Reconciling the Disagreement between Observed and ...

17
Boulder, CO Reconciling the Disagreement between Observed and Simulated Temperature Responses to Deforestation Liang Chen, Paul Dirmeyer George Mason University 06/19/2018 2018 CESM WORKSHOP

Transcript of Reconciling the Disagreement between Observed and ...

Boulder, CO

Reconciling the Disagreement between Observed and Simulated Temperature Responses to Deforestation

Liang Chen, Paul Dirmeyer

George Mason University

06/19/2018

2018 CESM WORKSHOP

Motivation

• Most of the LUCID and CMIP5 didn’t capture the “observed” JJA TX warming by deforestation.

(Lejeune et al. 2017)

land use only – piControlall forcings – all forcings but without land use

TX90P

(Li et al. 2018)

2 /16

Research question

Space-for-time analogy vs. coupled deforestation simulations

• Is it a fair comparison?

• Does scale matter?

space-for-time

3 /16

Data and methods

• Observations• MODIS LST, land cover type, albedo, NDVI and ET• Global Forest Cover (Hansen/UMD)• CERES surface fluxes analysis• Other satellite (and reanalysis) LST products

less treehigher tmp.

more treelower tmp.

more treehigher tmp.

less treelower tmp.

sensitivity of Ts to deforestation (-1*TreeCover)

Tree Cover

Daytime Ts

~ * -1

assume the pixels receive the similaratmospheric forcingswithin the analysis window…

Ts tree cover

4 /16

Data and methods

• CESM large-scale deforestation simulations (CAM4 and CLM4)

Replace all the trees PFTs with grass.

The type of grass is determined by its latitude (arctic C3, C3, or C4 grass)

• low-res offline (2 deg)

• Low-res subgrid (2 deg)

• high-res explicit (0.5 deg)

• high-res regular (0.5 deg)

low-res(1.9x2.5)

high-res explicit(0.47x0.63)

tree: 37.5%

grass: 62.5%

high-res regular(0.47x0.63)

slope (Ts~deforestation) grass–tree grass–tree

5 /16

Observations MODIS

Within the bigwindow (1° x 1° )

6 /16

Offline vs. coupled experiments low-res

7 /16

Ts change at PFT level

Local impacts:same atmosphere

different land cover

Non-local impacts:same land cover

different atmosphereCombined effects

before: the coupled run before deforestationafter: the coupled run after deforestation

8 /16

T, Q, RH, and cloud fraction in coupled simulations

a b

c d

9 /16

“Fair” comparisons

SW

offline experiments, PFT-level comparisons, or observationally

based assumptions

LW

share the same atmospheric background

coupled experiments (mid/high latitudes in the CESM

deforestation experiments)

different atmospheric background

coupled experiments (clear sky)

different atmospheric background

coupled experiments (clear sky condition

and the same LWdown)

“same” atmospheric background

10 2 3

10 /16

Filtered Ts changes in the coupled experiments

1 2

3

11 /16

Is it a valid assumption if at a broader scale?clear-sky downward LW

atmosphericbackground can bedifferent

cloud can alsoplay a role

Within the bigwindow (3° x 3° ):

cloud radiative effects net SW

clear-sky downward SW

cloud radiative effects net LW

CERES

W/(m2%)*10

12 /16

Does scale matter?MODIS

satellite observations and reanalysis

0.05 deg 0.05 deg 0.05 deg

25 km 0.5 deg 0.5 deg

0.75 deg 1 deg

1 deg window 1 deg window 1 deg window

125 km window 3 deg window 3 deg window

3 deg window 3 deg window

1 deg

3 deg window

13 /16

k/% * 10

Does scale matter? CESM simulations

“local” effects through different ways

“coupled” effects

14 /16

K

model-dependent?

subgrid PFT-level comparison in GFDL(courtesy of Li and Liao)

JJA temperature

Local vs. non local impacts of LCC in MPI-ESM

(Winckler et al. 2017)

(Lejeune et al. 2017)

15 /16

Conclusion

• CESM (CLM) agrees with the observations in a “fair” comparison.

• More attention should be paid to atmospheric models and the coupling between land and atmosphere.

• The scale matters, and summer cooling by deforestation is possible.

16 /16

Thank you!

Liang Chen, [email protected]