1

KK--StateState

Comparison of Soil Carbon

Measuring Techniques

Comparison of Soil Carbon Comparison of Soil Carbon

Measuring Techniques Measuring Techniques

R.C. Izaurralde, M.H. Ebinger, J.B. Reeves, C.W. Rice,L. Wielopolski, B.A. Francis, R.D. Harris, S. Mitra,

A.M. Thomson, J.D. Etchevers, K.D. Sayre, A. Rappaport, and B. Govaerts

Agriculture’s Role in the New Carbon EconomyCASMGS Forum

Manhattan, KS17-18 December, 2007

2

IntroductionIntroductionIntroduction

Changes in soil C stocks can be

measured directly through soil

sampling or estimated using eddy

covariance, stratified accounting,

or simulation models

Steps for measuring soil C include

soil sampling, sample preparation,

measurement by dry combustion,

and calculation of results on a soil-

mass basis

However, there is a need to

develop fast and accurate

procedures to measure soil C

changes at the field scale

Root C

Litter

C

Eroded C

Cropland C

Wetland C

Eddy flux

Sample

probe

Soil profile

Remote

sensor

Respired C

Captured C

Heavy

fraction

C

Woodlot C

Harvested C

Buried C

Light

fraction

C

Respired C

Soil organic C

Soil inorganic C

Simulation modelsDatabases / GIS

SOCt = SOC0 + Cc + Cb - Ch - Cr - Ce

Post et al. (2001) Climatic Change 51:73-99

2

3

ObjectivesObjectivesObjectives

Discuss issues related to the detection of soil C changes as a result of

changes in agricultural management

Compare the performance of three advanced technologies in their ability to

measure soil C under field conditions and use in developing countries:

� Laser Induced Breakdown Spectroscopy: LIBS

� Inelastic Breakdown Spectroscopy: INS

� Mid-Infrared Reflectance Spectroscopy: MIRS

4

Soil survey maps can be used to estimate the spatial Soil survey maps can be used to estimate the spatial

distribution of soil organic C stocksdistribution of soil organic C stocks

Long term experiments have been essential tools Long term experiments have been essential tools

to understand the temporal dynamics of soil Cto understand the temporal dynamics of soil C

The challenge consists in developing costThe challenge consists in developing cost--effective methods effective methods

for detecting changes in soil organic C that occur in fields as for detecting changes in soil organic C that occur in fields as

a result of changes in managementa result of changes in management

Measuring and monitoring soil C sequestration: Measuring and monitoring soil C sequestration:

a new challenge?a new challenge?

3

5

On baselines, soil C sequestration

and avoided C loss

On baselines, soil C sequestration On baselines, soil C sequestration

and avoided C lossand avoided C loss

0

20000

40000

60000

80000

100000

0 10 20 30 40 50 60 70 80 90 100 110 120 130

Years

C (kg ha-1)

Management ChangeSoil A

Soil B

McGill et al. (1996)

0

5000

10000

15000

20000

25000

30000

0 10 20 30 40 50 60 70 80 90 100 110 120 130

Years

C (kg / ha)

Management Change

Soils with different C stocks may Soils with different C stocks may

respond differently to changes in respond differently to changes in

managementmanagement

Time 0 baseline in a soil gaining C Time 0 baseline in a soil gaining C

may lead to an overestimation of soil may lead to an overestimation of soil

C sequestrationC sequestration

6

Soil organic matter affects soil bulk density and thus

temporal comparisons of soil C stocks

Soil organic matter affects soil bulk density and thus Soil organic matter affects soil bulk density and thus

temporal comparisons of soil C stockstemporal comparisons of soil C stocks

Ellert et al. (2002)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0

Soil organic C (g kg-1 x 10-1)

Soil bulk density (Mg m

-3)

