impact of large-scale climate extremes on biospheric ...2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 16...

Impact of Large-Scale Climate Extremes on BiosphericCarbon Fluxes:

An Intercomparison Based on MsTMIP DataAGU 2013

Jakob Zscheischler, Anna Michalak, Miguel Mahecha,Markus Reichstein, Christopher Schwalm, Yaxing Wei,

MsTMIP team and data providers

Carnegie Institution for Science, Department of Global EcologyMax Planck Institute for Biogeochemistry

December 12, 2013

Zscheischler et al. (DGE | MPI BGC) Extremes in MsTMIP December 12, 2013 1 / 17

A Multi-scale Synthesis and Terrestrial ModelIntercomparison Project

common experimental protocol for running > 20 Terrestrial BiosphereModels (TBMs)

same set of driver data

evaluate how structural differences in model design impact estimatesof carbon uptake and release

[Huntzinger et al., GMDD 2013; Wei et al., GMDD, 2013]


Models, variables and resolution

version: Global MsTMIP v1.0

models used in this study (PI)

biome-bgc (Weile Wang, [Thornton et al., 2002])clm (Dan Hayes, [Mao et al., 2011, 2012])clm4vic (Maoyi Huang, [Li et al., 2011])dlem (Hanqin Tian, [Tian et al., 2011, 2012])gtec (Daniel Riccuito)isam (Atul Jain)lpj (Benjamin Poulter, [Sitch et al., 2003])orchidee-lsce (Shushi Peng, [Krinner et al., 2005])vegas (Ning Zeng, [Zeng et al., 2005])visit (Akihiko Ito, [Ito, 2010])

Gross Primary Production (GPP), Total Respiration (TR), NetEcosystem Exchange (NEE)

0.5 degree spatial resolution, monthly temporal resolution, 1981-2010


Two perspectives: drivers and responses

“Forward assessment”:identify climate extremes and analyze their impacts

“Backward assessment”:identify extreme changes in carbon fluxes and analyze their causes


Definition of extreme events in climate drivers

compute standardized precipitation index (SPI) at each pixel (timescale 3 months)

compute standardized temperature index (STI, in the spirit of SPI) ateach pixel (time scale 1 month)

define a value as extreme if it exceeds ±2

4 categories of climate extremes:1 drought: SPI < −2 (-P),2 extremely wet period: SPI > 2 (+P),

3 cold spell: STI < −2 (-T), and4 heat extreme: STI > 2 (+T).


Definition of extreme events in carbon fluxes

Let x denote the carbon flux of a certain month. Then define x as extremeif

|x − x̄ | > 2σ(x)


Search for large spatiotemporally contiguous extremes

For each variable, combine extremes that are contiguous in time or spaceto extreme events.

lat

lon

Extreme events are sorted according to their size= integral of STI/SPI/C flux anomalies over spatiotemporal domain of theevent. [Zscheischler et al., Ecol. Inf., 2013; Zscheischler et al., ERL., 2013]


Forward analysis

Impact of climate extremes on carbon fluxes averaged across models.

−0.6

−0.4

−0.2

0

0.2

0.4

0.6

−P ("drought")ch

ange

in fl

ux in

Pg

C

0 20 40 60 80 100

−0.6

−0.4

−0.2

0

0.2

0.4

0.6

−T ("cold spell")

chan

ge in

flux

in P

g C

climate event rank

+P ("wet period")

GPPTRNEE

0 20 40 60 80 100

+T ("heat wave")

climate event rank


Forward analysis

Cumulative impact of 1000 largest climate extremes.


C flux sensitivity to extreme events

C flux response to SPI and STI values.

