IASI for Surveying Methane and Nitrous Oxide in the Troposphere: MUSICA products and its validation

Transcription

1 IASI for Surveying Methane and Nitrous Oxide in the Troposphere: MUSICA products and its validation Omaira García M. Schneider, B. Ertl, F. Hase, C. Borger, E. úlveda, T. Blumenstock, Uwe Raffalski and A.J. Gómez-Peláez. Remote Sensing of CH 4 and N O I. General retrieval characteristics II. A posteriori calculated difference between CH 4 and N O. Theoretical Characterisation I. Representativeness II. Error Assessment 3. Validation by using a Multi-Platform Reference Database I. HIPPO II. GAW in situ, and NDACC/FTIR 4. A posteriori CH 4 Correction 5. Summary and Outlook

2 . Remote sensing of CH 4 and N O General Retrieval Characteristics IASI processor developed during the ERC project MUSICA (MUlti-platfrom remote Sensing of Isotopologues for investigating the Cycle of Atmospheric water), based on the retrieval code PROFFITnadir [Schneider and Hase, ] Optimal estimation retrieval: combine a priori information with the measured IASI spectra and estimate the most likely atmospheric state. A priori information is kept constant (no variation in space and time), i.e., all the retrieved variability is introduced by the IASI spectra. Simultaneous retrieval of H 6 O, HD 6 O, CH 4, N O and HNO 3 temperature on log. scale. Key!! (+CO ), atmospheric and skin The spectral signatures of H O variations are more than an order of magnitude stronger than the signatures of CH 4 and N O Quality of CH 4 and N O products strongly depends on a correct interpretation of H O intereferences MUSICA IASI incorporates a sophisticated H O isotopologue retrieval and water continuum contributions (MT_CKD v.5.)

3 . Remote sensing of CH 4 and N O Long-term monitoring Example of continuous time series of MUSICA IASI CH 4 daily mean data retrieved at 4. km altitude in the surroundings of Tenerife Island between 7 and 7. Global coverage Example of global geographical distribution of the free tropospheric MUSICA IASI CH 4 product retrieved at 4. km altitude, filtered for c sen < 5% and averaged for an latitude x longitude area of ºxº. IASI has a great potential for CH 4 and N O monitoring and for greenhouse gas cycle studies 3

4 . Theoretical Characterisation Example for mid-latitude summer land pixel Representativeness: Averaging Kernels Moderate sensitivity in free troposphere But, the sensitivity varies with latitude, season, Altitude regions that well-detectable by MUSICA IASI products: c observation scenario (land/ocean) sen <5% and time of Good sensitivity in UTLS MUSICA IASI products can capture satellite atmospheric overpass variations (morning/evening) of CH 4 and N O between -6 km with a vertical resolution of 5-8 km Are the MUSICA IASI vertical resolution and the sensitivity good enough to detect typical CH 4 and N O variations (vertical MUSICA width of IASI 5 km)? CH 4 data offer a better sensitivity than N C, atmospheric covariance O data matrix 4

5 . Theoretical Characterisation Example for mid-latitude summer land pixel Error Assessment Statistical sources (Gaussian distribution): Measurement noise < % Emissivity <.5% HO cross-dependency <.5% Atmospheric temperature [-3]% Non-Gaussian sources: Spectroscopy [-, ]% Water continuum [-,]% Clouds [-4.5, ]% Latitudinal Cuts of Leading Errors (atmospheric temperature and measurement noise) 5

6 3. Validation by using a Multi-Platform Database Ground-based NDCC/FTIR In situ Picarro HIPPO (HIAPER Polo-to-Pole Observation) aircraft profiles High precision&accuracy, high vertical resolution, good latitudinal coverage (67ºS-8ºN, Pacific Ocean) Accuracy, precision, profiling capability and latitudinal gradients GAW (Global Atmosperic Watch) in situ and NDACC/FTIR (Fourier Transform Infrared Spectrometer) High precision&accuracy, continuous (GAW), quasi continuous (daytime, FTIR), long-term series Precision, profiling capability and temporal signals that are detectable [Details on collocation, data treatment and filtering are given in Extra Material slides] 6

