1. Objectives

• Last direct measurements of the absolute solar spectral radiation in wide UV, visible and IR region were performed at the beginning of 1960 (Labs and Neckel, Thekaekara et al., Arvesen et al.).
• The previous investigations fulfilled at the Peak Terskol have shown that the main uncertainties of our ground-based measurements were because of the uncertainty of standard radiation sources calibration, not because of the atmosphere extinction reduction procedure.
• The space platform oriented measurements has faced with the sufficient difficulties of calibration.
• The calibration for the ground-based measurements can be fulfilled much more precisely and reliably than for the space-based ones.

2. Summary

• The spectral radiation data (disk-center intensity, irradiance at the 1 AU and continuum radiation temperatures) are now available for the 310---1070 nm wavelength range.
• Figure 1 shows the measured spectral (continuum + lines) intensity of the solar disk-center radiation in 10¹² W m^-3 ster^-1. The dotted curve corresponds the data given by Neckel and Labs.
• Figure 2 shows the disk-center radiation temperatures for local maxima at 1-nm bands of the calibrated FTS Kitt Peak spectrum. Because of our integrals are free from atmosphere absorption but FTS spectrum is not, the telluric lines deviate up from continuum level while the solar lines deviate down. The continuum is located at the intervals free from the solar and telluric lines. Dashed line marks the data by Neckel and Labs, diamonds represent our results.
• The synthetic color-indices of the Sun are very close to U-B=0.10, B-V=0.62, which are in the high accordance with the data for the normal G2V star. These values are in the contradiction with the measured by Tug and Schmidt-Kaler ones (0.18 and 0.67). The synthetic value V=-26.73 is close to the commonly used -26.75.

FIGURE 1:Spectral intensity of the solar disk-center radiation in 10¹² W m^-3 ster^-1 versus wavelength in nm. Thewavelength range 645---685 is the common for both series of the measurements (UV and visible, 1989: Burlov-Vasiljev et al., 1995; IR, 1992: Burlov-Vasiljev et al., 1997). The dotted curve corresponds to the data given by Neckel and Labs (1984).

FIGURE 2:Disk-center radiation temperatures for local maxima at 1-nm bands of the calibrated FTS Kitt Peak spectrum, continuum of Neckel and Labs (1984) (dashed line), our results (diamonds). See explanations in Summary section.

3. Apparatus

• solar horizontal telescope with coelostat group (main mirror D = 230 mm, f = 3 m)
• grating spectrophotometer with collimator (D = 180 mm, f = 2 m), camera mirror (D = 230 mm, f = 2 m), and grating (140x150 mm, 600 or 1200 grooves/mm) mounted in the vertical plane. The exit slit, quartz field lens and receiver (photodiode) are mounted on the scanning carriage, which is moved within 60 mm from the central point by a stepping motor. For the division of the diffraction orders and decreasing of the scattered light level the proper colored glasses is mounted just behind the entrance slit of the spectrometer.
• absolute calibration system uses a ribbon tungsten lamp, which is placed at the focus of the collimating mirror (D = 230 mm, f = 3 m). Then the parallel beam is directed to the coelostat mirror and then along the same optical path as during observations of the Sun. The ribbon lamp was calibrated in units of the absolute spectral brightness
• atmosphere halo photometer is used for monitoring of the optical properties of the Earth's atmosphere
• auxiliary equipment to control the apparatus includes optical and electronical stuff for lamps feeding and control, for the investigations of the polarization properties of the coelostat group and spectrograph, for the measurements of spectral reflectivity of the lamp's collimator, etc.

4. Observations

The two main series of measurements were carried out: for the wavelength range 310---685 nm (1989) and for the wavelength range 645---1070 nm (1992). The result of 1989 was derived from measurements during 7 days in UV region (310---400 nm) and 9 days in visible (400---685 nm) during the period July---August. Totally during 1992 were conducted 3 series of observations (March 24---April 7, August 4---8, September 29---October 1). All together there are 19 days of observations. For final result in IR were used the observations on April 2, 3, 6, 7 and October 1. All the days in March, on April 1 and on September 29, 30 were extracted because of unsatisfactory stability of the atmospheric transparency.

The investigated spectral area was divided into few areas, each of which was measured at a certain position of the diffraction grating. The adjacent areas were overlapped among themselves on 1/2 length.

