Trends in Terrestrial Carbon Sinks Driven by Hydroclimatic Change since 1948: Data-Driven Analysis using FLUXNET Trends in Terrestrial Carbon Sinks Driven.

Trends in Terrestrial Carbon Sinks Trends in Terrestrial Carbon Sinks Driven by Hydroclimatic Change Driven by Hydroclimatic Change since 1948: Data-Driven Analysis since 1948: Data-Driven Analysis

using FLUXNETusing FLUXNET

Christopher Schwalm, Christopher Williams, Kevin Schaefer, Kusum Naithani, Jingfeng Xiao

Ameriflux Science Meeting & 3rd NACP All-Investigators Meeting 2011

January 31 – February 4, New Orleans, LA

• We ask– What are the carbon consequences of hydrologic

change?• We merge

– Global monitoring network (FLUXNET)– LUH time-varying land cover (IPCC AR5)– NCEP/NCAR Reanalysis

• We derive– Monthly time series (1948 – 2009)– 1° latitude/longitude resolution– Observationally-based estimates of carbon flux solely

attributable to hydrologic change

Outline

Global monitoring network

FLUXNET: Network of regional networksEddy covariance method: temporally dense in situ CO2 exchange including gross primary production and ecosystem respirationAncillary data: soil moisture, temperature, latent heat flux, LAI, etc.

Mapping points to pixels

Evaporative Fraction

Carb

on F

lux

Extract relationship between hydrologic change and carbon flux

Aggregate FLUXNET sites by IGBP land cover class

Calculate sensitivity: change in carbon flux to a unit forcing in evaporative fraction (z-score)

Sensitivity: g C m-2 month-1 σ-1

Map sensitivities to globe using

1) LUH [gridded land cover class]

2) NCEP/NCAR Reanalysis [gridded EF]

Schwalm et al. (2010) Global Change Biology

22 monm

Cg

monm

Cg

Spatial scaling: LUH land cover

1948

2009

62 annual snapshots of land cover from Land Use Harmonization (LUH)

Crosswalk: LUH → IGBP

http://luh.unh.edu/

IGBP maps18 IGBP land cover classes by pixel

FLUXNET sensitivitiesVegetated classes – observedNon-vegetated classes – set to zero

+

Pixel sensitivity [weighted average]

=

18

1,,,,,,,,

iyxtiyxtiyxt ysensitivitcoverage

Units: g C m-2 month-1 σ-1

“Points to pixels”

Temporal scaling: NCEP reanalysis

22 monm

Cg

monm

Cg

EF (σ)NEP sensitivity

(g C m-2 mon-1 σ-1 ) δNEP (g C m-2 mon-1 )

Example – Europe in June 1998

Global time series

5.3,7.4,2.2 RPNEP Sink (2000-2006) = +2.8 Canadell et al. (2007) PNAS

Global trends

Trend line (p > 0.44)

Visually the same as zero reference line

Grey envelope is ±2σ

Continental trends - δNEP

significant

not significant

More uptake

More uptake

Less uptake

Less uptake

Continental trends – δP & δR

Cumulative trend

TNEP [g C m-2 62yr-1]outgassing uptake

Differential response: Case study

Highest density of FLUXNET sites

Geography of hydrologic change

CRU vs. Willmott precipitation

Relating trend to background fluxFLUXNET + LUH + NCEP

δNEPδP

δR

NEPP

R

MODIS + CARBONTRACKER

Does the trend overpower the mean?

What spatial features are present?

Net effect on gross fluxes

P

T

Mean

Trend P

R

T

Mean

Trend R

|δP| > |δR| - color contrast

Median ratio 40% larger for |TP/P| than for |TR/R|

More clusters with |δP| > P

Fewer clusters with |δR| > R

Low productivity areas

Net effect on source/sink

Blue: source to sink [4%]Blue: source to sink [4%]

Red: sink to source [20%]Red: sink to source [20%]

Green: enhanced uptake [18%]Green: enhanced uptake [18%]

Yellow: enhanced outgassing [12%]Yellow: enhanced outgassing [12%]

20082000

kerCarbonTrac

T

Mean

Trend NEP

Summary

• Observationally-based estimates of carbon cycling solely attributable to hydroclimatic variability

• Range in del equals or exceeds terrestrial carbon sink magnitude or gross fluxes.

• Hydroclimatic variability has acted to flip sources to sinks and vice versa (25%) over the 62-yr record → “key player”

i

Net effect on photosynthesis

20092000

MODIS

T

Mean

Trend P

Net effect on respiration

2008200020092000

kerCarbonTracMODIS

T

Mean

Trend R

Net effect on gross fluxes

20092000

MODIS

T

Mean

Trend PLess assimilationTrend < 0

More assimilationTrend > 0