A&A 373, 730-745 (2001)
R. González-Riestra1, - A. Cassatella2,3, - W. Wamsteker4,
1 - Laboratorio de Astrofísica Espacial y Física Fundamental, VILSPA, PO Box 50727, 28080 Madrid, Spain
2 - Istituto di Astrofisica Spaziale, CNR, Area di Ricerca Tor Vergata, Via del Fosso del Cavaliere 100, 00133 Roma, Italy
3 - Dipartimento di Fisica E. Amaldi, Universitá degli Studi Roma Tre, Via della Vasca Navale 84, 00146 Roma, Italy
4 - ESA-IUE Observatory, VILSPA, PO Box 50727, 28080 Madrid, Spain
Received 15 February 2001 / Accepted 17 April 2001
This paper deals with the definition of the input fluxes used for the calibration of the IUE Final Archive. The method adopted consists in the determination of the shape of the detector's sensitivity curves using IUE low resolution observations with model fluxes of the DA white dwarf G191-B2B. A scale factor was then determined so that the IUE observations of some bright OAO-2 standards match the original measurements from Meade (1978) in the spectral region 2100-2300 Å. The ultraviolet fluxes of six standard stars used as input for the Final Archive photometric calibration together with the model fluxes of G191-B2B normalized to the OAO-2 scale are given. A comparison with the independent FOS calibration shows that the IUE flux scale for the Ultraviolet is 7.2% lower. We consider this mainly to be caused by the different normalization procedures. It is shown that the present flux calibration applies to spectra processed with the INES low resolution extraction software.
Key words: methods: data analysis - space vehicles: instruments -
astronomical databases: miscellaneous -
Many improvements have been made to the standard processing of IUE data over the years. The combination of a better understanding of the instruments and the rapid evolution of computing capabilities has allowed the use of the carefully planned calibration data, obtained under well controlled acquisition conditions over the 18 years of the IUE Project, to prepare a new calibration of the complete IUE data set.
The IUE Final Archive, IUEFA, was the end-product of the above process, which started to be defined in the late eighties. The processing software developed for this purpose was NEWSIPS (Garhart et al. 1997). The INES (IUE Newly Extracted Spectra) System is the final configuration of the IUE archive.
A detailed revision of the NEWSIPS output products indicated that there were still some problems which could be corrected. The most important deficiency was found in the NEWSIPS extraction and noise models for low resolution spectra (SWET), which e.g. caused emission line fluxes frequently to be wrongly extracted (Schartel & Skillen 1998). In high resolution data, a systematic mismatch of about 20 km s-1 between the velocity scales of short and long wavelength spectra was present. These, together with other errors, were corrected in the INES system developed by the ESA IUE Observatory (Wamsteker et al. 2000). A full description of the INES system and its data processing is given in Rodríguez-Pascual et al. (1999), Cassatella et al. (2000) and González-Riestra et al. (2000).
The INES data are available from the INES Principal Centre http://ines.vilspa.esa.es or from the INES National Hosts (Wamsteker 2000). For details on the instrumental history of IUE see Pérez-Calpena & Pepoy (1997).
In this paper we discuss the way the IUE absolute flux scale was redefined (Sects. 2 and 3). The specific algorithms needed to optimize the internal consistency of IUE spectra, such as those used to determine the effective exposure times and to correct for the time and temperature dependency of the sensitivity of the IUE cameras, are described in Sect. 3. In Sect. 4.2, a comparison is made between fluxes obtained through the present calibration and those derived from previous IUE calibrations and from other experiments (HST and HUT). In Sect. 4.3 we demonstrate the applicability of the our calibration to the data in the INES archive.
