PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING
April 2016
243
Q: I am evaluating a lidar-based surface model and
collecting checkpoints in each land category as follows:
Land Category
Number of Checkpoints
Bare Earth
21
Low Vegetation
30
Medium Vegetation
20
High Vegetation
20
Urban
21
Total
112
The surface model needs to meet an accuracy suitable
for the generation of 2-foot contour interval. After field
checking the vertical fit of 112 checkpoints, I found the
differences are ranging from -0.985 to 0.525 feet with an
RMSE value of 0.290 feet. Please see details in the tables
below. As an ASPRS member, I attended many of your
presentations over the years and would appreciate your
insight on the following questions.
1) What statement can be made about the RMSE value
of 0.290 feet?
2) Considering the deliverable data was for a contour
interval of 2 feet, what are the qualifying statements for
vegetated and non-vegetated terrain accuracy?
3) For the classes of vegetated and non-vegetated terrain,
what number in my results is compared to the ASPRS
report to make these qualifying statements?
Victor Murray, ASPRS Member, Ada, Oklahoma
Dr. Abdullah:
Looking at the accuracy requirements of the
final product, it is clear that the project laps two eras of map
accuracy standards, the old ASPRS standards of 1993 and
the new ASPRS Positional Accuracy Standards for Digital
Geospatial Data. This was obvious when Victor stated,
“The surface model needs to meet accuracy suitable for the
generation of 2-foot contour interval.” I called this era “the
transition period,” as users of the new standard are trying to
bridge the gaps and minimize confusions between the old and
new standards. Predicting such behavior was a top priority for
the drafting committee during the design and authorization of
All Categories Real (ft) ABS (ft)
Bare Earth Real (ft) ABS (ft)
Count
112
112 Count
21
21
Mean
-0.037
0.223 Mean
0.108
0.158
Median
0.021
0.172 Median
0.093
0.143
Minimum
-0.985
0.000 Minimum
-0.210
0.014
Maximum
0.525
0.985 Maximum
0.525
0.525
STD DEV
0.289
0.186 STD DEV
0.180
0.135
RMSEz
0.290
0.290
VEG LOW Real (ft) ABS (ft) VEG MED
Real (ft) ABS (ft)
Count
30
30
Count
20
20
Mean
0.010
0.191 Mean
-0.260
0.343
Median
0.009
0.164 Median
-0.302
0.322
Minimum
-0.608
0.010 Minimum
-0.648
0.023
Maximum
0.426
0.608 Maximum
0.335
0.648
STD DEV
0.235
0.132 STD DEV
0.307
0.205
VEG HIGH
Real (ft) ABS (ft) URBAN PAVED Real (ft) ABS (ft)
Count
20
20
Count
21
21
Mean
-0.136
0.277 Mean
0.061
0.170
Median
-0.017
0.190 Median
0.092
0.117
Minimum
-0.985
0.000 Minimum
-0.342
0.015
Maximum
0.384
0.985 Maximum
0.481
0.481
STD DEV
0.365
0.268 STD DEV
0.203
0.122
the new standard. We expected that users would eventually
need guidelines during such transition period. Rightfully so,
we provided a wealth of examples and tables to relate the new
standard to the legacy ones. Looking into Victor’s question, I
noticed the following:
1.
The Stated Accuracy Requirement
: Specifying that
product accuracy should meet 2-foot contours does not align
with the spirit of the new standard, as the new standard
Photogrammetric Engineering & Remote Sensing
Vol. 82, No. 4, April 2016, pp. 243–248.
0099-1112/16/243–248
© 2016 American Society for Photogrammetry
and Remote Sensing
doi: 10.14358/PERS.82.4.243