Soil bulk density is a function of the Soil bulk density is a function of the

soil mineral density and the soil soil mineral density and the soil

organic matter contentorganic matter content

Comparisons of soil C stocks across Comparisons of soil C stocks across

treatments should be done using the treatments should be done using the

equivalent soil mass methodequivalent soil mass method

m

bOMOM

ρ

ρ%100

244.0

%

100

−+

=

4

7

Spatial variability influences estimates of soil

organic C

Spatial variability influences estimates of soil Spatial variability influences estimates of soil

organic Corganic C

100 samples to detect 2-3% change in

SOC

16 samples to detect 10-15 % change

in SOC

Garten and Wullschleger (1999)

8

Sampling protocol used in the Prairie Soil

Carbon Balance (PSCB) project

Sampling protocol used in the Prairie Soil Sampling protocol used in the Prairie Soil

Carbon Balance (PSCB) projectCarbon Balance (PSCB) project

Use “microsites” (4 x 7 m) to reduce spatial variabilityThree to six microsites per fieldCalculate SOC storage on an equivalent mass basisAnalyze samples taken at different times togetherSoil C changes detected in 3 yr� 0.71 Mg C ha-1 – semiarid� 1.25 Mg C ha-1 – subhumid

2 m

5 m

initial cores (yr 1997)

subsequent cores (yr 2002)

initial cores (yr 1997) with buried marker (electromagnetic)

N

Ellert et al. (2001)

5

9

Random

Grid

Mean SD CV n

Grid 1 48.4 3.9 8.0 25

Grid 2 47.8 3.5 7.4 25

Grid 3 47.8 2.7 5.7 25

Average 48.0 3.4 7.0 75

Comp 1 48.5 1

Comp 2 44.7 1

Comp 3 46.7 1

Average 46.6 1.9 4.1 3

Grid and sampling schemes demonstrate that soil C Grid and sampling schemes demonstrate that soil C

stocks can be determined with low errorsstocks can be determined with low errors

Source: G. Watson and C. Rice

Soil C (Mg C haSoil C (Mg C ha--11 to 30to 30--cm soil depth)cm soil depth)

10

Determination of soil organic C concentration:

standard methods

Determination of soil organic C concentration: Determination of soil organic C concentration:

standard methodsstandard methods

Wet combustion� Soil sample treated with acid dichromate

solution� CO2 generated evaluated with titrimetric or

gravimetric methods� Recovery is incomplete (avg. 81%)

Dry combustion� High temperature (1000 – 1500 C)� CO2 generated assessed with

spectrophotometric, volumetric, titrimetric, gravimetric, or conductimetric techniques

� Very accurate, minimal variability, low operational errors

� Corrections needed when samples contain carbonates

National and international efforts needed to cross-calibrate methods against standard (soil) samples

y = 1.0216x + 0.0342

R2 = 0.9453**, n = 171

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0.0 0.5 1.0 1.5 2.0 2.5 3.0

Instrument 1 (g C kg-1 x 10-1)

Instrument 2 (g C kg

-1 x 10-1)

Izaurralde and Rice (2006)

Total soil C as measured by two dry

combustion instruments

6

11

Laser Induced Breakdown Spectroscopy: LIBSLaser Induced Breakdown Spectroscopy: LIBSLaser Induced Breakdown Spectroscopy: LIBS

Based on atomic emission spectroscopy

Portable

A laser pulse is focused on a soil sample, creating high temperatures and electric fields that break all chemical bonds and generating a white-hot plasma

The spectrum generated contains atomic emission peaks at wavelengths characteristic of the sample’s constituent elements

A calibration curve is required to predict soil C concentration Cremers et al. (2001) J. Environ. Qual. 30:2202-2206

12

Inelastic Neutron Scattering: INSInelastic Neutron Scattering: INSInelastic Neutron Scattering: INS

In situ, non-destructive technique

that consists in directing fast

neutrons (14 MeV) produced by a

neutron generator into the soil,

where they interact with the nuclei

of atoms including 12C and other

atoms (H, N, O, Si, K, Ca, P, etc.)