7 3. Validation by using a Multi-Platform Database HIPPO (uary 9) HIPPO (ember 9) HIPPO3 (ch/il ) HIPPO4 (e ) HIPPO5 (ust/ ) MUSICA IASI data capture well CH 4 latitudinal gradients, but not for N O (very small variations). +.±.7% -.±.% Latitudinal dependency of bias (according to IASI sensitivity). Precision:.5-.5% Accuracy: ±% -.5±.5% +.5±.% 7

8 3. Validation by using a Multi-Platform Database N O CH 4 CH 4 * CH 4 ' MUSICA IASI at 4. km [ppmv] N O N=7 R =3% N=63 R =3% GAW in situ [ppmv] N=563 R =% N=76 R =9% CH N=55 R =% CH 4 * N=55 R =43% CH 4 ' Year Tenerife, 8ºN, 7-7 Karslruhe, 49ºN, -7 Kiruna, 68ºN, 7-7 GAW and NDACC/FTIR comparison confirms the HIPPO results over time. The overall precision: -% for N N=745 O and -3% for CH 4 (free troposphere R 8% and UTLS)..9 N=745 Year R =44% But, what temporal signals 3 are really detectable by the MUSICA IASI products? NNDACC/FTIR O at 4. km [ppmv] CH Temporal CH 4 * decomposition CH 4 of ' a time series into signals belonging to different timescales: MUSICA IASI at UTLS [pmv] N=735 R =4% NDACC/FTIR at UTLS [[ppmv] N=769 R =37% Year.9 7 xx tt = xx mm ([tt, tt ]) + ss tt + 8 ll tt + dd(tt) 9 Reference.8 value 3 4 N=76 R =% N=76 R =3% Seasonal cycle Long-term signal Day-to-Day variations [Comparison for Karlsruhe and Kiruna is shown in Extra Material slides] 8

9 3. Validation by using a Multi-Platform Database Seasonal cycle relative to long-term background [%] Tenerife, 8ºN, 7-7 Karlsruhe, 49ºN, -7 Temporal omposition xx tt = xx mm ([tt, tt ]) + ss tt + ll tt + dd(tt) GAW in situ MUSICA IASI data capture well the CH 4 seasonal cycles (phase and amplitude) at free troposphere only at subtropical latitudes. NDACC/FTIR at 4. km NDACC/FTIR at UTLS Not for N O (very small variations in free troposphere). For all latitudes and for both N O and CH 4 MUSICA IASI data capture well the seasonal cycles (phase and amplitude) in the UTLS. MUSICA IASI red, GAW and NDACC/FTIR black 9

10 3. Validation by using a Multi-Platform Database Long-term signals: Deseanolised montlhly mean Temporal omposition Tenerife, 8ºN, 7-7 Karlsruhe, 49ºN, -7 xx tt = xx mm ([tt, tt ]) + ss tt + ll tt + dd(tt) MUSICA IASI at 4. km [ppmv] MUSICA IASI at UTLS [ppmv] GAW in situ [ppmv] NDACC/FTIR at 4. km [ppmv] NDACC/FTIR at UTLS [[ppmv] L PPF V4 V5./ V5./3 V6 MUSICA IASI data are affected by an inconsistency between the versions EUMETSAT L PPF v5./3 and V6 (used as a priori information). When this inconsistency is not considered (Karlsruhe) the agreement with GAW in situ suggests that MUSICA IASI data detects the long-term variations in the free troposphere up to midlatitudes. The colour code identifies the four different EUMETSAT L Products Processing Facility (L PPF) software versions. [Details on L PPF versions are given in Extra Material slides].

