Integrated Assessment Model Diagnostics – May 3, 2012 1

Land use, water, and carbon in the integrated Earth System Model (iESM)

WILLIAM COLLINS JAE EDMONDS, TONY JANETOS, ANDY JONES, ALLISON THOMPSON, AND PETER THORNTON

Lawrence Berkeley National Laboratory Pacific Northwest National Laboratory – Joint Global Change Research Institute Oak Ridge National Laboratory

Three Major Objectives

! Task 1: Create a first generation integrated Earth System Model (iESM) with both the human components of an IAM and a physical ESM

! Task 2: Further develop components and linkages within the iESM and apply the model to improve our understanding of the coupled physical, ecological, and human system

! Task 3: Add realistic hydrology, including freshwater demand, allocations and demands to hold stocks of water as well as representation of freshwater availability from surface water, ground water, and desalinization.

Why Is This So Important?

! In the present world emissions mitigation analysis is undertaken under the assumption that the climate is not changing. E.g. crop yields are unchanged, water resources are unchanged, and energy systems don’t have to cope with a changing climate.

! Climate impacts analysis is undertaken with the assumption that no resources are being diverted to address climate change. E.g. no land is being used for reforestation, no land is being used to grow bioenergy, and energy prices are unchanged.

! The changes in behavior of the coupled climate-energy-land modeling system are significantly different than in the un-coupled models. Modeling the system as a unit materially changes our understanding of the evolution of the physical climate system

! The development of an iESM means that fully consistent analysis of potential future climate change, emissions mitigation options, and impacts and adaptation options will be possible.

iESM Schematic

Four Modeling Systems: GCAM, GLM, CLM, & CCSM

1.  GCAM (Human

Dimensions Elements only;

15 ghgs, aerosols, SLS; 14

geopolitical regions; 151 Ecoregions)

2. GLM (½ x ½ degree grid

land-use-land-cover.)

ESM1 (3. CLM & 4. CESM)

Fossil Fuel & Industrial Emissions (Gridded)

Land Use

LU-LC

1. GCAM(Human

Dimensions Elements only;

15 ghgs, aerosols, SLS; 14

geopolitical regions; 151 Ecoregions)

2. GLM(½ x ½ degree

grid land-use-land-cover.)

ESM1(3. CLM & 4. CCSM)

Fossil Fuel & Industrial Emissions (Gridded)

Land Use

LU-LC

Two Experiments

! Experiment 0: Compare two cases ! RCP 4.5 ! An RCP 4.5 replication Wm-2 scenario without terrestrial

mitigation. ! (Sneaker-net).

RCP 4.5 Climate

Rep 4.5 Climate

Two Experiments

RCP 4.5 Climate

Rep 4.5 Feedback

! Experiment 1: RCP 4.5, with feedback from the CLM-CCSM calculations ! 15-year time steps; ! Lagged adjustment: ! Sneaker-net.

Carbon Density and Productivity

1. GCAM(Human

Dimensions Elements only;

15 ghgs, aerosols, SLS; 14

geopolitical regions; 151 Ecoregions)

2. GLM(½ x ½ degree

grid land-use-land-cover.)

ESM1(3. CLM & 4. CCSM)

Fossil Fuel & Industrial Emissions (Gridded)

Land Use

LU-LC

iESM multi-phase coupling strategy

GCAM CLM/CESM GLM

Climate

C stocks, productivity

Up/down scaling (space and time)

Fossil fuel emissions

Atm CO2

iESM Experiment 0 (RCP 4.5)

iESM Experiment 1 Terrestrial Feedbacks

iESM Experiment 0: Bioenergy Scenarios with the Simplest One-Way Coupling

! GCAM to GLM to CESM ! Does the one-way pass of information replicate the

original RCP4.5 simulation done in CMIP5? ! If we now add a different policy scenario, but keep the

same concentration pathway, does the evolution of the climate system differ?

! What happens when we replicate the approach of the Wise et al. Science paper? ! RCP4.5 – carbon price on all carbon (UCT) ! RCP4.5 – carbon price ONLY on fossil carbon (FFCT)

9

GLM

GCAM scenario LULCC

information

Historical LULCC

information

Transition matrix dynpft file CLM

translator

Experiment 0 Work Flow

CLM / CCSM

Carbon fluxes and

[CO2]

Carbon stocks

Climate change

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1990 2005 2020 2035 2050 2065 2080 2095

Other Unmanaged Land

Unmanaged Forests

Managed Forests

Pasture

Crops

Unmanaged Pasture

Desert

Bioenergy Crops

RCP 4.5 Land Use

No Policy

RCP4.5 (UCT) Rep 4.5 (FFICT)

RCP 4.5 and a Replication without terrestrial mitigation policy

Change in Landcover under the RCP4.5 FFICT Scenario

Jones et al, 2012

Effect of Land cover scenario on global mean temperature

Jones et al, 2012

Jones et al, 2012

Link between albedo and temperature

Jones et al, 2012 Jones et al, 2012

Linkage of water vapor and greenhouse effect

Jones et al, 2012

Jones et al, 2012

Major Findings of Experiment 0

! The two scenarios have the same radiative forcing from GHGs. ! Yet they are substantially different in the evolution of the climate:

the equivalent of 1.5 W/m2, or about 0.5°C global annual average. ! We can replicate RCP 4.5 with a one way pass of information. ! But it is also true that the actual policy chosen matters –

in this case the very large land-use change associated with FFICT. ! Radiative forcing by GHGs is not a complete metric for

evaluating the evolution of the climate system

Experiment  1:  Simplest  possible  feedback  from  CESM(CLM)  to  GCAM      

•  Send maps of carbon density, by PFT, from CLM to GCAM

•  GCAM updates internal carbon densities based on changes from CLM

•  GCAM regenerates RCP, including new LULCC trajectory, based on updated carbon densities.

