The solar chromospheric and coronal magnetic fields play a major role in
the physics of the solar atmosphere. So far, these magnetic fields were
determined by the extrapolation of
*observed photospheric*
magnetic fields upon assuming either (1) a *potential*,
current-free model and using (only) the *line of sight*
component or (ii) a *force-free field* (FFF) model
(,
), and using all three components
of the photospheric field. Various analytical or/and numerical methods
for the determination of the magnetic field in half the space above the
photosphere have been suggested or/and implemented. (See, e.g., the reviews
by Sakurai 1989 and Gary 1990; also, the papers
by Schmidt 1964;
Semel 1967, 1988;
Sturrock & Woodburg 1967;
Molodensky 1969;
Nakagawa & Raadu 1972;
Nakagawa 1974;
Levine 1975, 1976;
Chiu & Hilton 1977;
Seehafer 1978;
Sakurai 1981, 1982;
Alissandrakis 1981;
Low 1982, 1985;
Aly 1984, 1987, 1989, 1992;
Schmal et al. 1982;
Wu et al. 1985;
Wu et al. 1990;
Gary et al. 1987;
Cuperman et al. 1989a, 1989b, 1990, 1991a, 1991b, 1993;
Low & Lou 1991;
Amari & Demoulin 1992;
Faubert-Sholl et al. 1992;
Bruma & Cuperman 1993).

Obviously, in both the *potential* and *FFF*
cases, calculated chromospheric and coronal magnetic field configurations
suffer from the rather limited information used for their computation:
only at the *bottom* of the three-dimensional
integration domain are boundary conditions
(i.e., photospheric observations) used.

Referring to the more general FFF-case
(also considered in this paper), the pertaining
progressive vertical extrapolation method used in the past can be
summarized as follows: (a) Starting from the vector magnetograph data at
the photosphere (say, *z*=0), use the *z*-component of the FFF-equation
in order to calculate the
function ; (b) next, by the aid of the *x*- and *y*-components
of the FFF equation, as well as of the
divergence equation
and the obtained at step
(a), calculate the derivatives (*i* = *x*,
*y*, *z*); (c) finally, utilizing a Taylor series expansion with the first two
(or three)terms retained, obtain the values at a vertical
(height) position , etc. Various aspects related to the ill-posed
character (in the sense of Hadamard 1923) of the progressive
vertical method for the extrapolation of the photospheric magnetic field
have been addressed
by several authors and will not be repeated
here (see, e.g.,
Morse & Feshbach 1955;
Tikhonov 1963;
Antokhim 1968;
Bakushinsky 1968;
Kryanev 1973;
Tikhonov & Arsenin 1977;
Low 1982, 1985;
Cuperman et al. 1990). Based on the fact that many
ill-posed problems do occur in science and technology these authors
recommend the development and use of suitable
``regularization" methods leading to
satisfactory solutions.

Recently, it has been pointed out (Zhang 1993, 1994) that
physical situations may exist in which, besides measurements of
*three-component photospheric* magnetic fields, data on
*longitudinal* (line of sight) *chromospheric*
magnetic fields are also available; and that these two types of information
can be combined in order to determine the full three-component
chromospheric magnetic field. Thus, when NLTE effects are relatively
small (e.g., in the case of quiet and plage regions),
*longitudinal chromospheric magnetograms at different
wavelenghts in the blue wing of line are indicative of the
longitudinal magnetic fields at different chromospheric altitudes.*
For example, using a set of eight magnetograms in the blue wind of the
line at wavelengths , (monochromatic images of Stoke's parameter *V* of the
line), and considering that the core of the line forms at
about 1900 km in the solar atmosphere (Allen 1973) Zhang
(1993) estimated the formation height of the wavelength to be about 1200 km. It is anticipated that in the near future, many
more (40-50) multi-level longitudinal magnetic field measurements will be
possible (Ai 1994a, 1994b; Fang 1994).

In this work we develop - within the FFF framework -
a high order precision method for the
reconstruction of chromospheric magnetic configurations.
The method is based on the
utilization of both types of existing magnetic field observations: (i)
*vector* (three-component) *photospheric*
measurements and (ii)*longitudinal* (along the line of
sight) *chromospheric* measurements. The goal is to compute
the
full *three*-component magnetic field
configuration at all points *above* active photospheric regions.
A high order computational
algorithm was developed and utilized for this purpose. As it will be
shown, this algorithm enables one the reconstruction of the chromospheric
vector magnetic field with a maximum relative error of .

The paper is organized as follows: Sect. 2 presents the general formulation of the problem, including a discussion of the observations, the basic equations utilized and analytical solutions of FFF model equations used as a test case; Sect. 3 describes the general computational algorithm developed and its implementation; Sect. 4 presents the results of the calculation demonstrating the high performances of the algorithm; Sect. 5 elaborates on the elimination/minimization of measurements errors as well as on the fitting of the corrected data to a FFF-state; A summary is given in Sect. 6.

web@ed-phys.fr