Using Satellite Observations to Constrain Carbon Flux Estimates

Transcription

1 Using Satellite Observations to Constrain Carbon Flux Estimates Shaun Quegan (University of Sheffield) Centre for Terrestrial Carbon Dynamics & National Centre for Earth Observation Lecture content The global carbon cycle and its components Carbon cycle models Atmospheric observations of CO2 and CH4 Using satellite data to improve estimates of carbon fluxes from the land Using satellite data to improve estimates of carbon fluxes from the ocean Challenges

2 The Global Carbon Cycle Atmospheric CO 2 : the Keeling Curve CO 2 concentration (ppm) Charles David Keeling ( ) Mauna Loa Observatory Year

3 Natural and anthropogenic components of the carbon cycle Need to quantify and spatialise: - Reservoirs - Processes - Sources - Sinks - Change Black = preindustrial sizes of reservoirs Red = changes resulting from human activities Sarmiento, J.L. and Gruber, N. (2002). CO 2 : emissions vs atmospheric increase From: Sinks for Anthropogenic Carbon, Physics Today, August 2002, Jorge L. Sarmiento and Nicolas Gruber

4 Perturbation of Global Carbon Budget ( ) CO 2 flux (Pg C y -1 ) Sink Source fossil fuel emissions deforestation atmospheric CO 2 land ocean (GtC y -1 ) Time (y) Le Quéré et al The mean global carbon cycle for the 1990s 9 High High GtC/year Land use change flux Low Fossil fuel emissions Residual land sink Low Net ocean sink 2 1 Atmospheric increase Anthropogenic Sources Changes in C pools

5 Global distribution of sinks over the period (flask inversion method) Sources red and yellow Sinks green and blue Fossil fuels not included Roedenbeck et al. (2003) Atmos Chem Phys Discussions 3, Atmospheric carbon dioxide from space First global satellite observations of total atmospheric CO2 (SCIAMACHY/ ENVISAT) C-theme May 2009 Buchwitz et al, 2007

6 SCIAMACHY SWIR WFDOAS ASIAN CO, CO2 and CH4 in 2003 Four data products: Vertical columns of CH4, CO, CO2, and O2 from SCIAMACHY nadir observations using appropriate spectral windows in the near-infrared Data products: Methane VMR (XCH4 = CH4-column/aircolumn) Carbon monoxide column (molecules/cm 2 ) Carbon dioxide VMR (XCO2 = CO2-column/aircolumn) Details latest versions: de Beek et al., ACPD, 2006 GOSAT: dedicated C cycle satellite JAXA s next generation greenhouse gas satellite with improved precision and accuracy: launched 23/01/09 CH 4 Results GOSAT CH 4 More than 1 year of global observations of CO 2 and CH 4 columns are now available First comparisons to GEOSChem and Carbontracker show good correlation of large scale patterns GEOS -Chem CH 4 CO 2 Results TCCON Validation CH 4 Zonal Mean August 2009 GOSAT CO 2

7 Expected Improvement in CO 2 Flux Estimates from GOSAT large small (L. Feng, P. Palmer U Edinburgh) Models Carbon flux models developed mainly to investigate the response of the land and ocean to climate change Intended to be predictive, hence parameterised rather than data driven. Designed for a data-poor environment Recent extension of land models to full climateland surface coupling to take account of climate-carbon cycle feedbacks

8 The C4MIP comparison of coupled models Simplified structure of a carbon flux model Climate Model coupling S n Model S n+1 Other inputs

9 How can EO data affect a carbon flux model? Climate Parameters S n Model S n+1 Other inputs Processes Feedback? Testing The land component of the C cycle: natural fluxes photosynthesis Plant respiration Soil respiration Gross Primary Production (GPP) Net Primary Production (NPP) Net Ecosystem Production (NEP) Net global fluxes disturbance Net Biome Production (NBP) 2 Gt C

10 Model: 20 th century changes in productivity ERS coherence: forest age, biomass Seawifs fapar SPOT-VGT LC MODIS burnt area SPOT-VGT budburst Space Measurements of Carbon Emissions from Biomass Burning Vegetation releases fixed amount of energy when burned A proportion emitted as radiation detectable by satellite MSG SEVIRI Diurnal cycle in emissions NOAA,

11 Estimating C Emissions from Radiative Energy Fire Seasonality and Location Temporal Emissions Variation NH Africa Tg SH Africa Tg [Very strong seasonal cycle] Short-Term Emissions Estimation as Model Drivers Observed GeostationaryFRP [W/m 2 ] (red) Modelled (blue) 2007 Greek Fires J. Kaiser (ECMWF)

