Land-sea interaction and dynamic vegetation modelling

Ben Smith

Dept of Physical Geography and Ecosystem Science, Lund University, Sweden

• Land-sea-atmosphere connectivity in the global Earth system

• Carbon export from land to sea – possible impacts on coastal ecosystems and biogeochemistry

• Terrestrial vegetation and carbon cycling – recent trends and underlying causes

• Modelling climate change effects on vegetation and carbon cycling – introducing dynamic global vegetation models (DGVMs)

• Case study modelling land-sea carbon export and possible impacts on biogeochemistry of the Baltic Sea

• Future outlook: coupled models of regional Earth system dynamics

Lecture outline

Supporting exercise (in own time)

• Modelling ecosystem response to climate change using the LPJ-GUESS DGVM

Compartments and fluxes of the global carbon cycle

IPCC-AR5 2014

Emissions = 9 PgC/yr

Land-atmosphere = -3 PgC/yr

Ocean-atmosphere = -2 PgC/yr

Land-ocean = 1 PgC/yr

(data for 2000-2009)

Effects of riverine C input to coastal seas*

• Browner, more turbid water, steeper light attenuation

• Substrate for bacteria, increased respiration and pCO2

• Lower pH, decreased carbonate saturation state, represses calcifying organisms

• Cascading effects on overall biogeochemistry and ecology

• Few detailed system-level studies

*Riesebell 2004 J. Oceanography 60: 719-729

Effects of a BaU future emission scenario on global ocean CO2, carbonate and pH (Houghton et al. 1995)

Increased water CO2 concentration enhances production of some phytoplankton species

10.90 river input

point sources

fisheries

CO2 exchange with atmosphere

atmospheric deposition

input from North Sea

export to North Sea

input to sediment

return flux from sediment to water column

−11.58

−3.87 1.14

−1.05

−0.06

0.04 3.91

0.57

*Kulinski & Pempkoviak 2011 Biogeosciences 8: 3291-3230

Riverine input (all sources) 10 Tg/year 40% organic C

Ope

n oc

ean

Land Atmosphere

Sediments

River runoff is the main source of C to the Baltic Sea*

Likely causes: • increased primary production • longer growing season • expanded distribution of trees and shrubs

*Tucker et al. (2001) Int. J. Biometeorol. 45: 184

Change in land surface greenness 1982-1999*

increasing ’greenness’ →

normalised difference vegetation index

NDVI = NIR – R NIR + R

Vegetation cover and productivity have increased in concert with recent decades’ climate warming

*Ahas & Aasa (2004)

Trends in 644 plant phenological time series for Estonia 1948-1999*

non-significant trend

significant trend

Num

ber o

f spe

cies

/ ph

ases

Change in days

earlier spring later autumn

Earlier and longer growing season

*Walther et al. 2005 Proceedings of the Royal Society B 272: 1427

previous distribution

recent distribution

recent observations

0°C january isotherm

1931-1960 1981-2000

Changing distribution of holly, Ilex aquifolium*

Species distributions track climate shifts

*Harsch et al. 2009, Ecology Letters 12: 1040

1912 2009

Treeline advance, Mt Nuolja, Sweden (Van Bogaert et al. 2011. J. Biogeog.)

Par

amet

er e

stim

ate

3

2

1

0

–1

–2

Temperature trend

diffuse abrupt treeline

type

Effect of temperature change on tree limit elevation shift for 166 near-treeline sites + = warming associated with upslope shift*

Treelines are rising on average across the globe

Woody biomass is increasing in many shrublands, savannahs and extra-tropical forests

Eastern Cape Province South Africa (Welz 2013)

Above-ground biomass change 1993-2012

Satellite passive microwave measurements Liu et al. 2015 Nature Climate Change 5: 470-474

Likely causes include increased water-use efficiency under elevated CO2

Duke forest FACE experiment in North Carolina

NPP at ca. 350 ppm CO2 (gC m−2 år−1)

NP

P a

t 550

ppm

CO

2(g

C m

−2år

−1)

25% increase in net primary productionat 550 ppm CO2

NPP at ca. 350 ppm CO2 (gC m−2 år−1)

NP

P a

t 550

ppm

CO

2(g

C m

−2år

−1)

25% increase in net primary productionat 550 ppm CO2

CO2 is food for plants Increased atmospheric CO2 → higher NPP, at least in short term ...

