144
3 - Graphical representation of the errors
In order to provide complete comprehension of the results of each test, besides
the list of differences of the control points coordinates, we have represented
graphically the distribution of these differences on the points of the polygon. These
representations are contained in figures from 1 to 29.
The errors in X and Y, or planimetric errors, have been represented by means
of a vector oriented with the origin in the position of the corresponding point on the
planimetry of the polygon, whose length is given by
K V(KJ + (A/
where K is the scale of representation of the vectors.
The opportunity, simplicity and evidence of such a representation are remark
able. The total planimetric behaviour of the errors on the polygon is very clear.
The choice of the principle for the representation of the errors in height of the
control points in the different tests was not so easy. At first, the Secretary of the
Group proposed to indicate next to the point, on the planimetry of the polygon, the
numerical value of the error and to draw, between the points, the curves of equal
error.
However, this plan did not appear convenient, even though quite helpful for
purposes of interpretation.
As the I.G.N. of Paris suggested, we preferred to represent the errors in height
by means of vectors, starting from the planimetric position of the point, oriented to
the top or the bottom of the page according to the sign of the error and with length
proportional to the magnitude of the error itself.
The drawing of these graphs was assigned to the respective centres. In order to
make the representation uniform, a unique principle was chosen as a basis for data
rejection. We established that points with an error in the coordinates higher than
60 m should be removed from the graph concerning each test; moreover, a group of
14 points was suppressed because they were evidently anomalous for non photogram-
metric reasons. These are the points: 11, 12, 26, 83, 99, 174, 199, 209, 236, 249, 270,
282, 294, 300.
In part C of this chapter, where we deal with the statistical research on the
anomalous points, we shall justify this removal.
An important element for the interpretation of results is the number and the
position of the control points used for the planimetric and altimetric adjustment of the
blocks. These points are marked by a triangle on the graphic representation of the
errors.
In addition the points that have not been plotted or with an error in the coor
dinates higher than 60 m, have been removed from the graphs of each test, thereby
eliminating confusion between these points and the ones with error equal to 0.
4 - STATISTICAL DISTRIBUTION OF THE ERRORS
Even if not absolutely necessary, we thought it useful to add to the complete
list of the differences and to their graphic representation, a synthetic grouping, sim
plifying and helping eventual comparisons among the different results. Table 2
represents, in fact, the residual errors on the coordinates of the different tests, grouped
in classes of frequencies of variable amplitude.