G Hayman1, E Blyth1, D Clark1, F O'Connor2, M Dalvi2 and N Gedney2

1 Centre for Ecology and Hydrology (UK); 2 Met Office Hadley Centre (UK)

28th February 2012

Land Surface Modelling:The ALANIS Methane project

Contents and acknowledgements

2

Contents• Background

• ALANIS Methane project

• Land surface modelling with JULES

• Key Messages and future activities

Acknowledgements

• ALANIS Methane project partners

• European Space Agency

• iLEAPS

• To produce relevant Earth Observation products for land surface modelling of the methane emissions from wetlands

• To improve the emission estimate of methane from boreal Eurasian wetlands

ALANIS Methane: Project aims, objectives and outputs

3

Aims and Objectives

Outputs

• Earth Observation datasets relevant to methane emissions from wetlands • Modelled emission flux estimates from wetlands• Peer-reviewed papers

Source: http://www.riceweb.org/reserch/Res.issmethane.htm

CH4 wetland emissions by diffusion across the soil or water interface, by ebullition (bubbling), and by plant-mediated transport

Key parameters for land surface and climate modelling: wetland extent temperature soil carbon

US EPA, 2010

Wetlands are largest natural source but there are large uncertainties

Background

4

ALANIS Methane: EO products for large-scale land surface modelling

5

JULES (CEH)

Column CH4

(Bremen)

Wetland Extent(CNRS/Estellus)

Surface State(Vienna)

Test Site #1: Western Siberia (N)• Subarctic-Arctic• continuous to discontinuous permafrost• hotspot of lake change

Test Site #2: Western Siberia (S)• Boreal• Ob river floodplains• sporadic to discontinuous permafrost• extensive peatlands

Test Site #3: Lower Lena River floodplain and delta•Subarctic to High Arctic lowlands•key region for understanding the basic processes of the dynamic and development of permafrost in the Siberian Arctic•upstream basin with flood plains •extensive delta area with several terraces

ALANIS Methane: Focus on Northern Eurasia, 2007-2008

6

Existing product Use satellite data at different wavelengths (ERS scatterometer, SSM/I, AVHRR) Global coverage with spatial resolution compatible with climate studies Long time series (1993-2007)

Several publications [Prigent et al., GRL, 2001; JGR, 2007; Papa et al., JGR, 2010] For ALANIS methane, adjustments in methodology

Use MetOP ASCAT scatterometer data Higher temporal resolution (10 days from monthly)

Estellus: Regional Wetlands Extent and Dynamics

7

New product based on ENVISAT ASAR Wide Swath

Classification of open water surfaces, 10-day updates for maps of wetland dynamics

Cross-comparison with the regional wetland product planned

TU Wien: Surface Water Bodies

8

June 2007 September 2007

• Metop ASCAT

• Algorithm development based on ECMWF ERA-Interim soil temperature

• Post-processing to identify day of year

• Begin of thaw• End of thaw• Refreeze

TU Wien: Snowmelt and Ground freeze/thaw

9

IUP, Bremen: Sciamachy atmospheric column methane

SCIA WFMD v2.0.2Schneising et al., ACP, 2011

Noisier after Oct. 2005 due to detector degradation

Recent increase(~ 7-8 ppb/yr)

NOAA surface CH4 South Pole

FwCH4 = kCH4* fw * Cs * Q10(Tsoil)(Tsoil-T0)/10

FwCH4 = methane flux from wetlands

kCH4 = scaling factorfw = wetland fractionCs = “substrate”: fixed soil carbon content Q10 = temperature sensitivity

Gedney et al [2003, 2004] parameterisations of large-scale hydrology and wetland biogeochemistry

Modelled wetland fraction is based on soil moisture saturation

Current version has no overbank inundation

Can be used in different configurations:a. Point/Offlineb. Gridded/Offlinec. Coupled into atmospheric

chemistry model

JULES – Modelling methane emissions from wetlands

11http://www.jchmr.org/jules/

Use of EO products with JULES

Inundation products • Test ability of JULES to reproduce spatial and temporal patterns of ’wetland’

extent• Products can be to evaluate (’constraint’) or define (’driver’) wetland extent

Surface state• Test soil physics in JULES• Products can be to evaluate (’constraint’)

Sciamachy CH4 columns• Test of JULES wetland scheme

NOTE: Other products also being used (e.g., Land surface temperature, snow cover, ....)

Several versions of JULES, resolutions and input datasets have been used.