11 4. A posteriori CH 4 Correction A posteriori calculated difference between CH 4 and N O Day-to-day N O Variability [%] MUSICA IASI R=.58 NDACC/FTIR R=.5 GAW in situ R= Izaña Observatory Day-to-day CH 4 Variability [%] N O concentrations show a rather low short-term variability and their long-term increase is very stable Significant variations in the N O retrievals are due to errors Assuming that N O and CH 4 have correlated errors (e.g. temperature, clouds, ) and common signals (e.g. UTLS shift) Combining co-retrieved N O and CH 4 to generate a combined product with reduced errors and common signals, and then better representativeness of sources/sinks signals Retrieved states A priori Averaging kernels Difference of errors is smaller than individual erros We propose two approaches for calculating the combined products based on: () N O model simulations (CH 4 *) and () N O climatology (CH 4 ) (details on Extra Material slides). But, both can be characterised by the {(ln[ch 4 ] - ln[n O])} state.

12 4. A posteriori CH 4 Correction Sensitivity Similar vertical distribution, but slightly lower sensitivity for the combined product Leading Errors The a posteriori CH 4 correction with co-retrieved N O offers better precision!!!

13 4. A posteriori CH 4 Correction Better estimation of latitudinal gradient High precision xx tt = xx mm ([tt, tt ]) + ss tt + ll tt + dd(tt) -3.7±.7% -.±.% +.5±.% +.5±.6% 3

14 5. Summary and Outlook The MUSICA IASI data capture signals that are larger than -%, like the latitudinal gradients, the long-term increase and the seasonal cycles in the UTLS region. While for N O the sensitivity is mainly limited to the UTLS region, for CH 4 at low latitudes the MUSICA IASI processor can detect variations that take place in the free troposphere independently from the variations in the UTLS region. Room for improvements, but the benefits are limited (other sophisticated retrievals point out the same results, e.g. rough profile capability only for lower and middle latitudes). Our next plans: () Performing retrievals for many orbits (producing data globally for longer time periods) IASI A+B measures about.5 Millions spectra per day! We focus on the about % that are cloud-free (.5 Millions) () Examining the usefulness of IASI CH 4 and N O data for source/sink studies By combining IASI space-based observations and model estimates, we will investigate the kind of CH 4 and N O sink/source signals at a global scale that can be captured by high quality IASI observations, and the transport of the CH 4 and N O in the atmosphere. 4

15 Many Thanks for your Attention!! Acknowledgements: This work has benefit from funding by the European Research Council under FP7/(7-3)/ERC Grant agreement n 5696 (project MUSICA), by the Deutsche Forschungsgemeinschaft for the project MOTIV (Geschäftszeichen SCHN6/), by the Ministerio de Economía y Competitividad from Spain trough the projects CGL-3755 (project NOVIA) and CGL6-8688P (project INMENSE), by the Ministerio de Educación, Cultura y Deporte (Programa "José Castillejo", CAS4/8), and by EUMETSAT under its Fellowship Programme (project VALIASI). Furthermore, we would like to acknowledge the National Science Foundation (NSF) and the National Oceanic and Atmospheric Administration (NOAA), that supported the collection of the original HIPPO data. Izaña Atmospheric Observatory 5