During the operational life of IUE, and prior to the Final Archive processing, several photometric calibration algorithms have been applied as a consequence of the changes made in the processing software. In all cases the flux calibration was based on the UV absolute fluxes of the bright B3 V standard star UMa (V=1.84) as defined by Bohlin et al. (1980). However, evidence for systematic errors in this UMa flux scale made it necessary to find alternatives to be used as primary calibration standards for the IUE Final Archive. In this Section we will describe the basis of the early IUE photometric calibrations and the new flux scale.
The primary flux calibration for IUE data is done on the low resolution spectra, while the high resolution calibration is derived from this. The common basis of all early calibrations was the absolute flux of UMa defined by Bohlin et al. (1980), who took the OAO-2 data as the main reference for fluxes longward of 2000 Å, and the rocket data of Brune et al. (1979) for shorter wavelengths. UMa is too bright to be observed directly with IUE at low dispersion, and therefore a set of secondary standard stars was defined. These were chosen from the OAO-2 and TD1 Catalogues. The original OAO-2 and TD1 fluxes of these standards were reduced to the common UMa flux scale by applying the "correction factors'' given by Bohlin & Holm (1984).
With the growing observational material acquired over the years, it became clear that there were systematic differences between observations and models for objects of very different physical nature, such as white dwarfs (Greenstein & Oke 1979), BL Lac objects and sdO stars (Hackney et al. 1982). Finley et al. (1990) showed discrepancies of up to a 15% when comparing IUE observations and fluxes predicted by models of DA white dwarfs. The fact that these differences were maximum in the region of largest disagreement between the original OAO-2 and TD1 fluxes, pointed to the existence of systematic errors in the UMa flux scale.
A complete revision of the IUE flux calibration was therefore considered a primary requirement in the planning of the IUE Final Archive (Cassatella 1990). Rather than deriving the flux scale for the UV on a star which can not be observed with the instrumental setup supplying the bulk of currently available UV data, a different approach was taken, allowing the use of the IUE large data set and of new special purpose observations to derive an independent calibration. Hot DA white dwarfs were chosen as the most suitable objects to define the relative IUE flux scale. They were used to determine the shape of sensitivity curves by comparison of the IUE observations with model fluxes. A scaling factor was defined to bring the relative fluxes of the OAO-2 standards at an absolute scale. In the absence of other (and better) calibration sources for the space-UV, the absolute scale was defined by the original OAO-2 measurements from Meade (1978). The accuracy on computed fluxes for DA white dwarfs is discussed by Finley (1993).
To obtain the shape and the scale factor of the sensitivity curves, an intensive observing campaign was made in 1990 and 1991. These observations included not only the traditional TD1 and OAO-2 standards already in use, but also a selected sample of DA white dwarfs. Details of the procedure followed to obtain the input fluxes for the IUE calibration are given in the next section.
Two sets of data were used to derive the flux calibration for the IUE Final Archive. The first one consisted of a large number of observations of the IUE standard stars taken at the time of the acquisition of the 1984-85 Intensity Transfer Functions (hereafter ITFs). This set included spectra obtained in all the possible observing modes (high and low dispersion, large and small aperture, trailed, etc.).
A considerably more extended set of calibration data was taken in 1991, which included not only observations of the IUE standard stars, but also of several selected white dwarfs, in particular G191-B2B. The acquisition of these data was carefully planned to determine all parameters necessary for the calibration of the instruments, such as the size of the spectrograph apertures and the camera response times. The 1991 data were used to derive the absolute fluxes of the IUE standard stars. The use of close-in-time observations of both the white dwarfs and the standard stars avoided the need to correct for the cameras sensitivity loss.
Only point-source Large Aperture spectra were used for the derivation of the flux calibration. As part of the complete calibration of the IUE instrument, the factors necessary to calibrate other observational modes, e.g. trailed spectra, were redetermined.
Table 1 gives the number of standard star spectra used to derive the photometric calibration of the IUE cameras.
