The Improved Limb Atmospheric Spectrometer is a satellite-borne sensor developed by the Environment Agency (now the Ministry of the Environment) of Japan to monitor the stratospheric ozone layer based on the solar occultation technique [Sasano et al., 1999a]. ILAS was launched from Tanegashima Space Center, Japan, by the Advanced Earth Observing Satellite (ADEOS) of the National Space Development Agency of Japan (NASDA) on 17 August 1996. The ADEOS satellite was carried into a Sun-synchronous subrecurrent orbit with an inclination angle of 98.6° and an altitude of 800 km. After a 3-month initial checkout period, continuous operation of ILAS started on 30 October, lasting until 29 June 1997. On 30 June, all the instruments on board ADEOS stopped operating owing to the failure of the satellite's solar paddle. For the period of about 8 months, ILAS made measurements over high-latitude regions from 57° to 71° north and from 64° to 88° south. Including the trial measurements on 18 September and 14, 16, and 18 October during the initial checkout phase, the latitudinal coverage ranged from 57° to 72° north and from 64° to 89° south. The latitudinal coverage varied according to the seasonal changes in geometrical relationship among the Sun, the Earth, and the satellite since the measurements were based on the solar occultation technique (see Figure 1).
 ILAS consists of an infrared spectrometer that covers the wavelength region from about 6 to 12 μm with a detector array of 44 elements, and a visible spectrometer from 753 to 784 nm with a detector of 1024 elements. It also is equipped with a mirror mounted on two-axis gimbals to track the brightness center of the Sun, and a Sun-edge sensor to detect angular differences between the top edge of the Sun and the directions of instantaneous field of view (IFOV) of the spectrometers. See Suzuki et al.  and Nakajima et al. [2002a].
 The main target parameters of ILAS measurements were vertical mixing ratio profiles of ozone, nitric acid, nitrogen dioxide, nitrous oxide, methane, and water vapor from the infrared spectrometer as well as vertical profiles of aerosol extinction coefficient at 780 nm from the visible spectrometer. The altitude ranges for which the profiles are retrieved are from cloud-top height (or tracking starting height) to 7 km in altitude. Values are derived every 1 km with an effective vertical resolution of 1.9 km at 15 km, 2.5 km at 25 km, and 3.0 km at 35 km of tangent height, respectively, as a result of combined effects of IFOV size, data smoothing, and atmospheric refraction.
 For the purposes of obtaining data for validation, field campaigns employing large balloons and various remote sensing instruments, as well as in-situ measurements, were conducted. One of them was conducted in Esrange, Kiruna, Sweden, in February and March 1997 [Kanzawa et al., 1997], and the other (NASDA ADEOS campaign) in Fairbanks, Alaska, in May 1997. Some validation studies were done for Version 3.10 aerosol extinction coefficient [Burton et al., 1999], ozone [Lee et al., 1999; Sasano et al., 1999b], and nitric acid [Koike et al., 2000], and preliminary scientific analyses using Version 3.10 data were published previously [e.g., Kondo et al., 2000; Sasano et al., 2000; Hayashida et al., 2000].
 Applying both the onion-peeling method for vertical profiling and the nonlinear least squares method for spectral fitting to the multi-layer and multi-spectral data, vertical profiles of concentrations were simultaneously derived for various gas species [Yokota et al., 2002]. The contribution of attenuation by aerosols must be corrected (nongaseous component correction) in advance to retrieving gas profiles, and was first estimated at four window spectral elements where the contribution of gas absorption could be neglected or regarded as comparatively small. The gas contribution at the window channels was calculated from climatological gas profiles as the first approximation. Aerosol extinction coefficients at other spectral elements were given by linear interpolation of those four values.
 Yokota et al.  describes the details of the data processing algorithm (version 5.20) and error estimates using real ILAS data products. In addition, they discuss in detail the possible bias errors that might be caused by the nongaseous component correction. Their analysis shows that profiles of some species are a little too biased to neglect, especially when attenuation due to polar stratospheric clouds (PSCs) is very large. It also shows that the magnitudes of bias error are dependent on gas species and type of PSCs.
 Figure 1 shows an overview of ILAS data products by a cross-sectional display of each parameter as a function of altitude and date. Figure 1 indicates temporal variations of major products (ozone, nitric acid, nitrogen dioxide, nitrous oxide, methane, water vapor, and aerosol extinction coefficient at 780 nm), separately for the Northern and Southern Hemispheres. The UKMO temperature [Swinbank and O'Neill, 1994] is also shown. The horizontal scale is day number starting from 0, which corresponds to 1 January 1997. Figure 1 enables us to grasp general features of temporal variations of parameters measured for both hemispheres.
 The special section that follows includes papers on the ILAS instrument and its characteristic [Nakajima et al., 2000], Version 5.20 data processing algorithm [Yokota et al., 2002], algorithm for tangent height determination [Nakajima et al., 2002b], comprehensive validation analysis results for individual parameters of Version 5.20 ozone [Sugita et al., 2002], nitrogen dioxide and nitric acid [Irie et al., 2002], and water vapor [Kanzawa et al., 2002] using various data sources for validation. Three more papers describe the results of a validation analysis that used JPL/MkIV [Toon et al., 2002] and SAO/FIRS-2 [Jucks et al., 2002] data for ILAS multi-parameters, and compare the column amount of gases derived from ground-based remote sensors [Wood et al., 2002]. Four other papers [Choi et al., 2002; Pan et al., 2002; Saitoh et al, 2002; Terao et al., 2002] deal with scientific analyses using ILAS data. All the papers concern version 5.20 products, which have been made available to the public through the Internet or on CD-ROM upon request.
 As discussed in detail by Nakajima et al. [2002b], errors in tangent height determination might be a major cause of bias errors in derived profiles. In version 5.20 data processing software, the spectroscopic data of the oxygen A band for the visible spectrometer was from HITRAN1996 [Rothman et al., 1998], while new data [Brown and Plymate, 2000] has been published recently. Using this new data, the tangent height is registered lower by about 300 m. In general, this causes underestimation of concentration, but its magnitude is dependent on the parameters (their profile shapes). A revision is planned which will incorporate minor changes in the algorithm as well as the spectroscopic database for the oxygen A band.
 The ILAS project was sponsored by the Ministry of the Environment. We greatly appreciate all those who have been involved in the project, including NASDA, ILAS Science Team members, ILAS Validation Team members, the United Kingdom Meteorological Office, the instrument manufacturers, and the data processing software developers.