16 References Barthlott, S., et al.: Using XCO retrievals for assessing the long-term consistency of NDACC/FTIR data sets, Atmos. Meas. Tech., 8, , 5. Barthlott, S., et al., : Tropospheric water vapour isotopologue data (H6O, H8O, and HD6O) as obtained from NDACC/FTIR solar absorption spectra, Earth Syst. Sci. Data, 9, 5-9, doi:.594/essd-9-5-7, 7. Borger, C., et al.: Evaluation of MUSICA MetOp/IASI tropospheric water vapour profiles by theoretical error assessments and comparisons to GRUAN Vaisala RS9 measurements, Atmos. Meas. Tech. Discuss., in review, 7. Christner, E.: Messungen von Wasserisotopologen von der planetaren Grenzschicht bis zur oberen Troposphäre zur Untersuchung des hydrologischen Kreislaufs, Dissertation, Karlsruhe Institute of Technology (KIT), Germany, 5. Christner, E., et al.: The influence of snow sublimation and meltwater evaporation on δd of water vapor in the atmospheric boundary layer of central Europe, Atmos. Chem. Phys., 7, 7-5, doi:.594/acp-7-7-7, 7. Dyroff, C., et al.: Airborne in situ vertical profiling of HDO/H6O in the subtropical troposphere during the MUSICA remote sensing validation campaign, Atmos. Meas. Tech., 8, 37-49, 5, doi:.594/amt García, O. E., et al.: Upper tropospheric CH4 and NO retrievals from MetOp/IASI within the project MUSICA, Atmos. Meas. Tech. Discuss., doi:.594/amt-6-36, in review, 7. González, Y., et al.: Detecting moisture transport pathways to the subtropical North Atlantic free troposphere using paired HO-δD in situ measurements, Atmos. Chem. Phys., 6, , doi:.594/acp , 6. Rodgers, C.: Inverse Methods for Atmospheric Sounding: Theory and Praxis, World Scientific Publishing Co., Singapore,. Rodgers, C. and Connor, B.: Intercomparison of remote sounding instruments, J. Geophys. Res., 8, 46 49, doi:.9/jd99, 3. Schneider, M. and F. Hase: Optimal estimation of tropospheric HO and δd with IASI/METOP, Atmos. Chem. Phys.,, ,. Schneider, M., et al.: Ground-based remote sensing of tropospheric water vapour isotopologues within the project MUSICA, Atmos. Meas. Tech., 5, 37 37, doi:.594/amt-5-37-,. Schneider, M., et al.: "Fourier Transform Infrared Spectrometry", Chapter 6 of "Monitoring Atmospheric Water Vapour - Ground-Based Remote Sensing and In-situ Methods", ISSI Scientific Report Series, Vol., Kämpfer, Niklaus (Ed.), ISBN , 3. Schneider, M., et al..: Empirical validation and proof of added value of MUSICA's tropospheric δd remote sensing products, Atmos. Meas. Tech., 8, , doi:.594/amt , 5. Schneider, M., et al.: Accomplishments of the MUSICA project to provide accurate, long-term, global and high-resolution observations of tropospheric {HO,δD} pairs a review, Atmos. Meas. Tech., 9, , doi:.594/amt , 6 Schneider, M., et al.: MUSICA MetOp/IASI {HO,δD} pair retrieval simulations for validating tropospheric moisture pathways in atmospheric models, Atmos. Meas. Tech.,, 57-55, doi:.594/amt--57-7, 7. Wiegele, A., et al.: The MUSICA MetOp/IASI HO and δd products: characterisation and long-term comparison to NDACC/FTIR data, Atmospheric Measurement Techniques, 7, 79 73, doi:.594/amt , 4. 6

17 Extra Material AEMET, Agencia Estatal de Meteorología 7

18 Project MUSICA MUlti-platfrom remote Sensing of Isotopologues for investigating the Cycle of Atmospheric water, European Reseach Council, -6. Objetive: Generating robust tropospheric {H O, δd} pairs using ground-based NDACC/FTIR (Fourier Transform Infrared Spectrometer) and space-based MetOp/IASI (Infrared Atmospheric Sounding Interferometer). () ground-based FTIR Mexico Addis () space-based IASI (3) in-situ, surface + aircraft MUSICA integrates ground- & space-based remote sensing, and in-situ observations Publications: Schneider and Hase, ; Schneider et al., ; Schneider et al., 3; Wiegele et al., 4; Barthlott et al., 5; Christner, 5; Dyroff et al., 5; Schneider et al., 5; González et al., 6; Schneider et al., 6; Barthlott et al., 7; Borger et al., 7; Christner et al., 7; García et al., 7; Schneider et al., 7. And more than publications related with MUSICA activities. 8

19 Theoretical Characterisation (Continuation) Assumed Uncertainties in the Error Assessment Uncertainty Source Uncertainty Value Uncertainty Source Uncertainty Value Measurement noise Pequignot et al. (8) Line intensity and pressure broadening NO % Temperature -km K Line intensity and pressure broadening CH4 -% Temperature -5km, 5-km K Opaque cumulus cloud % fractional cover with cloud top at.3, 3. and 4.9 km Temperature above km K Cirrus cloud Particle properties OPAC Cirrus3 km thickness, 5% fractional cover with cloud top at 6, 8, and 4 km Surface emissivity % at 85, 4, 95, 35, and 45 cm - Mineral dust cloud Particle properties OPAC Desert Homogeneous coverage for layers: ground- km, -4 km and 4-6km Water vapour continuum % underestimation {H O,δD} variations {%,%}, with.5km of correlation length Error vertical profiles for a mid-latitude summer land pixel 9

