Comparison of XCO2 Total Column Retrieved from IASI/MetOp-A Using KLIMA Algorithm and TANSO-FTS/GOSAT Level 2 Products
Cortesi, Ugo; Del Bianco, Samuele; Laurenza, Lucia; Gai, Marco
IFAC-CNR, ITALY

Carbon dioxide is the primary greenhouse gas released into the Earth's atmosphere by human activities. The balance of natural sources and sinks of carbon dioxide, leading to stable values of about 280 ppm for atmospheric CO2 concentration in pre-industrial times, has been significantly perturbed by anthropogenic forcing - mostly from fossil fuel burning and cementification and from deforestation and other land use change - since the beginning of the industrial era in the late eighteenth century. A crucial task to achieve reliable predictions of future levels of atmospheric CO2 is, along with monitoring of CO2 emissions, an accurate determination of the geographic distribution of surface carbon fluxes on a variety of spatial and temporal scales.
In this study, one year of thermal infrared spectral radiances observed by the Infrared Atmospheric Sounding Interferometer (IASI) spectrometer, on-board the MetOp-A satellite, are used to retrieve the columnar abundance of atmospheric carbon dioxide, for a global geographical coverage and in clear-sky conditions. Our work represents an attempt to gain greater insight in the capabilities of passive remote sounding from space in the thermal infrared spectral region. This is achieved by investigating the potential of measurements acquired by the IASI to provide CO2 total column values with precision and accuracy necessary to compare with Thermal And Near-infrared Sensor for carbon Observation Fourier Transform Spectrometer (TANSO-FTS), onboard of the Greenhouse gases Observing SATellite (GOSAT) products. TANSO-FTS is the first satellite-based sensor dedicated to the measurement of atmospheric CO2 using spectral bands in the SWIR and in the TIR regions. and by the inter-comparison between CO2 total columns retrieved from the TIR spectra of IASI and from the SWIR spectra of TANSO-FTS. The dedicated software is based on the KLIMA inversion algorithm originally proposed by IFAC-CNR for cycle 6 of ESA Earth Explorer Core Missions and now adapted into a non-operational inversion code to process IASI Level-1 data. It aims at retrieving the CO2 total column with a target accuracy of 0.3%, on a regional scale (1000x1000 km) at monthly intervals, which is consistent with the requirements of TANSO-FTS SWIR products.
In order to obtain a reasonable capacity to bulk processing IASI data, we chose to integrate the KLIMA code into the ESA grid based operational environment G-POD system (Grid Processing On-Demand). We introduced a series of approximations in the radiative transfer code, with the aim to achieve adequate features in term of program size and computing time necessary for the integration into G-POD system. On the base of sensitivity study of IASI channels, a channel selection was also made. The channels sensitive to carbon dioxide features are located in two spectral bands: around 700 cm-1 and 2000 cm-1 respectively. Considering that the IASI radiometric noise is higher in the shortwave region and the use of one band only permits to decrease largely the computing time, only channels located around 700 cm-1 have been considered for the retrieval. At the end of the integration of the KLIMA-IASI code and considering the capacity of G-POD computing resources, we decided to process one week per month of a complete year of IASI measurements - from March 1, 2010 to February 28, 2011 - for global geographical coverage. In this selected temporal range, TANSO-FTS SWIR Level-2 data were obtained from the GOSAT User Interface Gateway (GUIG) and data from selected stations of the Total Carbon Column Observing Network (TCCON), covers a different latitudes, were collected from TCCON Data Archive. After applied selection criteria to assess the data quality and performance of retrieval, KLIMA provided about 240000 values in the selected annual range while from TANSO-FTS only the retrieval of about 3000 observations are available in the same period. Using this dataset, an extensive comparison of column-average CO2 dry air mole fraction (XCO2) has been performed between values retrieved from IASI measurements by using the KLIMA/G-POD inversion code and operational Level-2 SWIR products (Version 01 and Version 02) from collocated TANSO-FTS observations. We performed also a comparison of KLIMA products against collocated data provided by several TCCON ground stations.
Several strategies have been adopted for the purpose of inter-comparison between KLIMA L2 and EUMETSAT operational L2 products and between KLIMA L2 and TANSO-FTS L2 products. As a first step, we compared the XCO2 total columns as a time and spatial averaged products (every month has been averaged over the globe on a grid of 2°x2°). The second step was the comparison of XCO2 total column retrieved only from co-located observations of IASI and TANSO-FTS. The third step consisted in the comparison of the seasonal cycle of XCO2 from March 2010 to February 2011 in the Northern hemisphere and in the Southern hemisphere and the seasonal trends on selected macro-areas (Northern/Southern hemispheres, Pacific Area, some latitude bands).
For what concerns the seasonal cycle in the Northern hemisphere, TANSO-FTS and KLIMA-IASI show similar amplitude and trends, with a maximum in Spring (April-May) and a minimum in Fall (September-October). The seasonal cycle resulting from EUMETSAT data shows very little variation from March to September and a decrease of approximately 5 ppm from October to February. We obtain similar behavior for southern hemisphere, with a maximum in March and a decrease started from October. Further results from the different steps of the comparison are presented and discussed and the overall outcomes of our investigation are reported in the concluding remarks, along with proposed improvements and new applications of the KLIMA/G-POD inversion code.