5. Calibration

During 1989---1990 years we had conducted numerous comparisons of our lamps with the standards of some other laboratories of former Soviet Union and with the standard of the PMOD/WRC (Davos, Switzerland). These measurements showed that during this period of several years (namely from the moment of the first calibration in 1988) the lamps saved the brightness within the limits of a parts of a percent.

The lamps were calibrated in 1988, 1989 and 1990. The differences between calibrations were within the accuracy bar stated above. During measurements of a the solar spectrum the lamps were used in different modes. The lamp #359 was used for calibration daily. The lamp #92 was used not so often -- only for mutual comparisons with the lamp #359 in the very beginning and at the end of a series of observations. In result the time of burning of a lamp #92 is less in factor 10 than for a lamp #359. We believe that in this case the constancy of the ratio of two lamps brightness can be a criterion of preservation of an internal laboratory scale of measurements. It is very difficult to assume that both lamps will in coordination change the brightness at so different modes of use.

The measurements 1992 and result of calibration 1990 differ not more, than on 1%. We believe, that the scale of our measurements was constant within that limit.

6. Sources of Uncertainty

• Uncertainty of the lamp's collimator reflectivity. For the forming of the image of a calibrated area of a lamp ribbon was used the spherical collimator mirror. Factor of reflection of the mirror was defined experimentally. Uncertainty of these measurement equals to 1% at lambda = 300 nm, 0.3---0.5% at wavelengths 400---700 nm, and 1% at lambda = 1200 nm.
• Transmittance changes of the optical system. The telescope transmittance changes during the day due to reflectivity changes of the coelostat mirror with the angle of incidence of the solar beam. Moreover, the light reflected by the coelostat is partially polarized, and the amount and character of polarization depends on the incidence angle. All of the following optical elements are sensitive to polarization. Thus their transmittance changes during measurements. It should be noted that the mirror reflectivity is influenced by dust: the decrease of reflectivity with an increase of incidence angle becomes more steep. The results of the special investigation can be summarized follows:
  • 1. degree of linear polarization after the coelostat mirrors may change by 1---5% during the day
  • 2. polarization properties of the whole spectrometer are mainly due to the grating
  • 3. absorption caused by dust increases by 0.5% with increasing angle of incidence from 0° to 40°, but influence of the dust upon the polarization properties is almost zero.
  The maximum error caused by variation of the optical system transmittance can amount to 3%. But this error can be reduced substantially by choosing the coelostat position and time interval. By this means and with the use of grating with a small polarization effect, the uncertainty can be reduced to 0.1---0.2%.
• Detector's stability between calibrations. It was monitored by radioluminiscent emitters. Trend did not exceed 0.1---0.2% during several hours and the reproducibility is within the errors of measurements (about 0.2%).
• Detector's linearity.} Because of differences of solar and lamp brightness, which is up to 4 order near lambda = 300 nm, detector must have a high linearity. That is why we used silicon photodiode (spectral sensitivity ranges from 190 to 1100 nm).It was investigated with a set of neutral density filters calibrated at the State Optical Institute (St. Peterburg, Russia). Results demonstrated a high linearity over four orders of magnitude.
• Lamp adjustment and feeding. The transfer of the lamp calibration to the spectrometer is influenced by the following factors: accuracy of the lamp adjustment inhomogeneity across the ribbon surface, stability of the power supply, anisotropy of the radiation and possible temperature effects. Our measurements have shown that the resulting uncertainty is about 0.5% at lambda = 300 nm and 0.2% at lambda = 1100 nm.

