10
The approaches with inherent data rejection are applicable to samples with redundant
points. Examples are digitised profiles, contour lines, multiple heights generated by
electronic correlators,etc. From the original set only data pertinent for further pro
cessing are extracted according to appropriate selection rules. The main shortcomings
of the approaches are:
- the sampled data are usually less accurate due to continuous (dynamic) operation;
- the point density differs substantially along the lines traced and perpendicular to them
Sampling is carried out in a single run. These approaches are objective.
The progressive approaches implement two or more successive sampling runs, starting
with coarse sampling and proceeding gradually towards finer sampling. Before initia
ting each next sampling run, the data of the preceding runs are analysed and a list of
further points to be sampled is set up. The progressive approaches are objective.
' ; . - , s -...
The basic approaches to sampling, outlined above, may also occur in different combi
nations.
The sampled points can be arranged in different patterns . These may be:
1. Regular sequences or grids (i. e. coherent samples) which are convenient for
automation;
2. Irregular sequences or grids of characteristic points. Procurement and processing
of such data is difficult to automate;
3. Regular sequences or grids supplemented by morphologic lines and/or points. Pro
curement and handling of the supplementary data is difficult to automate;
4. Sets of regular sequences or grids having adapted densities (i .e. heterogenous
square grids). The allotment of densities to different terrain areas can be subjec
tive (by interpretation) or objective (by instrumental sampling and computer analy
sis).
The patterns listed under 4. may be supplemented by morphologic lines and/or points.
Such D, T. M. representation appears to be very comprehensive. Their processing
can to a great extent be automated.
Amopg the different sampling approaches listed above, progressive sampling of
heterogenous square grids [ 10 ] seems to be the most feasible. This requires,
however, an on-line mini-computer for data analysis, for control of recording, and
for locating the instrument tracking device (X, Y, Z). The principal merit of the pro
gressive sampling method is the matched density of a locally regular point grid to
variability in the terrain relief. This is achieved by alternative sampling and analysis