Simulating terrestrial ecosystems: current progress …2014/11/26  · Importance of ecosystem Cox et al. (2000) Friedlingstein et al. (2001) uncertainties Sink Source fully coupled

Simulating terrestrial ecosystems: current progress and future perspectives

Takeshi IseKyoto University

About myself

Forest ecology Climate change Simulation modeling

University of Wyoming

University of Wyoming Intensive field ecologyHow can I use my knowledge

to solve global environmental problems?

Harvard UniversityModeling forest

dynamicsPrediction concerning

environmental problems

JAMSTECThe Earth SimulatorBuilding terrestrial

ecosystem submodel as a component of the Earth System Model

“An ecologist among experts in physics”

University of Hyogo (2011-2014)

京コンピュータ兵庫県立大学大学院シミュレーション学研究科

Your next door!

Kyoto University (2014-)

Climate change:importance of terrestrial ecosystems

ecosystem

changingclimate

Changes in Biomass? Fire? Soil carbon?

Positive feedback?

Negative feedback?

Carbon cycle Changes in Temperature Precipitation

Fig. 7.3Carbon cycle

Atmosphere

6.4 GtC/yr120 GtC/yr

Importance of ecosystem

Cox et al. (2000) Friedlingstein et al. (2001)

uncertainties

Sink

Source fully coupled

only CO2 fertilization,no CO2 greenhouse effect

only CO2 fertilization,no CO2 greenhouse effect

fully coupled

Modeling terrestrial ecosystems

Types of simulation models

Big-leaf models(bucket models)

Individual-basedmodelsvs.

Phenomenon-basedModels (“regression”)

Process-basedmodelsvs.

Static models Dynamic modelsvs.

Modeling terrestrial ecosystems Lots of variables and

parameters Temperature Precipitation Soil type Time since disturbance Species competition Species characteristics

Suitable climate Suitable soil Suitable time after

disturbance Physiology

Photosynthetic rates Wood density Leaf thickness deciduousness

Modeling terrestrial ecosystems

Our challengesLots of parameters!Heterogeneity!Weak theories

(comparing against physics)!

Abrupt changes (i.e., cusp)!

vs.

2 topics about soil organic carbon

Physics-based simulation

Parameter estimation using annealing and maximum likelihood

Takeshi IseKyoto University

High temperature sensitivity of peat decomposition due to physical-biogeochemical feedback

Ise, T., A.L. Dunn, S.W. Wofsy, and P.R. Moorcorft. 2008. High temperature sensitivity of peat decomposition due to physical-biogeochemical feedback

Ise and Moorcroft (2006)Global Soil Data Task (2000)0

10

20

30

> 40 [kgC/m2]

Soil carbon•1500 GtC (2x in the atmosphere)•up to 30% in northern peatlands

Why peatland?

B1

A1B

A2

Boreal region under climate change

Peatland carbon cycleContinental bogFen

Peatland biogeochemistryContinental bogFen

Bog Disconnected from

regional hydrology

terrestrialization paludification

Forested bog, northern Manitoba

http

://gs

c.nr

can.

gc.c

a/la

ndsc

apes

/

(Anderson, Foster, & Motzkin 2003)

http://www.na.fs.fed.us/spfo/pubs/n_resource/wetlands/wetlands9_organic.htm

Mature spruce bog

SOC in peatland

humic layer

bedrock

fibrous layerhumic layer

fibrous layer

years

Young spruce bog

litter & moss

Peat column gains height

Rise in water table

mineral soil

bedrock

mineral soil

Mature spruce bogHow to model water table?

humic layer

fibrous layer

humic layer

fibrous layer

years

Young spruce bog

watertabledepth

watertabledepth

Hypothesis:Constant from surface(Clymo 1984)

humic layer

fibrous layerwatertabledepth

bedrock

mineral soil

bedrock

mineral soil

Strong positive feedback(paludification)

Mature spruce bog

humic layer

fibrous layer

humic layer

fibrous layer

years

Young spruce bog

watertableheight

Constant from bedrock

Strong negative feedback

watertableheight

bedrock

mineral soil

bedrock

mineral soil

Null hypothesis

How to model water table?

null?

?

fibrous layer

Mature spruce bog

humic layer

humic layer

fibrous layer

years

Young spruce bog

Which hypothesis ?

Somewhere in between

Needs for mechanisticsimulation! water balance

soil properties

bedrock

mineral soil

bedrock

mineral soil

How to model water table?

ED2

• Process-based land-surface model

• Fast timescale fluxes carbon water energy

Tatm

Tcanopy

Tsurface

Tsoil_1

Tsoil_i

Tsoil_n

eatm

ecanopy

esurface

esoil_1

esoil_i

esoil_n

CO2

CO2

CO2

CO2

Rin Rout

Input data update in 30 minutes meteorological variables

(SW, LW, air temperature, precipitation, humidity, wind speed/direction, and [CO2])

ED1: Moorcroft et al. 2001. Ecological Monographs 71:557-585.ED2: Medvigy et al. 2006. Ph.D Thesis. Harvard University.