The ITFs are used to linearize the IUE raw Data Numbers (DNs) by transforming them into Flux Number (FNs). The ITFs are constructed from graded exposures of lamps under well controlled thermal spacecraft conditions and radiation background. For historical reasons these ITFs have been derived through linear interpolation between 12 selected exposure levels spaced over the dynamic range of the IUE Cameras (from 0 to 255 DN). This has resulted in some small linearity errors for the highest and lowest exposure levels persisting in the IUE data (González-Riestra 1998). Since the ITFs define the linearity of the cameras, any calibration is linked to a specific ITF. In what follows, we describe the ITFs used for the derivation of the IUE Final Archive flux calibration.
LWP: The original ITF for this camera was based on data obtained in 1984-1985. It was decided to acquire a new ITF in May 1992. Although some anomalies were found in the cross-correlation behavior of this ITF, its effects were limited and it was decided to maintain the 1992 ITF for the IUEFA processing. The existence of two well differentiated groups of zero level ("NULL'') images, presented an additional anomaly. Although the cross-correlation behavior of one of these two (the "NULL-A'') was worse, this was selected for the complete processing, since it avoided strong negative extrapolations at the short wavelength end of the camera.
LWR: The LWR camera was declared as the non-primary long wavelength camera in October 1983 (Pérez-Calpena & Pepoy 1997). A new ITF was acquired one month later. It was found that this ITF gave a poor correlation with science images, especially with those taken before the camera was declared non-operational. Two ITFs were constructed for this camera. ITF-A is the original 1983 ITF with its own NULL level. It is appropriate for most of the images taken after 1983. ITF-B has as NULL level the average of all the NULL images with similar geometric characteristics taken in the period 1978-1983. The upper levels are the same as in ITF-A, but resampled to match the geometric characteristics of this modified NULL level. The NEWSIPS processing cross-correlates every science image with both ITFs, choosing for the processing the one with the highest correlation coefficient. The use of two different ITFs in the photometric correction required us to derive two inverse sensitivity curves for this camera.
SWP: The ITF acquired in 1985 was used for the processing of all SWP images.
The white dwarf G191-B2B was selected as the primary standard to define the relative fluxes of the other IUE standard stars due to its brightness (V=11.8), pure hydrogen atmosphere, high effective temperature (implying a narrow Lyman absorption line), and negligible interstellar absorption ( cm-2, Kimble et al. 1993). The model used was provided by D. Finley (private communication, 1991), and was computed using the code of D. Koester (see a detailed description in Finley et al. 1997). The model has the following characteristics (see Fig. 2):
The particular choice made here for the model parameters of G191-B2B has little effect on the IUE calibration in the sense that, if an improved model becomes available in the future, it would be straightforward to derive a suitable correction from the ratio between the new model and the one used here (see Appendix B).
For very short exposures, the effective exposure time of the IUE cameras (
is different from the commanded one (
), due to the
quantization of the clock (0.4096 s) and the so-called "Camera Rise/Fall
time'' (CRFT). New data were obtained to re-derive the rise/fall times in 1991
and the values used in the IUEFA production are given in Table 2
(González-Riestra 1991). The effective exposure time is:
The sensitivity of the IUE cameras depends substantially on the temperature
of the Camera Head Amplifier (THDA):
The zero epoch of the IUE calibration was defined to be 1985.0, because at this time the higher quality ITF observations were performed. This epoch was also taken as reference to correct for the loss of sensitivity of the IUE detectors. The procedure to derive the time sensitivity correction is fully described in Garhart et al. (1997). In short, fluxes in steps of 5 Å were derived for several hundreds of spectra of the standard stars covering the whole spacecraft lifetime and normalized to the average of spectra obtained near the reference epoch 1985.0. The ratios were binned into time steps of six months, and then fitted to polynomials over different time periods. For the LWP camera there are two approaches: after 1984.5, a linear fit is used. Before that epoch, there are few data points available, and a linear interpolation between each pair of points is used. For the LWR, a fourth order polynomial is used. For SWP, after 1979.5 a linear fit is used. Prior to this date the same approach as for the early LWP data is used. These corrections were all derived from pre-1990 data. The corrections were updated after the end of orbital operations to avoid the need for extrapolation.
