8.536397T$^{2}$-0.124397T$^{3}$ The ocean surface temperature is approximated by air surface temperature. It is assumed that if the temperature is less than -20 C sea ice prevents DMS emissions. A maximum ocean temperature of 28 C is imposed. We could use SST and sea ice cover here. The above is inconsistent with the treatment of CO fluxes from the ocean, which uses another formulation!

Depending on the TM resolution a scaling factor of 1, 1.32 (jm=24) or 1.065 (jm=48) is also applied in TM3. Note the dependency on specific values of jm, 24 and 48.

A review of different DMS ocean flux data sets can be found on the GEIA website
http://www.geiacenter.org/reviews/dimethsulfide.html
The different emission data sets were also evaluated in a model by [Boucher et al., 2003]. They suggest that [Nightingale et al., 2000] is a better choice for the seawater-air flux calculation.

/em In TM5 it is suggested to use the 2m wind for the calculation, we should rather use the 10 m wind!

Land

DMS emissions over land (kg/month) from [Spiro et al., 1992] are read in from DMSland.d and pre-processed by program rspiro following [Jeuken, 2000]. DMSland.d contains the monthly mean emission of DMS and H2S in kg(S)/month. Interpolation by rspiro is done by calculating the fractional overlap of the areas of the cells of the source and target grid. The original data files are

File name	Description
inv.sol.dms	DMS emission by soils
inv.veg.dms	DMS emission by vegetation
inv.sol.h2s	H2S emission by soils
inv.veg.h2s	H2S emission by vegetation

These files seem to originate from Harvard ftp io.harvard.edu. However, this site seems closed down.

Natural Isoprene emissions

GEIA VOC emissions from the biosphere [Guenther and et al., 1995] are used for isoprene and read in from fisop.d from GEIA [Houweling et al., 1998] valid for 1990. They have been reduced by 100 Tg(C)/yr compared with the GEIA recommendations as described in [Houweling et al., 1998]. fisop.d is a calculated from the original file isop90mn1.1a by program geia2tm3 called from script non_annual. This original file is available from the GEIA website at:
http://www.geiacenter.org/emits/nvoc.html

The total natural NMHC emission in subroutine sources_sinks of TM3, hcisopyr, is 400 Tg C. This accounts for 220 Tg isoprene, 130 Tg terpenes and and 50 Tg acetone.

Natural VOC emissions

Natural VOC emisisons from [Guenther and et al., 1995] and available at the GEIA website at:
http://www.geiacenter.org/emits/nvoc.html (files: nvoc90mn1.1a and terp90mn1.1a. However, I do not know if these have been used by Houweling to construct input files, e.g. for foth.d. See also isoprene above.

Natural ammonia emissions

Emissions from soils under natural vegetation

The TM input file is NATNH3.d. They are regridded from nh3nat.1x1 by program interp_nh3. nh3nat.1x1 is the official EDGAR2-GEIA emission data described in [Bouwman et al., 1997]. Note that there is also a NAT category in EDGAR2 which are emissions from microbes in the soil. NH3NAT is missing in the NH3 directory of EDGAR2 original data files for TM3 (no historical timeline), this category is only used for NO$_x$ in EDGAR2. The EDGAR2-GEIA natural NH3 emissions are available from
http://arch.rivm.nl/env/int/coredata/geia/index.html (click on NH3) or by anonymous ftp to
ftp://info.rivm.nl (directory pub/lae/EDGARV20/DATA/details/NH3). It should contain 2.4 Tg N/yr.

Ocean emissions

Emissions from the ocean are taken care of in the module dry_deposition of TM3. TM input file NH3_conc.d is read in subroutine inisurf. The program to make this file at the right TM resolution seems absent. Note that natural oceanic emissions of NH$_3$ are also available from EDGAR-GEIA,
http://arch.rivm.nl/env/int/coredata/geia/index.html.

Production of NO$_{x}$ by lightning

This is simulated directly in the TM3 model following the parameterisation described in [Meijer et al., 2001]. We do not use an emission data set because the source strongly depends on actual meteorology. The total source is set to approximately 5 Tg(N)/yr (depends slightly on the simulate year of meteorology). The geographical distribution is proportional to convective precipitation from the ECMWF model. Marine lightning is assumed to be 9 times less active as lightning over land. The fraction of cloud-to-ground over total flashes is determined by a 4th order polynomial fit of the cold cloud thickness, D, [Price and Rind, 1993]:
cloud-to-ground=total/( $0.021 D^{4}-0.648 D^{3}+7.493 D^{2}-36.54 D+64.09)$
intracloud = 1-cloud-to-ground
The NOx production for intracloud flashes is ten times less than that for cloud-to-ground flashes, according to [Price et al., 1997], who assume that intracloud flashes dissipate one tenth of the energy of cloud-to-ground flashes. Recent insights contradict this, see for example [Zhang et al., 2003].

The vertical distribution of the lightning source was chosen to closely follow the profiles from [Pickering et al., 1998]: The intracloud lightning NO$_{x}$ and 70 percent of the cloud-to-ground lightning NO$_{x}$ are distributed between T=-15 degrees C and the cloud top proportionally to air density. 10 percent of the cloud-to-ground lightning NO$_{x}$ is distributed between the ground and the T=-15 degrees C level and 20 percent of the cloud-to-ground lightning NO$_{x}$ is distributed between the ground pressure and 0.8 times the ground pressure (about 1000-800 hPa), both again proportionally to air density.

Lightning at polar latitudes is prohibited (j $\geq$ jm/8 or j $\leq 1+7jm/8$). The cloud top has to be higher than 5 km in geopotential height.

The lightning source is calculated in subroutine noxlight_cvp of module sources_sinks.