FORMAT-EO SCHOOL GREENHOUSE GAS REMOTE SENSING PROFESSOR JOHN REMEDIOS DR. HARTMUT BOESCH EOS-SRC, Dept. of Physics and Astronomy, University of Leicester
What is covered in this lecture? Climate, radiation and greenhouse gases Principles of atmospheric radiative transfer Greenhouse gases in the SWIR Tutorial 1: Understanding radiative transfer (Thermal infra-red) Tutorial 2: CO 2 mixing ratios and fluxes
Section 1 Climate radiation and greenhouse gases
TRENDS IN SURFACE TEMPERATURE [From the IPCC reports, http://www.ipcc.ch] T ANOMALY = T - <T> IPCC 2007 SST rise: Avg(2001/2005) Avg(1850/1899) = + 0.76 C. Last 50 yrs rise = +0.13 C/decade. Projected is 0.2 C/decade for next 2 decades. IPCC 2001 Left panel: global-mean surface temperature changes since about 1850 as observed and as simulated by a climate model forced with estimated anthropogenic, solar and volcanic aerosol external forcings. Right panel: Global-mean radiative forcing, relative to 1750, due to a variety of mechanisms with an indication of their uncertainties (IPCC, 2001). IPCC 2007 updates. Note the separation into land and ocean Black= smoothed mean obsservations Blue=simulations, natural only (solar, volcanoes) Red=simulations, natural and anthropogenic factors Prof. J. Remedios, Leicester. 3653, Lecture 4 1 IPCC 2007
Climate, radiative balance and space observations Climate and radiative balance The first essential balance of climate is energy, i.e., energy into a planet = energy lost from planet. This balance controls temperature. The instantaneous balance is power in Wm -2. A fraction of energy from the sun is absorbed (heating) in the atmosphere and at the surface; the remainder is scattered/reflected to space. Energy is also lost from the Earth by emission of radiation from the surface and atmosphere to space (cooling). This is fundamental. Space Observations The spectrum of energy lost to space from the Earth can be observed by instruments in space looking back at the Earth. Reflected radiation from the Sun = ultra-violet/visible radiation (strictly 0.3 to 4.0 mm) Emitted radiation from the Earth = infra-red and microwave radiation (typically 4 to 200 mm). 5
METEOSAT-7 VISIBLE COLOUR 6
METEOSAT-7 INFRA_RED 7
RADIATIVE TRANSFER AND RADIATION BUDGET ULTRA-VIOLET/VISIBLE HEATING INFRA-RED COOLING Prof. J. Remedios, Leicester. 3653, Lecture 1 8
BREAKING DOWN THE SPECTRUM VISIBLE/SWIR INFRA-RED 9
Greenhouse gases (GHGs) I Natural greenhouse gases: water vapour ozone Not affected directly by human activity but can be indirectly influenced as we have seen [This is the reason why ozone appears in the radiative forcing tables] Anthropogenic gases include principally: carbon dioxide methane nitrous oxide, CFCs (and a number of other less abundant species). Gases like carbon dioxide and methane are not only human-produced but are also natural i.e. they existed thousands of years ago. What is different are their concentrations now and their rates of change. [One could argue that they have changed so rapidly that they have changed the definition of climate to apply to a minimum time period of a decade] Gases like CFCs are essentially human-produced. 10
Radiative forcing and surface temperature change Need a defined term: Radiative forcing (RF) can be considered as The effective change in the radiative energy balance of the Earth system between modern and preindustrial times. Example Units are Wm -2 (power) IPCC 2007(4 th assessment report) looked at 2005 relative to pre-industrial times (1750) A simple linear model to estimate surface temperature, Ts, change: ΔTs = λ x RF where λ is a climate sensitivity parameter and is calculated with a model (the amount of surface T change per unit change in power). RF (CO 2 ) 1.5 Wm -2 ; λ 0.8 K/(Wm -2 ). Hence expect ΔTs 1.2 K which is a bit high compared to observations. 11
RADIATIVE FORCING III [From the IPCC 2007 report, http://www.ipcc.ch; new rpt 2013] Updated radiative forcing diagram from the IPCC 2007 report RED(ish)= + Surface T tendency (heating) BLUE = - Surface T tendency (cooling) Prof. J. Remedios, Leicester. 3653, Lecture 1 12
