Retrieval of trace gas concentrations from the IASI nadir radiances spectra

1. Retrieval of trace gas concentrations from the IASI nadir radiances spectra C. Clerbaux(1,2), J. Hadji-Lazaro(1), S. Turquety(1), P.-F. Coheur(2), D. Hurtmans(2) (1) Service d’Aéronomie, Institut Pierre-Simon Laplace, Paris, France ccl@aero.jussieu.fr (2) Service de Chimie Quantique et de Photophysique: Atomes, Molécules et Atmosphères, Université Libre de Bruxelles, Brussels, Belgium Introduction IASI, the Infrared Atmospheric Sounding Interferometer, is a new tropospheric remote sensor to be carried for a period of 15 years on the Metop-1, 2, and 3 weather satellites deployed as part of the future European Polar System starting from 2006. The instrument consists of a Fourier Transform spectrometer (spectral resolution = 0.5 cm-1) designed to measure the spectrum emitted by the Earth in the thermal infrared, using a nadir geometry. The IASI mission will provide accurate measurements of the temperature profiles in the troposphere and lower stratosphere, as well as moisture profiles in the troposphere in order to improve the quality of numerical weather forecasts. It will also allow the probing of some of the chemical components playing a key role in the climate monitoring, the global change issues and the atmospheric chemistry. In the framework of the IASI mission, we have developed processing tools to analyze Level 1 data (geolocated radiance spectra) to retrieve Level 2 products (trace gas contents at the measurement locations) and enhanced Level 3 products (global distributions). We are working on two complementary approaches for the trace gas retrieval: one is based on neural network techniques to inverse columns of ozone (O3), carbon monoxide (CO) and methane (CH4), and is working in operational (near-real time) mode [Turquety et al., JGR 2004]. Another algorithm [Hurtmans et al., ASSFTS 2005], based on the Optimal Estimation method [Rodgers, 2000], is used in research mode to derive profile information for the same species and to measure weak absorbers. This poster illustrates some of the results we have obtained using the IASI trace gas algorithm package on existing data. IMG spectra Channels selection The Interferometric Monitor for Greenhouse gases (IMG) launched aboard ADEOS in August 1996 was the first high resolution nadir infrared instrument allowing the simultaneous measurement from space of a series of trace gases [Clerbaux et al., ACP 2003]. Unfortunately, in June 1997, the ADEOS satellite ceased to collect and transmit data due to a power failure in its solar panel. However, the measurements performed during 9 months provide a unique opportunity to demonstrate the potential scientific return to be expected from scheduled missions, as IASI and TES. For each retrieved gas, the spectral channels corresponding to strong absorption features and minimizing the interferences due to other absorbing species have been selected [Clerbaux et al., JGR 1998]. SA-NN Atmosphit The core of the SA-NN retrieval code is composed of a multi layer perceptron with two hidden layers, an input layer, comprising the measured radiances along with temperature profile information, and an output layer for the retrieved concentrations. The network is trained using the output profiles calculated from an atmospheric CTM, and it is operating in near real time. Details of the performance for each gas may be found in Turquety et al. JGR 2004. In Atmosphit, a synthetic spectrum is computed using the lineparameters and absorption cross-sections for the heavier molecules ascollected in spectroscopic databases (e.g HITRAN [Rothman et al., JQSRT 2003] or GEISA [Jacquinet-Husson et al., JQSRT 2005]). The retrievals are done using the Optimal Estimation Method (OEM) [Rodgers, 2000] with a priori profile and covariance matrices information build as in SA-NN. Details and illustrations may be found in the poster of Hurtmans et al. Global distributions Vertical information and detection of weak absorbers Uccle Vertical profiles have been retrieved for ozone [Coheur et al., JGR 2005] and CO [Barret et al., ACPD 2005] using the same IMG periods. Examples of ozone profiles retrieved from IMG data are provided in the plots on the left, for co-located sondes measurements above the Uccle and Ny-Alesund stations. In bothcases the retrievals provide avery good representation of typical structures in the ozone profile, such as the lower height ofthe stratospheric ozone maximum in Ny-Alesund and the presence of a secondary ozonemaximum in the upper troposphere – lower stratosphere at Uccle. Finally, the detection of weak absorbing species such as CFCs [Coheur et al., JGR 2003] was undertaken, and recent results have been obtained for SO2 and HNO3. CO and CH4 global distributions obtained from the analysis of IMG spectra for 10 consecutive days [Hadji-Lazaro et al. , JGR 2001; Turquety et al., GRL 2002, Clerbaux et al., ACP 2003] using cloud-filtered data [Hadji-Lazaro et al., GRL 2001]. Available NDSC ground based measurements are represented by dots, using the same color scale. Ny-Alesund The CO map may be compared with the CO distribution as observed by the MOPITT instrument [Deeter et al., JGR 2003] onboard TERRA, for the same month (April) but a different year (2000). Conclusion and future work As part of the ongoing studies in the framework of the preparation of the IASI mission, we have exploited existing information provided by IMG onbard ADEOS to analyze to which extent information on trace gases can be derived from nadir radiance spectra recorded in the thermal infrared. From the combined use of the two approaches (neural network and optimal interpolation) we hope to maximize the scientific return to be expected from IASI measurements, with the possibility to retrieve global distributions (columns) and vertical information (profiles). An improved inversion algorithm that couples SA-NN and Atmosphit is under construction.