This section briefly summarizes how EUV spectroscopic measurements and visible polarized radiance observations can be used to determine the basic plasma parameters of the solar wind source region. The diagnostic methods, which are used, are well documented (Kohl and Withbroe, 1982; Withbroe et al., 1982a, 1985; Noci, Kohl and Withbroe, 1987 and van de Hulst, 1950) and will therefore not be discussed in detail here. The feasibility and utility of the spectroscopic diagnostics for proton random velocity, and outflow velocity in both coronal holes and quiet coronal regions have been demonstrated with rocket and Spartan 201 observations of HI Ly- and O VI 1032 Å, 1037 Å (Kohl et al., 1980, 1984; Withbroe et al., 1982b, 1985, 1986; Strachan, 1987; Kohl et al., 1995; Kohl et al., 1994; Strachan et al., 1994). Examples of electron density determinations from white light coronagraphs are described in the literature as well (e.g., Munro and Jackson, 1977).
Proton and ion velocity distributions can be determined from measurements of the spectral line profiles of HI Ly-, Fe XII 1242 Å, O VI 1032 Å, Mg X 610 Å and Si XII 499 Å. The shape of a spectral line depends on the velocity distribution of the particles emitting or scattering the measured photons. The velocities are produced by thermal motions, nonthermal motions (due, for example, to waves), and bulk outflow velocities in the line of sight. The coronal HI Ly- profile is strongly affected by thermal motions (130 km s at 10^6 K), while the more massive particles tend to be affected most strongly by non-thermal motions. Expected line widths are provided in Figure 1 along with measurements of the width of HI Ly- made with a rocket UV coronagraph in a polar coronal hole observed in 1980 (Withbroe et al., 1985).
The velocity distribution of neutral hydrogen in the corona reflects the proton distribution, because the characteristic lifetime for a hydrogen atom at typical coronal densities is significantly shorter than typical coronal expansion times (Withbroe et al., 1982a). For example, at in a low density coronal hole, a newly created HI atom will move about 0.02 solar radius before it is ionized.
The directly measurable quantity is the velocity distribution along the line of sight. The use of resonance line profiles provides a direct method for measuring coronal velocity distributions. Because the corona is optically thin for the selected lines, the profiles are, effectively, direct measurements of velocity distributions along the line of sight. UVCS provides synoptic observations that can be used to help determine three dimensional models of stable structures such as coronal holes.
Electron temperatures and velocity distributions are determined from the coronal line profile of the electron scattered component of HI Ly-. The electron component of coronal HI Ly- is produced by Thomson scattering of chromospheric HI Ly- by electrons in the extended corona. This means of measuring coronal electron temperatures was first suggested by Hughes (1965). The electron component has the approximate form:
where is the Doppler width and x is along the line of sight. Because of the high thermal velocity of electrons at coronal temperatures (nearly 7000 km s for T K) the effects of solar wind flows and nonthermal motions on the shape of the profile can be neglected. This profile is much wider ( Å) than the resonantly scattered component ( Å). An empirical model must be used to take into account small effects on the profile shape due to the Thomson scattering process (see Withbroe et al., 1982a).
Electron densities can be determined by measuring the polarized radiance of the visible corona with the visible light section of the UVCS. The percent polarization of the observed intensity can be determined by measuring the polarized radiance at 0, +60 and -60 (cf. Billings, 1966). This eliminates the F corona which is expected to be unpolarized at heights up to 5 R and also eliminates unpolarized stray light. The Thomson scattering process that gives rise to the K-corona is well understood and the techniques enabling the electron density to be determined from the polarized radiance have been employed with eclipse and satellite data for a number of years. These techniques require that the coronal polarized radiance be determined relative to the spectral irradiance of the solar disk.
Electron densities can also be measured using the intensity of the electron-scattered component of HI Ly-. Since the Ly- emission from the F corona is much narrower than the broad electron-scattered component of Ly-, the Ly- emission from the F corona can be subtracted without having to use a polarizer.
Outflow velocity determinations in the extended corona will be based on observations, off the limb, of Doppler dimming (Hyder and Lites, 1970) and on spectral line shifts due to the mean value of the outflow component in the direction of the observer. Examples of Doppler dimming calculated for an isothermal corona with a temperature of K are shown in Figure 2 .
There are several ways of determining Doppler dimming. One method originally used by G. Noci to analyze Ly- data from the 1970 eclipse, makes use of the intensity ratio between the resonant component of a line such as HI Ly-, O VI 1032 Å, or Mg X 610 Å and electron scattered visible light. The ratio
where A is the elemental abundance, is the ionization balance term and is the Doppler dimming term plotted in Figure 2 . Measurements of the intensity ratio as a function of radius can be used to determine the amount of Doppler dimming and, hence, the bulk outflow velocity of the observed ions. Spectral lines of different ions can have different sensitivities to outflow velocity.
A definitive analysis requires an empirical model which uses as inputs, measured electron temperatures and electron densities. This information is used in the model to calculate the ionization balance terms Ri and predict the observed EUV intensity as a function of the outflow velocity. Since the outward particle flux must be conserved, the models can provide self consistent predictions of the Doppler dimmed EUV lines as a function of height.
It is highly desirable to use several complementary methods to determine outflow velocities. For example, the required electron densities can be determined from both visible light and Thomson scattered Ly- (the latter eliminates the radiometric calibration uncertainty from Doppler dimming analyses of Ly-). An indication of supersonic outflow in a coronal structure from Doppler dimming would usually imply that the observed spectral line should be shifted by a measured amount (this provides a self consistent check).
The dimming of the O VI line at 1037.613 Å is particularly useful because pumping by C II 1037.018 Å extends its velocity sensitive range to include values from 90 km s to 250 km s (Noci, Kohl and Withbroe, 1987). Pumping by C II occurs when the O VI outflow velocity is large enough that the C II profile of the incoming light from the lower atmosphere is red shifted onto the coronal O VI scattering profile in the rest frame of the outflowing O VI. The dashed line in Figure 2 shows the effect.
Because Doppler dimming only affects the resonantly scattered component of spectral lines and lines of O VI and Mg X have collisional components, it is necessary to determine the relative intensity contributions of the two mechanisms. Kohl and Withbroe (1982) and Noci, Kohl and Withbroe (1987) have shown that the intensity ratio of the resonance line doublet for a lithium-like ion such as O VI or Mg X, determines the relative intensities of the collisionally excited and resonantly scattered components. The collisional component can also be predicted with the empirical model.
In view of the above discussion and determinations of supersonic outflow in a coronal hole within 3 solar radius of Sun-center (Kohl et al., 1984; Strachan, 1987), it is apparent that Doppler dimming and Doppler shifts can provide a sensitive determination of the magnitude and location of solar wind acceleration in the solar wind source region of the corona.
The intensities of the EUV lines, in combination with the electron temperatures and densities determined from UVCS measurements, can be used to derive abundances of the parent elements (O, Mg, Si, Fe). For ions with resonant doublets measured by UVCS (e.g., O VI, Mg X, Si XII) the value of the collisional component of the line intensity can be determined with the technique described by Kohl and Withbroe (1982) and Noci, Kohl and Withbroe (1987). The result can be used to determine the O, Mg, and Si abundances independently of the outflow velocities. This approach is useful in regions where the collisional component is a significant fraction of the total intensity. The UVCS measurements of electron temperatures, densities and outflow velocities can provide the necessary data for constructing a reliable model on which to base the abundance determinations.