FLOOD01751

• County: Fremont
• Community: Canon City
• Stream Name: Arkansas River
• Title: Canon City Community File
• Date: 1/1/2001
• Prepared For: CWCB
• Prepared By: CWCB
• Floodplain - Doc Type: Community File

.'

13

( PRACTICAL INTERPRETATION OF

MAP ACCURACY SPECIFICATIONS
. MORRIS M. THOMPSON
Atlantic R~gion Engineer

.'

I have chnscn to discuss the interpretation.
of map-accuracy specifications because there Is
a gro~ing tendency to evaluate maps in precise
mathprnatical te~s ond to refine these terms so
that they reflect the most elegant statistical
theory. Now I am all for using correct scientific
trrminology.to express map accuracy, as I think
my previous papcr$ on the subject clearly in-
dicate; but I do want to s~und a note of .caution:
lest we fall into such practices os first-order
manipulation of fourth-order data, or to use a
cliche, putting a .lPic.TClmetcr on the end of .0.
).ardstick.

I
"

It is my c~nvictio" that map-accuracy
spccificatiCJn~ and testing proccdure~ cannot br
so clear :lnd matllC'Nticalty incC'lntrovcrtible th t
they will give the exact and only answer to tht"
problem of pvaluating the accuracy of a given
Dap~ There is an area of interpretation. whqse
existence we mu~! recognize if ~e wish to avoid
penalizing ourselve~ by rigidly applying narrow
rules in a way that does not reflect the spirit
or intent 9f the $pecifieations~

I
'-

tn a sense. accuracy specifications are
akin to laws in our civil community. A law can be
c~carly ~ritten 'and apparently unmistakable in
aeaning, yet a case involving the application
of that law may go through court after court with
aany variations in its interpretation~ A de-
fendant may be declared guilty in onc court and
innocent in another, under the same."body of law.
In the same vny. .a map may pass or Cail. accord-
ing to bow the accurncy sp~cificQtions arc applied
or intcrprct~d~ }Iy di5cus~i~~ r~ ~A~Trs5ed to
som. of the fnC'ets of th~ in[erpn"tntion function
10 applYlng mnp-nccuracy speC'ltlC~[lOnS~

RELIABILITY OF TEST DATA

. The first point to consider is: ~'hat
credcnct' shall be- give-n to the data obtained
from the nccuracy-t~st survey? Every survey
embodies some error in it. In a test survey
of good ~~ality. accidental errors of relatively
~inor magnitude are expected, and these errors
. ~.will tend J.9.r.give lavorable results about as
"

Presented at a panel discussion on }mp
Accuracy. sponsored by the Potomac Region. ASP,
and the National Capital Area. ACSM, Dec~ 16,
1965, at Washin~ton, D~C.

(

'.

'.

often a9 un~avorable. Truly accidental errors
in the test survey are not a cause (or concern,
nor do they call for interpret2tion~

. Apart from the consideration of'accidental
errors 1n the test survey, ther~ is need to
take account at another type of error: the
blunde~ in applying test data. These blunders
can often be detected by simple exa~ination of
test data and the application of common sensr.
The signal flag for det~ctjng such blunders is
the sudden occurrence of large errors 1n a scries
of smaller errors. Consider the C3Sf' of 'an ac-
curacy survey done by transit or planetable
trav~rse in which the reported ~osition errors
in feet for well-defined points along [he traverse
arc 22, 24, 18. 26. 28. 35, 184, 39, 48. 38~
Assuming that the specifications require that
9r.t of the points must be correct within 4C fect.
it appears that the map would "fail. as only 80% :.
are vithin 40 feet~ But if we look at the series.
of errors again, we notice something strange and
inconsistent. ~~at is that value of 184 ~oing
among all those other values ranging from 18 to
.4G? In one actual case of this kind. of which 1
have persona.l knowledge, we noted this abnormal.
error and said "Let ts .look at that point .," Here .
is \lhat we found (f ig~~ 1): The point 'Was describ~d
as tT-road vest" and the map showed aT-road
west; but ~hen wr looked at the aerial photo-
&.raphs, we saw that there were two T-roads wc'st.
about 200 feet apart, only one of which had been
comp1led~ The T-road circled on the photograph.
as tbe test point was not the one compilcd on
the map. . Yh~n &111 this vas straightened out.
the actual error was found to be well vithin 4C
feet and the map passed. But if we had insis[ed
on blind acceptance of the test data at face
value, that map would have been listed as f~iling.

\

In another instance, a test point was
described as "South end of bridge~" The bridge.
was 80 feet ~ong, but there was no indication
of this on the map. which showed only a road
crossing a single-line stream. Yc therefore had
a built-in error of some'40 feet to start with.
We believe ~ were ~ntirely justified in discard-
ing this point. for it was ambiguous~

1 am not trying to make it seem a virtue tha~
some of our people make mistakes and others find
them. ~at 1 am trying to point up is tha: we
~st recognize that people arc human and viII
~ke mistakes and that we have got to be alert

,.

.'
,.,
"j
.,

-: