1 of 33

Biogenic emissions from �tropical ecosystems

Michael Barkley & Paul Palmer

University of Edinburgh

2 of 33

biogenic, pryogenic, anthropogenic

pryogenic

anthropogenic

biogenic

anthropogenic

pryogenic

pyrogenic

anthropogenic

biogenic

anthropogenic

Thomas Kurosu, Harvard-Smithsonian

HCHO August 2006 (Ozone Monitoring Instrument)

Isoprene is the main driver of variability in Ω_HCHO

(Palmer et al., JGR, 2003,2006)

3 of 33

Inverting HCHO columns for isoprene emissions

GOME slant columns (July 96)

So if we have convinced ourselves that isoprene is the main contributor to the formaldehyde column, how do we then derive the isoprene emissions from the satellite column measurements?

Well it is essentially a four step process:

The slant columns are determined from the satellite spectral measurements

The slant columns are then converted into vertical columns by dividing them by the air mass factor which is calculated using a radiative transfer model and the global chemistry transport model GEOS-CHEM

The air mass formulation is important as it accounts for scattering processes within the atmosphere both from clouds and aerosols, and also Rayleigh scattering, and importantly the vertical distribution of the formaldehyde with the vertical column.

It is the product of the geometrical air mass factor with the integral of the scattering weights (as calculated by the radiative transfer model) multiplied by the shape factor or the formaldehyde vertical profile from GEOS-Chem where the blue line here represents what GOME ‘sees’.

Typical errors on the AMF are about 30%

This process ensures consistency between model and satellite comparisons

Once the vertical columns have been obtained, we then determine a linear transfer function, using GEOS-CHEM, that relates the model formaldehyde columns, sampled along the satellites ground track, to the model isoprene emissions which are based on the MEGAN emission inventory data.

The intercept represents the formaldehyde background.

The slope is dependent on model yield of formaldehyde from isoprene and the loss rate constant for formaldehyde.

We can invert the transfer function and use the GOME, or OMI, formaldehyde columns to then estimate the isoprene emissions.

4 of 33

Instrument

sensitivity w(s)

(“scattering

weight”)

Vertical shape

factor S(s)

(normalized mixing ratio)

what

GOME

sees

Apply AMF

Inverting HCHO columns for isoprene emissions

GOME slant columns (July 96)

GOME vertical columns (July 96)

VCD = SCD / AMF

5 of 33

Ω_HCHO=SE_isop+B

GOME slant columns (July 96)

GOME vertical columns (July 96)

Apply AMF

Model Transfer Function

6 of 33

Ω_HCHO=SE_isop+B

GOME slant columns (July 96)

GOME vertical columns (July 96)

Apply AMF

E_isop~ (Ω_HCHO-B) / S

Model Transfer Function

7 of 33

Ω_HCHO=SE_isop+B

GOME slant columns (July 96)

GOME vertical columns (July 96)

GOME

isoprene

emission

inventory

Apply AMF

E_isop~ (Ω_HCHO-B) / S

Model Transfer Function

8 of 33

The tropics: GOME observations: Jan-June 1997

Plots smoothed with 3x3 box-car filter!

9 of 33

The tropics: GOME observations: July-Dec 1997

Plots smoothed with 3x3 box-car filter!

10 of 33

Significant pyrogenic HCHO source

Plots smoothed with 3x3 box-car filter!

Grey diamonds = fires detected by ATSR

11 of 33

Partition Amazon into West & East regions

Plots smoothed with 3x3 box-car filter!

12 of 33

East region: Contribution of wild fires

Biomass burning is the main source of HCHO

Maximum in dry season

13 of 33

Partition Amazon into West & East regions

Plots smoothed with 3x3 box-car filter!

14 of 33

Isolate West Amazonian + remove fires

Plots smoothed with 3x3 box-car filter!

15 of 33

5 yr time series over western Amazonian

If we look at the GOME data over the tropics, focussing over South America, during the wet season January to May we find there doesn’t seem to be a lot happening.

However, when we move into the dry season there is a significant increase in the formaldehyde slant columns

This is increase is mostly attributable to both biomass burning emissions of formaldehyde, which is assumed to be dominant, and an increase in leaf-level isoprene emission rates from the wet to dry seasons.

By over plotting the positions of wild fires as detected by the ATSR instrument, we can see that the formaldehyde enhancements, are spatially closely correlated with the biomass burning events.

The ATSR wild fire atlas is useful in that, if we over plot it onto the GOME data over several years, we find that we can identify two distinct regions. An eastern region where biomass burning events are very common and a western area in which wild fires are less frequent

Focussing on the eastern Amazonian, if we plot the time series of GOME monthly means over this region against the number of fire counts then a clear correlation exists indicating that biomass burning is the main source of the increase formaldehyde.

If we want to determine biogenic isoprene emissions we need to remove the biogenic component.

We can attempt this by isolating the western region and removing all grid boxes where fire occur.

When we now plot the formaldehyde time series over this region, we see notice that the columns are lower with a

16 of 33

5 yr time series over western Amazonian

Transport of biomass burning ?

Water availability

In situ isoprene data @ 2.8°S, 59.4 ° W (Trostdorf et al., ACPD, 2004)

Yr = 2002

Rainfall