TERN Ecosystem Surveillance Plots Kakadu National Park
Vanessa Haverd_Multiple observation types reduce uncertainty in Australia's terrestrial carbon and water cycles
1. The Australian Terrestrial Carbon
Budget
V. Haverd, M.R. Raupach, P.R. Briggs, J.G. Canadell, S. J.
Davis, P. Isaac, R. M. Law, C. P. Meyer, G. P . Peters, C.
Pickett-Heaps, S. Roxburgh, B. Sherman, E. van Gorsel ,
R. Viscarra Rossel, Z. Wang
Vanessa Haverd | Research Scientist
14 February 2012
CMAR
Acknowledgements: TERN OzFlux and ACCSP
2. Introduction Australian Continent
First attempt at full carbon budget (C-CO2) of
• 1990-2011 period
Contribution to RECCAP (REgional Carbon Cycle Assessment and
Processes) project (Canadell et al., 2011)
Places contributions from the terrestrial biosphere and
anthropogenic emissions along side each other
Territorial C = Biospheric C + Fossil Fuel C + Harvested Wood Products C
−dCT / dt = − dCB / dt − dCFF / dt − dCHWP / dt
= ( FNPP + FRH + FFire + FLUC + FTransport +FHarvest )
+ ( FFF + FFF , Export ) − dCHWP / dt
6. Net Primary Production (NPP) and Net Ecosystem
Production (NEP = NPP – Heterotrophic Respiration)
Obtained using BIOS2 = CABLE-SLI-CASAcnp
in AWAP operational framework
SLI = Soil-Litter- CASAcnp = AWAP = Australian Water
CABLE = Community Availability Project
Atmosphere-Biosphere-Land Iso Biogeochemical
Exchange model model Continental processing
Soil hydrology,
soil evaporation Soil and plant framework
Water, energy, carbon fluxes
C, N, P dynamics Met and soil data
Wang et al. (2011) Haverd et al.
(2010) Wang et al. (2010) Model-Data Fusion
Raupach et al. (2009)
Non-zero NEP results from climate variability
and rising CO2
Disturbance effects (particularly fire and LUC)
accounted for by other methods
18. Effects of variable climate and rising CO2
on mean (1990-2011) NPP and NEP
Net
Net Primary
Ecosystem
Production
Production
19. Fire: Comparison of GFED3 and NGGI CO2-C
Gross Fire Emissions estimates. (i) Annual fluxes
per state per vegetation class; (ii) Total annual
Australian continental fire emissions 1997-2009.
20.
21. Gross Fire emissions of C-CO2 by
region
• Tropics and savanna account for 80% of
the total emissions
• South-eastern temperate region
contributes significantly in years with
extreme fire events such as 2003 (28 %)
and 2006 (15 %).
• Total gross fire emissions (127 Tg C y-1)
are comparable to Australian territorial
emissions from the burning of fossil fuels
(96 Tg Cy-1)
• The net C-CO2 emissions from biomass
burning are much smaller (26 TgCy-1)
23. Net effect of emissions from traded
fossil fuels (2004)
24. Key Findings
•Australia’s NBP of 36 ± 35 TgC y-1 offsets fossil fuel emissions
(95 ± 6 TgC y-1) by 38 ± 36 %.
•The interannual variability in NEP and hence NBP exceeds
Australia's total carbon emissions by fossil fuel consumption
•Gross fire emissions account for 6% of continental NPP,
approximately the same as the 1σ interannual variability in NPP
•Net fire emissions account for only 1 % of NPP.
•Land use change emissions similar to net fire emissions,
accounting for 1% of NPP.
•Fossil fuel export ~1.5 times territorial emissions (1990-2011).
•Fossil fuel export ~2.5 times territorial emissions (2009-2010).
25. Vanessa Haverd
Thank you
Research Scientist
t +61 2 6246 5981
e vanessa.haverd@csiro.au
CMAR
26. BIOS2 results: water balance and net
1500
primary production Precip (mm y-1) i 1500
ET (mm y )
-1
ii
Parameter + forcing error
Parameter error
1000 1000 Best estimate
Forcing error
Best estimate
500 500
0 0
Soil evaporation fraction iii Runoff / precip fraction iv
1.0
0.4
0.5
0.2
0.0 0.0
NPP (g m-2 d-1) v NPP recurrent fraction vi
1.0
2
0.5
1
0 0.0
Global
Global
Desert
Desert
Tropics
Tropics
Mediterr
Mediterr
Australia
Australia
Savanna
Savanna
Cool Temp
Cool Temp
Warm Temp
Warm Temp
1 2 3 4 5 6 A 1 2 3 4 5 6 A
31. NPP and ET: NPP (g m-2 d-1)
comparing with 3 i
3pg
AussieGrass
BiosEquil
other estimates 2
Century
CenW
dLdP
Miami-oz
12 mean NPP Miami
1 Olson
estimates for Australia RFBN
from (Roxburgh et al TMS
Vast
0
2004) ET (mm y-1)
ii AWAP
1000 AWRA
Guerschman
NDTI
etlook
7 mean ET 500 MODIS
estimates for
Australia
0
(King et al 2012)
Desert
Tropics
Mediterr
Australia
Savanna
Cool Temp
Warm Temp
1 2 3 4 5 6 A
32. Multiple constraints on Australian
terrestrial Net Primary Production : Eddy
flux data provide the tightest constraint
error bars = uncertainty from propagated parameter uncertatinties (1σ)
Prior estimate
Eddy fluxes
Streamflow
Litterfall
Eddy fluxes + Litterfall
Streamflow + Litterfall
Streamflow + Eddy fluxes
Eddy fluxes + Litterfall + Streamflow
0 1 2 3 4
NPP (GtC y-1)
33. BIOS2 evaluation: Gross Primary
Production from 12 OzFlux sites
Map: long term
mean (BIOS2)
Boxes: annual
cycle (Ozflux sites)
34. Flux components of Net Biosphere
Production
(i) Fluxes out of Aust. (ii) Process contributions (iii) Fluxes into
biosphere to accumulation of atmosphere from Aust.