As mentioned above, no correction for time-dependent sensitivity degradation was needed for the single epoch data used for the derivation of the flux calibration, and therefore no additional uncertainty was introduced in the calibration by the time dependent sensitivity correction algorithm.
The direct use of white dwarf atmospheres to define the absolute flux scale was discarded a priori by the IUE Project due to the possible errors implied in the determination of the stellar parameters. Normalization to optical photometry and/or spectrophotometry was also excluded to avoid the extrapolation over a wide spectral range, which could amplify substantially the errors.
We derived the zero point of the Final Archive absolute flux scale directly from ultraviolet observations. To this purpose we used as reference the OAO-2 fluxes in the 2100-2300 Å band. The reason for the selection of this window is that in this particular wavelength region the OAO-2 and the TD1 measurements show the best agreement (Beeckmans 1977).
The procedure used to obtain this scale factor was as follows:
Therefore, the model fluxes provided by Finley, after normalization to Massey et al. (1998) optical spectrophotometry, still had to be be divided by a factor of 1.042 to agree with the OAO-2 absolute flux scale. This scaling factor defines the UV absolute flux scale of the IUE instruments.
|Star||G191-B2B scale/OAO-2 scale|
The procedure to derive the relative Inverse Sensitivity Curves
(hereafter ISCs) of the SW and LW cameras in the low resolution mode was
very similar. It can be summarized in two major steps:
a) Determination of the absolute fluxes of the IUE standard stars from the 1991 data:
All the spectra were individually inspected, rejecting those presenting any anomaly. All the exposure times were corrected for OBC quantization, THDA sensitivity and CRFT (see Sect. 3.4.1). The mean spectrum was computed by averaging all the available spectra and weighting each point by its associated error;
The WD model was divided by the mean spectrum, and the ISC was derived via a bi-cubic spline, excluding the region around Lyman and the spurious 1515 Å absorption (de la Peña 1992). The resulting curve was resampled in bins of 10 and 15 Å for the SW and LW cameras, respectively. Finally, the scaling factor of 1.042 (see above) was applied to the curves;
The final ISCs were applied to the mean 1991 spectra of the standard stars in order to derive their absolute fluxes. The fluxes so obtained define the absolute flux scale of IUE data (see Appendix A).
For each of the standard stars, an ISC was computed from the average 1985 spectra and the absolute fluxes of the standard stars derived as explained above. Average ISC were derived from the OAO and TD1 stars. The OAO curve was scaled to the TD1 one, and both were averaged weighting by the number of spectra used in each set. In the case of the LWR camera, separate ISC were derived for both ITFs.
The ISCs derived following this procedure are only applicable to low resolution Large Aperture Point spectra. Suitable scaling factors for Small Aperture and Trailed Low Resolution spectra are given by Garhart et al. (1997).
|Figure 1: Comparison of the fluxes of the IUE standard stars derived for the present calibration with those provided by Bohlin & Holm (1984) Shown are the ratios for stars observed with TD1 (upper panel) and with OAO-2 (middle panel). The bottom panel shows the average ratio for the two groups of stars.|
|Open with DEXTER|
As already pointed out in Sect. 2, there was growing evidence that systematic errors were present in the calibrations prior to the IUEFA. In the following we make a comparison between the present flux calibration and the previous one by ratioing the fluxes of the standard stars used to derive them. The results are shown in Fig. 1, where we represent the ratio between the new and old fluxes for the individual standard stars and, separately, the average ratios for faint (TD1) and bright (OAO-2) standards. To compute these ratios we have used the "corrected'' fluxes of the standard stars as given in Bohlin & Holm (1984). We stress that these fluxes are not the original ones provided by the TD1 and OAO-2 experiments, but are corrected using the factors derived by these authors to transfer them into the UMa scale defined by Bohlin et al. (1980). The discontinuities in the flux ratios shown in the figure clearly indicate the errors in the previous flux scale. These were most likely introduced by the "correction factors'' themselves.
