greenhouse gas observations from space: why and how? - home - earth … · 2014-09-10 · michael...
TRANSCRIPT
Michael Buchwitz Institute of Environmental Physics (IUP)
ESA Earth Observation Summer School, ESRIN, 4-14 August 2014
Greenhouse gas observations from space: Why and how?
1 www.iup.uni-bremen.de
Greenhouse gas observations from space 1. Why and how ?
2. Results from ESA‘s GHG-CCI project
3. Proposed future mission CarbonSat
Overview Talks 1 - 3
2
Greenhouse gas observations from space: Why and how?
• Why ?
• … including at least a little bit of the most relevant background information …
• How ?
• From satellite radiances -> atmospheric GHG concentrations -> GHG surface fluxes (sources & sinks)
Overview Talk 1
3
Why ?
Greenhouse gas observations from space
4
„System Earth“
Source: http://energy.lbl.gov/AQ/smenon/images/earth_system.jpg (via IPCC)
Large and increasing human influences We are living in a new geological epoch (TBC), the
„Anthropocene“ (Crutzen, 2000)
A complex system with many positive and negative feedback cycles.
Outgoing long-wave radiation
Incoming solar radation
5
Greenhouse Effect
6 Trenberth, K. E., J. T. Fasullo, and J. Kiehl, 2009: Earth's global energy budget. Bull. Amer. Meteor. Soc., 90, No. 3, 311-324, doi: 10.1175/2008BAMS2634.1.
http://www.bis.gov.uk/go-science/climatescience/greenhouse-effect
1370 W/m2
340 W/m2 100 W/m2 240 W/m2
„Solar constant“
Climate Change: Observations
7
Source: IPCC 2013, AR5, Approved „Summary for Policy Makers“
Mean temperature increase (1880-2012): 0.85 [0.65 to 1.06] °C
Radiative Forcing
8 Source: IPCC 2013, AR5, Approved „Summary for Policy Makers“ 2.3 W/m2 [1.1-3.3] corresponds to
observed 0.85 oC [0.65-1.06]
Solar energy input and long-wave outgoing radiation
9
O3
CO2
H2O
Wavelength in 10-6 meter = µm
0.5 x10-6 meter
= 500 nm
15 x10-6 meter
= 15 µm
Solar Input Heat Output
Doubling CO2: Radiative transfer simulations
10
Source: pload.wikimedia.org/wikipedia/commons/9/9c/ModtranRadiativeForcingDoubleCO2.png
15 µm CO2 Band
Transparent „window“ to space
9.6 µm O3 H2O, CH4, …
H2O, …
+270C
-530C
Rule of thumb (neglecting feedbacks): ΔF = 5.3 ln(CO2-ratio=2) = 3.7 -> 3.7/3 = ~1.2 K = ΔT
260 K
240 K
S 240
L 236
CO2 x2
Ts = 15oC
S 240
L 240
Ts = 15oC
Climate sensitivity: Temperature change for doubling CO2:
Doubling?: Pre-industrial = 280 ppm x2 -> 560 ppm 400 ppm reached in May 2013 (= maximum of seasonal cycle, not yearly average)
„2 degree goal“: ~450 ppm should not be exceeded (Controversial !?: e.g. Hansen argues for < 350 ppm) 11
S 240
L 240
CO2 x2
Ts = 15+1.2 oC
S 240
L 240
CO2 x2 With feedbacks: H2O, íce-albedo,
clouds, ocean, …, ???
Ts = 15+(1.5-4.5) oC
(AR5 likely range equilibrium clim.sensit.) Houghton, 2004
Kohlenstoffkreislauf
12
GtC GtC/yr
Earth‘s breathing
13
Courtesy: D. Crisp (NASA/JPL), GEO-X Meeting, Geneva, 13 Januar 2014
Atmospheric CO2 2002 -2008
Carbon cycle: Sources and sinks
Emissions Atmospheric
growth
Ocean sink
Land sink
Le Quere et al., 2013
NOAA/Scripps CO2 Time series
15
Charles Keeling, 1958
Oil crisis 1973
El Nino 1997
Pinatubo 1992
The more accurate and the longer the observational time series, the more we can learn from analyzing the observations …
Observed Emissions and Emissions Scenarios
Emissions are on track for 3.2–5.4ºC “likely” increase in temperature above pre-industrial Large and sustained mitigation is required to keep below 2ºC
Linear interpolation is used between individual data points Source: Peters et al. 2012a; CDIAC Data; Global Carbon Project 2013
Emissions from fossil fuels and cement
16
RCP: Represenative Concentration Pathway; 8.5: RF(2100) = 8.5 W/m2
Why bother ? Just a few degrees more !