12 Tropical peatland fires in Borneo Haze from Indonesian peat-land fires blankets SE Asia billion tonnes of carbon were emitted from peatland fires during 1997/98. 16th August 2005: Smoky haze chokes Southeast Asia. Again hundreds of fires burn deep into the underlying peat spreading smoke across the region ALOS PALSAR image of Sumatra

13 Improving estimates of tropical deforestation flux and REDD Tropical cloud obscures optical imagery Landsat 5 Challenges: 1. Improve area estimates 2. Improve biomass estimates 3. Provide REDD methodology 4. More realistic flux model Deforestation Same area: radar images are cloud-free Detections JAXA - ALOS Deforestation Deforestation Deforestation 10 km Annual carbon flux from land-use change Year Flux (Gt C/year) Tropical Temperate Total Source: CDIAC-ORNL IPCC 2007 range of estimates Carbon emitted = Biomass x Area deforested x 0.5 (IPCC Good Practice Guide) Use ALOS long wavelength radar The ESA BIOMASS mission Link to: ecosystem models fire observations GOSAT atmos. measurements soil processes Much combined science to be done Kilometers N Sumatra Malaysia Singapore Sea Riau Dry Land Riau Peat Land 102 East 0 Equator %[ Pekanbaru Dryland Forest Peatland Forest 2005 Forest Cover 1997 Hot Spot * NOAA Satellite 1998 Hot Spot * NOAA Satellite 1999 Hot Spot * NOAA Satellite 2000 Hot Spot * NOAA Satellite 2001 Hot Spot * NOAA Satellite 2002 Hot Spot * NOAA Satellite 2003 Hot Spot * NOAA Satellite 2004 Hot Spot * MODIS Rapid Response 2005 Hot Spot * MODIS Rapid Response 2006 Hot Spot * MODIS Rapid Response Singapore Riau has burned almost everywhere

14 Integration of EO with models Models include processes, interpolate beyond view (space, time) Data Assimilation: Uses observations to constrain/correct model variables & parameters Test model processes Improve model forecasts Courtesy of Ricardo Torres Data assimilation to improve estimates of Net Ecosystem Production No assimilation Assimilating MODIS (red/nir)

15 Processes influencing air-sea carbon dioxide fluxes CO 2 Ice melt stability & exposure Temperature Coastal Resuspension Salinity? Sea state Circulation - horizontal - vertical Pigments & Primary Production Subsurface X CO 2 X CDOM How can we measure CO 2 exchange with the ocean? Solubility pump Directly by satellites? Not yet Indirectly by satellites? Temperature Sea state/winds Algal biomass Models Also need direct measurements only available in situ Biological pump

16 Physical gas exchange processes Gas transfer velocity (K) directlyrelated to sea state processes indirectlyrelated to wind Wave height/slope reflects history (direction, fetch) and damping (slicks) Whitecapping reflects turbulence, bubbles, microbreaking From Wade McGillis Biological carbon reservoir & Primary Production Bigger cells (>20µm) Medium cells (2-20 µm) CHLa [mg/m 3 ] Size of algal cells regulates ecosystem processes: Primary production Length of food web Whole ecosystem production & respiration Carbon dioxide drawdown Hirata et al., 2008 RSE; Brewin et al., 2010 Eco Mod Smaller cells (<2 µm)

17 Analysis & Optimization of Plankton Ecosystem Models Weighting model-data misfit to allow for uncertainty in environmental inputs Robust multi-site calibration of models Intercomparison of models on the basis of structure & formulation Role of satellite EO data: Ocean Surface pco 2 JAN-DEC at 40 N 20 W Provide many observations in all surface environments Test / optimise model parameters for applicability at basin-scalesscales Provide contextual information for site-based time-series Courtesy of John Hemmings Assimilation of Ocean Colour Data in Met Office Carbon Cycle Models Chlorophyll-a Concentration DAILY 2-D CHLOROPHYLL ANALYSIS Local 1-D balancing scheme for carbon and nitrogen pools ACRI & the GlobColour Team Courtesy of John Hemmings Tests with synthetic data show potential for improving C fluxes Ocean Surface pco 2 JAN-DEC at 40N 20W Dissolved Inorganic Carbon R.M.S. Error (N. Atlantic) TRUTH FREE RUN ASSIMILATION APR MAY JUN JUL AUG

