0.300
E 0.200
a M |
= 0.100 | ;
tome | p
0 0 000 an n I ni n APR | | | |
c . Hep ddl à. T | | J
Hi LP RIRE UNE
= -0.100
=
>
2 -0.200
-0.300
e o v eo e oN qe
— ce eo eo e e ©
Ww» e N o - eo wa
<< eo e = = o Ne
+ + w 0 uo + 0
Ne ~~ te m he he te
EASTING (m)
Figure 3. Differences in Y (Northing) between the
WisDOT and Leica AT values, as a function of Easting
along the strip.
3.2 Surface Generation
Two DEMs were generated using the DPW 770 with a
grid spacing of 5 meters and 2 meters, respectively.
Only the 2 meter DEM was compared to the WisDOT
DTM file MIL23X. The 23,799 X, Y, Z values in the
MIL23X were treated as check points. Elevations from
the softcopy generated DEM were then interpolated us-
ing a bilinear interpolation at each DTM point, and then
compared to its corresponding value in MIL23X. The
differences are summarized in the first column of Table
2. Many of the points in MIL23X were derived from
break lines.
The second column in Table 2 presents statistics of the
comparison between the Leica DEM and the 4,405
points in MIL23X that were not derived from break
lines (points marked *REG"). The "Difference" in this
table refers to the value of WisDOT elevations minus the
Leica elevations at the same spatial location.
Table 2. Comparison Statistics Leica's DEM and Wis-
DOT's DTM.
Elevation Points within WisDOT DTM
All Points “REG” Points Only
Ave. Difference (m) -0.08 -0.10
Variance (m) 0.57 0.51
S.D. (m) 0.75 0.72
RMS (m) 0.75 0.72
Number of Points 23,799 4,405
In order to investigate the large root mean square (RMS)
differences that are evident in Table 2, a histogram of
the number of points with a given difference was calcu-
lated. The histogram, shown in Figure 5, displays the
frequency of points with a given difference versus the
differences.
342
Frequency
Difference in Elevation (m)
Figure 4. The frequency of points within the WisDOT
DTM with a given difference in elevation between the
WisDOT DTM and the Leica DEM.
The histogram shows that the differences are nearly
equally distributed on either side of the mean difference.
This may indicate that the models are tipped relative to
each other. To further investigate, the Leica DEM and
the interpolated DEM, that is derived from the WisDOT
DTM, are visualized in Figures 5 and 6, respectively.
The differences between these two DEMs are not evident
from these figures.
In order to determine which of the DEMs is most likely
correct, a further comparison was made between the
elevations derived from the WisDOT AT values and the
elevations from the two DEMs. Fifteen WisDOT AT
elevation points (from the pugs) were found within each
of the DEMs. Two of the points (73 and 78) were lo-
cated toward the edge of a photograph. Table 3 shows
statistics derived from comparing elevations within the
Leica DEM to the WisDOT AT derived elevations. The
first column presents the results from comparing all 15
WisDOT AT elevation values to the corresponding ele-
vations derived from the Leica DEM. The second col-
umn contains the same comparison for the 13 WisDOT
AT points away from the photo edges. Column 3 pres-
ents statistics of comparing the 15 Leica derived AT
elevations to the corresponding elevations derived from
the Leica DEM. Column 4 contains the same compari-
son in column 3, excluding points 73 and 78.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996