1 2 3 4 5 6 7 8 9 10−5

0

5

lpj

dlem

orchidee−lsce

clm4vic

clm

gtec

isam

visit

biome−bgc

vegas

−4

−3

−2

−10

1

2

3

4

ch

an

ge

in

g C

/ m

−2 p

er

un

it S

PI

pe

r u

nit S

TI

−4

−3

−2

−10

1

2

3

4

NEE

i) hot & dry j) hot & wet

k) cold & dry l) cold & wet

TR

e) hot & dry f) hot & wet

g) cold & dry h) cold & wet

GPP

a) hot & dry b) hot & wet

c) cold & dry d) cold & wet


Backward analysis – all models – globe

−GPP

28.5 % 10 %

59.4 % 2.1 %

T

cold

−2

−1

0

1

2

hot

+GPP

2.3 % 43.7 %

20 % 34 %

T

Pdry −2 −1 0 1 2 wet

cold

−2

−1

0

1

2

hot

−TR

45.6 % 13.8 %

38.9 % 1.7 %

+TR

1.1 % 24.1 %

24.4 % 50.4 %

Pdry −2 −1 0 1 2 wet

−NEE

13.5 % 51.8 %

20.1 % 14.6 %

+NEE

12.8 % 8.4 %

68.5 % 10.3 %

Pdry −2 −1 0 1 2 wet

−GPP

28.5 % 10 %

59.4 % 2.1 %

T

cold

−2

−1

0

1

2

hot

+GPP

2.3 % 43.7 %

20 % 34 %

T

Pdry −2 −1 0 1 2 wet

cold

−2

−1

0

1

2

hot

biome−bgc

clm

clm4vic

dlem

gtec

isam

lpj

orchidee−lsce

vegas

visit

Range of mean SPI and STI values during the 100 largest C flux extremes.


Compound versus additive impact – NEE

I = Impact

heatwave ⇒ IHdrought ⇒ ID

heatw .&dr .⇒ IHDIHD = IH + ID?

2 4 6 8 10 12 14 16

2

4

6

8

10

12

14

16

additive impact (gC m−2

month−1

)

com

pound im

pact (g

C m

−2 m

onth

−1)

NEE | hot & dry

biome−bgc

clm

clm4vic

dlem

gtec

isam

lpj

orchidee−lsce

vegas

visit

The compound impact of heat waves and droughts is larger than theiradded impact.


Compound versus additive impact – TR

I = Impact

heatwave ⇒ IHdrought ⇒ ID

heatw .&dr .⇒ IHDIHD = IH + ID?

−18 −16 −14 −12 −10 −8 −6 −4

−18

−16

−14

−12

−10

−8

−6

−4

additive impact (gC m−2

month−1

)

co

mp

ou

nd

im

pa

ct

(gC

m−

2 m

on

th−

1)

TR | cold & dry

biome−bgc

clm

clm4vic

dlem

gtec

isam

lpj

orchidee−lsce

vegas

visit

The compound impact of cold spells and droughts is equal to their addedimpact.


Forward: NEE response to heat waves in regions

global tropical temperate boreal

−2

−1

0

1

2

3

4

5

6

7

change in g

C m

−2 m

onth

−1 p

er

unit S

TI

NEE | heat waves

Large uncertainties in the response to heat waves in boreal forests.


Backward analysis – biome dependent

−GPP

T

cold

−2

−1

0

1

2

hot

+GPP

T

Pdry −2 −1 0 1 2 wet

cold

−2

−1

0

1

2

hot

−TR

+TR

Pdry −2 −1 0 1 2 wet

−NEE

+NEE

Pdry −2 −1 0 1 2 wet

−GPPT

cold

−2

−1

0

1

2

hot

+GPP

T

Pdry −2 −1 0 1 2 wet

cold

−2

−1

0

1

2

hot

tropical

temperate

boreal

global

Extremes in GPP and TR are P driven in tropical forests but T driven inboreal forests.Zscheischler et al. (DGE | MPI BGC) Extremes in MsTMIP December 12, 2013 15 / 17

Conclusions

1 at global scale droughts and heat waves lead to large net carbonrelease or decrease in carbon C sink

2 models agree on direction of response but magnitude of impactlargely differs across models

3 hot and dry conditions compound each other

4 disagreement between models for NEE response to climate extremesin boreal forests


Introduction into MsTMIPExtreme eventsForward analysisBackward analysisCompound events

Conclusions

impact of large-scale climate extremes on biospheric ...2 4 6 8 10 12 14 16 2 4 6 8 10 12 14 16...

Documents