Fast neutrons collide with C, H,

and N atoms and release gamma

rays with energies of 4.4, 2.2, and

10.3 MeV

Soil mass interrogated: >200 kg

The INS was tested in stationary

and scanning modesWielopolski et al. (2000) IEEE Trans. Nuclear Sci. 47:914-917

7

13

The INS in action…The INS in actionThe INS in action……

14

Mid-Infrared Reflectance Spectroscopy: MIRSMidMid--Infrared Reflectance Spectroscopy: MIRSInfrared Reflectance Spectroscopy: MIRS

Unlike LIBS and INS, MIRS probes

the bond identities of a sample's

molecules, offering the possibility

of directly distinguishing inorganic

from organic C, thus eliminating

the need for acid pretreatment to

remove inorganic C

Yet for the same reason,

quantifying soil C must be done

indirectly, by recourse of advanced

data-fitting routines that require

libraries of soil spectra vs. soil C

dataMcCarty et al. (2002) Soil Sci. Soc. Am. J. 66:640-646

8

15

First Field Test: Beltsville, MD; October 2006First Field Test: Beltsville, MD; October 2006First Field Test: Beltsville, MD; October 2006

Conducted on a 25-ha field known as OPE3 (Optimizing Production Inputs for Economic and Environmental Enhancement)

Three 30 m x 30 m plots containing 9 sampling points were sampled at three depth intervals (0-5, 5-15, 15-30 cm)

Soil samples were processed in the field for LIBS and MIRS analysis

The INS instrument estimated soil C density via soil scanning

All samples were analyzed for C content at Kansas State Univ. by dry combustion and the results reported as soil C density using Db determined by the soil core method

The dry-combustion team provided selected C values to the other teams for calibration and then collected the C estimates for Plot No. 3 (R3) from the other teams for comparison

16

Mean soil C density (g C cm-2) to a depth of 30 cm

in Plot No. 3 of OPE3 field

Mean soil C density (g C cmMean soil C density (g C cm--22) to a depth of 30 cm ) to a depth of 30 cm

in Plot No. 3 of OPE3 fieldin Plot No. 3 of OPE3 field

A subset of C concentration values

determined by dry combustion (DC)

was provided to all teams

Soil C density estimates for Plot No. 3

are the results of a blind test

MIRS produced the closest estimates

of soil C density but required the

largest amount of information

LIBS estimates could be improved by

including more data points into the

universal calibration curve

INS estimates should be further

explored with regards to uncertainties:

� Calibration procedures

� Variability of soil C density

+6-37-20%Diff

9-99n

0.0610.0520.0810.055σ

0.4320.2570.3270.407µ

MIRSINSLIBSDC

9

17

Re-visiting Plot No. 3 of OPE3 with INS to

estimate soil C density (g C cm-2)

ReRe--visiting Plot No. 3 of OPE3 with INS to visiting Plot No. 3 of OPE3 with INS to

estimate soil C density (g C cmestimate soil C density (g C cm--22))

The two repeated INS measurements gave similar values (µINS = 0.257 g C cm

-2) but the mean value was different from that determined by DC (µDC = 0.407 g C cm-2)

Two hypotheses are possible:

� The true mean soil C density of the field is lower than predicted from a finite number of grid points

� The INS calibration was based on too few points; thus, more calibration points are needed to improve the prediction of soil C density of the Plot No. 3 of the OPE3 field

0.4070.2620.392Visit II

0.4070.2520.388Visit I

Dry

Comb.

ScanningStatic

Meas.

Wielopolski et al. J. Environ. Qual. (accepted for public.)

18

Second Field Test: CIMMYT, Mexico; April 2007Second Field Test: CIMMYT, Mexico; April 2007Second Field Test: CIMMYT, Mexico; April 2007

This test did not include the INS instrument

Conducted at CIMMYT on a 17-year old crop rotation, tillage, residue study

Treatments sampled:

� Maize (m) and wheat (w) grown in monoculture (M) or in rotation (R)

� Grown with conventional (CT) or no tillage (ZT), and with (+) or without (-) removal of crop residues

� Each treatment is replicated twice

A composite soil sample made of 12 subsamples per soil depth (0-5, 5-10, and 10-20 cm) was taken from each of the 22 x 7.5 m plots

Soil samples were processed and analyzed as in the Beltsville test.

