LINEAR is the most optimistic (see Tables 6 and 7). Percent
differences for long profiles are within + 34% except with program
LINEAR.
For short profiles RF is the most conservative while LINEAR is the
most optimistic (see Tables 6 and 7). All programs have percent
differences exceeding +50%.
For the Raymond profiles of varying sample spacings, LOGKV and RF have
no increasing trends in percent differences (see Table 6). RF is the
most conservative, while LINEAR is the most optimistic. Percent
differences are within + 40%.
For the Crowsnest Pass profiles of varying sample spacings, all
Programs have increasing trends in percent differences (see Table 7).
LOGKV is the most conservative, while SPECTRA is the most optimistic.
Only program RF has percent differnces within + 33%.
6. RECOMMENDATIONS
Based upon the results of relative and absolute comparisons, the most
suitable program(s) is/are recommended for each sample profile
depending on its roughness, length and sampling rate (see Table 8) as
follows:
a) Roughness
- rough (R.F. 257)
- flat (R.F. <5%)
Length
- long (length >1.75 model base)
- short (length < 1.75 model base)
c) Sampling rate
= high (Ad <1.0mm at image scale)
- low (Ad >1.0mm at image scale)
The recommended programs are based upon percent differences within
+50% obtained from the tables referenced and profiles used. If more
than one program is recommended, the programs are arranged in
increasing degree of optimism. In other words, the first program is
the most conservative, while the last is the most optimistic based
upon their means of percent differences for more than one profile. In
case of contradiction between relative and absolute comparisons the
latter prevails. There were no data for (c) and (8).
It can be noted that short profiles can be used only if they are rough
and of high sampling rate. For more reliable result, the sample
profile shall be as long and of high sampling rate as possible, and
shall be truly representative of the model's roughness.
ACKNOWLEDGEMENTS
The following companies are gratefully acknowledged for supplying the
data under contracts, used in this paper:
a) Terrain Aerial Surveys Ltd, Edmonton, Alberta for measuring
- l6
— Tm