-4oC
? +4oC
17
Ice & snow
Medieval warm period (Vikings in Greenland with cattle & sheep, etc.)
Major crop failures northern Europe, UK, …, „Great Famine“ ~1300
CO2 emissions -> Temperature change
Emitted until 2011: 531 (446-616) GtC
IPCC 2013, AR5 „Summary for Policy Makers“
18
Future: Large uncertainty, e.g., response
of terrestrial biosphere to changing climate ?
Natural CO2: Terrestrial C sinks
19
Houghton, Biologist, 2002: “Strangely, the difference between the net terrestrial sink and the emissions from land-use change suggests that there is a residual terrestrial sink, not well understood, that locked away as much as 3.0 PgC/yr during the last two decades. … The exact magnitude, location and cause of this residual terrestrial sink are uncertain, …”
Gurney et al., Nature, 2002
TransCom 3 regional CO2 flux inversions
Observartions: Very accurate but sparse
Information content sources & sinks (excluding fossil fluxes):
Large regions only (continents, ocean basins)
Large uncertainties (often +/- 100%)
Natural CO2 fluxes from in-situ (mostly surface) obs.
20
A priori land
Within model uncertainty
Inversions:
Left / right: different inversions
Mean flux x
Uncertainty reduction using satellite data - I
21
Rayner and O‘Brien, 2001
Natural CO2 fluxes from space?: • Yes ! If …
Precision < 2.5 ppm for monthly 8ox10o
Uncertainty reduction using satellite data - II
22
Hungershöfer et al., 2010
Weekly mean error reduction
Altitude sensitivity Prior uncertainty
Methane
23
Coal mining
Landfills
Termites
Wastewater
Wetlands Rice
Ruminants
Hydrates
Natural gas
Energy
Methane • Second most important
anthropogenic GHG (directly after CO2)
• Many anthropogenic and natural sources; large uncertainties
Kirschke et al., 2013
? ?
How will sources and sinks behave in a changing climate?
CO2 and CH4 sources and sinks: !! ??
25
Are the reported emissions correct?
How will today's CO2 sinks behave in a changing climate?
How will today's CH4 sources (e.g., wetlands) behave in a changing climate?
Will sinks turn into sources?
Will sources be amplified?
How much is emitted where, when and by what?
How much CO2 is absorbed by land and oceans? Where and when?
How strong are the various sources and sinks ?
Essential Climate Variable (ECV) Greenhouse Gases (GHG)
26
ECV GHG (GCOS-154*)): “Retrievals of greenhouse gases, such as CO2 and CH4, of sufficient quality to estimate regional sources and sinks.”
*) „SYSTEMATIC OBSERVATION REQUIREMENTS FOR SATELLITE-BASED DATA PRODUCTS FOR CLIMATE“
Reliable climate prediction requires a good understanding of the natural and anthropogenic (surface) sources and sinks of CO2 and CH4.
Important questions are, for example:
• Where are they ? • How strong are they ? • How do they respond to a changing climate ?
A better understanding requires appropriate global observations and (inverse) modelling.
CO2 and CH4 are the two most important anthropogenic greenhouse gases and increasing concentrations
result in global warming.
Observed and predicted temperature change (AR5)
Future?
Why ?
• The greenhouse gases (GHGs) carbon dioxide (CO2) and methane (CH4) are „Essential Climate Variables“ (ECVs)
Greenhouse gas observations from space
27 (#) See Talk 2: GHG-CCI results
• Increasing atmospheric concentrations result in global warming
• Reliable prediction of the future climate of our planet requires a good understanding of their (variable) natural and anthropogenic sources and sinks
• Our understanding has large gaps and global satellite data help to improve our understanding, i.e., they help reducing uncertainties on GHG sources and sinks (#)
How ?