|Figure 2: Comparison of the models of the WD G191-B2B used for the calibration of IUE and HST-FOS. The model used for IUE is normalised as to agree with the OAO-2 scale at 2200 Å, i.e. the flux of the model originally normalised to the spectrophotometry of Massey et al. (1998) has been divided by a factor 1.042 (see text).|
|Open with DEXTER|
In the short wavelength range, there are large fluctuations in the ratios
of up to 20% over intervals less than 100 Å wide. The ratio is more
uniform between 2000 and 3100 Å and then decreases abruptly, with the
new fluxes being lower by up to a 15%. It is interesting to remark that
the largest discrepancies are present in the region 1500-1700 Å where,
probably not accidentally, the differences between TD1 and OAO-2 fluxes
The broad features visible in the flux ratio shown in Fig. 1 are remarkably similar to those of the "correction factor'' derived by Finley et al. (1990) from the comparison of atmosphere models and observations of DA white dwarfs, in particular shortward 2000 Å.
White dwarf models have been used for the flux calibration in the UV range of other space experiments. This is the case of the Hubble Space Telescope and the Hopkins Ultraviolet Telescope (HUT, Kruk et al. 1997, 1999). In the following we will compare the IUE and HST-FOS flux calibrations. The comparison of HUT (both ASTRO-1 and ASTRO-2) and FOS is described in Kruk et al. (1997, 1999).
|Figure 3: Comparison of the IUE, HST-FOS and HUT fluxes of the standard star BD+75 325. Shown for comparison in the upper panel is the ratio of the models of the WD G191-2B2 used for the calibration of IUE and FOS (the same as shown at the bottom panel of Fig. 2). This shows that although the relative calibrations are quite consistent, the absolute scale is still rather uncertain.|
|Open with DEXTER|
The absolute calibration of the HST Faint Object Spectrograph is based on a slightly different model of the DA WD G191-B2B (Bohlin et al. 1995) with a pure Hydrogen atmosphere, an effective temperature of 61300 K, , and normalised to V=11.781 (Colina & Bohlin 1994).
The difference in effective temperature of the models used for the calibrations of IUE and FOS results in a slightly different slope in the UV range (approximately 1%, see Fig. 2). The model fluxes used for the IUE calibration (with the original scaling to optical spectrophotometry) are lower than the model used for FOS by 1.1% at 5500 Å due to the different normalisation. The additional 4.2% scaling factor makes this difference 5.3% at V (in the sense that the FOS model is brighter. This normalisation implies a V magnitude of 11.84 for G191-B2B, in contrast with the recent revision by Bohlin (2000) which derived a value of (1).
The slightly different slope of the models increases this discrepancy in the IUE range. The average ratio of the models used in the IUE and the FOS calibrations in the range 1150-3350 Å (excluding the region around Lyman ) is 0.933, i.e. model used for the IUE calibration is lower by a 7.2%.
We have compared the IUE, FOS (Bohlin 1996) and HUT (Kruk, private communication) absolute fluxes of the standard star and BD+75 325 in the spectral region of overlap of the three experiments, as shown in Fig. 3. The continuous line in the upper panel of the figure represents the ratio between the models used for the IUE and FOS calibrations. The average ratios IUE/other over the common wavelength range is and for FOS and HUT, respectively. The figure shows that the overall agreement between IUE and FOS flux and model ratios is good, although there are some broad features, which are thought to be induced by the effects of the residual non-linearities of the IUE cameras on the spectra used for the calibration. On average, the flux ratio IUE/FOS is within a 3% of the model ratio, except for the region 2250-2450 Å where it is lower by a 4%.