Section 2 Principles of radiative transfer
Recap: Principle of Remote Sensing Observations Sensor (absorbing gases, clouds, surface ) Measurement Data Analysis Information on atmosphere/ surface
Interaction of EM radiation with atmosphere and surface sun Rule of thumb: Shortwave (Solar): Emission can be ignored Longwave (IR): Scattering can be ignored Absorption Scattering Scattering from a cloud Emission from a cloud Atmosphere Cloud Aerosol / Molecules Scattering / reflection from a cloud Scattering within a cloud Transmission through a cloud Emission Absorption on the ground Reflection on the ground Emission from the ground
Recap: Interaction of EM radiation with atmosphere and surface sun Rule of thumb: Shortwave (Solar): Emission can be ignored Longwave (IR): Scattering can be ignored Absorption Scattering Scattering from a cloud Emission from a cloud Atmosphere Aerosol / Molecules Absorption on the ground Scattering / reflection from a cloud Reflection on the ground Cloud Scattering within a cloud Transmission through a cloud Emission How can we use molecular absorption to quantify concentration in atmosphere? How do we separate molecular absorption from all other processes? Emission from the ground
Radiative Transfer (RT) Equation I di( ) ds -k s ( ) I( ) - k loss by scattering a ( ) I( ) ( ) B(, T) loss by absorption e gain by emission S MS ( ) gain by multiple scattering
Radiative Transfer (RT) Equation II Only for shortwave Only for longwave di( ) ds -k s ( ) I( ) - k loss by scattering a ( ) I( ) ( ) B(, T) loss by absorption e gain by emission S MS ( ) gain by multiple scattering
RT Equation III Purely Absorbing Case di( ) ds -k s ( ) I( ) - k loss by scattering a ( ) I( ) ( ) B(, T) loss by absorption e gain by emission S MS ( ) gain by multiple scattering
TRANSMISSION (Purely Absorbing Case) I o () 0 L GAS: k a (), c I t () Lambert-Beer Law: di() = k a () dl T() = exp [ σ() c L ] with c = the density of molecules per unit volume. Three factors matter: Spectroscopy: absorption cross section σ() [cm 2 /molecule] Composition/density: c = c air [molecules/cm 3 ] Photon pathlength: geometrical distance = L [km] If we want to infer c (or ), we need to know speciesspecific absorption cross section σ() and photon path L
Absorption Spectroscopy Main assumption: no scattering Lambert Beer law: L I( L) I exp - ( L') c( L') dl 0 and thus Optical depth 0 ' t ( L) -ln( I / I0) ( T, p) SCD absorption cross section for mean temperature and pressure SCD For atmospheric measurements, a more useful quantity is the vertical column density VCD: number of molecules per area along the vertical direction for the whole atmosphere VCD 0 c( L') dl' Slant column density gives the number of molecules per area along the path L How can you obtain VCD from SCD? 0 c( L') dl'
Example: Direct-Sunlight Observations The principal of absorption spectroscopy is valid for directsunlight observations: No surface effects High intensity and thus scattering into observerdirection is not important Scattering will lead to broad reduction of total intensity but will not change optical depth Solar Zenith Angle SZA Ground-based Fourier Transform Spectrometer Direct sunlight observations: Relation between slant path and vertical path is given by geometric factor of 1/cos(SZA)
Section 3 Greenhouse gases in the SWIR
RT in the Shortwave (solar) di( ) ds -k s ( ) I( ) - k loss by scattering a ( ) I( ) ( ) B(, T) loss by absorption e X gain by emission S MS ( ) gain by multiple scattering RT with scattering and absorption: Emission term can be omitted Atmospheric scattering will have a strong effect on light path which we need to know In general: RT equation is system of coupled differential equations, ie. both, absorption and scattering at on altitude z will impact the intensity all other altitudes