Greenhouse gas observations from space
28
Viewing Geometries
Satellite Observation Geometries
Nadir Occultation Limb
SCIAMACHY, AIRS, IASI, TES, GOSAT, OCO-2, CarbonSat, A-SCOPE, MERLIN,
ASCENDS, …
SCIAMACHY,
MIPAS, …
SCIAMACHY, ACE-FTS, …
29
Measurement Techniques
Measurement Techniques
Active
SCIAMACHY,
GOSAT, OCO-2, CarbonSat, …
TOVS, IMG, AIRS, IASI, TES,
MIPAS, …
A-SCOPE, MERLIN,
ASCENDS, …
Passive
Solar Thermal Laser
Reflected solar (NIR/SWIR) vs thermal (TIR) Thermal emission NIR absorption
Passive Active
Sensitive to mid/upper
troposphere
Sensitive (also) to near-surface
GHG concentration
variations ->
Important to get source/sink
info
31
Note:
32
In the following I will focus on • nadir observations in the • solar spectral region using • passive satellite observations
GHG-CCI project www.esa-ghg-cci.org
SCIAMACHY/ENVISAT
Global satellite observations Global information on near-surface CO2 & CH4
TANSO/GOSAT
Upper layer CO2 & CH4
IASI, MIPAS, SCIA/occ, AIRS, ACE-FTS, …
Global observations
Calibrated radiances Calibration (L 0-1)
Reference observations Validation
Inverse modelling
(L 2-4)
Improved information on GHG sources & sinks
Atmospheric GHG distributions
Retrieval (L 1-2)
33
SCIAMACHY SCIAMACHY on ENVISAT SCIAMACHY Scanning Imaging Absorption Spectrometer for Atmospheric CHartographY
Nadir mode: Swath width: 960 km XCO2 & XCH4 from 1.6 & 0.76 µm bands Horizontal resolution: 30 x 60 km2
Tropospheric data products from SCIAMACHY/nadir
... and more.
35
Greenhouse gases via SCIAMACHY/ENVISAT
36
CO2 CH4
XCO2 = CO2 column / Air column
37
H L
CO2
O2
XCO2
XCO2 := CO2 column-averaged dry air mixing ratio (mole fraction)
Buchwitz et al., ACP, 2005
+/- 10% +/- 1.5%
XCO2 ? Counting molecules …
38 XCO2 := CO2 column-averaged dry air mixing ratio (mole fraction)
O2 Vertical Column
Buchwitz et al., 2008
XCO2 = CO2-Column / Air-Column
CO2 Vertical Column
Measurement principle - I
39
Measurement principle - II
40
Measurement principle - III
41
Light path issues
From: SCIAMACHY – Monitoring the changing Earth‘s atmosphere
42
SCIAMACHY nadir spectrum
Buchwitz, 2000; Schneising et al., 2008, 2009
Radiative Transfer
44
e.g., plane-parallel RTE
Change of radiance I, dI, along path ds due to sources (+εS) and/or losses (-εI).
SCIA: IUP-UB retrieval algorithms
45
• XCO2 & XCH4 • Focus on speed and data
volume • Tabulated RT • Least squares fitting • References:
• Buchwitz et al., 2000, 2005 • Heymann et al., 2012 • Schneising et al., 2011, 2012, 2013
Details see ATBDs at http://www.esa-ghg-cci.org/
• XCO2 • Focus on accuracy and
precision • Online RT • Optimal estimation • References:
• Reuter et al., 2010, 2011
WFM-DOAS (WFMD) BESD
46
Measured radiation -> CO2 emissions
Measured Radiation
Atmospheric CO2 or CH4
CO2 or CH4 Emissions A
B
C
Forward- Models
Transport-Chemistry-Model
Radiative-Transfer-Model
Inversion-Models
„Inversion“
„Retrieval“ =
„Inversion“
47
Inversion ? • Different methods
• (Nearly all are) Based on „adjusting“ model parameters until model data „optimally“ agree with the observations
• Requires sufficiently accurate and fast forward models
• Often based on „Bayesian Inference“ or „Optimal Estimation“
Thomas Bayes [beɪz] (~1701 - 1761). • English statistician,
philosopher and Presbyterian minister.
• Bayes‘ Theorem: he suggested using this theorem to update beliefs considering new knowledge. http://en.wikipedia.org/
Bayes‘ Theorem
𝑃𝑃 𝑋𝑋 𝑌𝑌 = 𝑃𝑃 𝑌𝑌 𝑋𝑋 𝑃𝑃(𝑋𝑋)
𝑃𝑃(𝑌𝑌)
P(X = atmos.CO2 | Y = radiation) P(X = CO2 emission | Y = atmos.CO2)
To be minimized cost function (Optimal Estimation (Rodgers, 2000)): + Gaussian statistics
48
Illustration
Y Radiation
or atmospheric CO2, …
X Atmospheric CO2, or CO2 emission, …
xi
yj
P(X, Y)
P(Y) = Σi P(Y|Xi) x P(Xi)
𝑃𝑃 𝑋𝑋 = 𝑥𝑥𝑖𝑖 𝑌𝑌 = 𝑦𝑦𝑗𝑗 = 𝑃𝑃 𝑌𝑌 = 𝑦𝑦𝑗𝑗 𝑋𝑋 = 𝑥𝑥𝑖𝑖 𝑃𝑃(𝑋𝑋 = 𝑥𝑥𝑖𝑖)∑ 𝑃𝑃 𝑌𝑌 = 𝑦𝑦𝑗𝑗 𝑋𝑋 = 𝑥𝑥𝑖𝑖 𝑃𝑃(𝑋𝑋 = 𝑥𝑥𝑖𝑖)𝑖𝑖
P(X)
P(Y)
49
Does God exist ?
http://www.colorado.edu/philosophy/vstenger/Briefs/Bayes.pdf
50
Bayesian inference: Example: Monty Hall problem 1 2 3
3 closed doors
1 car 2 goats
p(C=1) = 1/3
A priori probabilities to win the car: p(C=i): p(C=1) = 1/3 p(C=2) = 1/3 p(C=3) = 1/3
What is the probability to win the car ?