biospheric C territory
35. Flux components of Land-Atmosphere-
Exchange
(i) Fluxes out of Aust. (ii) Process contributions (iii) Fluxes into
biosphere to accumulation of atmosphere from Aust.
biospheric C territory
36. BiosEquil
Long-term NPP: 2
Century
CenW
comparing with
dLdP
Miami-oz
Miami
1
previous estimates Olson
RFBN
TMS
0 Vast
12 mean NPP NPP (g m-2 -1 -1)
d
ET (mm y ) 3pg
3 i
estimates for Australia ii AussieGrass
BiosEquil
AWAP
from (Roxburgh et al 1000 Century
AWRA
2 CenW
2004) Guerschman
dLdP
NDTI
Miami-oz
etlook
Miami
MODIS
500
1 Olson
RFBN
TMS
0 Vast
0
ET (mm y-1)
ii AWAP
1000
Desert
AWRA
Tropics
Mediterr
Australia
Savanna Guerschman
NDTI
Cool Temp
Warm Temp etlook
MODIS
500
0
Desert
Tropics
Mediter
Austra
Savan
Coo
Wa
37. Temperature
Precip NPP
NPP NEP
NEP
Rising CO2 Preindustrial CO2
Editor's Notes
The impact of each of three data sets (leaf –NPP (litter-fall), streamflow and eddy flux data) and combinations thereof on the long-term mean Australian continental NPP estimate and its uncertainty. Each data set individually leads to a reduction in uncertainty compared with the prior estimate, although with quite different values, reflecting possible biases in the model and/or observations for the particular observable. The estimates are more convergent when 2 data sets are used simultaneously, and the estimate constrained by all three is a compromise between the results obtained using each data set individually. The error bars in Figure 3 indicate that eddy flux data provide a stronger constraint than leaf-NPP, even though leaf NPP observations more widely distributed (Figure 2). This reflects the high precision of the eddy flux measurements, compared with disparate litterfall observations which do not share a common methodology and are subject to large errors from fine scale heterogeneity. Long-term evaporation from streamflow provides a relatively weak constraint because in most regions of Australia, it is largely driven by rainfall (continentally, evaporation accounts for 90% of precipitation).
BIOS2 paper, Figure 6: Continental long term GPP (map) and mean annual cycle (averaged over years of obs) of GPP’ (BIOS2 and obs, left axis) and LAI (right axis), at 12 OzFlux sites.
BIOS2 paper, Figure 6: Continental long term GPP (map) and mean annual cycle (averaged over years of obs) of GPP’ (BIOS2 and obs, left axis) and LAI (right axis), at 12 OzFlux sites.
Here, the top two panes of the previous slide (NPP and ET) are shown with all elements greyed out. Superimposed in colour on the NPP pane are the various Roxburgh results. Superimposed on the ET pane are most of the WIRADA ET estimates. The old CABLE results from WIRADA will be removed for publication. Also left out of the WIRADA results are those of Yongqiang, which have some problems. All ET results are for the period Jan 2000-Dec 2005, except etlook which is for 2002/07-2005/06.
The impact of each of three data sets (leaf –NPP (litter-fall), streamflow and eddy flux data) and combinations thereof on the long-term mean Australian continental NPP estimate and its uncertainty. Each data set individually leads to a reduction in uncertainty compared with the prior estimate, although with quite different values, reflecting possible biases in the model and/or observations for the particular observable. The estimates are more convergent when 2 data sets are used simultaneously, and the estimate constrained by all three is a compromise between the results obtained using each data set individually. The error bars in Figure 3 indicate that eddy flux data provide a stronger constraint than leaf-NPP, even though leaf NPP observations more widely distributed (Figure 2). This reflects the high precision of the eddy flux measurements, compared with disparate litterfall observations which do not share a common methodology and are subject to large errors from fine scale heterogeneity. Long-term evaporation from streamflow provides a relatively weak constraint because in most regions of Australia, it is largely driven by rainfall (continentally, evaporation accounts for 90% of precipitation).
BIOS2 paper, Figure 6: Continental long term GPP (map) and mean annual cycle (averaged over years of obs) of GPP’ (BIOS2 and obs, left axis) and LAI (right axis), at 12 OzFlux sites.
Here, the top two panes of the previous slide (NPP and ET) are shown with all elements greyed out. Superimposed in colour on the NPP pane are the various Roxburgh results. Superimposed on the ET pane are most of the WIRADA ET estimates. The old CABLE results from WIRADA will be removed for publication. Also left out of the WIRADA results are those of Yongqiang, which have some problems. All ET results are for the period Jan 2000-Dec 2005, except etlook which is for 2002/07-2005/06.