

In Fig. 2, we plot the electron density as derived from Si VIII, coupled with N_{e} from UVCS (Kohl et al. 1998) and LASCO (Lamy et al. 1997). The LASCO data was acquired in February 1996, while the UVCS images were timeaveraged from November 1996 to April 1997.
Guhathakurta & Fisher (1995, 1998) estimated the electron density
distribution in the coronal hole from observations made with the space based
SPARTAN 20101,03 instrument. At that phase of the descending solar magnetic activity
cycle the coronal conditions were different from our observing period (during
our observing period the solar activity was in the minimum phase), so we
don't compare our results with SPARTAN, but the essential density profile
has a striking similarity. The above authors obtained the density profile from
polarized brightness (pB) measurements between 1.16  6 .
They
give an analytic prescription of the density profile as,
(1) 
where N is the number density (= ), is mean atomic weight (= 0.62), m_{p} the proton mass, k_{B} the Boltzmann constant, the solar gravity and the effective temperature is given by T_{eff}=(T_{e}+T_{p})/2, where T_{e} and T_{p} are the electron and proton temperature, respectively. The line fit as a function of for all densities out to 8 is shown as the solid line in Fig. 2 for T_{eff} = 1.2 10^{6} K. Thus for a locally isothermal fully ionized twofluid coronal plasma solution we can either estimate T_{e} or T_{p}. Using T_{e} of Si VIII as 8 10^{5} K (i.e. its formation temperature in ionization equilibrium), we find an estimate of T_{p} as 1.6 10^{6} K. Furthermore this formula predicts an electron density of cm^{3} at 8 in excellent agreement with observations.
Our results for the ion temperature are in agreement with observations obtained by UVCS (Kohl et al. 1997; Cranmer et al. 1999) who find that at larger distances from the limb the ions in a coronal hole are extremely `hot' and the electrons are much `cooler'. Thus our observations reconfirms that in the coronal hole plasma, particularly at larger distances from the limb the assumption of collisional ionization equilibrium can not be used any more. This is also consistent with a study of several ions by Tu et al. (1998) who found ion temperatures perhaps as high as 23 times the formation temperature in ionization equilibrium even at heights of only a few tens of arcsecs. Recently, Esser et al. (1999) have combined data from SPARTAN, White light coronagraph (WLC) from 1.5  5.5 in April '93, with Mauna Loa observations between 1.16  1.5 (Fisher & Guhathakurta, 1995). We are unable to fit the data in Fig. 2 with their polynomial as our dataset is more complete, extending from 1.02  8 . Infact, the expression given by Esser et al. predicts a rather high electron density at the limb and a density three orders of magnitude smaller at 1AU. Instead, the data in Fig. 2 suggests a falloff in density proportional to r^{8} from 1 to 2 , then r^{4} from 2 to 4 and finally as r^{2}. A polynomial fit of the form
(2) 
provides an excellent fit to the data as shown by the dotted line in Fig. 2. The first coefficient in the R.H.S of Eq. (2) is determined by the electron density close to the limb (i.e. from our Si VIII SUMER data), the last coefficient determined by the electron density at 1 AU, while only the middle coefficient was varied in order to provide the best fit.
In Fig. 3, we plot the half width at 1/e of the peak
intensity (V_{1/e}) in km s^{1} as measured from Si VIII SUMER
data plus O VI UVCS data (Kohl et al. 1998).
The effective (ion kinetic) temperature (T_{k}) can be obtained from
(Kohl et al. 1998),
(3) 
where M is the ion mass and is the observed 1/e half width. We have applied a mass correction factor of 1.32 to the Si VIII data to make it consistent with O VI. Note that the kinetic temperature include contributions both from microscopic thermal motions and unresolved transverse wave motions. As shown in Fig. 3, this data suggests a small increase from 1 to 1.2 , then a plateau up to 1.5 , followed by a sharp increase up to 2 , then a more gradual increase further out.
Now we turn our attention to the question of the nonthermal velocity.
The analysis of DBP and BTDW showed that the
nonthermal lineofsight velocity as derived from the Si VIII line
widths increases above the limb while the electron density decreases.
On a closer inspection, the observations revealed that the nonthermal velocities
were inversely proportional to the quadratic root of the electron density, in
excellent agreement with that predicted for undamped radially propagating
Alfvén waves. In the WKB approximation the rms wave velocity
amplitude and density are related by (Hollweg 1990),
(4) 
The nonthermal velocity can be deduced from the standard
equation for an optically thin line broadened by thermal broadening caused by
the ion temperature T_{i} and broadening caused by nonthermal motions as given
by,
(5) 
Measuring the FWHM of the Si VIII 1445 line and using Eq. (5) with T_{i} K (as suggested from the line fitting to N_{e}), the diamond symbols in Fig. 4 represents the observed nonthermal velocities at different heights. Now If we assume the same ion temperature as Si VIII for the 721.23Å line (as suggested by the intensity falloff compared to Si VIII), the Si VIII nonthermal widths implies an ion mass of 53 for this new feature. This latter assumption is probably valid as different emission lines are generated in the same column mass as seen by the spectrograph. Although, this can not be taken as absolute proof it does suggest that the line may be due to Fe, probably ionization stages VIIVIII.
In Fig. 4, the (+) represents the velocities calculated on the basis of the Eq. (4), using the measured N_{e} from Si VIII. We have used a magnetic field strength of B= 6G and a proportionality constant so as to match the energy flux density (see BTDW for details). In both cases, we find excellent agreement in the inner corona, but for the outer corona 200 arc sec (see Fig. 4), the (+) symbols starts to deviate from those calculated from the observed FWHM(diamond symbols in Fig. 4). It is also possible to estimate the errors in the calculated nonthermal velocity (+); for the last 4 points, errors of 3, 3, 5 and 15 km s^{1} have been estimated. Despite these large errors we have no overlap with the errorbars of the observed nonthermal velocities, suggesting an effective breakdown of Eq. (4) above 1.2 .
Lou & Rosner (1994) have already pointed out that waves can be reflected against gradients in the Alfvén speed and the WKB approximation fails. The reflection is important as it increases the momentum transfered from the Alfvén wave to the medium. Recently Torkelsson et al. (1998) have shown that the Alfvén waves steepen and produce current sheets in the nonlinear regime. These waves are strongly damped by nonlinear steepening. They have also reported density oscillations in their simulations, which are consistent with observations by Ofman et al. (1997) who reported quasi periodic variations in the polarized brightness in the polar coronal holes between 1.92.45 , as observed from the white light channel (WLC) of UVCS. It would be interesting to see whether these density oscillations are also present in the inner corona which can be due to the presence of nonlinear high amplitude compressional waves as reported by Ofman et al. (1997). For that one needs a high cadence dataset. We propose that somewhere around 1.2  1.3 this nonlinearity becomes important (as indicated by Figs. 3 & 4). The Alfvén waves with an amplitude of 3050 km s^{1} (as observed) at the base of the coronal hole can generate nonlinear solitary type of waves, which can contribute significantly to solar wind acceleration in open magnetic field structures.