20 Collocation, Data Filtering and Data Treatment Collocation and data filtering HIPPO: I. All IASI observations within ºxº latitude/longitude box around the mean location of each HIPPO profile, ±h around the mean time of each HIPPO profile. II. Altitude of HIPPO profiles: free troposphere (4.8 km) up to 8 km, for ULTS (9.8 km) up to.5 km GAW&NDACC/FTIR: I. All IASI observations within ~ km box south of GAW&NDACC/FTIR stations II. Night-time means (GAW in situ), ±8h (NDACC/FTIR), evening and morning overpass (IASI) MUSICA/IASI: c sen <5%, cloud free pixels, residual-to-signal in the fitted window<.4, skin temperature>75k HIPPO Data Treatment I. The vertically highly-resolved HIPPO profiles are degraded applying the IASI averaging kernels: II. The HIPPO profiles are extended by the a priori data used by the MUSICA IASI processor. Comparison of MUSICA IASI and NDACC/FTIR remote sensing data I. The NDACC/FTIR data are adjusted to the unique MUSICA a priori data by adding (A FTIR - I)(x a - x a;ftir ) to the NDACC/FTIR retrieval results (being A FTIR and x a;ftir the averaging kernels and a priori statecorresponding to the NDACC/FTIR retrieval, respectively) [Rodgers and Connor, 3]. II. MUSICA IASI N O data are compared with smoothed NDACC/FTIR data at all altitudes. The free tropospheric MUSICA IASI CH 4 and CH 4 data are also compared to smoothed NDACC/FTIR data. However, in the UTLS region the MUSICA IASI and NDACC/FTIR CH 4 and CH 4 data are directly compared (no prior smoothing of the NDACC/FTIR data).

21 Validation by using a Multi-Platform Database (Continuation) gfraujoch / Karlsruhe Seasonal cycle signal re Kiruna, 4. km Karlsruhe, 4. km Seasonal cycle signal relative to long-term ba (e) x (e) N O CH 4 CH 4 * CH (f) (g) (h) (f) (g) (h) - (i) (j) (k) (l) Temporal omposition xx tt = xx mm ([tt, tt ]) + ss tt + ll tt + dd(tt) Seasonal cycle relative to longterm background [%] Karslruhe, 49ºN, -7 Kiruna, 68ºN, 7-7 Kiruna, 9.8 km Karlsruhe,.9 km Seasonal cycle signal relative to long-term ba (e) - - (f) (g) (h) - (i) 3 (j) 3 (k) 3 (l) MUSICA IASI red, GAW and NDACC/FTIR black

22 Validation by using a Multi-Platform Database (Continuation) gfraujoch / Karlsruhe MUSICA IASI [ppmv] MUSICA (e) N=46 R =3% (f) N=484 R =4% (g) N=379 R =7% (h) N=379 R =6% Year Daily Observations GAW in situ [ppmv] Karlsruhe, 4. km MUSICA IASI [ppmv] (e) N=54 R =6% (f) N=566 R =% (g) N=56 R =44% (h) N=56 R =3% Year Karslruhe, 49ºN, -7 Kiruna, 68ºN, Kiruna, 4. km (i) N= R =% (j) N=6 R =% (k) N=6 R =% (l) N=6 R =6% Year NDACC / FTIR [ppmv].34 (e). (f). (g). (h) Year Karlsruhe,.9 km MUSICA IASI [ppmv].3.3 N=557 R =9% N=569 R =4% N=563 R =3% N=563 R =% (i). (j). (k). (l) Year Kiruna, 9.8 km.3.3 N= R =6% N=6 R =55% N=3 R =% N=3 R =9% NDACC / FTIR [ppmv]