NPP at 350 ppm (gC m−2 yr−1)

NP

P at

500

ppm

(gC

m−2

yr−1

)

Net reaction of photosynthesis

CO 2 + H 2 O + sunlight CH 2 O + O 2 carbon water dioxide

carbohydrate oxygen

CO 2 + H 2 O + sunlight CH 2 O + O 2 carbon water dioxide

carbohydrate oxygen

Norby et al. 2005 PNAS 102: 18052-18056

H2O

CO2

H2O

CO2

Reduced stomatal conductance under elevated CO2 may enhance water-use efficiency (WUE)

reduced stomatal conductance in response to higher ambient CO2 reduces

transpirational water loss

WUE = CO2 uptake by photosynthesis

H2O loss by transpiration

*Keenan et al. 2013 Nature 499: 324-372

mean of 13 models W

UE

ano

mal

y (g

C /

kg H

2O h

Pa)

Forests around the world show a positive trend in water-use efficiency of growth*

Carbon balance of woody ecosystems responds slowly to changes in structure, cover

or distribution

uptake

release

*Hyvönen et al. 2007 New Phytologist 173: 463-480

Net

eco

syst

em e

xcha

nge

of C

O2

(gC

m−2

yr−1

)

Young forests are a sink for CO2, storing assimilated carbon in the stems of growing trees. Over time, phenology and mortality augment soil carbon pools, causing decomposition losses to balance CO2 uptake

*Ciais et al. 2013. IPCC-AR5

Glo

bal C

flux

(Pg

C y

r−1)

25% of C emissions are taken up by terrestrial ecosystems due to surplus of photosynthesis relative to respiration and emissions from wildfires

Land ecosystems are a sink for atmospheric CO2 — a temporary anomaly not expected at steady state

atmosphere → plant

GPP

plants (→ heterotrophs) → soil

litter production

soil → atmosphere RH

plant → atmosphere

RA

net flux from atmosphere

to plants

Soil organic matter

NPP = GPP−RA net flux from atmosphere

to ecosystem

NEE = NPP−RH

Climate response of terrestrial ecosystems is intricately tied to their role in the global carbon cycle

CO2 CO2

CO2

net ecosystem exchange

net primary production

gross primary production

heterotrophic respiration

autotrophic respiration

chloroplast, mitochondrion

resp

onse

dis

tanc

e (1

0y m

)

response time (10x years) 1 sec 1 hour 1 min 1 day 1 year 100 years 10,000 years

tree seedling

stomate

adult tree

1 km

1000 km

mesophyll cell

Scales of terrestrial ecosystem processes

-6

-5

-4

-3

-2

-1

0

1

2

3

4

5

6

7

-8 -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5

photosynthesis respiration

gas exchange

growth

population dynamics

competition phenology

speciation genetic adaptation

disturbance

soil organic matter turnover

management

global change impacts research

• link structure to function, accounting for

feedbacks between them

• link ’fast’ (physiology, biogeochemistry)

and ’slow’ (demography, composition) ecosystem processes

• account for transient ecosystem dynamics when driving conditions (climate, CO2) are changing rapidly

StructuresFunctions

space

time

Dynamic Global Vegetation Model (DGVM) globally-applicable coupled model of

terrestrial vegetation dynamics and ecosystem biogeochemistry

Soil organic matter

Soil organic matter

soil bio- geochemistry

population dynamics

& disturbance

plant biogeography

primary production & growth

Vegetation

climate CO 2 CO 2

• link structure to function, accounting for

feedbacks between them

• link ’fast’ (physiology, biogeochemistry)

and ’slow’ (demography, composition) ecosystem processes

• account for transient ecosystem dynamics when driving conditions (climate, CO2) are changing rapidly

