New Measurement and Mapping of SOC in Australia supports national carbon accounting and monitoring

R.A. Viscarra Rosselwith contributions from C. Chen, P. Campbell, C. Lobsey, C. Sharman, J. Baldock

GSP-FAO, Rome. 21–23 March 2017

Agriculture

FullCAM

Forest

Land cover change

Climate

Agricultural management

Crop yields

Forest management

Forest growth

Estimates of change in C stock and GHG emissions

Soil

The Australia’s GHG inventory: pre-2014

Richards (2001)Richards and Evans (2004)

Inventory2000–2013

CS /tha-1CS t/ha

Baseline map of soil organic C stock 0–30 cm

Viscarra Rossel et al. 2014 GCB

Spatial modelling and mapping

•  SCaRP: 2009–2012•  vis–NIR predictions

(Viscarra Rossel & Webster, 2012)

•  Cs = (C x ρ ) x (1–g)•  Harmonised 0–30 cm

What did this map say about soil C in Australia?•  The average stock in the 0–30

cm is 29.7 t/ha (22.6–37.9 t/ha)(cf. global average of 40–72 t/ha)

•  Cropping soil has on average 35 t/ha (30–42 t/ha). The total stock is 0.90 Gt (0.75–1.1 Gt).

•  Agricultural land (inc. grazing land) holds ~ 51% of the total C stock, i.e. 12.8 Gt (9.9–16.0 Gt)

•  Even small increases in the soil organic C stock across the vast area of agricultural land could sequester a significant amount of organic C and thereby play an important role in the 4 per mille initiative.

Maps soil organic C composition 0–30 cm

•  Physical fractionation & NMR ~500 samples (red points) (Baldock et al., 2014)

•  vis–NIR prediction (grey points) (Viscarra Rossel & Hicks, 2015)

Inventory2000–2013

SCF /t ha-1

Viscarra Rossel et al. 2017 in prep.

Particulate 2–0.05 mm

Humus <0.05 mm

Resistant <2 mm

Decom

posableN

utrient poorR

esistantN

utrient rich

Australia-wide 90 x 90 m 3D soil attribute maps in 6 layers

Viscarra Rossel, et al. (2015)

Data downloadable via:

• CSIRO data access portal• Google Earth Engine Project, data and methods described in Special Issue of Soil Research Vol. 53

Inventory1950–2014•  Site data•  vis–NIR

predictionsViscarra Rossel & Webster (2012)

Agriculture

FullCAM

Forest

Land cover change

Climate

Agricultural management

Crop yields

Forest management

Forest growth

Estimates of change in C stock and GHG emissions

Soil

Informing Australia’s GHG inventory: post-2014

Total net emissions2016

523.1 Mt CO2-eAgriculture

72.4 Mt CO2-e

Crediting of GHG abatement in the land sectorThe Carbon Credits Act 2011 enables crediting of GHG abatement in the land sector.

Emissions Reduction Fund (ERF) – a voluntary offsets scheme provides opportunity for farmers to earn income by reducing GHG emissions and/or storing C in soil and vegetation though changes to agricultural land management.

Legislated methods to quantify changes in soil Cunder the Australian Emissions Reduction Fund

Based on measurementsSequestering carbon

in soil in grazing systems

ΔC t/ha yr-1 = (Ctn – Ct0)/(tn – t0)

Based on modellingSequestering carbon in soil using

modelled abatement estimates

ΔC t/ha yr-1 = ΔCGain – ΔCLoss

•  Uses geo-stratification of CEA and random sampling with compositing across strata (Chappell et al., 2014) and laboratory analysis of composite samples

•  Cost for sampling and lab measurements but provides more confidence

•  Simulates changes using default values for different management actions derived from FullCAM.