Where is the light coming from for a satellite looking downward? A: Simplest situation: plane parallel, no extinction B: plane parallel, with absorption SZA L O S SZA L O S H A A SZA and LOS independent of altitude! 0 s I I 1 1 cos( ) cos( ) 0 A => Clearly wrong for low sun! Lambert-Beer Law: I I I 0 0 e Ae H 0 H 0 -k ( h)/ cos( ) dh a -k ( h)/ cos( ) dh -( k ( h)/ cos( ) k ( h)/ cos( )) dh a a H 0 Ae a
Where is the light coming from for a satellite looking downward? C: plane parallel, with absorption and single scattering D: spherical, with absorption and multiple scattering L O S SZA Offset for clarity only! A Intensity measured at top of atmosphere is sum of Light reflected from surface Scattered light from different altitudes Can still be written analytically as extension of Lambert Beer Law Valid for small angles and little scattering A Reality is more complicate: Multiple scattering Spherical geometry RT equations needs to be solved
DOAS DATA ANALYSIS Lambert Beer Law: Transmitted intensity (absorption and scattering) after traveling path L Sum over all absorbers I( ) I0( ) exp - ( i(, T, p) ci - kr( ) - km ( )) dl L i Measured Spectrum (Solar) Reference Spectrum Molecular Absorption Rayleigh and Mie Scattering Define Model Function: F() Fit model function F to the log(i(λ)): 2 n (ln i1 I( i) - F( i)) ( i) 2 min. ε: noise of measurement i: index for spectral point Fitted parameters are: Slant column densities SCD i for each gas and coefficients of polynomial a p
SCanning Imaging Absorption spectrometer for Atmospheric CHartographY (SCIAMACHY) Multi-channel grating spectrometer that covers spectral range from UV (~240 nm) to near-infrared (~2380 nm) with moderate spectral resolution between 0.2-1.5 nm Limb, nadir and occultation geometry Good spatial resolution (typically 30 x 60 km 2 ) Global coverage within 3-4 days Launched March 1st, 2003 on ENVISAT Successor of GOME instrument launched April 1995
SCIAMACHY Spectrum A large number of species as well as cloud and aerosol information can be retrieved from SCIAMACHY spectra Ozone NO 2 H 2 O H 2 O H 2 O SCIAMACHY Channels Channel Range (nm) Resolution (nm) / Resolving power O 2 CO 2 CH 4 1 214-334 0.24 / ~1000 2 300-412 0.26 / ~1400 H 2 O CO 2 3 383-628 0.44 / ~1200 4 595-812 0.48 / ~1500 CO 2 N 2 O, CH 4 CO, CH 4 5 773-1063 0.54 / ~1700 6 971-1773 1.48 / ~1000 7 1934-2044 0.22 / ~9000 8 2259-2386 0.26 / ~9000 Similar instruments: GOME and GOME-2, OMI, Sentinel 5 precursor (>2015)
OCO Measurement Approach Measurement of SWIR CO 2 and O 2 bands to retrieve information on scattering (aerosol/clouds) together with CO 2 : 1.61 mm CO 2 band: Column CO 2 2.06 mm CO 2 band: Column CO 2, clouds/aerosols, H 2 O, Temperature 0.76 mm O 2 A-band: Surface pressure, clouds/aerosols, Temperature OCO has been specifically designed for CO 2 column observations High spectral resolution Large number of key parameters can be retrieved independently Enhanced sensitivity and minimized biases due to interferences O 2 A band Spectral Bands of OCO 2.06 mm CO 2 band 1.61 mm CO 2 band Aerosols, p surf, T CO 2 Column CO 2, H 2 O, Aerosols
Accurately retrieving CO 2 (and CH 4 ) is extremely difficult and time-consuming: The OCO-1 Full-Physics Retrieved CH 4 and CO 2 will depend on assumptions of retrieval algorithm (retrieval biases) Forward Model needs to describes accurately physics of measurement: Multiple-scattering RT Polarization Correction Spherical Geometry Surface (polarized) BRDF Instrument Model Solar Model Retrieval Algorithm Inverse Method estimates state: Rodger s optimal estimation technique (Boesch et al., 2006, 2011, O Dell et al., 2011)
Greenhouse gases Observing SATellite (GOSAT) launched January 23rd 2009 Mission objectives: 1) To monitor the density of greenhouse gases precisely and frequently worldwide. 2) To study the absorption and emission levels of greenhouse gases per continent or large country over a certain period of time. 3) To develop and establish advanced technologies that are essential for precise greenhouse-gas observations.