• based on generalised global (or biome-specific) parameterisations and plant functional types (PFTs)

StructuresFunctions

space

time

Average individual for plant functional type cohort in patch

Modelled area (stand) 10 ha - 2500 km2

replicate patches in various stages of development

Patch 0.1 ha

tree or shrub grass

crown area

height

fine roots

leaves

LAI

sapwood heartwood

0 - 50 cm 50 - 100 cm

leaves / LAI

fine roots

stem diameter

crown area

height

fine roots

leaves

LAI

sapwood heartwood sapwood heartwood

0 - 50 cm 50 - 100 cm

leaves / LAI

fine roots

stem diameter

Vegetation representation in the LPJ-GUESS DGVM*

*Smith et al. 2001 Global Ecology & Biogeography 10: 621-637

Parameter

max establishment rate (ha−1 yr−1)

max longevity (yr)

survival in shade

optimal temp for photosynthesis (°C)

bioclimatic distribution

allocation to stem growth

leaf:sapwood area ratio (m2 cm−2)

leaf phenology

crown spreading

boreal

10-25

evergreen

0.3

150

0.05

high

900

1250

temperate

15-25

summergreen

0.4

250

0.05

high

900

1250

boreal-temperate

10-25

summergreen

0.4

250

0.1

low

300

2500

no limits

10-30

summergreen-raingreen

-

-

-

low

-

-

Trait differences influence functioning and interactions among plant functional types (groups of similar species)

Smith et al. 2014 Biogeosciences 11: 2027-2054

Potential vegetation change in Sweden*

Rel

ativ

e co

ver (

leaf

are

a in

dex)

Norway spruce Scots pine other conifer beech elm ash oak alder birch other broadleaved herbaceous

* Smith et al. 2007 SOU 2007:60

LPJ-GUESS

RCA3

A2 emissions

ECHAM4

Conifer forest NPP 1996-2000*

*Smith et al. 2008. Forest Ecology & Management 255: 3985-3994

T+P+R+C T+P+C T P

T = change i temperature P = change in precipitation R = change in incoming SW radiation C = change in CO2

Scenario

R C

Year in climate model scenario RCA3-ECHAM4/OPYC3-A2

Net

prim

ary

prod

uctio

n (k

gC m

−2 y

r−1)

N Sweden

S Sweden

* Smith et al. 2007 SOU 2007:60

Climate and CO2 effects on NPP*

*Yurova et al. 2007 J. Geophysical Research 112

Net ecosystem CO2 exchange (NEE) Degerö mire, northern Sweden*

Water table depth (cm)

observed modelled

−2.0 −1.5 −1.0 −0.5 0.0

2.0 1.5 1.0 0.5

1900 1920 1940 1960 1980 2000

Glo

bal N

EC

B (G

tC)

flux to atmosphere

flux to biosphere

IPCC-AR4 estimates

Globally, past net ecosystem C balance (NECB) reflects land use and CO2 trends

atmospheric [C

O2 ]

cropland cover

Future land ecosystem carbon balance*

Eco

syst

em C

poo

l (G

tC)

LPJ-GUESS

Business-as-usual emissions (RCP8.5)

GCM

1900 2100 23000

8.5

W/m2

1900 2100 23000

8.5

W/m2

sink

source

neutral

* Ahlström et al. 2012 Environmental Research Letters 7

increased sink / reduced source

reduced sink / increased source

Robust sign of change for some regions

kgC m−2 yr−1

−0.150

0.150

0

∆ Net biome C exchange (2071-2100)−(1961-1990)

latitude

earlier leaf-out → ↑photosynthesis

milder autumn → ↑respiration

intensified seasonality in some

tropical regions

Some robust seasonal and regional trends

month

increased sink / reduced source

reduced sink / increased source

J F M A M J J A S O N D

increased sink / reduced source

reduced sink / increased source

Department of Earth Sciences

www.oceanclimate.se

Carbon in runoff from land is a major input to Baltic Sea biogeochemistry

Organic carbon (dissolved, particulate)