•  No sampling or laboratory analysis, small cost but also less confidence in the magnitude of change – conservative estimates

TWG updating the measurement based method to enable: •  Use of prior information for stratification in the sampling design

•  Use of sensors (visible–infrared; gamma-ray attenuation) to measure organic C stocks (England & Viscarra Rossel, 2016; Viscarra Rossel, 2017)

Method being developed to enable future updates as technologies develop

Based on measurements

Sequestering carbon in soil in grazing systems

ΔC t/ha yr-1 = (Ctn – Ct0)/(tn – t0)

Based on modelling

Sequestering carbon in soil using modelled abatement estimates

ΔC t/ha yr-1 = ΔCGain – ΔCLoss

Legislated methods to monitor changes in soil Cunder the Australian Emissions Reduction Fund

Sensing of soil organic C stocks

•  Visible and thermal cameras•  vis–NIR spectrometer•  Gamma attenuation sensor

(Lobsey & Viscarra Rossel, 2016 - EJSS)

•  Embedded computing•  3G communication

•  Can measure 1.5 m soil core •  Measurements at z cm

intervals. 1 m soil core with z = 2.5 cm measured in around 15–20 min, i.e. ~ 35 s per measurement

Automated field deployable soil core sensing platform

Viscarra Rossel et al. (2017) submitted

Rapid, accurate measures of soil C stockSensor measurements on 150 soil cores from three CEAs in a 300 ha farm in NSW

R2 = 0.84 R2 = 0.76 R2 = 0.74 R2 = 0.75 R2 = 0.80 R2 = 0.76

Spectroscopic modelling using RS-LOCAL method (Lobsey & Viscarra Rossel, 2017)

Estimating the stocks of C and 3D mapping

Depthcm

Meant/ha

S.E.t/ha

0–10 8.05 0.23

10–20 7.12 0.31

20–30 4.85 0.25

30–40 4.03 0.19

40–50 3.64 0.16

50–60 3.63 0.17

60–70 3.55 0.14

70–80 3.32 0.11

Design-based estimates of the mean organic C stock for discrete depth layers

Mean 20.0 S.E. 0.6

Mean 18.6 S.E. 0.7

Estimating the stocks of C and 3D mapping

Design-based estimates of the mean organic C stock for discrete depth layers

Depthcm

Meant/ha

S.E.t/ha

0–10 8.05 0.23

10–20 7.12 0.31

20–30 4.85 0.25

30–40 4.03 0.19

40–50 3.64 0.16

50–60 3.63 0.17

60–70 3.55 0.14

70–80 3.32 0.11

Design-based estimates of the mean organic C stock for discrete depth layers

Mean 20.0 S.E. 0.6

Mean 18.6 S.E. 0.7

Final remarksWe have many of the ‘pieces’ needed for a statistically sound soil organic C monitoring system: putting it together is our current aim.

Data-modelassimilation

Best prediction of soil C + s at time t

Best parameter estimates + s at time t

Carbon Model

Temporal downscaled

updates

Soil properties + s

Land use & management…

Climate + s

Spatialmodelling &

estimation + sProximal

sensing + s

Remotesensing + s

ts

Analytical data + s

Spatially explicit data Model Assimilation Prediction

Final remarksWe have many of the ‘pieces’ needed for a statistically sound soil organic C monitoring system: putting it together is our current aim.

Data-modelassimilation

Best prediction of soil C + s at time t

Best parameter estimates + s at time t

Carbon Model

Temporal downscaled

updates

Soil properties + s

Land use & management…

Climate + s

Spatialmodelling &

estimation + sProximal

sensing + s

Remotesensing + s

ts

Analytical data + s

Spatially explicit data Model Assimilation Prediction

Final remarksWe have many of the ‘pieces’ needed for a statistically sound soil organic C monitoring system: putting it together is our current aim.

Data-modelassimilation

Best prediction of soil C + s at time t

Best parameter estimates + s at time t

Carbon Model

Temporal downscaled

updates

Soil properties + s

Land use & management…

Climate + s

Spatialmodelling &

estimation + sProximal

sensing + s

Remotesensing + s

ts

Analytical data + s

Spatially explicit data Model Assimilation Prediction

Final remarksWe have many of the ‘pieces’ needed for a statistically sound soil organic C monitoring system: putting it together is our current aim.

Data-modelassimilation

Best prediction of soil C + s at time t

Best parameter estimates + s at time t

Carbon Model

Temporal downscaled

updates

Soil properties + s

Land use & management…

Climate + s

Spatialmodelling &

estimation + sProximal

sensing + s

Remotesensing + s

ts

Analytical data + s

Spatially explicit data Model Assimilation Prediction

Thank you.

For more information on any of the content please come see me or contact me:

Raphael VISCARRA [email protected]+61 467 769 364