The GOSAT Payload TANSO - FTS Provides spectrallyresolved 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 TANSO - CAI 4 broadband channels from UV to SWIR with high spatial resolution Provides aerosol and cloud information required for the greenhouse gas retrieval Sampling Pattern Used for CO 2 and CH 4 retrieval
The Aerosol and Cloud Problem Aerosol and clouds are highly variable in amount, vertical distribution and optical properties Poses a critical problem for the retrieval Incorrect description can bias retrieval already for relatively small AODs What are the options? Use the available information in the measurements Use external information (models, other satellites) Alternative retrieval approaches without explicit descriptions of aerosols e.g. CH 4 /CO 2 proxy approach O 2 A band 2.06 mm CO 2 band 1.61 mm CO 2 band Provides some constraint on vertical distribution and spectral characteristics of aerosols/clouds
The X CO2 and X CH4 Retrieval Full physics CO 2 retrieval (Not yet done global FP CH 4 ) Simultaneous 3-band fit to retrieve CO 2 together with additional aerosol, surface and atmospheric variables CH 4 proxy retrieval: CO 2 column from spectrally-close window is used as proxy for the unknown light path for the CH 4 retrieval (Frankenberg et al., 2008): XCH4 proxy XCH4 XCO2 retrieved retrieved XCO2 Both datasets are generated for the ESA CCI program (GHG-CCI) model Very simple, fast retrieval Reduced sensitivity to aerosols/clouds and instrument calibration Requires accurate model for atmospheric CO 2 (see Schepers et al., 2012)
Validation against ground-based TCCON TCCON (Total carbon column observing network) network of groundbased Fourier Transform Spectrometers Provides precise, accurate total columns of CO 2, CH 4 and others gases Ideal for satellite validation, but there are potential differences in Spectroscopy A priori assumption for CO 2 and CH 4 Vertical sensitivity (averaging kernels) But, we need more sites in Africa, S-America and Asia Upcoming site in Korea TCCON calibration against in-situ data from aircraft profiles Wunch et al., AMT, 2010
FP CO 2 (v4.0): Validation against TCCON Strict quality filtering (clouds, AOD, cirrus, χ 2 etc.) Bias-correction scheme has been applied Update from A. Cogan, H. Boesch et al, JGR, 2012
FP CO 2 (v4.0): Validation against TCCON Strict quality filtering (clouds, AOD, cirrus, χ 2 etc.) Bias-correction scheme has been applied Update from A. Cogan, H. Boesch et al, JGR, 2012
Next Steps: A Continuous Presence Geostationary GHG Mission? A coordinated global network of surface and space-based CO 2 monitoring systems are needed for long-term monitoring of atmospheric sources and sinks and to provide insight into processes controlling atmospheric CO 2 As a first step, existing and planned CO 2 monitoring satellites could be coordinated into an ad-hoc network to improve coverage and resiliency
TUTORIAL 1 Illustrating the atmosphere remote sensing problem with thermal infra-red
Absorption and transmission in the IR Total Signal I o () GAS a(), T() I t () 2 nd signal Total Signal, I() is given by > I() = T() I o () + ( 1- T() ) x B(, T) Transmittance T = exp-{k a L} Transmitted intensity Emitted (and transmitted) intensity Note: in IR, Scattering can usually be omitted