Dissolved CO2 (mainly respiration)

Alkalinity (mainly weathering products)

• Dissolved organic carbon (DOC) comprises a relevant part of the carbon transported to the Baltic Sea

• Main source: wetlands • DOC is generated by decay of soil

organic matter and transported by runoff and rivers to the Baltic Sea

Urban areas

Bare areas

Cultivated land

Pastures and natural grassland

Open herbaceous vegetation with shrubs

Lichens and mosses

Cropland-woodland mosaic

Wetlands

Snow and ice

Sparse vegetation

Broadleaved deciduous closed forest

Broadleaved deciduous open forest

Mixed closed forest

Mixed open forest

Needleleaved closed forest

Needleleaved open forest

Water

Land cover of the Baltic Sea Basin

Ledwith (2003), GLC2000 project

Urban areas

Bare areas

Cultivated land

Pastures and natural grassland

Open herbaceous vegetation with shrubs

Lichens and mosses

Cropland-woodland mosaic

Wetlands

Snow and ice

Sparse vegetation

Broadleaved deciduous closed forest

Broadleaved deciduous open forest

Mixed closed forest

Mixed open forest

Needleleaved closed forest

Needleleaved open forest

Water

Land cover of the Baltic Sea Basin

Urban areas

Bare areas

Cultivated land

Pastures and natural grassland

Open herbaceous vegetation with shrubs

Lichens and mosses

Cropland-woodland mosaic

Wetlands

Snow and ice

Sparse vegetation

Broadleaved deciduous closed forest

Broadleaved deciduous open forest

Mixed closed forest

Mixed open forest

Needleleaved closed forest

Needleleaved open forest

Water

Land cover of the Baltic Sea Basin

Urban areas

Bare areas

Cultivated land

Pastures and natural grassland

Open herbaceous vegetation with shrubs

Lichens and mosses

Cropland-woodland mosaic

Wetlands

Snow and ice

Sparse vegetation

Broadleaved deciduous closed forest

Broadleaved deciduous open forest

Mixed closed forest

Mixed open forest

Needleleaved closed forest

Needleleaved open forest

Water

Land cover of the Baltic Sea Basin

Urban areas

Bare areas

Cultivated land

Pastures and natural grassland

Open herbaceous vegetation with shrubs

Lichens and mosses

Cropland-woodland mosaic

Wetlands

Snow and ice

Sparse vegetation

Broadleaved deciduous closed forest

Broadleaved deciduous open forest

Mixed closed forest

Mixed open forest

Needleleaved closed forest

Needleleaved open forest

Water

Land cover of the Baltic Sea Basin

high veg cover (more conifers & wetlands) high veg cover (less conifers & wetlands) large lakes/reservoirs open highland alpine + subalpine birch forest glacier

Mechanisms?

• Vegetation effect on water residence time

• Decomposition influence on soil water CO2 concentration and pH

• Root exudate effect on soil pH

• Direct weathering by DOC

Weathering reaction (plagioclase):

2CO2 + 3H2O + Ca2Al2Si2O8 → Al2Si2O5(OH)4 + Ca2+ + 2HCO3−

alkalinity

Vegetation cover/type affects export of organic carbon and weathering products*

* Humborg et al. 2004 Limnology & Oceanography 49: 1871-1883.

