186 
In the first section a numeration is given to the 
name of all network points. Thus the points are 
identified by the same integer number in the file 
"misnrete.dat" (in which the tridimensional Carte 
sian geocentric coordinates with the e.q.m. of the 
I.G.S./A.S.I. and of the known local stations are 
reported, if these must be assumed as observables, 
and the GPS measured baseline vector compo 
nents with the correspondent cofactor matrix), in the 
file "xinirete.dat" (in which there are the Cartesian 
coordinates of the permanent I.G.S./A.S.I., the 
known local stations and the unknown points) and 
the file "nppirete.dat" (where is reported the nume 
ration, the name, the unknown parameters of the all 
points, besides the unknown pointer array). 
In the second section, there are other input data 
files : the file "emisrete.dat" (which reports the initial 
scale factors for the GPS vectors cofactor matrices) 
and the file "parellis.dat" (in which there are the 
geometric parameters of the reference ellipsoid). In 
this section, it is performed the least-squares adjust 
ment as in 3., the blunder detection as in 4. , and 
finally the BIQUE estimates of the two variance 
components of the prefixed stochastic model (14) 
as in 5. 
The input and output of both sections are reported 
in the following Figure 1 : 
Figure 1 : Input - Output flow chart of the files 
MISPAR.for and PARPESI.for 
The output of the section parpesi.for is the file 
TEMP which contains all questioned in the file 
parpesi.opz. Particularly, the output of the least- 
squares adjustment can be the adjusted tridimen 
sional Cartesian geocentric coordinates (X,Y,Z) 
with the e.q.m., the geodetic coordinates (cp, X., h) 
with the e.q.m. (c v , Ox, c?h) and (a e , a n , a h ) of the 
points. The output of the internal reliability can be 
the a posterior global variance factor, the norma 
lised weighted LS residuals and the magnitude of 
one gross error. 
In the file "pesirete.dat" we have, for each iteration 
in the BIQUE estimate, the value of the variance 
factors of updating and the BIQUE of the variance 
components. 
In the file "xitprete.dat" we have, for each iteration 
of the BIQUE estimate, the value adjusted tridimen 
sional Cartesian geocentric coordinates (X, Y, Z). 
Finally, in the output file TEMP, we can have for 
each GPS vector: the names of station-forward 
points, the final BIQUE of the variance components, 
the length of the baseline and scale factor of the 
cofactor matrix. 
3. LEAST SQUARES ADJUSTMENT 
3.1 Functional model 
The functional model of Gauss-Markov is assu 
med : 
Ax = I + v (1) 
where 
A is the (n , u) design matrix of known coeffi 
cients with rank(A) = u<n ; 
x is the (u , 1) vector of the unknown tridimen 
sional Cartesian geocentric coordinates ; 
I is the (n, 1) random vector of the obser 
vations ; 
v is the (n , 1) vector of true errors of the obser 
vations ; 
£| the variance-covariance (n , n) matrix of the 
observations. 
The design (n , u) matrix A is a (3,3) block struc 
tured matrix, as the (u , 1) vector I is partitioned in 
(3,1) uncorrelated subvectors. 
In this section, I denotes the (3,3) identity matrix. 
The i-th block row of A is : 
a i =[0 I ... o] (2) 
- if the tridimensional Cartesian geocentric coor 
dinates of a known (I.G.S./A.S.I. or local) station 
are assumed as observables, then the corre 
spondent (3,1) observation subvector is