1 Introduction

The International Ultraviolet Explorer (IUE) collected more than 104000 spectra of all types of astronomical objects during its more than 18 years of operations. The IUE Project considered it desirable to make available to the astronomical community a "Final Archive'' holding all the IUE data processed in an uniform way and with improved reduction techniques and calibrations. For this purpose a new processing system (NEWSIPS) was developed and the full IUE archive was re-processed with a newly derived linearization and wavelength scale. Also an adapted optimal extraction scheme [Horne1986, (Horne 1986),] SWET, was used to derive the low resolution absolutely calibrated output spectra. A full description of the NEWSIPS system is given in [Nichols & Linsky1996, Nichols & Linsky (1996)] and in Nichols (1998). Technical details can be found in the NEWSIPS Manual [Garhart et al.1997, (Garhart et al. 1997).]

One of the main goals of the system is to obtain the maximum signal-to-noise ratio in the final data. For this purpose the geometric and photometric corrections are performed through a new approach, based on cross-correlation techniques to align science and Intensity Transfer Functions (ITF) images [Linde1990, (Linde & Dravins 1990).] The application of this new approach reduces substantially the fixed pattern noise, and leads to improvements in the signal-to-noise ratio between 50 and 100% in low dispersion spectra and between 50 and 200% in high resolution data [Nichols1998, (Nichols 1998).]

The intrinsic non-linearity of the detectors (SEC VIDICON cameras) makes the photometric correction one of the most critical tasks in the processing of IUE data. The correction is performed through the Intensity Transfer Functions (ITFs), which are derived from series of graded lamp exposures. These functions transform the raw Data Numbers (DN) of each pixel in the Raw Image into linearized Flux Numbers (FN) in the Photometrically Corrected Image. Specifically for the Final Archive, a new set of ITFs images were obtained for the three cameras under well controlled spacecraft conditions and through improved algorithms. However, the final extracted spectra still show some residual non-linearities, most likely due to the breakdown at the extreme ITF levels of the assumption that over small differential flux ranges the relation between FNs and DNs can be approximated by a linear interpolation.

Further modifications implemented in NEWSIPS include the improvement in the wavelength calibration, the revision of the flux scale, the derivation of noise models and the optimal extraction of spectra (only for low resolution). The existence of noise models has allowed to estimate the errors on IUE fluxes for the first time. A special effort has also been made to ensure the correctness of all the information referring to the specific observation attached to the data.

The quality control procedures applied by the IUE Project have shown that the NEWSIPS reprocessed spectra are superior to the IUESIPS spectra in all cases (Nichols 1998). For the high resolution spectra the new methods to estimate the image background [Smith1998, (Smith 1998)] and the ripple correction algorithm [Cassatella et al.1998, (Cassatella et al. 1998)] result in a much higher quality high resolution spectra for this data. However, it was found that the low resolution data extraction still contained some serious shortcomings which would affect significantly the usefulness of the extracted spectra. [Talavera et al.1992, (Talavera et al. 1992;] [Nichols1998, Nichols 1998).] Most of these shortcomings and drawbacks in the IUEFA products were related to the method for the final extraction of the 1-D spectra (SWET) from the bi-dimensional, spatially resolved, rotated images (SILO files).

Within the framework of the ESA IUE Data Distribution System, it was decided to correct all the low dispersion spectra through the application of new extraction algorithms that significantly improve the quality and reliability in the final data products. A completely different philosophy is behind these new algorithms. The model-dependent strategy followed in SWET is abandoned, with the aim of retaining as much information as possible concerning the data. We anticipate that the results of both techniques are essentially identical, when the model parameters used by SWET are well suited, namely, for well exposed continuum sources. The method chosen has assured that the improvements achieved with the NEWSIPS geometric and photometric corrections are preserved since the new algorithms work on the SILO files. In this paper we describe the main features of the INES extraction procedures: background and spatial profile determination, quality flags handling, solar contamination removal, homogenization of the wavelength scale (Sect. 2).

In Sect. 3 the repeatability, errors reliability and linearity of INES low dispersion data are evaluated. Finally, the major improvements achieved by INES are summarized in Sect. 4.