Dissolved organic carbon (DOC) production and export in the LPJ-GUESS DGVM*

*Yurova et al. 2008 Water Resources Res. 44

DOC production and export Degerö mire, N Sweden*

*Yurova et al. 2008 Water Resources Research 44

modelled DOC storage

modelled DOC export

observed DOC export

modelled DOC production

modelled DOC export

Production Export gm−2

Export mg l−1

Storage gm−2

runoff basis mg/l

catchment area basis g/m2

Simulated DOC production 1961-2005 Wetland sources

DOC in runoff gC m−2 yr−1 Wetland fraction of landscape

−0.1 0.0 0.1 0.2 0.3

Råne älv Niemisel Torne älv Mattila

Töre älv Infl.Bölträsket Kalix älv Karlsborg

Lule älv Luleå Skellefte älv Slagnäs

Alterälven Norrfjärden Pite älv Bölebyn

Rickleån Utl Ume älv Stornorrfors

Öre älv Torrböle Lögde älv Lögdeå

Gide älv Gideåbacka Ångermanälven Sollefteå

Ljusnan Funäsdalen Indalsälven Bergeforsen Ljungan Skallböleforsen

Delångersån Iggesund Gavleån Gävle

Dalälven Älvkarleby Forsmarksån Johannisfors

Norrström Stockholm Nyköpingsån Spånga

Motala Ström Norrköping Göta Älv Trollhättan

Botorpström Brunnsö Viskan Åsbro

Emån Emsfors Ätran Falkenberg Nissan Halmstad

Ljungbyån Ljungbyholm Lagan Laholm

Lyckebyån Lyckeby Mörrumsån Mörrum

Rönneån Klippan Helgeån Hammarsjön

TOC concentration trend (mgC l−1 yr−1)

Ljungbyån

0

5

10

15

20

25

30

1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

TOC

(mgC

l−1)

∆ DOC production (1996-2005)−(1976-1988)

LPJ-GUESS

gC m−2 yr−1

∆ runoff DOC concentration (1996-2005)−(1976-1988)

LPJ-GUESS

mgC l−1

∆ precipitation (1996-2005)−(1976-1988)

mm yr−1

Simulated versus observed DOC concentration trends*

*Measurements: T. Wällstedt Modelling: Guy Schurgers (wetlands only)

Scaling up: root zone to catchment

CSIM

climate atmospheric CO2 acid deposition ...

LPJ-GUESS

vegetation DOC

DOC, DIC alkalinity

Forest

Herbacous

Lake and streams

Cultivated areas

Ground water compartment 1

Ground water compartment 2

PrecipitationEvapotranspiration

Forest

Herbacous

Lake and streams

Cultivated areas

Ground water compartment 1

Ground water compartment 2

PrecipitationEvapotranspiration

runoff

• CSIM catchment hydrochemistry model*

• based on generalised watershed loading functions (GWLF)

• basic model assumption → water flow path is the most

important factor regulating river chemistry

*Mörth et al. 2007. Ambio 36: 124

y = 0,9991x R² = 0,9712

0

1

2

3

4

0 1 2 3 4

Obs

erve

d

Modelled

DIC (CT, mmol/L)

y = 1,0154x R² = 0,8997

0

0,5

1

1,5

2

2,5

0 0,5 1 1,5 2 2,5

Obs

erve

d

Modelled

DOC (mmol/L)

y = 0,9978x R² = 0,9719

0

1

2

3

4

0 1 2 3 4

Obs

erve

d

Modelled

Alk (AT, mmol/L)

y = 0,986x R² = 0,8341

0

20

40

60

80

0 20 40 60 80

Obs

erve

d

Modelled

Runoff (mm)

Calibration results 1996-2000 point = watershed

boreal forest and peatlands of the north

cultivated south

south

north

*T. Wällstedt, C.M. Mörth, C. Humborg et al. unpublished

~ +15%

~ +25%

~ +25%

Future climate scenarios

1975 2000 2025 2050 2075 2100

mean scenario (A1B) business-as-usual (A2) best case (B1)

DO

C

Alk

alin

ity

DIC

mol yr−1

CSIM

LPJ-GUESS

RCA3

ECHAM5

emissions scenario

Runoff to Bothnian Sea

*T. Wällstedt, C.M. Mörth, C. Humborg et al. unpublished

BB

BS

BP

GF

GR

DS

KA

runoff

DIC

alkalinity

DOC

Change in riverine export × sub-basin (2069-2098)-(1996-2005)

business-as-usual (A2) scenario

flux

chan

ge (%

)

*Omstedt et al. 2012 Tellus 64B: 19586

n.s.

