Preliminary results of the validation of IASI CO and CH4 retrievals HYMN Final Meeting October 2009, Brussels, Belgium Tobias Kerzenmacher 1, Bart Dils 1, Nicolas Kumps 1, Ann Carine Vandaele 1, Thomas Blumenstock 2, Cathy Clerbaux 3,4, Pierre-François Coheur 3, Philippe Demoulin 5, Pierre Duchatelet 5, Maya George 4, Frank Hase 2, Nicolas Jones 6, Emmanuel Mahieu 5, Justus Notholt 7, Katinka Petersen 7, Ariane Razavi 3, Theo Ridder 7, Mathias Schneider 2, Cindy Senten 1, Gauthier Vanhaelewyn 1, Corinne Vigouroux 1, Martine De Mazière 1 1 Belgian Institute for Space Aeronomy, 2 IMK-ASF, Forschungszentrum Karlsruhe and Universität Karlsruhe, 3 Université Libre de Bruxelles, 4 LATMOS-IPSL, Paris, 5 University of Liège, 6 University of Wollongong, 7 University of Bremen Summary: 1. G-b FTIR measurements 2. Validation of CO 3. Validation of CH4 4. Summary

Kiruna Bremen Jungfraujoch Izaña Wollongong Reunion Island Ny Ålesund Validation at NDACC sites Validation of IASI retrieved CO and CH 4 using the ground-based FTIR network from the Network for the Detection of Atmospheric Composition Change (NDACC), data were taken using either Bruker or Bomem FTS and were analysed using SFIT2 or PROFITT (Izaña and Kiruna). The CH 4 data is harmonised within the HYMN project.

H 2 O, CH 4 presented by A. Razavi  Near real time processing based on the OEM:  FORLI-CO Offline (very) slow processing NRT processing CO, HNO 3, O 3 C. Wespes M. George S. Turquety spectra/day for processing BUFR - L1c - L2 (T,cld) Output A,G,K,,… Output Continua Spectra Fitting OEM SaSa SS X-sections look up tables HNO3 (P,T), CO (P,T) H2O (P,T,RH) Synthetic spectra Lineshapes ILS Direct RT Model Model P, T, HNO 3, H 2 O, CO xa profiles Level 2 Z gnd, P L2, T L2, H 2 O L2 ε, Cloud, Θ Retrievals of CO profiles from IASI Colocation: same day IASI pixels, 100 km from ground- based stations.

CO validation: averaging kernel differences There are large differences in the sensitivity profiles for the CO retrievals. The IASI sensitivity has a maximum in the upper troposphere and is larger than 1, whereas those for the FTIR retrievals are constant at about 1. Therefore, to compare the gb and IASI data, we must apply a smoothing of the gb FTIR data with the IASI kernels and also an altitude correction to take into account altitude differences between gb and IASI measurements.

CO validation: averaging kernel smoothing The averaging kernel smoothing increases the original total columns.

CO validation: averaging kernel smoothing and altitude correction As expected from the averaging kernels we see an increase of the CO total columns after the smoothing. Also expected for the mountain site Izaña (2400 masl) when corrected to mainly sea level the columns increase even more. The altitude correction is still crude.

CO validation: comparisons for 6 NDACC stations The approach presented for the comparisons seems to work well for all stations (r=0.67) except Izaña.

IASI Methane profile retrievals are performed using the line-by-line retrieval code ASIMUT developed at BIRA. ASIMUT uses OEM or Tikhonov regularisation The IASI retrievals are done in a two stage process: first the surface temperature, H2O profile in seven layers and the O3 column in the following 2 windows, — cm -1 mainly CO2, H2O and O3, noise = W/cm 2 /Sr/cm — cm -1 mainly H2O, noise = W/cm 2 /Sr/cm -1 The results from the first stage are used to retrieve the CH4 profile in 13 layers and the N2O column in the following 3 windows: — cm -1 mainly H2O and N2O, noise = W/cm 2 /Sr/cm — cm -1 mainly CH4 and H2O, noise = W/cm 2 /Sr/cm — cm -1 mainly CH4, N2O and H2O, noise = W/cm 2 /Sr/cm -1 Retrievals of methane profiles from IASI

