Modeling Approaches for Understanding and Predicting Soil Carbon Sequestration:

Field to Landscape to Region

RC Izaurralde (JGCRI), K Paustian (CSU), JD Atwood (USDA), J Brenner (USDA), R Conant (CSU), M Easter (CSU), S Ogle (CSU), S Potter (TAMU), AM Thomson (JGCRI), JR Williams (TAMU)

Third Annual Conference on Carbon Sequestration4-6 May 2004 – Alexandria, VA

BackgroundSoil carbon sequestration (SCS) has significant potential to attenuate the increase of atmospheric CO2• Global: 0.4 – 0.8 Pg C y-1; 50 – 100 y (IPCC 1996)• USA: 0.08 – 0.21 Pg C y-1; 30 y (Lal et al. 1999)

Near-term SCS potential in croplands• Global: 0.12 Pg C y-1 by 2010; 0.26 Pg C y-1 by 2040 (Sampson et

al. 2000)Current estimates of SCS• USA: 0.021 Pg C y-1 during 1982 – 1997 (Eve et al. 2001)

The deployment of SCS practices will require robust methods for monitoring soil carbon changesSimulation models with ability to simulate soil carbon and nitrogen dynamics can play major role in monitoring SCS at field, landscape and regional scales

Detecting and scalingchanges in soil carbon

Detecting soil C changes• Difficult on short time scales• Amount changing small

compared to total CMethods for detecting and projecting soil C changes (Post et al. 2001)• Direct methods

– Field and laboratory measurements

– Eddy covariance• Indirect methods

– Accounting- Stratified accounting

– Remote sensing– Models

Root C

LitterC

Eroded C

Cropland C

Wetland C

Eddy flux

Sampleprobe

Soil profile

Remotesensor

Respired C

Captured C

HeavyfractionC

Woodlot C

Harvested C

Buried C

Lightfraction

C

Respired C

Soil organic C

Soil inorganic C

Simulation modelsDatabases / GIS

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

Post et al. (2001)

Objectives

Review components and drivers of the carbon balance in agroecosystemsDiscuss modeling approaches in Century and EPICPresent examples of applications of these models at various scales of spatial resolution• Field• Landscape• Region

Litterfall

Soil erosionSoil deposition

Harvest

Photosynthesis

Decomposition & respiration

Dissolved organic C

Annual Carbon Balance in an Agroecosystem

CO2

CO2

0.1 Mg C ha-1 ??

1 Mg C ha-1

3.5 Mg C ha-1

2 Mg C ha-12.5 Mg C ha-1

?Soil C: 60 Mg C ha-1 + 0.4 Mg C ha-1

Humification

Environmental Variables and Management DetermineCarbon Flux Between Soils and the Atmosphere

CO2

Soil C

Cropping Rotation Practice:Type of Crop, Use ofWinter Cover Crops,Hay in Rotation, Legumes, Cropping Intensification

Tillage Management:Conventional, Reduced or No-till

Residue Management Fertilizer Management

Land-Use Change

Soil Characteristics

Topography

Climate

Irrigation Management

Two terrestrial ecosystem models

Century• Century• DayCent• C-STORE

EPIC• EPIC• APEX

Soil ProcessesWater movement Erosion

Temp & Moisture

Density Changes

Above Gr. LiveAbove Gr. DeadBelow Gr. LiveBelow Gr. Dead

Harvest

Plant Growth

Leaching

Soil Properties, Management, Weather, CO2

PesticidesSurface residuesSubsoil residues

Humus

OrganicTransformations

CO2

NitrificationNH3 Volatilization

DenitrificationPi reactions

InorganicTransformations

NH3, N2O, N2

Processes and drivers

Residue C

Metabolic Litter Biomass C Passive C

Slow C Leached C

Carbon and nitrogen flows

Structural Litter

The C-Store® Soil Carbon ModelThe C-Store® Soil Carbon Model

Crop Yields

Tillage

Crop Growth Timing and Distribution

Weather

Land Use History

Soil Physical Properties

Residue and Manure Mgmt

C-Store® Soil C

C-STORE: Predicting soil C changes at

the field level

Select State and CountyDescribe Land Use History

Specify Soil Type

SpecifyTillageSystem

PastManagement System

(1900-2000)

SpecifyYields

A similar screen is used for2001-2020

Run the Model

Graphic output

Detailed tabular output

Measured vs. Modeled Soil Carbon StocksMeasured Initial C Values

y = 1.0293xR2 = 0.8976

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

modeled (g/m^2)

mea

sure

d (g

/m^2

)

all sitesAkron wf nt 1Arlington LTN1 CCC 0 CTBushland wgf NT 1elansing ccc ctHoytville 1 CB CTIndian Head F-(SW)-SW 1 F-SW-SW NP CTlamberton ccc ct 0nlethbridge2 f-sw 0N ct 1lethbridge8 f-sw blade SM 1Manhattan 1 BW CTmead ccc hfi ctPendleton1 fw 0N CT 1Swift Current F-(SW)-SW 1 F-SW-SW NPWooster 1 CB CTLinear (all sites)

C-Store® Development Status

Copyright 2003The Natural Resource Ecology Laboratory

Colorado State University

Initial Model

Development

InitialTesting

BuildValidation & Verification

DatasetIntegrate with

Economic Models

Web InterfaceDevelopment, Beta Testing

Final Testing

ReleaseFall, 2004

Verification andUncertainty

Process

APEX, a watershed model to simulate plant growth, hydrology, soil erosion and nutrient cycling on multiple fields Routing runoff & sediments

(soil, nutrients, pesticides)

