to deal with the stability of monument settings. Four control
points were found to be outliers, while one was obviously
far from the mean, and this control point was removed from
analysis requiring physical inspection. While there were no
differences in easting, the observed difference in northing (at
0.05 significance) diminished when the confidence level was
increased to 99 percent. This slight discrepancy does not in
any way indicate a problem with the stability of the monu-
ments as the
t
-test performed on the total horizontal difference
indicated the two datasets were not different in terms of their
horizontal components. After removing three outliers, the
mean difference of vertical coordinates was 7 cm (within 2σ
measurement errors) and is not considered excessive relative
to the
GPS
techniques used and the short term of observations
to acquire these data. Monument stability analysis revealed
that monuments made of cement post provided comparably
better long-term accuracy than monuments made of stone.
While the geocenter and length of the axes of the coor-
dinate frame in the first realization of
NAD
83(1986) were
computed based on Doppler data, subsequent realizations
(e.g.,
NAD 83
(
HARN
)) included improved knowledge of axis
lengths and orientations based on advancements of
GPS
and
other technologies. Although the geocenter remained un-
changed, a new scale factor was introduced in order to be
consistent with
ITRF89
(Snay and Soler, 2000a). These changes
made the newer realizations different from the first realization
of
NAD 83
(1986). It is expected that
NAD 83
(
HARN
) coordinates
may differ by nearly 1 m from corresponding
NAD 83
(1986)
coordinates (Snay and Soler, 2000a).
NAD 83
(
NSRS2007
) is a
refinement of
NAD 83
(
HARN
) or
NAD 83
(
CORS96
) as it included
better ellipsoid heights and implemented better computa-
tional techniques. The difference in coordinates between
NAD
83
(
HARN
) versus
NAD 83
(
CORS
96) realizations were expected
to be <10 cm (Snay and Soler, 2000a). However, because of
insufficient samples within the
MCPD
database, we could not
make this comparison. In our analysis, comparisons between
NAD 83
(1986) and
NAD 83
(
NSRS2007
) resulted in only +/-6 cm
difference; however, we believe a larger difference would
likely follow with a more representative dataset. The changes
in coordinates between
NAD 83
(2011) epoch 2010.0 and
NAD
83
(
CORS96
) epoch 2002.0 were expected to be approximately 2
cm (Sickle, 2013) if no velocity factors were involved.
A summary of horizontal datum comparisons (Table 2) in-
dicates the largest changes or differences in coordinates exist
between
NAD 83
(1986) versus
NAD 83
(2011) realizations.
Since successive geoid models have better resolutions
and improved gravimetric models,
GPS
-derived orthometric
heights using the latest geoid model provides better results
(Table 3). The development of
GEOID03
was better than that of
GEOID99
, partly because there were more satellites available in
the
GPS
constellation, additional airborne based gravimetric
data existed, better global gravity models had been developed,
far more
GPS
baselines were available, and better ellipsoid
heights, and better computational techniques were available.
As a result, there are larger difference in elevation reported
between
GEOID99
and
GEOID12A
, relative to the differences
reported between
GEOID03
and
GEOID12A
.
The velocity map for Idaho published by
NGS
is an indica-
tion of probable changes in coordinates due to the velocity of
crustal motion across a region. The datasets analyzed near the
IDBY
CORS
clearly indicate changes in horizontal coordinates
and generally conform to changes in position due to velocity
suggested by
NGS
. The variable horizontal differences around
IDBY
suggest east Idaho is subject to substantial crustal mo-
tion.
Conclusions
The techniques used in this research present a case study and
approach to analyze the quality of a passive control database.
This study deals with factors that contribute to changes in
horizontal and vertical positioning over time and the three
strategies applied were successful in delineating the essential
information necessary to understand the effect of time on
space. Stability analysis on control points in Bannock County
provided a good picture of monument stability. Survey marks
of cement post represent relatively consistent positional ac-
curacy relative to other monument types.
The comparison of coordinates between
NAD 83
(1986) and
NAD 83
(2011) highlights the importance of using the latest
realization of the horizontal datum. In essence, there is more
than one meter difference between these datums across this
case study. Comparisons between vertical coordinates derived
from realizations of different geoid models revealed a sub-
stantial difference (63 cm) between
GEOID99
and
GEOID12A
. As
expected, this difference was greater than vertical differences
between
GEOID03
and
GEOID12A
.
As the phenomenon of regional tectonic velocity varies
across the western US, the results from the comparison of
coordinate displacement between
NAD 83
(2011) epoch 2010.0
and
NAD 83
(
CORS96
) epoch 2002.0 should also vary and be
dependent on crustal velocity of the area. Since the impact of
velocity on coordinates has been incorporated into the cur-
rent realization of
NAD 83
(2011), it is considered best practice
to use this datum for coordinate determination especially in
Idaho due to its active tectonics. Since position coordinates
are only relevant with the datum in which they were mea-
sured, the
NGS
recommends using the latest national datum.
Table 2. Results of horizontal datum comparison.
NAD 83(1986) vs
NAD 83(2011)
NAD 83(1986) vs
NAD 83(NSRS2007)
NAD 83(CORS96) vs
NAD 83(2011)
Statistically different in easting, northing and Cartesian horizontal
Yes
Yes
Yes
Mean difference (easting)
98 cm
4 cm
0.9 cm
Mean difference (northing)
68 cm
4 cm
1.4 cm
Mean difference (Cartesian Horizontal)
120 cm
6.2 cm
2.2 cm
Table 3. Results of geoid model comparison.
GEOID03 vs GEOID12A GEOID99 vs GEOID12A
Statistically different in GPS-derived vertical coordinates
Yes
Yes
Mean difference
30 cm
63 cm
224
April 2018
PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