18 Summary & Challenges Carbon cycle basic component of Earth system: atmosphere, land & ocean a fundamental element of global change & climate warming intimately related to water cycle, atmospheric chemistry & biodiversity A crucial issue is credibility of models. Need an integrated approach to using satellite EO with in situ observations and modelling systems Recent advances in data assimilation provide route for this Exploit co-located sensors to improve local flux estimates: Envisat Sentinel-3 IR and OC radiometers, Altimeters, Scatterometers Build & exploit new sensors for carbon cycle monitoring: Atmospheric greenhouse gases Biomass Global Carbon Data Assimilation System Geo-referenced Geo-referenced emissions emissions inventories inventories Atmospheric Atmospheric measurements measurements Remote sensing of atmospheric CO 2 Data assimilation link Climate and weather fields Atmospheric Transport Model Optimised fluxes Optimised model parameters Ocean time series Biogeochemical pco 2 Surface observation pco 2 nutrients Water column inventories Ocean Carbon Model Coastal studies Ocean remote sensing Ocean colour Altimetry Winds SST SSS Terrestrial Carbon Model rivers Lateral fluxes Remote sensing of vegetation properties Growth cycle Fires Biomass Radiation Land cover/use Ciais et al IGOS-P Integrated Global Carbon Observing Strategy Eddy-covariance flux towers Biomass soil carbon inventories Ecological studies

19 Radiative forcing IPCC, Climate Change, 2001

20 Greenhouse gases Gas Radiative efficiency (Wm -2 ppb -1 ) Lifetime (years) Global Warming Potential Time horizon 20 yrs 100 yrs 500 yrs CO CH N 2 O IPCC, Climate Change, 2001 The GOSAT Satellite GOSAT satellite is first dedicated greenhouse gas sensor which carries 2 instruments: TANSO Fourier Transform Spectrometer (FTS): Provides spectrally-resolved radiances for 4 shortwave-ir (polarized) and thermal-ir bands Covers several absorption bands of CO 2, CH 4, O 3 and H 2 O (and others) and O 2 Cloud aerosol imager (CAI): 4 broadband channels from UV to SWIR with high spatial resolution Provides aerosol and cloud information required for the GHG retrieval Coverage of TANSO FTS NCEO Carbon PI Meeting, 11 February, Leicester

21 GOSAT Data GOSAT sampling strategy: 3 day repeat cycle Isolated soundings with ~100 km distance in-between Relatively large footprints with 10.4 km diameter Nadir sampling over land and sunglint sampling over ocean between 35 o N and 35 o S Data availability and quality: Sampling of GOSAT sample product over Japan University of Leicester GOSAT FTS spectra, CAI data and L2 for SWIR CO 2 and CH 4 data are available Initial calibration and validation has been carried out: Improvements to calibration are still on-going and L1B data versions can differ significantly No Muller Matrix available so far to treat polarizationsensitivity of the instrument -> problem for aerosol retrieval! Problem with phase correction for thermal-ir channel NCEO Carbon PI Meeting, 11 February, Leicester Global trends in CO2

22 Current estimates of fluxes of anthropogenic CO2 Seasonal variability in atmospheric CO2

23 Data EO focus has been on producing geophysical products (fapar, LAI, snow water equivalent, soil moisture, etc.) Implications: Deriving products often involves inverting a complex model, with no unique solution. Error properties of products often poorly characterised. EO interactions with a Land Surface Model Phenology Snow water Burnt area Climate Parameters Processes Fire emissions fapar LAI Snow cover Land cover Forest age S n LSM S n+1 Soils Observable Possible feedback Testing: Radiance fapar

24 Comparing models and data SDGVM MODIS fapar SeaWIFS fapar Are model and data representations consistent? Derived product Assumptions Observations MODEL Assumptions The real world

25 Resolving the inconsistency Observations Testing and-or Data Assimilation MODEL The real world Assumptions The C cycle: 1990s budget of anthropogenic CO2 Land uptake? (2.6 by difference) Accumulation in atmosphere 3.2 Gt C yr -1 Fossil fuel release 6.4 Gt C yr -1 Deforestation 1.6 Gt C yr -1? Ocean uptake 2.2 Gt C yr -1

26 Ocean Colour signal Ocean Colour satellites NEVER measure: e.g. Phytoplankton CDOM SPM They measure light (electromagnetic wave) Absorption by seawater a Scattering by seawater b Absorption By atmosphere absorption Scattering By atmosphere Sea surface roughness Observation geometry scattering Etc. (e.g. adjacency effects) Optical proxy of POC Water constituents (e.g. Phytoplankton,CDOM, SPM etc ) a = a w + a ph + a d + a CDOM Courtesy of Takafumi Hirata Optical proxy of phytoplankton * Primary Production (organic carbon) * PFTs (carbon export into deep ocean, DMS, etc) Optical proxy of DOC Optical proxy of SPM, POC & PIC Nothing to do with seawater (also size distribution) b = b w + b p * Same for backscattering