7. Extrapolation to Zero Air Mass

• Monochromatic radiation. In UV region the observations were performed with a resolution of 0.5 nm, and 0.1 nm in visible and IR. The error of the extrapolation to zero air mass because of nonmonochromatic radiation observed in UV and visible for these cases is less then 0.2%, but in some narrow regions of the strong oxygen and water vapour telluric bands in IR it can be extremely large: for wavelengths 890---990 nm (H₂O) - up to 30% (for real water abundance at Peak Terskol); for wavelengths 759---771 nm (O₂) - 75%! In the regions with the strong telluric bands we had to do the added reduction which was derived from the synthetic atmospheric absorption spectra computed on the basis of HITRAN molecular parameters data and standard model of the atmosphere. The uncertainty of these reductions needed in a few narrow bands is about few percent.
• Accuracy of the air mass computation. Usually the air mass of Bemporad is used. Straightforward analysis shows that Bemporad's air mass at some wavelength in UV (with telluric ozone bands) and IR (water vapour bands) even for observations with zenith distances less then 75° may cause a relative error in the atmosphere transparency coefficient of about 3%. We performed a more elaborate calculation of air mass using real model of atmosphere over Peak Terskol and taken into account the dependence of air mass on lambda because of different contribution of Rayleigh scattering, telluric bands and aerosol extinction to air mass value at different lambda. Obtained values is valid to within 0.1% for zenith distance less 80°.
• Stable optical atmospheric conditions. Instability of atmospheric extinction in wide wavelength range is mainly due to variations of the aerosol portion of the atmosphere. Moreover, special investigation fulfilled for Peak Terskol show that the atmospheric water vapour abundance is correlated with abundance of the scattering aerosol while the absorbing aerosol over Peak Terskol is mostly absent at all. For monitoring of the scattering aerosol stability we used atmosphere halo brightness measurements, which let us to find time with stable atmospheric transparency. As a rule it was only from sunrise to 10-11^h of local time.
• Absence of the scattered atmospheric radiation. Multiple scattering of the light in the atmosphere would affect on the results obtained with Bouger's method at the high values of the atmospheric optical thickness (in strong telluric bands only at high air masses). We have carryed out the model computations to avoid the observations at these conditions.

8. Results

FIGURE 3:Disk-center radiation temperatures for local maxima at 1-nm bands of the calibrated FTS Kitt Peak spectrum, continuum of Neckel and Labs (1984) (dashed line), our results (diamonds). See explanations in Summary section.

9. Synthetic UBV

Different determinations give the value of 0.14---0.20 for (U - B)_sun, and 0.63---0.69 for (B - V)_sun (Makarova et al. (1991)). The values (U - B)_sun = 0.18 and (B - V)_sun = 0.67 are adopted. These values differ from the standard meanings for G2V star (0.10 and 0.62 respectively) and are more close G5V (0.20 and 0.68). The absolute magnitude of the Sun V_sun = -26.70 now is commonly used. To compute the solar color-indices and absolute magnitude we used the relative reaction curves for UBV system given by Buser (1978) calibrated us by spectral energy distribution for Vega. We consider here the following data set for Vega: by Hayes (1985), Archarov et al. (1989), Terez (1983), Knyazeva and Kharitonov (1988). In the case of necessary the absent data at 300 nm was added.

One can see that the values of the color-indices more close to G5V star are in the contradiction with the solar and star absolute spectrophotometry. Moreover, as the both of the color-indices are not independent, they must lie on the normal stars Main Sequence in the two colors UBV diagram We could find two close pairs of the spectrophotometer curves for the Sun and Vega, respectively (present work --- Archarov et al.), and (Makarova et al. --- Terez), which satisfy this requirement. In both cases the solar color-indices are typical for G2V star. The solar absolute magnitude V_sun obtained from spectrophotometry is close to the commonly-used value.

References

Archarov A.A. et al. 1989, Tables of Standard Reference Data. Spectral density of energetic illumination from stars on the boundary atmosphere Earth at the range 0,32...1.08 mcm, GSSSD, Moskow (in Russian).

Arvesen,J.C., Griffin,R.N., & Pearson,B.D. 1969, Appl. Optics. 8, 2215. Burlov-Vasiljev,K.A., Gurtovenko,E.A., & Matvejev,Yu.B. 1995, Solar Phys. 157, 51.

Burlov-Vasiljev,K.A., Vasiljeva,I.E., & Matvejev,Yu.B. 1997, Solar Phys. (in press)

Buser,R. 1978, A&A 62, 411.

Hayes,D.S. 1985, in Hayes,D.(eds.), Calibration of fundamental stellar quantities, IAU symp. 111, p. 225.

Knyazeva,L.N. & Kharitonov,A.V. 1988, Standard star News. 13, 22.

Lockwood,G.W., T\"ug,H., & WhiteN.M. 1992, ApJ, 390, 668.

Makarova,E.A., Kharitonov,A.V., & Kazachevskaya,T.V. 1991, Solar Flux, Nauka, Moskow (in Russian).

Neckel,D. & Labs,H. 1984, Solar Phys. 95, 229.

Shaw,G.E. 1982, Appl. Optics. 21, 2006.

Shaw,G.E. & Frö hlich,C. 1979, in B.McCormac and B.Seliga(eds.), Solar-Terrestrial Influences on Weather and Climate, D.Reidel Publ. Co., Dordrecht, Holland, p. 69.

Terez,E.I. 1983, Standard star News. 2, 10.

Thekaekara,M.P. 1974, Appl. Optics. 13, 518.