Runs presented here use: • JULES v3.0 (the latest version)

• 0.5x0.5° global grid (some runs on test areas only)

• CRU-NCEP 6-hourly meteorological data

• New input fields (“ancillary files”) for this grid including topographic index data from Hydro1k

• TOPMODEL-based parameterisation of wetlands (and runoff generation)

• Prescribed vegetation (no dynamic vegetation model)

JULES Offline model configuration

13Background Boreal Wetlands African Wetlands Future/Summary

14

Offline JULES vs. EO Regional Wetland product

EstellusJULES

Wes

tern

Sib

eria

Lena

1st -10th September 2007

Regional Wetland product: Area averages

Western Siberia (North+South)North

South

Offline JULES vs. EO Surface State product: area averages

Western Siberia (North+South)Lena

• Timeseries of fraction of area with frozen ground

• Currently no direct equivalent diagnostic in JULES

• Use frozen and/or melting fraction• ‘Autumn melt’ events in JULES.

16

Comparison with TU Wien Surface Soil Moisture product

Circumpolar product, regridded to 0.5°.

Soil moisture (as fraction of maximum) on 15th April 2007.

Comparison with Surface Soil Moisture product: Area averages

Western Siberia (North+South) Lena

Some differences in seasonality. Amplitude good.

Surface Soil Moisture product: 1st-10th September 2007

Western Siberia (North+South)

TU Wien JULES

20

Intention to use latest generation Hadley Centre climate model (HadGEM3) : Can be nudged to observed meteorology Regional configurations available

Initial runs showed that the configuration of the model used was not ‘reactive’ enough; this could not be corrected by tuning of parameters

Reverted to HadGEM2-ES Used for IPCC AR5 Assessment Paper in GMD [Collins et al., 4, 1051, 2011]

Nudged version of ‘optimised’ AR5 HadGEM2-ES run created

JULES as land surface module in HadGEM2/3-A

UKCA tropospheric chemistry scheme

Emitted species:CO, NOx , CH4, C2H6, C3H8, HCHO,

CH3CHO and CH3COCH3

21

All sources (526 Tg CH4 per annum)

JULES: HadGEM2 AR5 CH4 emission inventories

All sources except wetlands (345 Tg CH4 per annum)

Wetlands from Fung et al (1991) (181 Tg CH4 per annum)

22

JULES: Wetland emissions for ALANIS

23

Modelled 2000-2005 period using different wetland emission scenarios: ‘AR5’ distribution based on

Fung et al. (total =181 Tg pa) JULES masked by EO (total

=181 Tg pa) JULES driven with EO (total

=181 Tg pa) JULES driven with EO (total

=200 Tg pa) JULES driven with EO (total

=165 Tg pa)

JULES unconstrained not used

Wetland emissions

24

HadGEM2-ES Year-on-year sea surface

temperatures and sea ice distribution

Nudged using ECMWF ERA40

Created year-on-year emission inventories EDGAR v4.2 for

anthropogenic, shipping and aviation

GFED v3.1 for biomass burning (scaled to give AR5 decadal mean)

Natural sources as used before

NOTE increase in anthropogenic CH4 (and other) emissions (*)

Model runs for ALANIS methane

(*) See Monteil et al, ACP, 2011

25

• Atmospheric Methane Dry Air Mole Fractions, 1983-2009• NOAA ESRL Carbon Cycle Cooperative Global Air Sampling Network• Used 15 sites

Comparison with surface measurements

AR5 emissions JULES (masked) JULES+EO

26

Comparison with Sciamachy: Annual mean (preliminary)

xgxja: AR5 emissions based on Fung et al. [1991] xfzal: JULES wetland emissions masked with EO xfzam: JULES wetland emissions with EO wetland fraction

27

Comparison with Sciamachy: Zonal means (preliminary)

• Black: Sciamachy

• Red: AR5 wetlands [Fung et al., 1991]

• Green: JULES (masked)

• Blue: JULES (EO)

• Blue lines: JULES (EO) sensitivity runs

28

Check the spatial patterns of the wetland emissions – rice/paddy fields: in progress

Investigate column profiles (fall off of methane too strong in UKCA??): Sciamachy stratospheric CH4 products

Further model runs, potentially including those from inverse modelling (to 2008)

HadGEM Next Steps and Summary

Next steps

Summary Nudged version of HadGEM2-ES setup to investigate wetland methane

emissions First use of Sciamachy to test model performance; preliminary analysis

indicates that the model underestimates the methane column, greater than differences in inventories

Value of EO inundation product (increases extra-tropical contribution)

http://www.alanis-methane.info

Website and Data Dissemination

29

30

New EO products developed relevant to methane emissions from wetlands EO and emission datasets produced for boreal Eurasia for 2007-2008 Land surface modelling using the UK JULES model in 3 configurations EO products have identified areas for improvement in the model in hydrology

and biogeochemistry

Summary and future activities

Summary

Future activities Ongoing activities to model wetlands (through NERC grants on African and

high latitude wetlands) Established links with UKCA team (Cambridge/Met Office) Developing links with NCEO atmosphere (Palmer, Chipperfield) Bring insight gained from NCEO Atmosphere and Land themes into forward

modelling Longer term – benchmarking datasets