Future impacts on biogeochemical export to the Baltic Sea*

Northern basins: DOC increases more than runoff

Southern basins: fluxes mainly follow runoff

pH c

hang

e (2

069-

2098

)−(1

971-

2000

) total pH change additional pH change

S = surface water - summer W = surface water - winter D = deep water

atm CO2

+ climate

+ nutrient loads

+ land use

atm CO2

+ climate

+ nutrient loads

+ land use

atm CO2

+ climate

+ nutrient loads

+ land use

*Omstedt et al. 2012 Tellus 64B: 19586

Change factor combination

Simulated effects on seawater pH* PROBE-Baltic oceanography-biogeochemistry model

Soil organicmatter

Soil organicmatter

soil C dynamics

populationdynamics

& disturbance

migrationphenology& growth

Vegetation

climateCO2CO2

Rossby Centre Atmospheric-Ocean

Model RCAO

temperature radiation soil water

Future prospect: regional Earth system model coupling land-atmosphere-ocean*

albedo leaf area index

fractional cover - broadleaved forest - needleleaved forest - open land vegetation

daily

LPJ-GUESS dynamic vegetation model

*Smith et al. 2011 Tellus 63A: 87-106

CO2 CH4

climate at domain boundaries

CO2 CH4

open loop for biogeochemical

feedback

grass – tundra IBS BNS BINE BNE

forest

100 yrs →

rela

tive

cove

r (%

LA

I) Vegetation change across the boreal zone

and Arctic*

LPJ-GUESS

RCA4

RCP forcing

EC-EARTH

*W. Zhang et al. in prep.

• Seasonality shift – longer growing season, earlier temperature peak • Evaporative cooling evens out growing season temperature profile

→ favours further shrub encroachment and treeline advance

-3

-2

-1

0

1

2

3

J F M A M J J A S O N D

(b)∆T

(°C

) albedo feedback

evapotranspiration feedback

Additional temperature change (2071-2100)−(1961-1990)

*W. Zhang et al. in prep.

max +1σ mean among yrs −1σ min

RCP 2.6 RCP 4.5 RCP 8.5

Albedo warming in spring, evapotranspiration cooling in summer

Vegetation-atmosphere feedback reduces sea ice cover*

*W. Zhang et al. in prep.

−25 −20 −15 −10 −5 0 5 10 15 20 25 %

Change in SIC 1992-2011 (feedback)−(no-feedback)

∆SLP (Pa)

(fb)−(no-fb)

∆LR (Wm−2)

(fb)−(no-fb)

DJF MAM JJA SON

more trees/ shrubs

more grass/ tundra

Veg. change

(2007-11) −(1992-96)

Causality?

• Land, sea and atmosphere are part of the Earth system, coupled via biogeochemical (especially carbon), hydrological and energy fluxes. River runoff carbon fluxes are in the ballpark 10% of anthropogenic emissions.

• Vegetation patterns and ecosystem functions are changing in response to climate change and elevated CO2

• Models that resolve processes at a wide range of scales are needed to describe potential future changes. DGVMs are built for this purpose.

• Increasing temperatures and CO2 will likely lead to vegetation distributional shifts, effects on carbon cycling will vary by climate zone and uncertainties due to forcing are large.

• System models accounting for land-sea carbon exports and impacts on marine biogeochemistry are emerging. One example for the Baltic Sea suggests 21st Century changes in climate, vegetation and CO2 concentrations will lead to lower pH in the Baltic Sea

• Regional Earth system models that fully couple land-sea-atmosphere matter and energy fluxes may be needed to resolve complex responses to multiple drivers

Summary of main points