IASI methane retrievals using ASIMUT (cont.) Monthly MODIS surface emissivities on a 0.5° lat/lon grid (MYD11C1) are used for the retrievals. Retrievals were filtered using χ² y < # of spectra, χ² < # of levels and RMS< median(RMS)+1σ. Nadir and offaxis pixels are used for the retrievals. Cloudy pixels were filtered by comparing the retrieved surface temperature with the ECMWF surface temperature (given on a 0.25° (initially 0.5°) lat/lon grid). Coincidence criteria IASI observations on the same day; within a 100 km radius around the ground based stations (60 km for Kiruna), for St Denis, pixels only over ocean were initially analysed (cf. Ile de La Réunion is very mountainous)

CH4 profile retrievals: Averaging kernels Kiruna, Sweden (5 June 2007) Saint-Denis, Reunion (5 August 2007) The colored horizontal dashed lines indicate the height for which the corresponding colored kernel is active. The sum of the rows of the averaging kernels is indicated in black. The numbers of degrees of freedom for signal (DOF) are shown in the plots Ground-based DOF 2.5 to 2.4 IASI DOF 2 to 1.7 Troposphere only

Time series of total vertical columns (VCD) of methane and percent differences Saint-Denis, Réunion June to September 2007 Kiruna, Sweden May to September 2007 Percent differences between the daily means. The differences are calculated with respect to the ground-based measurements. Daily means of methane VCD, the error bars represent the total random error. The horizontal yellow lines indicate ±10% in all panels. Std=4%Std=5%

CH4 validation: comparisons for 7 NDACC stations IASI retrievals have a higher variability than the ground-based measurements IASI retrievals overestimate at high altitude sites and underestimate at low altitude sites. Cloud lead to an overestimation at high altitudes

Structure functions for CH4: Saint-Denis (Reunion), Kiruna (Sweden) and Wollongong (Australia), courtesy of Nikita Pougatchev Based on 3 example cases of the structure function for the CH4 retrievals from IASI, namely at St Denis, Kiruna, and Wollongong, the precision of the retrievals is in the range 0.06 to 0.1 E19 mol/cm2 or <2 to 3%. This is within the 5% required precision for meaningful CH4 data. Structure function= sum of spatial non-coincidence error and state independent error

Retrieval of IASI trace gas products and their validation has started. For Carbon monoxide (retrieved with FORLI-CO): There are strong differences, especially in the upper troposphere/lower stratosphere, between the sensitivities of ground-based and IASI retrievals. The preliminary comparisons of the ground-based and IASI total columns look promising. For methane (retrieved with ASIMUT): We have more than 2 degrees of freedom for signal for the ground-based retrievals and 1.5 to 2 for the IASI retrievals. But there is a difference between the sensitivities of the ground-based and the IASI retrievals. Preliminary comparisons of the ground-based and IASI methane time series show reasonable agreement, but they need to be refined. CONCLUSION

Perspectives More work is necessary to refine the comparison methodology and to evaluate the precision of the IASI data products for trace gases. Production of level 2 data for our cases are being analysed at EUMETSAT to compare our cloud filter with the clouds present in L2 data. Column VMR w.r.t. N2O are being investigated for use in the CH4 comparisons to minimise the altitude effect. Averaging kernel smoothing has to be applied for CH4. Use of Tikhonov regularisation for the CH4 retrievals is being investigated.

Thank you!

IASI The Infrared Atmospheric Sounding Interferometer (IASI) is a Fourier transform spectrometer onboard METOP-1 observing at nadir the radiance from the Earth’s surface and atmosphere. Launched on 19 October 2006 into a sun-synchronous polar orbit, at 817 km altitude, with a 9:30 local equator crossing time. Spectral range : cm -1 (15.5 – 3.62 μm) ; spectral resolution: 0.5 cm -1 ; radiometric resolution: 0.5 K. Swath up to ± 48.3° off-axis; swath width: ±1200 km; each cell is analyzed simultaneously by a 2 x 2 array of detectors, with an instantaneous FOV diameter of mrad, each pixel diameter is 12 km at nadir. Global coverage twice daily; measurements per day (day and night) ‏. First IASI measurements available: 30 November Since May 2007 the L1c data are operationally distributed; reception at BIRA via EUMETCAST.

IASI FOV Pixels P1 up to P4 have a diameter of mrad, corresponding to 12 km at nadir.

IASI radiance spectrum Radiance split into different contributions (ASIMUT)

CH4 validation: time series for 7 NDACC stations The comparisons of the retrievals from the ground-based measurements (+) and the IASI measurements (o) with an altitude correction, but without averaging kernel smoothing. The high altitude sites Jungfraujoch and Izaña have smaller columns. There is a large scatter for the IASI retrievals compared to the ground- based CH4 retrievals.