Experiment  1:  A  few  other  details…      

•  15-year time step, first departure from control occurs in period 2020-2034.

•  Feedback modifies spatial/temporal pattern of GCAM carbon density as integration proceeds, but doesn’t modify baseline carbon cycle.

•  GLM has its own internal carbon density representation, which is not being modified in this experiment.

Transla'on  of  GCAM/GLM  coupling  response  to  CLM  vegeta'on  types  Control (2020 to 2034) Expt 1 – Control (2020 to 2034)

Trees

Grass

Crop

Percent cover Percent cover change

Transla'on  of  GCAM/GLM  coupling  response  to  CLM  harvest  Control (2020 to 2034) Expt 1 – Control (2020 to 2034)

Harvest

Percent cover Percent cover change

Summary of coupling influence on CLM forcing: •  Tree cover higher

•  Grass cover higher

•  Crop cover lower

•  Pattern reversed for Sahel, India, southern boreal forest in Asia

•  Regional modifications to harvest rate

CLM/CESM  response  to  Expt  1  coupling  (2020-­‐2034)  

No climate signal after first coupling time step

Expt1 Control

CLM/CESM  response  to  Expt  1  coupling  (2020-­‐2034)  

Clear signal in land cover change flux emerging signal in atmospheric CO2 concentration.

Expt1 Control

iESM Coupling Status

First Generation iESM

CESM

ATM (cam, datm)

LND (clm/iac, dlnd)

OCN (pop, docn)

ICE (cice, dice)

CLM IAC (giac, diac, siac)

GCAM GLM

CAM POP CICE

Driver Component Model Coupler

Status: •  iESM code is written. •  iESM code is running at NERSC. •  Validation against sneaker-net version of

experiment 1.1 is underway.

Sneaker Net Implementation

run GCAM at JGCRI => regional land use change

run matlab script at JGCRI => ½ degree land use change

run GLM at JGCRI => ½ degree land cover change

run PL code at ORNL run mksurfdat at ORNL

=> clm surface dataset on clm grid run CESM at ORNL

=> carbon stocks on clm grid run 01_pft_to_grid_h1.ncl at JGCRI run sneaker.r at JGCRI

=> carbon stocks on gcam grid

Issues: •  Diversity of languages

•  Large amount of effort to conduct this relay race

•  Human effort scales directly with coupling frequency.

Active Coupling Implementation

run CESM call lnd_run()

=> carbon stocks on gcam grid call gcam()

=> regional land use change call gcam2glm()

=> ½ degree land use change call glm()

=> ½ degree land cover change call glm2clm() ! (mksurfdt)

=> clm surface dataset on clm grid (to file) call clm()

=> carbon stocks on clm grid call clm2gcam

=> carbon stocks on gcam grid

iESM Coupling: Emulation of Sneaker-net Coupling

Status: •  We emulate sneaker-net using 15-year timesteps.

Emulation of pasture distributions using iESM

gpast

longitude (degrees_east)

latit

ude

(deg

rees

_nor

th)

Range of gpast: 0 to 1 (null)Range of longitude: -179.75 to 179.75 degrees_eastRange of latitude: -89.75 to 89.75 degrees_northCurrent time: 0 day as %Y%m%d.%fFile glm_state_2020.orig.nc

jet T

hu A

pr 5

13:

50:0

8 20

12�

gpast

longitude (degrees_east)

latit

ude

(deg

rees

_nor

th)

Range of gpast: -5.00679e-06 to 2.39909e-05 (null)Range of longitude: -179.75 to 179.75 degrees_eastRange of latitude: -89.75 to 89.75 degrees_northCurrent time: 0 day as %Y%m%d.%fFile glm_state_diff_2020.nc

jet T

hu A

pr 5

13:

44:5

8 20

12�

Status: •  We can reproduce the distributions of

pasture to 1 part in 100,000.

Overview of plans through 2014 ! Remainder of 2012

! Experiments ! Complete Experiment 1 ! Validate iESM code with Experiment 1 ! Meeting of working group on experiments to plan next phase

! Model development ! Integrate emissions coupling process into iESM ! Scoping of RTB integration into CESM

! Model testing ! Test CESM feedbacks to GCAM buildings sector

! Goals for 2013-2014 ! Integrate GCAM and CESM (RTB) water models ! Explore the consequences of the basic science uncertainties in CESM for

energy systems in GCAM ! Further develop and explore carbon cycle coupling with iESM ! Additional experiments with a wider range of climate, mitigation and

sustainability scenarios.

Acknowledgements to the Project Team

! PNNL – JGCRI: Ben Bond-Lamberty, Kate Calvin, Jae Edmonds, Mohamad Hejazi, Tony Janetos, Sonny Kim, Page Kyle, Pralit Patel, Allison Thomson, Marshall Wise, Yuyu Zhou

! PNNL- Richland: Maoyi Huang, Ruby Leung, Hongyi Li, Nathalie Voisin,

! ORNL: Peter Thornton, Marcia Branstetter, Jiafu Mao, Xiaoying Shi

! LBNL: Bill Collins, Andy Jones, John Truesdale, Tony Craig, Lisa Murphy, Alan Di Vittorio, Alan Sanstad, Margaret Torn

! UMD: George Hurtt, Louise Chini

30

iESM Project Briefing – April 12, 2012 31

Discussion