SINGLE LAYER GAS MODEL We can use the single gas layer model to understand the Physics (infra-red) involved again. I() = s () x B(,T s ) x T g(l) + (1- T g()) x B(,T g ) I t () g () B(, T g ) Atmosphere GAS T g () Surface (T s, ε s ) I o () = s () x B(,T s ) Observed signal will depend on both T g(, z) and B(,T g ) (and surface term) and hence on both composition and temperature More realistic description of atmosphere by combining multiple single layers
CONTROLLING TRANSMISSION So how is T() controlled? Remember: - T (, z ) exp( k ( z') dz' ) where k=σ*c with the spectroscopic term σ, concentration c and height z Imagine the spacecraft looking down at the surface at a wavelength where only 1 gas absorbs. Then for the strongest lines, k() is largest and T() reaches zero at higher altitude z (or shorter path) We end up with (T=temperature): toa z a z STRONGEST LINE INTERMEDIATE LINE WEAKEST LINE T(Z1) T(Z2) T(Z3)
EXAMPLE: NADIR IR RADIATIVE TRANSFER Sahara In the presence of absorber, we see atmospheric temperatures of different heights (depending on k a ) Mediterrean Antarctica 15 mm 10 mm 7 mm In each of diagrams: Surface T is different (180-320 K). Signal at centre of CO 2 band is always at 220K because T() is zero at same height in atmosphere
INFRARED NADIR RADIATIVE TRANSFER Schwarzschild s eqn: I(λ) = s B(λ,T s ) T(λ,) + B(λ,T z ) dt(λ,z)/dz dz + R S I Down (atm) τ(λ,) 1 st term: Surface emission, s B(,T s ), modified by transmission of atmosphere between ground and space, T(l,) 2 nd term: Emission from Planck fn. at each height in atmosphere, B (,T z ), modified by dt(λ,z)/dz 3 rd term: Reflection, R s, of radiation emitted from atmosphere back towards ground (often omitted) Schwarzschild s eqn. is consistent with single layer gas model. Indeed you can build up atmosphere as a series of single layers Transmission is now between altitude z and top of atmosphere toa - T (, z ) exp( k ( z') dz' ) z a z=toa T=1 z=z T= T z=0 T= T min T(z )
WEIGHTING FUNCTIONS: NADIR SOUNDING I The contribution of layer to Schwarzschild eqn. is given by 2 nd term: I() = s B(,T s ) T(,) + B(,T z )dt(,r,t z )/dz dz + R S I Down (atm) T(,) Surface contribution Atmospheric contribution The term W(z) = dt(,r,t z )/dz weights the Planck function at each height, z, by change of T() at z The term W(z) = dt(,r,t z )/dz is known as a weighting function and its peak and width determine which part of the atmosphere is sensed Often, we are interested in the sensitivity of the radiance to changes in the gas concentration at each layer, not the contribution to the total radiance from each layer. Gas weighting functions are more complicated because they depend on the potentially variable gas concentration itself.
WEIGHTING FUNCTIONS: NADIR SOUNDING II Weighting functions for nadir sounding are broad (8-15km) and bellshaped Maximum of weighting function: dw(z)/dz = 0 -> T() = exp(-1) (for well mixed gas) Weighting functions for lines of different optical depth provide information from different height NADIR, MULTI- Each curve = different T() = exp(-1) wavenumber Weaker (more transmitting lines) have peak weighting lower in atmosphere Question: What happens for very weak lines where T() is smaller than 0.36? Temperature Transmissivity T Weighting Function dt/dz