Fibrous

Two peat types Fibrous Humic

Simulation of SOC

Real-time conversion to peat depth

Simple, but powerful Reproduce feedbacks

Humic

fluctuatesaccording towater balance

Biogeochemical model

bedrock

mineral soil

metmetmetmet CrI

dtdC

strstrstrstr CrI

dtdC

humhumstrstrmhum CrCrh

dtdC

MdTdkr ii

Fibrous

Humic

0.0 0.2 0.4 0.6 0.8 1.0

0.0

0.2

0.4

0.6

0.8

1.0

volumetric water content

Md

(b)

-10 0 10 20 30

0.0

0.1

0.2

0.3

0.4

soil temperature [ oC]

Td

(a)Decompositiontemperaturedependence

moisturedependence

Frolking et al. 2002

bedrock

mineral soilTd

Md

Biogeochemical model

2 simulations

Fibrous

Humic

Dynamic peat depth model

strmetfib CCC

fibfib

fibfib DF

CZ

humhum

humhum DF

CZ

Static model (mineral soil model)

0 500 1000 1500 200022

2426

2830

year

soil

orga

nic

carb

on [k

gC m

2]

z = 32cmz = 28cmz = 24cm

Equilibrium SOC: self-regulatory Equilibrium SOC: sensitive to initialization

bedrock

mineral soil

Results

0 500 1000 1500 2000

2224

2628

year

soil

orga

nic

carb

on [k

gC m

2]

Dynamic ModelStatic Model

positivefeedback

negativefeedback

1000 1020 1040

28.4

28.7

29.0

Results

0 500 1000 1500 2000

2224

2628

year

soil

orga

nic

carb

on [k

gC m

2]

Dynamic ModelStatic Model

positivefeedback

negativefeedback

1000 1020 1040

28.4

28.7

29.0

2224

2628

Dynamic ModelStatic Model

positivefeedback

negativefeedback

Resultshydrology

peat depth insulation

2224

2628

Dynamic ModelStatic Model

positivefeedback

negativefeedback

0 500 1000 1500 2000

0.36

50.

370

0.37

50.

380

0.38

5

year

fract

ion

belo

w w

ater

tabl

e

Dynamic ModelStatic Model

0 500 1000 1500 2000

0.28

0.30

0.32

0.34

year

peat

dep

th [m

]Dynamic ModelStatic Model

fract

ion

belo

w w

ater

tabl

e

year

year

Comparison: BOREAS NOBS, 2003(Dunn, Barford, Wofsy, Goulden, & Daube 2007)Results

wat

er ta

ble

dept

h [m

]

0.3

0.2

0.1

0

observationsimulation

100 150 200 250

-50

510

1520

Julian day

soil

tem

pera

ture

[ o C]

observationsimulation

wat

er ta

ble

dept

h [m

]so

il te

mpe

ratu

re [°

C]

Julian day

Climate change:equilibrium

40% loss

Extrapolate overnorthern peatlands,

72-182 PgC34-87 ppm

year

soil

orga

nic

carb

on [k

gC m

2]

Dynamic ModelStatic Model

1520

2530

0 500 1000 1500 2000

undercurrentclimate

under4 oC rise

year

soil

orga

nic

carb

on [k

gC m

-2]

HadCM3 SRES A2 at 2099+ 4.3 °C+ 42.1 mm

~10% loss

Extrapolate overnorthern peatlands,

18-46 PgC9-22 ppm

Climate change:transient

1950 2000 2050 2100

2627

2829

year

soil

orga

nic

carb

on [k

gC m

2]

undercurrentclimate

underHadCM3SRES A2

Dynamic ModelStatic Model

year

soil

orga

nic

carb

on [k

gC m

-2]

Summary: continental bog

Both positive and negative feedbackprocesses are important determinants of peatland dynamics

Effects of climate change and on climate change will be more pronounced than previously thought

Acknowledgements

AdvisorsPaul R. MoorcroftDavid R. FosterJames J. McCarthySteven C. Wofsy

Moorcroft LabMarco Albani, Mike Dietze, Yeonjoo Kim, Gil Bohrer, David Medvigy, Heather Lynch, Shirley Dong, Jackie Hatala, Daniel Lipsitt

Significant othersAli Dunn, Jennifer Harden, Susan Trumbore, Mike Goulden, Yuko Hasegawa, Hugo Veldehuis, Motoko Inatomi, Tomomi Isono, BOREAS Project, Saskatchewan Environment Prince Albert Office

James Mills Peirce Fellowship Organismic and Evolutionary Biology