Global Carbon Cycle Model-Data Fusion Britton Stephens, EOL and TIIMES
Outline: 1.Why carbon-cycle model-data comparisons are important 2.Why carbon-cycle model-data comparisons are hard 3.Vertical profiles of CO 2 and continental flux partitioning 4.HIPPO Global field campaign 5.Seasonal O 2 and CO 2 cycles and Southern Ocean ventilation
It’s all about trajectories March, 1957 (IGY)4 October, 1957 (IGY)
How hot is it going to get? Climate projections are sensitive to human decisions and carbon cycle feedbacks... [IPCC, 2007]
... and human decisions are sensitive to scientific knowledge feedbacks How much can we burn? [IPCC, 2007] 3 or 7? 6 or 12?
How well can we predict human factors? Fossil-fuel emissions have already exceeded highest scenario used for IPCC projections [Raupach et al. 2007, PNAS; (PgCyr -1 )
How well can we predict climate feedbacks? Arctic summer sea ice levels have already exceeded the lowest model estimates [Stroeve, et al., GRL, 2007; NSIDC, 2008]
How well can we predict carbon-cycle feedbacks? In 2050, combined offsets have a range of 2.5 to 15 PgC/yr At $32/ton CO 2, ± 6 PgC/yr = ± $700 Billion/yr C 4 MIP Projections [Friedlingstein, et al., J. Climate, 2006] (PgCyr -1 )
Annual fluxes are small relative to balanced seasonal exchanges and to standing pools The global carbon cycle for the 1990s, showing the main annual fluxes in PgC yr –1. [IPCC, 2007] Annual residuals Land-Based Sink Net Oceanic Sink Pools and flows Uncertainties on natural annual-mean ocean and land fluxes are +/- 25 to 75 % (PgCyr -1 )
Polar vs. Tropical Oceans Where is carbon uptake currently occurring? Tropical vs. Northern Forests
Global atmospheric inverse models and surface data can be used to make regional flux estimates Forward: Flux + Transport = [CO 2 ] Inverse: [CO 2 ] – Transport = Flux
Model-data fusion is hard because: 1.Models often don’t predict something that can be measured 2.Observations don’t measure something that can be predicted 3.A cultural divide 360 m 120 m 800 m
[Gurney et al, Nature, 2002] Annual mean TransCom3 Level 1 results “For most regions, the between-model uncertainties are of similar or smaller magnitude than the within-model uncertainties. This suggests that the choice of transport model is not the critical determinant of the inferred fluxes.” Measurement uncertainty ≈ 0.2 ppm Continental site “data error” ≈ 2.2 ppm
12 model results from the TransCom3 Level 2 study ModelModel Name 1CSU 2GCTM 3UCB 4UCI 5JMA 6MATCH.CCM3 7MATCH.NCEP 8MATCH.MACCM2 9NIES ANIRE BTM2 CTM3 Systematic trade off between northern and tropical land fluxes
Regional land flux uncertainties are very large All model average and standard deviations: Northern Land = -2.4 ± 1.1 PgCyr -1 Tropical Land = +1.8 ± 1.7 PgCyr -1
A helpful discovery about the nature of the model disagreements ModelModel Name 1CSU 2GCTM 3UCB 4UCI 5JMA 6MATCH.CCM3 7MATCH.NCEP 8MATCH.MACCM2 9NIES ANIRE BTM2 CTM3 Systematic trade off is related to vertical mixing biases in the models Tropical Land and Northern Land fluxes plotted versus annual-mean northern- hemisphere vertical CO 2 gradient
12 airborne sampling programs from 6 international laboratories Northern Hemisphere sites include Briggsdale, Colorado, USA (CAR); Estevan Point, British Columbia, Canada (ESP); Molokai Island, Hawaii, USA (HAA); Harvard Forest, Massachusetts, USA (HFM); Park Falls, Wisconsin, USA (LEF); Poker Flat, Alaska, USA (PFA); Orleans, France (ORL); Sendai/Fukuoka, Japan (SEN); Surgut, Russia (SUR); and Zotino, Russia (ZOT). Southern Hemisphere sites include Rarotonga, Cook Islands (RTA) and Bass Strait/Cape Grim, Australia (AIA).