The IUE fluxes seem to agree better with the HUT data, but this might simply be an accidental artifact, due to the complex calibration of the different HUT instrumental configurations and the large uncertainties involved (Kruk, private communication).
Although the comparisons in Fig. 3 show a general agreement in the three calibrations (IUE independent of HST and HUT, which are based on the same absolute fluxes), to 3% in the relative calibration, it is clear that the absolute UV scale is still uncertain to 10%.
|Figure 4: Average ratio of NEWSIPS and INES fluxes for SWP spectra of the standard star BD+28 4211 (28 spectra with exposure times between 20 and 30 s and 22 spectra with exposure times between 50 and 60 s). The thin line corresponds to the average ratio for non-saturated spectra, and the thick one to saturated spectra, with only the non-saturated region shown. The dashed lines mark the 5% limits.|
|Open with DEXTER|
The flux calibration described in this paper has been derived from IUE low resolution spectra processed with NEWSIPS and the SWET optimal extraction procedure (Garhart et al. 1997). IUE low resolution data have been re-extracted from the NEWSIPS bi-dimensional SILO files using a different algorithm for the INES archive. The INES extraction is described in detail by Rodríguez-Pascual et al. (1999). It includes, among other features, new noise models and an improved extraction procedure. Both extraction algorithms (SWET and INES) use the same inverse sensitivity curve, therefore any differences in the flux calibrated spectra would also appear in NET spectra. Differences of this kind could in principle arise from the different procedures used to estimate the background level, to evaluate the spatial profile, and from the adopted noise model.
|Figure 5: Same as Fig. 4, but for the LWP camera. Shown are the average ratios corresponding to 24 spectra with exposure times between 45 and 50 s and 20 spectra with exposure times between 95 and 150 s of the standard star BD+28 4211.|
|Open with DEXTER|
In order to check the applicability of the Final Archive calibration to INES-extracted data we have taken low resolution spectra of the IUE standard star BD+28 4211 and compared the INES and the NEWSIPS fluxes. For this purpose we have divided the spectra into two groups according to their level of exposure: the first group containing non-saturated spectra, and the second one containing spectra saturated in the region of maximum sensitivity of the cameras. We have computed the mean ratio for each group of spectra. The results are shown in Figs. 4-6.
In the SWP camera, the average flux ratio INES/SWET for the short-exposure time spectra is longward 1250 Å, with a slight slope along the full wavelength range (i.e. the INES flux is slightly lower than the SWET flux shortward 1400 Å and slightly higher longward 1600 Å). Shortward of Lyman the INES flux is up to a 8% lower. The INES flux is also lower in this spectral range for the longest exposure spectra, but only by less than 4%. Longward 1400 Å the flux ratio is independently on the level of exposure of the spectra.
In the case of the LWP camera, the INES and SWET fluxes agree within 1% along most of the spectral range (2200-3000 Å). The largest differences are found at the edges of the range. While at the short wavelengths the INES extraction provides fluxes up to a 10% lower than SWET, the contrary occurs longward 3000 Å where INES fluxes can be a 10% higher. It must be noted that in both cases the differences are larger for short-exposure time spectra, suggesting that the discrepancy can originate from non-linearity effects at low exposure levels.
The largest discrepancies are found in the LWR camera (for images processed with ITF-B). In the region 2500-3000 Å the ratio INES/SWET is , while at shorter wavelengths (2100-2500 Å) it is closer to unity: . As in the case of the LWP camera, it is longward 3200 Å where the difference becomes larger, with the INES flux larger by up to a 20%. In this camera there is no significative difference in the behaviour of short and long exposure time spectra.