23 Validation by using a Multi-Platform Database (Continuation) Kiruna, 4. km Karlsruhe, 4. km gfraujoch / Karlsruhe Deseasonalised monthly mean MUSICA IASI [ppmv] (e) (e) (i).3.33 N=65 R =8% N=35 R =3% N O CH 4 CH 4 * CH 4 N=5 R =65% (f) N=49 R =4% (f) N=7 R =8% (j) N=35 R =% Deseasonalised monthly mean GAW in situ [ppmv] (g) N=45 R =4% (g) N=69 R =6% (k) N=35 R =% Deseasonalised monthly mean NDACC / FTIR [ppmv] (h) N=45 R =3% (h) N=69 R =8% (l) N=35 R =5% L PPF.5 v4 v5./ 3 v5./3 4v6 5 L PPF.5 v4 v5./ 3 v5./3 4v6 5 L PPF.5 v4 v5./ 3v5./3 4v6 5 Temporal omposition xx tt = xx mm ([tt, tt ]) + ss tt + ll tt + dd(tt) Long-term signals: Deseanolised montlhly mean Karslruhe, 49ºN, -7 Kiruna, 68ºN, 7-7 Kiruna, 9.8 km Karlsruhe,.9 km (e) (i) N=7 R =% N=34 R =4% (f) (j) N=7 R =7%.8.9 N=36 R =34% (g).9.8 (k).9.8 N=7 R =%.8.9 N=34 R =% Deseasonalised monthly mean NDACC / FTIR [ppmv] (h) (l) N=7 R =7%.8.9 N=34 R =3% L PPF.5 v4 v5./ 3 v5./3 4v6 5 L PPF.5 v4 v5./ 3v5./3 4v6 5 3

24 A posteriori CH 4 Correction (Continuation) () N O Model Simulations Approach: As N O is very stable, the horizontal, vertical and temporal N O variations can be captured well by model simulations. Then, the corrected CH 4 state is: A posteriori calculated difference Model simulations But, we are interested in quantifying the errors in the corrected CH 4 products only due to errors in the MUSICA IASI retrievals, not in model simulations. Thereby, we assume () N O Climatology Approach: Temporal decomposition of a time series into signals belonging to different timescales xx tt = xx mm ([tt, tt ]) + ss tt + ll tt + dd(tt) Reference value Seasonal cycle Long-term signal Day-to-Day variations Assuming that seasonal and long-term CH 4 variations are well captured by MUSICA IASI CH 4 products, we only correct for short-term signals Assuming again Co-retrieved N O product N O climatology The correction () strongly depends on the quality of N O model simulations, while () not. 4

25 Time Series Model and Long-term Data Consistency The modelled time series, for estimating the mean values for a reference period and for a first guess separation of seasonal cycle and long-term signals, is obtained by a multi-regression fit of different coefficients that consider variations on different timescales: Reference value Linear trend Inter-annual variation Intra-annual variation Example of continuous time series of MUSICA IASI CH 4 and CH 4* daily mean data retrieved in the surroundings of Tenerife Island between 7 and 7. The yellow line is the a priori data used. The results of the multi-regression fit of the time series model are depicted as black line and the time series model components describing the long-term behavior are represented by the blue line. The periods with different EUMETSAT L PPF software versions that can affect the MUSICA IASI products are indicated by magenta colour. 5

26 Time Series Model and Long-term Data Consistency History of EUMETSAT Level PPF software modification and EUMETSAT Level data usage that can potentially affect the MUSICA IASI products. 6

27 Geographical Coverage Examples of global geographical distribution of the free tropospheric MUSICA IASI products retrieved at 4. km altitude, filtered for c sen < 5% and averaged for an latitude x longitude area of ºxº. (a) and (b) for CH 4 ; (c) and (d) for CH 4. The maps are shown separately for mid ruary 4 (a and c) and mid ust 4 (b and d). 7

28 Geographical Coverage Same as previous figure, but for the.9 km retrieval altitude. 8