Choose one of the doors. If you pick the right one, the car is yours.
51
Bayesian inference: Example: Monty Hall problem 1 2 3
p(C=1) = p(stay) = 1/3
52
Bayesian inference: Example: Monty Hall problem 1 2 3
p(C=1) = p(stay) = 1/3
I will now open one of the two remainig doors
53
Bayesian inference: Example: Monty Hall problem 1 2 3
p(C=1) = p(stay) = 1/3
p(C=2) = p(switch) = ???
stay or switch ?
Stay or switch ?
54
Bayesian inference: Example: Monty Hall problem 1 2 3
p(C=1) = p(stay) = 1/3
p(C=2 | ???) =
55
Bayesian inference: Example: Monty Hall problem 1 2 3
p(C=1) = p(stay) = 1/3
p(C=2 | D=3) = p(D=3 | C=2) x p(C=2)
p(D=3 | C=1) x p(C=1) + p(D=3 | C=2) x p(C=2) + p(D=3 | C=3) x p(C=3)
1/3
What is the probability that the car is behind door 2 (C=2) given that door 3 (D=3) has been opended
A priori probability
Probability that door 3 has been opended (D=3) given that the car is behind door 2 (C=2)
1.0
0.5 x 1/3 0.0 x 1/3 1.0 x 1/3 + +
56
Bayesian inference: Example: Monty Hall problem 1 2 3
p(C=1) = p(stay) = 1/3
p(switch) = 2/3
So you better switch as this doubles your chance to win the car.
Except if you want a goat …
Retrieval: Optimal Estimation (Rodgers, 2000)
57
CarbonSat: BESD/C algorithm
58
Via pre-processing: Surface albedo
VCF / SIF @ 755 nm Cirrus Optical Depth (COD)
SCIAMACHY / BESD: Example fit
Reuter et al., JGR 2011 59
Jacobian matrix (K): Example
Details: Buchwitz et al., AMT, 2013
Col
umns
of K
Wavelength 60
Jacobian matrix (K): Example
Details: Buchwitz et al., AMT, 2013
Col
umns
of K
Wavelength 61
Zoom into Vegetation Chlorophyll / Solar Induced Fluorescence (VCF / SIF) Jacobian: Spectral region with clear solar Fraunhofer lines for accurate VCF / SIF retrieval
Jacobian matrix (K): Example
Details: Buchwitz et al., AMT, 2013
Col
umns
of K
Wavelength 62
Zoom into Spectral Squeeze Jacobian
Jacobian matrix (K): Example
Details: Buchwitz et al., AMT, 2013
Col
umns
of K
Wavelength 63
Zoom into Water Vapor Jacobian
Jacobian matrix (K): Example
Details: Buchwitz et al., AMT, 2013
Col
umns
of K
Wavelength 64
Zoom into Albedo NIR Jacobian
One could say much more …
Retrieval algorithms: • DOAS (WFM-DOAS, IMAP-DOAS, …), …
• Full Physics (FP) versus Proxy (PR), …
Modelling & inverse modelling: • … Other topics: • …
65
Satellite XCO2 retrieval algorithms … From „First ever“ to „recent“
Buchwitz et al., 2000
Buchwitz et al., 2005
Bovensmann et al., 2010
O‘Dell et al., 2012
Butz et al., 2011
Reuter et al., 2010, 2011
Oshchepkov et al., 2008
Schneising et al., 2011, 2012, 2013, 2014 WFMD
WFMD
Remote-C
WFMD
ACOS
BESD
BESD/C
PPDF
BESD/C
Buchwitz et al., 2013
Reuter et al., 2013 EMMA
66
Satellite XCH4 retrieval algorithms …
Buchwitz et al., 2000
Buchwitz et al., 2005
Bovensmann et al., 2010
Butz et al., 2011
Schneising et al., 2011, 2012 WFMD
WFMD
Remote-C
WFMD BESD/C
Frankenberg et al., 2005 IMAP Parker et al., 2011
Schepers et al., 2012
UoL-FP/PR
Remote-C
67
From „First ever“ to „recent“
Many thanks for your attention !
68