Belowground flow

Carbon and nutrient cycling

Study S ubareaDrainage Path

< 7580859095100105110115> 120

Slope from DEM01

23

456

5m Contours from DE M

Study SubareaDrainage Path

Crop type from imageryCornSoybeansWinter WheatDouble-Cropped WW/SBOther Row CropsOther GrassWoods

Digital Elevation Model help delineate drainage area

Remote Sensing imagery helps

determine crop types and management

Seven fields on a hillslope in Frederick Co., MD

7

4

5

6

3

1

2

DtB

DtB

AfB

DtB

DuB

DuB

HcBDrainage Path

Deliniated farm fieldsCornCorn/SoybeanHaySoybean

Field NumberDominant Soil Type

Tributary Main Channel TributaryField Routing Number

1 2 4 3 5 6 7

Deposition of sediment (tons/ha) 1.26 ----- 0.94

4.91 ----- 0.15 ----- Deposition of carbon (tons/ha) 0.104 ----- 0.070

0.345 ----- 0.012 -----

Soluble Nutrients in Q

Nutrient loss in sediment Soil C Change Subarea Runoff (Q) Sediment in Q

P N P N Top 0.25m Profile mm t/ha ------------------------------kg/ha---------------------------- --------------t/ha-------------- 1 164 2.1 1.4 6.2 1.5 10.5 -1.8 7.5 2 61 5.5 1.1 1.5 5.3 12.3 -6.3 1.1 3 162 4.1 1.6 2.3 2.2 7.4 -2.3 8.5 4 129 3.0 0.9 3.6 1.4 12.2 3.0 19.5 5 129 2.2 2.1 2.0 1.8 8.4 7.5 7.4 6 58 2.8 1.1 5.5 2.8 7.5 -3.7 10.1 7 126 1.5 0.9 3.3 0.8 7.3 3.2 18.9

Runoff, sediments, nutrients and soil C changes

Routing and deposition of sediments and carbon

12

3

4

5

6

7

Landscape modeling is helping our understanding of

the role of erosion in the carbon cycle

The EPIC model: validation and application to estimate soil carbon sequestration under no till at the national scale

Long term experiments

NA

0

10

20

30

40

50

60

70

80

90

0 20 40 60 80 100SIMULATED SOIL CARBON (MG C/HA)

OB

SER

VED

SO

IL C

AR

BO

N (M

G C

/HA

)

ALBERTANEBRASKAKANSASTEXASILLINOISOHIOARGENTINA

Adj r2 = 0.94

Overview of the data assembly and modeling system in EPIC

National Resource Inventory

Cropping Practice Survey

Populated Model Run Databases--"The Library of Runs"

RunBuilder

Census of Agriculture

EPIC Field-Scale

Model

Tillage Systems

Nutrient Applications

Soil and Topography

Weather and Climate

Irrigation Schedules

Manure Applications

Conservation Practices

Crop Growth Data

Data Reduction Processes

I_Epic Interface

Model Run Input Databases

Analytical Framework-National Datasets

Notill C Benefit for Dryland Corn by Soil Cluster for Selected States(Soils in each state sorted in descending order of NT benefit)

-20

-10

0

10

20

30

40

50

60

70

80

90

100

110

Car

bon

Sequ

este

red

(g/m

2/yr

)

GAIAMSOHOKPASD

Truncated, one in IA at -176.7, and one in OH at -108.9

NT Benefit (g/m2/yr) = ((NT30-NT1)- (Till30-Till1))/30

Soil Carbon Over Time for Soybeans on 2 Iowa Soils by Tillage Type

100

125

150

175

200

225

250

275

300

1999 2002 2005 2008 2011 2014 2017 2020 2023 2026 2029 2032 2035 2038

Met

ric

Ton

s/Hec

tare

S1_PlowS1_MulchS1_NotillS2_PlowS2_MulchS2_Notill

The Century model: validation and application at the national scale

Background image shows all MLRA’s that have more than 5% agricultural cropland

Climatic DataMonthly TemperatureMonthly Precipitation

Historical Cropping Practices

Recent Cropping PracticesCrop RotationTillageFertilizerIrrigation

Native Vegetation

Soil CharacteristicsTextureDrainage

INPUT DATAMODELED

SCENARIOS

Tillage ChangesNo TillReduced TillageConventional Tillage

Land Use ChangesCRPAbandoned farmlandConverted to grassland

Rotation Changes

MLRA 108

Kansas wheat-fallow test

0

20

40

60

80

100

120

140

1860 1880 1900 1920 1940 1960 1980 2000

Gra

in Y

ield

Car

bon

(g/m

2)

NASS KSCentury

WN1 N=0 W0 N=0

WN2 N=2

WN2 N=4

Manhattan, KS site data used in simulation. Average monthly weather data used for 1866-1894. Measured monthly precipitaion used for 1895 onward, along with mean monthly tmax, tmin.

N = gN/m2

75% straw removal------->50% straw removal----------------------->grain only removed------------------->

Net Carbon Gains - 1997 rates

0

1

2

3

4

5

6

7

8

MM

T C

/ ye

ar

on conventional-till annual cropland

on reduced-tillannual cropland

on no-till annual cropland

on hay-pasture

on CRP

Aggregate Century results

Century

IPCC

18.4 MMTC yr-1

on 168 Mha cropland

21.2 MMTC yr-1

on 149 Mha cropland

SummarySimulation modeling plays a fundamental role in predicting and understanding soil carbon sequestration at different scales of resolutionC-STORE promises to be a useful tool to develop field-estimates of soil carbon sequestrationLandscape modeling with APEX should help understand the role of erosion in the carbon cycleThe use of Century and EPIC at the regional and national scales will provide independent estimates of soil carbon sequestration