General view of plotsGeneral view of plotsGeneral view of plots

No Till w/o residuesNo Till w/o residuesNo Till w/o residues

No Till with residuesNo Till with residuesNo Till with residues

10

19

Mean soil C density (kg C m-2) by treatment and

summary statistics in the CIMMYT experiment

Mean soil C density (kg C mMean soil C density (kg C m--22) by treatment and ) by treatment and

summary statistics in the CIMMYT experimentsummary statistics in the CIMMYT experiment

0.00

0.50

1.00

1.50

2.00

2.50

3.00

3.50

4.00

4.50

5.00

ZTMw+r

ZTMw-r

ZTMm+r

ZTMm-r

CTRw-m+r

CTRw-m-r

ZTRw-m+r

ZTRw-m-r

CTMw+r

CTMw-r

CTMm+r

CTMm-r

ZTRm-w+r

ZTRm-w-r

CTRm-w+r

CTRm-w-r

Soil C density (kg C m

-2)

DryComb

LIBS

MIRS

112112112n

0.6251.7001.500Range

1.1660.6000.814Min

1.7912.3002.315Max

0.1340.3930.301σ

1.4131.4401.306µ

MIRSLIBSDC

Although LIBS and MIRS followed the C density trends detected by DC method

Correlation between methods was low� LIBS vs. DC: R2 = 0.174

� MIRS vs. DC: R2 = 0.329

20

Further calibration of LIBS and re-estimation of

CIMMYT data

Further calibration of LIBS and reFurther calibration of LIBS and re--estimation of estimation of

CIMMYT dataCIMMYT data

Partial Least Squares method was

used to improve calibration curves

A calibration curve was developed

using 31 samples run 3 times each

(1 missing value)

Re-estimation of data points

improved significantly (see graph

on the right)

Software issues need to be

addressed by Australian

developers

New software (The Unscrambler),

is being tested

y = 1.003x

R2 = 0.919

0.5

1.0

1.5

2.0

2.5

0.5 1.0 1.5 2.0 2.5

LIBS

DC

11

21

Further calibration of MIRS and re-estimation of

CIMMYT data

Further calibration of MIRS and reFurther calibration of MIRS and re--estimation of estimation of

CIMMYT dataCIMMYT data

Original estimation of CIMMYT data using MIRS was developed with the calibration curve based on OPE3 samples and 8 samples from Mexico

Eleven samples from the set of 112 were added to the calibration curve

Prediction of the remainder 101 points improved significantly with the revised calibration curve that used the OPE3 data points plus the 19 Mexican data points

With the MIRS method, the greatest difficulty in predicting the correct values seems to be associated with high C samples

A calibration using only the 112 samples had high R2 (~0.95) and revealed nothing wrong with the DC data� Recently, the Colegio de Postgraduados

(CP) in Texcoco (Mexico) ran dry combustion in all of the soil samples

� The R2 between CP and KSU data was >0.9

y = 0.7x + 0.4

R2 = 0.8

0.5

1.0

1.5

2.0

2.5

0.5 1.0 1.5 2.0 2.5

MIRS

DC

22

SummarySummarySummary

This study compared for the first time the side-by-side performance of three advanced technologies to measure soil C under field conditions: LIBS, INS, and MIRS

The LIBS and MIRS results compare directly with those obtained by Dry Comb. These methods require soil sampling and need soil bulk density information to convert soil C concentrations to soil C density

The INS instrument interrogates large volumes of soil to generate mean soil C values for the site measured or field scanned. The INS measurements require calibration with mean values obtained from soil sampling measurements. Comparison between INS and discrete soil sampling measurements requires further study

The results obtained indicate acceptable performances of the advanced instruments but they also show the need for improvement in terms of calibration

The three instruments demonstrated their portability and their capacity to perform under field conditions

Import / export issues to developing countries should be examined in order to facilitate the transport of these instruments across international borders