Dynamical Prediction ofthe Ecosystems and the Global Carbon Cycle

Ning Zeng1, Jinho Yoon1, Augustin Vintzileos2, G. James Collatz3, Eugenia Kalnay1, Annarita Mariotti1,

Arun Kumar2, Antonio Busalacchi1, Stephen Lord3

Thanks: M. Heimann, C. Roedenbeck, P Wetzel, Brian Cook, R. Joseph, H. Qian, R. Iacono, E. Munoz, W. Higgins, K. Mitchell

1 University of Maryland2 NOAA/NCEP3 NASA/GSFC

TargetsTargets• Prediction (seasonal-interannual), not projection (climate

change)

• Predict atmospheric CO2 concentration and growth rate. Atmospheric CO2 can be a ‘eco-climate index’ indicating anomalies in the global ecosystem as a whole, just like NINO3 can be used as an index for climate anomalies associated with ENSO

• Predict spatial patterns and temporal variability of carbon fluxes and pool size Examples: forest productivity, agricultural harvest, fire danger

‘‘BreathingBreathing’’ of the biosphere: COof the biosphere: CO2 2 as a major as a major indicator of ecosystems (and climate) indicator of ecosystems (and climate)

Carbon model forced by observed climate variabilityENSO, Pinatubo, drought and other signals

Modeled land-atmosphere flux vs. MLO CO2 growth rate

El Nino 97/98

VEGAS

InversionRoedenbeck 2003

SeasonalSeasonal--interannual Prediction interannual Prediction of Ecosystems and Carbon Cycleof Ecosystems and Carbon Cycle

Two strands of recent research made this a real possibility• Significantly improved skill in atmosphere-ocean prediction

system, such as CFS at NCEP• Development of dynamic ecosystem and carbon cycle models

that are capable of capturing major interannual variabilities, when forced by realistic climate anomalies

A pilot study:Feasibility study using a prototype eco-carbon prediction system dynamical vs. statistical (Statistical forecasts for crop yield, infectious

disease exist)

The NCEP Climate Forecast System The NCEP Climate Forecast System (CFS, (CFS, SahaSaha et al. 2006)et al. 2006)

CFS captures major ENSO and other seasonal-interannual variability

PhotosynthesisAutotrophic

respiration

Carbon allocation

Turnover

Heterotrophicrespiration

4 Plant Functional Types:Broadleaf treeNeedleleaf treeC3 Grass (cold)C4 Grass (warm)

3 Vegetation carbon pools:LeafRootWood

3 Soil carbon pools:FastIntermediateSlow

Atmospheric CO2

The VEgetation-Global Atmosphere-Soil Model (VEGAS)

NPP=60 PgC/y

Rh=60 PgC/y

NEE = Rh – NPP = + 3 (Interannual)

Fire, wetland/CH4, 13C etc.

A 25A 25--year year hindcasthindcast experiment using a experiment using a prototype prediction systemprototype prediction system

CFS (9mon, 15 members)

VEGAS

Output9mon, 15 members

Month 2

CFS (9mon, 15 members)

VEGAS

Output9mon, 15 members

Month 1

1 mo forecastensemble mean

IInitialization

ClimatePredition

Ecosystem+Carbon Model

PredictedEco-carbon

SpinupPrecipTemp

PrecipTemp

Forecasting procedure IIForecasting procedure II

L=1

L=0

L=3

L=2

tt-1 t+1

Ensemblemean

Simple initialization to avoid ‘shock’; eco-carbon data assimilation?

First look: Productivity (NPP) First look: Productivity (NPP)

Anomaly Correlation Anomaly Correlation FtaFta

High skills in• South America• Indonesia• southern Africa• eastern Australia• western US • central Asia

Beyond ENSO: Beyond ENSO: Drought during 1998Drought during 1998--20022002

Fire during the 2002 DroughtFire during the 2002 Drought

Model: dynamical

Model: dynamical+ Statistical

Observation

Beyond ENSO: Fire in the US Beyond ENSO: Fire in the US Natural and anthropogenic factorsNatural and anthropogenic factors

Observation

Model

CASA (satellite fire, climate) VEGAS (climate only)

Input: climate only

Fire carbon flux during 1997-98 El Nino

Mean

1997-98 El Nino Anomalies

Input: satellite fire counts, climate

2007 Southern California Fire2007 Southern California FireAn operational trial An operational trial

•Ecosystem and carbon cycle prediction is feasible: encouraging results (better than expected) • Memory in the hydro-ecosystem is important in the enhancement of skill• several issues such as overestimation at mid-latitude regions

Some major development needs•Initialization: eco-carbon data assimilation?•Preprocessing/downscaling/postprocessing•Dynamical + statistical•Operational

ConclusionsConclusions

• Applications to ecosystem and carbon cycle• Identifying more clearly society-relevant aspects• A useful framework for studying eco-carbon

response and feedback to climate• Identifying ways to incorporate eco-carbon

dynamics in the next generation of climate prediction models (European GEMS)

Implications for climate service

Thank you!Thank you!

Reference: Zeng et al. (2008) Global Biogeochemical Cycles, in press