In summary, the differences between the INES and SWET extractions are within 2% over most of the spectra range, with largest differences at the edges of the cameras (in the SWP only at the shortest wavelengths).
|Figure 6: Same as Fig. 4, but for the LWR camera. Shown are the average ratios for 27 spectra with exposure times between 55 and 65 s and 4 spectra with exposure times between 75 and 110 s of BD+28 4211. All the spectra have been processed with ITF-B (see Sect. 3.2) and were taken in the period during which the camera was still operational (1978-1983).|
|Open with DEXTER|
In this paper we have described the definition of the flux scale which has been adopted for the flux calibration of the IUE Final Archive data. After having discussed the inadequacy of the UMa flux scale as in Bohlin et al. (1980), we have shown that a more pertinent method which optimizes the internal and external consistency of IUE fluxes is to use the DA white dwarf G191-B2B as the primary standard star. The procedure followed consisted in using the IUE observations of this star obtained in 1991 together with model atmosphere fluxes normalized to the data from optical spectrophotometry in Massey et al. (1998) to obtain the shape of the inverse sensitivity curves for the three IUE operational cameras. At this point, the many IUE observations of Aur, Lep, 10 Lac and Dra, also obtained in 1991, were used to find a suitable scaling factor of the sensitivity curves such that, after calibration, the scaled fluxes best fitted the corresponding OAO-2 original measurements from Meade (1978) in the range 2100-2300 Å. The choice of this wavelength interval to set the scaling factor was to link the IUE calibration to ultraviolet instead of optical observations. Also, the 2100-2300 Å range is the one where the agreement between TD1 and OAO-2 fluxes is best. As shown in Sect. 3.5, the 2100-2300 Å OAO-2 fluxes of the quoted four standards are on average a factor of 1.042 lower than those from the G191-B2B model normalised to the data from optical spectrophotometry (Massey et al. 1998).
The sensitivity curves for the 1991 epoch, together with the very many observations of the standard stars secured in this epoch, were used to define the absolute fluxes of the standard stars. These fluxes were then used as input to derive the sensitivity curves for the 1985 calibration epoch (i.e. the epoch the ITFs were obtained for the IUE cameras).
The absolute fluxes of six IUE standard stars and the model fluxes of G191-B2B in the OAO-2 scale are given in Appendices A and B, respectively.
As shown in Sect. 4.2, the fluxes obtained with this calibration are on average 7.2% lower than the ones provided by the Faint Object Spectrograph on board the Hubble Space Telescope in the range 1150-3350 Å. This discrepancy can be ascribed to the different choice for scaling the G191-B2B model fluxes and, to a minor extent, to the slightly different stellar parameters adopted for the G191-B2B model.
Rodríguez-Pascual et al. (1999) have discussed the INES system and its advantages over NEWSIPS to remove the systematic errors found in this latter package. In view of the different extraction software used in the two systems, a specific test has been carried out in this paper to verify the applicability of the present flux scale to low resolution data processed with INES. The conclusion is that, in spite of the different extraction algorithms used, the application of the present calibration to INES spectra is fully justified. We stress that the present paper has a direct application to the absolute calibration of low resolution spectra. The method used to obtain absolute calibration of high resolution spectra has been discussed elsewhere (Cassatella et al. 2000).
We would like to thank the attendees of a Meeting on the IUE flux calibration which we organized at the ESA IUE ground station of Villafranca del Castillo in December 1990, and in particular, Prof. V. Weidemann and Drs. D. Finley, D. Husfeld, D. Koester and D. Massa. We are grateful to Dr. J. Kruk for his comments about the HUT calibration. We would also like to acknowledge the large effort made by the NASA and ESA IUE Observatories in supporting the calibration activities. This paper is dedicated to the memory of Andy Michalitsianos, who has strongly contributed to the success of the IUE project.
The table in Appendix B gives the model fluxes of the white dwarf G191-B2B
from Finley (private communication, 1991). These fluxes are scaled to the
OAO-2 flux scale, i.e. the fluxes provided - originally scaled to the
optical spectrophotometry of Massey et al. (1998) - have been divided by
1.042. Wavelength is in Å and flux in erg cm s Å.