NOAA ESRL data 12 airborne sampling programs from 6 international laboratories
Comparing the observed and modeled gradients ModelModel Name 1CSU 2GCTM 3UCB 4UCI 5JMA 6MATCH.CCM3 7MATCH.NCEP 8MATCH.MACCM2 9NIES ANIRE BTM2 CTM3 Most of the models overestimate the annual- mean vertical CO 2 gradient observed value 3 models that most closely reproduce the observed annual-mean vertical CO 2 gradients (4, 5, and C): Northern Land = -1.5 ± 0.6 PgCyr -1 Tropical Land = +0.1 ± 0.8 PgCyr -1 All model average: Northern Land = -2.4 ± 1.1 PgCyr -1 Tropical Land = +1.8 ± 1.7 PgCyr -1 Northern Land Tropical Land
Interlaboratory calibration offsets and measurement errors Diurnal biases Interannual variations and long-term trends Flight-day weather bias Spatial and temporal representativeness Observational and modeling biases evaluated All were found to be small or in the wrong direction to explain the observed annual-mean discrepancies
[figure courtesy of Scott Denning] Seasonal vertical mixing
[Stephens et al., Science, 2007] Northern forests, including U.S. and Europe, are taking up much less CO 2 than previously thought Intact tropical forests are strong carbon sinks and are playing a major role in offsetting carbon emissions However, large (O ~ 2 PgCyr -1 ) flux uncertainties associated with modeling atmospheric CO 2 transport remain Airborne measurements suggest:
ppm pressure NSNSNSNS ppm latitude Transcom3 fossil fuel response
HIPPO (PIs: Harvard, NCAR, Scripps, and NOAA): A global and seasonal survey of CO 2, O 2, CH 4, CO, N 2 O, H 2, SF 6, COS, CFCs, HCFCs, O 3, H 2 O, black carbon, and hydrocarbons HIAPER Pole-to-Pole Observations of Atmospheric Tracers 5 loops over 3 years, first was in January 2009
measures O 2 concentration using a vacuum-ultraviolet absorption technique designed specifically for airborne use to minimize motion and thermal sensitivity and with a pressure and flow controlled inlet system switches between sample and reference gases every 2 seconds and has a precision of +/- 2 per meg on a 4-second measurement consists of pump, cylinder, and instrument modules, and a dewar NCAR Airborne Oxygen Instrument (AO2)
HIPPO Profile at 80 N, 12 January, 2009 CO 2 O 2 O 3 CO ppm per meg ppb ppb C kft
CO 2 O 2 O 3 CO ppm per meg ppb ppb C kft HIPPO Profile at 65 S, 20 January, 2009
HIPPO CO 2 Cross Section, January, 2009 AO2 Instrument
HIPPO O 2 Cross Section, January, 2009
HIPPO APO Cross Section, January, 2009 Atmospheric Potential Oxygen: APO = O *CO 2
Harvard and NOAA Data
Planned CO 2 comparisons AO2 Instrument [data courtesy A. Jacobson and the CT Team, as compiled by S. Mikaloff-Fletcher]
[TransCom model runs from T. Blaine dissertation. Fossil fluxes not included] Planned APO comparisons
Solubility (thermal) and biological processes have strong and reinforcing effects on atmospheric O 2 Southern Ocean air-sea CO 2 and O 2 fluxes 75 S 65 S55 S
ORCA-PISCES-T UVic-ESCM Direction and magnitude of biogeochemical response to increased circumpolar winds is uncertain [Le Quéré et al., Science 2007] [Zickfeld et al., Science 2008] sea to air = positive CO 2 (ppm) GlobalView: Seasonal peak, trough, and mean over the Southern Ocean
SeaWiFS Summer Chlorophyll a SIO Gradient = PSA – (CGO+SPO)/2 SPO CGO PSA [PCTM runs courtesy David Baker] SIO O 2 Stations
Solubility (thermal) processes may be overestimated during winter or anthropogenic uptake may be overestimated year round What could be wrong with the models? 75 S 65 S55 S
SeaWiFS Summer Chlorophyll a SPO CGO PSA Include NIWA Baring Head continuous O 2 data ( ) to extend coverage and temporal resolution Atmospheric transport model runs to translate fluxes to concentrations Include multiple models and dissolved-gas climatologies Underway shipboard O 2 measurements BHD Future work:
Global carbon cycle model-data fusion is a major challenge, but also presents many opportunities for advancing the science Recent efforts at comparing TransCom3 models and light-aircraft CO 2 data suggest a significant revision to continental carbon budgets The first phase of the HIPPO project was successful, with excellent instrument and aircraft performance and a rich data set to be analyzed Atmospheric O 2 measurements are uniquely suited for validating ocean biogeochemistry models and diagnosing processes in the Southern Ocean Conclusions
Thank you